From baldrick at free.fr Mon Oct 1 02:47:21 2007 From: baldrick at free.fr (Duncan Sands) Date: Mon, 1 Oct 2007 09:47:21 +0200 Subject: [llvm-commits] Size of long double (X86_FP80Ty) In-Reply-To: <3AF2C9B1-D9B0-4CCD-B60F-5A1BB7E91E8A@apple.com> References: <200709282204.17074.baldrick@free.fr> <3AF2C9B1-D9B0-4CCD-B60F-5A1BB7E91E8A@apple.com> Message-ID: <200710010947.21712.baldrick@free.fr> > By the way, to be consistent, getTypeSize for an APInt > should return the number of bits rounded up to a multiple > of 8, rather than rounded up to a power of 2 as it is now. > Shall I make this change? Due to the way ExpandOp works for loads and stores, it looks like the size needs to be the smallest power of 2 bigger than or equal to the bitwidth. For example, consider i36 on a 32 bit machine. The type to expand to is i32, and a store gets expanded into two i32 stores, i.e. 64 bits. This suggests that the size needs to be at least 8 bytes, since otherwise you may write outside the allocated memory. Note that this applies regardless of alignment (I say this because getTypeSize seems to round the size of APInt's up to the next multiple of the alignment). For example, consider i128. This first gets expanded to i64, result on two 64 bit writes, then i64 to i32, resulting in four 32 bit writes. In any case, you write 128 bits so the size should presumably be 16 bytes, and alignment didn't come into this anywhere... Am I wrong? Thanks, Duncan. PS: Is it the same for floating point types? Can a 128 bit fp type be expanded into two 64 bit fp types? From neil at daikokuya.co.uk Mon Oct 1 08:07:48 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Mon, 1 Oct 2007 22:07:48 +0900 Subject: [llvm-commits] [llvm] r42484 - in /llvm/trunk/lib: CodeGen/SelectionDAG/LegalizeDAG.cpp CodeGen/SelectionDAG/SelectionDAG.cpp ExecutionEngine/ExecutionEngine.cpp VMCore/ConstantFold.cpp In-Reply-To: <200709301819.l8UIJ3wE003992@zion.cs.uiuc.edu> References: <200709301819.l8UIJ3wE003992@zion.cs.uiuc.edu> Message-ID: <20071001130748.GN19035@daikokuya.co.uk> Dale Johannesen wrote:- > Author: johannes > Date: Sun Sep 30 13:19:03 2007 > New Revision: 42484 > > URL: http://llvm.org/viewvc/llvm-project?rev=42484&view=rev > Log: > Constant fold int-to-long-double conversions; > use APFloat for int-to-float/double; use > round-to-nearest for these (implementation-defined, > seems to match gcc). Seems sensible. Ultimately FENV_ACCESS pragma should be taken into account, for C at least. Neil. From djg at cray.com Mon Oct 1 11:01:24 2007 From: djg at cray.com (Dan Gohman) Date: Mon, 01 Oct 2007 16:01:24 -0000 Subject: [llvm-commits] [llvm] r42487 - in /llvm/trunk/include/llvm: Analysis/CallGraph.h Analysis/LoopInfo.h Support/CallSite.h Message-ID: <200710011601.l91G1OQr011316@zion.cs.uiuc.edu> Author: djg Date: Mon Oct 1 11:01:23 2007 New Revision: 42487 URL: http://llvm.org/viewvc/llvm-project?rev=42487&view=rev Log: Add empty() member functions to a few container-like classes. Modified: llvm/trunk/include/llvm/Analysis/CallGraph.h llvm/trunk/include/llvm/Analysis/LoopInfo.h llvm/trunk/include/llvm/Support/CallSite.h Modified: llvm/trunk/include/llvm/Analysis/CallGraph.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/CallGraph.h?rev=42487&r1=42486&r2=42487&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/CallGraph.h (original) +++ llvm/trunk/include/llvm/Analysis/CallGraph.h Mon Oct 1 11:01:23 2007 @@ -189,6 +189,7 @@ inline iterator end() { return CalledFunctions.end(); } inline const_iterator begin() const { return CalledFunctions.begin(); } inline const_iterator end() const { return CalledFunctions.end(); } + inline bool empty() const { return CalledFunctions.empty(); } inline unsigned size() const { return CalledFunctions.size(); } // Subscripting operator - Return the i'th called function... Modified: llvm/trunk/include/llvm/Analysis/LoopInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/LoopInfo.h?rev=42487&r1=42486&r2=42487&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/LoopInfo.h (original) +++ llvm/trunk/include/llvm/Analysis/LoopInfo.h Mon Oct 1 11:01:23 2007 @@ -79,6 +79,7 @@ typedef std::vector::const_iterator iterator; iterator begin() const { return SubLoops.begin(); } iterator end() const { return SubLoops.end(); } + bool empty() const { return SubLoops.empty(); } /// getBlocks - Get a list of the basic blocks which make up this loop. /// Modified: llvm/trunk/include/llvm/Support/CallSite.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/CallSite.h?rev=42487&r1=42486&r2=42487&view=diff ============================================================================== --- llvm/trunk/include/llvm/Support/CallSite.h (original) +++ llvm/trunk/include/llvm/Support/CallSite.h Mon Oct 1 11:01:23 2007 @@ -110,6 +110,7 @@ return I->op_begin()+3; // Skip Function, BB, BB } arg_iterator arg_end() const { return I->op_end(); } + bool arg_empty() const { return arg_end() == arg_begin(); } unsigned arg_size() const { return unsigned(arg_end() - arg_begin()); } bool operator<(const CallSite &CS) const { From dalej at apple.com Mon Oct 1 11:03:14 2007 From: dalej at apple.com (Dale Johannesen) Date: Mon, 01 Oct 2007 16:03:14 -0000 Subject: [llvm-commits] [llvm] r42488 - in /llvm/trunk: include/llvm/Target/TargetData.h lib/Target/TargetData.cpp Message-ID: <200710011603.l91G3EoQ011433@zion.cs.uiuc.edu> Author: johannes Date: Mon Oct 1 11:03:14 2007 New Revision: 42488 URL: http://llvm.org/viewvc/llvm-project?rev=42488&view=rev Log: Add getABITypeSize, getABITypeSizeInBits Modified: llvm/trunk/include/llvm/Target/TargetData.h llvm/trunk/lib/Target/TargetData.cpp Modified: llvm/trunk/include/llvm/Target/TargetData.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetData.h?rev=42488&r1=42487&r2=42488&view=diff ============================================================================== --- llvm/trunk/include/llvm/Target/TargetData.h (original) +++ llvm/trunk/include/llvm/Target/TargetData.h Mon Oct 1 11:03:14 2007 @@ -159,10 +159,23 @@ /// type. uint64_t getTypeSize(const Type *Ty) const; + /// getABITypeSize - Return the number of bytes allocated for the specified + /// type when used as an element in a larger object, including alignment + /// padding. + uint64_t getABITypeSize(const Type *Ty) const { + unsigned char Align = getABITypeAlignment(Ty); + return (getTypeSize(Ty) + Align - 1)/Align*Align; + } + /// getTypeSizeInBits - Return the number of bits necessary to hold the /// specified type. uint64_t getTypeSizeInBits(const Type* Ty) const; + /// getABITypeSizeInBits - Return the number of bytes allocated for the + /// specified type when used as an element in a larger object, including + /// alignment padding. + uint64_t getABITypeSizeInBits(const Type* Ty) const; + /// getABITypeAlignment - Return the minimum ABI-required alignment for the /// specified type. unsigned char getABITypeAlignment(const Type *Ty) const; Modified: llvm/trunk/lib/Target/TargetData.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/TargetData.cpp?rev=42488&r1=42487&r2=42488&view=diff ============================================================================== --- llvm/trunk/lib/Target/TargetData.cpp (original) +++ llvm/trunk/lib/Target/TargetData.cpp Mon Oct 1 11:03:14 2007 @@ -471,7 +471,12 @@ return getTypeSize(Ty) * 8; } - +uint64_t TargetData::getABITypeSizeInBits(const Type *Ty) const { + if (Ty->isInteger()) + return cast(Ty)->getBitWidth(); + else + return getABITypeSize(Ty) * 8; +} /*! \param abi_or_pref Flag that determines which alignment is returned. true returns the ABI alignment, false returns the preferred alignment. From dalej at apple.com Mon Oct 1 11:08:54 2007 From: dalej at apple.com (Dale Johannesen) Date: Mon, 1 Oct 2007 09:08:54 -0700 Subject: [llvm-commits] Size of long double (X86_FP80Ty) In-Reply-To: <200709301023.44406.baldrick@free.fr> References: <200709282204.17074.baldrick@free.fr> <3AF2C9B1-D9B0-4CCD-B60F-5A1BB7E91E8A@apple.com> <200709301023.44406.baldrick@free.fr> Message-ID: <4753238B-3DAB-454B-BE0D-491C1E815E23@apple.com> The attached patch, plus the bits I've checked in, seems to work. There are no problems in the llvm testsuite but there's something funny going on in the gcc testsuite, although I'm not convinced it's the fault of this patch, so I'm not checking it in until I get that figured out. You could see if it works for you. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: ldarray.txt Url: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20071001/c7abc41d/attachment.txt -------------- next part -------------- From evan.cheng at apple.com Mon Oct 1 13:12:49 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 18:12:49 -0000 Subject: [llvm-commits] [llvm] r42492 - /llvm/trunk/lib/Target/X86/X86InstrSSE.td Message-ID: <200710011812.l91ICnk0017551@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 13:12:48 2007 New Revision: 42492 URL: http://llvm.org/viewvc/llvm-project?rev=42492&view=rev Log: Typo. X86comi doesn't read / write chain's. Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=42492&r1=42491&r2=42492&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Mon Oct 1 13:12:48 2007 @@ -32,8 +32,7 @@ def X86frsqrt : SDNode<"X86ISD::FRSQRT", SDTFPUnaryOp>; def X86frcp : SDNode<"X86ISD::FRCP", SDTFPUnaryOp>; def X86fsrl : SDNode<"X86ISD::FSRL", SDTX86FPShiftOp>; -def X86comi : SDNode<"X86ISD::COMI", SDTX86CmpTest, - [SDNPHasChain]>; +def X86comi : SDNode<"X86ISD::COMI", SDTX86CmpTest>; def X86ucomi : SDNode<"X86ISD::UCOMI", SDTX86CmpTest>; def X86s2vec : SDNode<"X86ISD::S2VEC", SDTypeProfile<1, 1, []>, []>; def X86pextrw : SDNode<"X86ISD::PEXTRW", SDTypeProfile<1, 2, []>, []>; From dpatel at apple.com Mon Oct 1 13:12:58 2007 From: dpatel at apple.com (Devang Patel) Date: Mon, 01 Oct 2007 18:12:58 -0000 Subject: [llvm-commits] [llvm] r42493 - in /llvm/trunk: lib/Transforms/Scalar/LICM.cpp test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll Message-ID: <200710011812.l91ICwiS017575@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 13:12:58 2007 New Revision: 42493 URL: http://llvm.org/viewvc/llvm-project?rev=42493&view=rev Log: Relax unsafe use check. If there is one unconditional use inside the loop then it is safe to promote value even if there is another conditional use inside the loop. Added: llvm/trunk/test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll Modified: llvm/trunk/lib/Transforms/Scalar/LICM.cpp Modified: llvm/trunk/lib/Transforms/Scalar/LICM.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LICM.cpp?rev=42493&r1=42492&r2=42493&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LICM.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LICM.cpp Mon Oct 1 13:12:58 2007 @@ -800,24 +800,15 @@ break; } - // Do not promote null values because it may be unsafe to do so. - if (isa(V)) - PointerOk = false; - - if (GetElementPtrInst *GEP = dyn_cast(V)) { - // If GEP base is NULL then the calculated address used by Store or - // Load instruction is invalid. Do not promote this value because - // it may expose load and store instruction that are covered by - // condition which may not yet folded. - if (isa(GEP->getOperand(0))) - PointerOk = false; - } - - // If value V use is not dominating loop exit then promoting - // it may expose unsafe load and store instructions unconditinally. - if (PointerOk) + // If one use of value V inside the loop is safe then it is OK to promote + // this value. On the otherside if there is not any unsafe use inside the + // looop then also it is OK to promote this value. Otherwise it is + // unsafe to promote this value. + if (PointerOk) { + bool oneSafeUse = false; + bool oneUnsafeUse = false; for(Value::use_iterator UI = V->use_begin(), UE = V->use_end(); - UI != UE && PointerOk; ++UI) { + UI != UE; ++UI) { Instruction *Use = dyn_cast(*UI); if (!Use || !CurLoop->contains(Use->getParent())) continue; @@ -825,16 +816,25 @@ ExitI = LoopExits.begin(), ExitE = LoopExits.end(); ExitI != ExitE; ++ExitI) { Instruction *Ex = *ExitI; - if (!DT->dominates(Use, Ex)){ - PointerOk = false; + if (!isa(Use) && DT->dominates(Use, Ex)) { + oneSafeUse = true; break; } + else + oneUnsafeUse = true; } - - if (!PointerOk) + + if (oneSafeUse) break; } - + + if (oneSafeUse) + PointerOk = true; + else if (!oneUnsafeUse) + PointerOk = true; + else + PointerOk = false; + } if (PointerOk) { const Type *Ty = cast(V->getType())->getElementType(); Added: llvm/trunk/test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll?rev=42493&view=auto ============================================================================== --- llvm/trunk/test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll (added) +++ llvm/trunk/test/Transforms/LICM/2007-10-01-PromoteSafeValue.ll Mon Oct 1 13:12:58 2007 @@ -0,0 +1,21 @@ +; RUN: llvm-as < %s | opt -licm | llvm-dis | grep promoted +; Promote value if at least one use is safe + + +define i32 @f2(i32* %p, i8* %q) { +entry: + br label %loop.head + +loop.head: ; preds = %cond.true, %entry + store i32 20, i32* %p + %tmp3.i = icmp eq i8* null, %q ; [#uses=1] + br i1 %tmp3.i, label %exit, label %cond.true + +cond.true: ; preds = %loop.head + store i32 40, i32* %p + br label %loop.head + +exit: ; preds = %loop.head + ret i32 0 +} + From clattner at apple.com Mon Oct 1 13:38:39 2007 From: clattner at apple.com (Chris Lattner) Date: Mon, 1 Oct 2007 11:38:39 -0700 Subject: [llvm-commits] [llvm] r42484 - in /llvm/trunk/lib: CodeGen/SelectionDAG/LegalizeDAG.cpp CodeGen/SelectionDAG/SelectionDAG.cpp ExecutionEngine/ExecutionEngine.cpp VMCore/ConstantFold.cpp In-Reply-To: <20071001130748.GN19035@daikokuya.co.uk> References: <200709301819.l8UIJ3wE003992@zion.cs.uiuc.edu> <20071001130748.GN19035@daikokuya.co.uk> Message-ID: On Oct 1, 2007, at 6:07 AM, Neil Booth wrote: >> Author: johannes >> Date: Sun Sep 30 13:19:03 2007 >> New Revision: 42484 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=42484&view=rev >> Log: >> Constant fold int-to-long-double conversions; >> use APFloat for int-to-float/double; use >> round-to-nearest for these (implementation-defined, >> seems to match gcc). > > Seems sensible. Ultimately FENV_ACCESS pragma should be taken into > account, for C at least. Unfortunately, LLVM doesn't propagate this down into the optimizers (yet). -Chris From sabre at nondot.org Mon Oct 1 14:22:35 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 14:22:35 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/index.cgi Message-ID: <200710011922.l91JMZhf020616@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: index.cgi updated: 1.75 -> 1.76 --- Log message: only log and mail if the code is not the default code. --- Diffs of the changes: (+17 -11) index.cgi | 28 +++++++++++++++++----------- 1 files changed, 17 insertions(+), 11 deletions(-) Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.75 llvm-www/demo/index.cgi:1.76 --- llvm-www/demo/index.cgi:1.75 Sun Sep 30 17:23:13 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 14:21:45 2007 @@ -18,10 +18,10 @@ my $LOGFILE = "$ROOT/log.txt"; my $FORM_URL = 'index.cgi'; +my $MAILADDR = 'sabre at nondot.org'; my $CONTACT_ADDRESS = 'Questions or comments? Email the LLVMdev mailing list.'; my $LOGO_IMAGE_URL = 'cathead.png'; my $TIMEOUTAMOUNT = 20; -$ENV{'LLVM_LIB_SEARCH_PATH'} = '/home/vadve/gaeke/llvm/Release/runtime/'; $ENV{'LD_LIBRARY_PATH'} = '/home/vadve/shared/localtools/fc1/lib/'; my @PREPENDPATHDIRS = @@ -29,6 +29,12 @@ '/home/vadve/shared/llvm-gcc4.0-2.1/bin/', '/home/vadve/shared/llvm-2.1/Release/bin'); +my $defaultsrc = "#include \n#include \n\n" . + "int power(int X) {\n if (X == 0) return 1;\n" . + " return X*power(X-1);\n}\n\n" . + "int main(int argc, char **argv) {\n" . + " printf(\"%d\\n\", power(atoi(argv[0])));\n}\n"; + sub getname { my ($extension) = @_; for ( my $count = 0 ; ; $count++ ) { @@ -166,14 +172,8 @@ my $source = $c->param('source'); -# Start the user out with something valid. -if (!defined($source)) { - $source = "#include \n#include \n\n" . - "int power(int X) {\n if (X == 0) return 1;\n" . - " return X*power(X-1);\n}\n\n" . - "int main(int argc, char **argv) {\n" . - " printf(\"%d\\n\", power(atoi(argv[0])));\n}\n"; -} +# Start the user out with something valid if no code. +$source = $defaultsrc if (!defined($source)); print $c->textarea( -name => "source", @@ -448,6 +448,12 @@ try_run("llvm2cpp","llvm2cpp $bytecodeFile -o $l2cppFile 2>&1", $l2cppFile); } + + # Get the source presented by the user to CGI, convert newline sequences to simple \n. + my $actualsrc = $c->param('source'); + $actualsrc =~ s/\015\012/\n/go; + # Don't log this or mail it if it is the default code. + if ($actualsrc ne $defaultsrc) { addlog( $source, $pid, $UnhilightedResult ); my ( $ip, $host, $lg, $lines ); @@ -455,7 +461,7 @@ $lg = $c->param('language'); $ip = $c->remote_addr(); chomp( $host = `host $ip` ) if $ip; - mailto( 'sabre at nondot.org', + mailto( $MAILADDR', "--- Query: ---\nFrom: ($ip) $host\nInput: $lines lines of $lg\n" . "C++ demangle = " . ( $c->param('cxxdemangle') ? 1 : 0 ) @@ -465,7 +471,7 @@ . ( $c->param('showstats') ? 1 : 0 ) . "\n\n" . "--- Source: ---\n$source\n" . "--- Result: ---\n$UnhilightedResult\n" ); - + } unlink( $inputFile, $bytecodeFile, $outputFile, $disassemblyFile ); } From sabre at nondot.org Mon Oct 1 14:30:20 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 14:30:20 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/index.cgi Message-ID: <200710011930.l91JUKGW020943@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: index.cgi updated: 1.76 -> 1.77 --- Log message: fix typo --- Diffs of the changes: (+1 -1) index.cgi | 2 +- 1 files changed, 1 insertion(+), 1 deletion(-) Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.76 llvm-www/demo/index.cgi:1.77 --- llvm-www/demo/index.cgi:1.76 Mon Oct 1 14:21:45 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 14:29:53 2007 @@ -461,7 +461,7 @@ $lg = $c->param('language'); $ip = $c->remote_addr(); chomp( $host = `host $ip` ) if $ip; - mailto( $MAILADDR', + mailto( $MAILADDR, "--- Query: ---\nFrom: ($ip) $host\nInput: $lines lines of $lg\n" . "C++ demangle = " . ( $c->param('cxxdemangle') ? 1 : 0 ) From sabre at nondot.org Mon Oct 1 14:36:27 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 14:36:27 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/index.cgi Message-ID: <200710011936.l91JaRbi021251@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: index.cgi updated: 1.77 -> 1.78 --- Log message: don't compile before they hit submit --- Diffs of the changes: (+1 -1) index.cgi | 2 +- 1 files changed, 1 insertion(+), 1 deletion(-) Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.77 llvm-www/demo/index.cgi:1.78 --- llvm-www/demo/index.cgi:1.77 Mon Oct 1 14:29:53 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 14:36:08 2007 @@ -351,7 +351,7 @@ close $fh; } -if ($source) { +if ($c->param('source')) { print $c->hr; my $extension = $suffixes{ $c->param('language') }; barf "Unknown language; can't compile\n" unless $extension; From isanbard at gmail.com Mon Oct 1 15:02:28 2007 From: isanbard at gmail.com (Bill Wendling) Date: Mon, 01 Oct 2007 20:02:28 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42496 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Message-ID: <200710012002.l91K2SY6022595@zion.cs.uiuc.edu> Author: void Date: Mon Oct 1 15:02:27 2007 New Revision: 42496 URL: http://llvm.org/viewvc/llvm-project?rev=42496&view=rev Log: GCC uses "i32 0" for null pointers. LLVM requires a pointer type. Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/objc/objc-act.c?rev=42496&r1=42495&r2=42496&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Mon Oct 1 15:02:27 2007 @@ -2566,7 +2566,8 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2576,7 +2577,8 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2586,7 +2588,8 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2667,7 +2670,8 @@ /* struct _prop_list_t *properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -4974,7 +4978,8 @@ /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ if (flag_next_runtime || ! sel_ref_chain) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, @@ -9234,7 +9239,8 @@ (objc_protocol_template)), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + refs_expr = null_pointer_node; /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9304,7 +9310,8 @@ if (newabi) { /* 'isa' is NULL in the new ObjC abi */ - expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + expr = null_pointer_node; } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9312,7 +9319,8 @@ else { if (!objc_protocol_or_opt_ins_meth) - expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + expr = null_pointer_node; else expr = convert (build_pointer_type (objc_protocol_extension_template), build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); @@ -9326,7 +9334,8 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -9337,7 +9346,8 @@ } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -9352,7 +9362,8 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9360,7 +9371,8 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!opt_cls_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9369,7 +9381,8 @@ } /* APPLE LOCAL end radar 4695109 */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -10389,7 +10402,8 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - NUL pointer */ + ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); /* Set type */ encode_field_decl (field_decl, @@ -10446,7 +10460,8 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - NUL pointer */ + ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); /* Set type. */ encode_field_decl (field_decl, @@ -11142,7 +11157,8 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -11152,7 +11168,8 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -11164,7 +11181,8 @@ /* protocol_list = */ if (!protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ @@ -11186,7 +11204,8 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -11253,7 +11272,8 @@ /* objc_ivar_list = */ if (!ivar_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -11265,7 +11285,8 @@ /* objc_method_list = */ if (!dispatch_table) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin mainline */ @@ -11277,22 +11298,27 @@ if (flag_next_runtime) /* method_cache = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* dtable = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* subclass_list = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* sibling_class = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); } /* protocol_list = */ if (! protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (build_pointer_type @@ -11307,18 +11333,21 @@ { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ if (!objc_class_ext) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (build_pointer_type (objc_class_ext_template), @@ -11328,7 +11357,8 @@ } else /* gc_object_type = NULL */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11490,13 +11520,15 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11540,7 +11572,8 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11550,7 +11583,8 @@ /* baseMethods */ if (!baseMethods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, @@ -11560,31 +11594,34 @@ /* baseProtocols */ if (!baseProtocols) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { tree protocol_list_t_p = build_pointer_type ( xref_tag (RECORD_TYPE, get_identifier (UTAG_V2_PROTOCOL_LIST))); expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); + build_unary_op (ADDR_EXPR, baseProtocols, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* ivars */ if (!ivars) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11592,11 +11629,12 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11854,7 +11892,8 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - class_superclass_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + class_superclass_expr = null_pointer_node; flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -12043,7 +12082,8 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - super_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + super_expr = null_pointer_node; root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13303,7 +13343,8 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + refs_expr = null_pointer_node; /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From isanbard at gmail.com Mon Oct 1 15:09:53 2007 From: isanbard at gmail.com (Bill Wendling) Date: Mon, 01 Oct 2007 20:09:53 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42497 - /llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Message-ID: <200710012009.l91K9rUA022838@zion.cs.uiuc.edu> Author: void Date: Mon Oct 1 15:09:53 2007 New Revision: 42497 URL: http://llvm.org/viewvc/llvm-project?rev=42497&view=rev Log: GCC uses "i32 0" for null pointers. LLVM uses pointer types. Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c?rev=42497&r1=42496&r2=42497&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Mon Oct 1 15:09:53 2007 @@ -2576,7 +2576,8 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2586,7 +2587,8 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2596,7 +2598,8 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2677,7 +2680,8 @@ /* struct _prop_list_t *properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -4871,7 +4875,8 @@ /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ if (flag_next_runtime || ! sel_ref_chain) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, @@ -9134,7 +9139,8 @@ (objc_protocol_template)), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + refs_expr = null_pointer_node; /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9204,7 +9210,8 @@ if (newabi) { /* 'isa' is NULL in the new ObjC abi */ - expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + expr = null_pointer_node; } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9212,7 +9219,8 @@ else { if (!objc_protocol_or_opt_ins_meth) - expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + expr = null_pointer_node; else expr = convert (build_pointer_type (objc_protocol_extension_template), build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); @@ -9226,7 +9234,8 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9235,7 +9244,8 @@ } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9248,7 +9258,8 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9256,7 +9267,8 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!opt_cls_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9265,7 +9277,8 @@ } /* APPLE LOCAL end radar 4695109 */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -10291,7 +10304,8 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - NUL pointer */ + ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); /* Set type */ encode_field_decl (field_decl, @@ -10348,7 +10362,8 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - NUL pointer */ + ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); /* Set type. */ encode_field_decl (field_decl, @@ -11042,7 +11057,8 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, @@ -11050,7 +11066,8 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, @@ -11060,7 +11077,8 @@ /* protocol_list = */ if (!protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ @@ -11082,7 +11100,8 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, @@ -11149,7 +11168,8 @@ /* objc_ivar_list = */ if (!ivar_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_ivar_list_ptr, @@ -11159,7 +11179,8 @@ /* objc_method_list = */ if (!dispatch_table) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, @@ -11169,22 +11190,27 @@ if (flag_next_runtime) /* method_cache = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { /* dtable = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* subclass_list = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* sibling_class = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); } /* protocol_list = */ if (! protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (build_pointer_type @@ -11199,18 +11225,21 @@ { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ if (!objc_class_ext) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (build_pointer_type (objc_class_ext_template), @@ -11220,7 +11249,8 @@ } else /* gc_object_type = NULL */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11382,13 +11412,15 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11432,7 +11464,8 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11442,7 +11475,8 @@ /* baseMethods */ if (!baseMethods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_method_list_ptr, @@ -11452,31 +11486,34 @@ /* baseProtocols */ if (!baseProtocols) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { tree protocol_list_t_p = build_pointer_type ( xref_tag (RECORD_TYPE, get_identifier (UTAG_V2_PROTOCOL_LIST))); expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); + build_unary_op (ADDR_EXPR, baseProtocols, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* ivars */ if (!ivars) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11484,11 +11521,12 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11735,7 +11773,8 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - class_superclass_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + class_superclass_expr = null_pointer_node; flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -11927,7 +11966,8 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - super_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + super_expr = null_pointer_node; root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13190,7 +13230,8 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - NUL pointer */ + refs_expr = null_pointer_node; /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From evan.cheng at apple.com Mon Oct 1 15:44:09 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 20:44:09 -0000 Subject: [llvm-commits] [llvm] r42499 - in /llvm/trunk: include/llvm/CodeGen/LinkAllCodegenComponents.h include/llvm/CodeGen/ScheduleDAG.h lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp Message-ID: <200710012044.l91Ki9Fu024385@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 15:44:07 2007 New Revision: 42499 URL: http://llvm.org/viewvc/llvm-project?rev=42499&view=rev Log: Remove simple scheduler. Removed: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp Modified: llvm/trunk/include/llvm/CodeGen/LinkAllCodegenComponents.h llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h Modified: llvm/trunk/include/llvm/CodeGen/LinkAllCodegenComponents.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/LinkAllCodegenComponents.h?rev=42499&r1=42498&r2=42499&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/LinkAllCodegenComponents.h (original) +++ llvm/trunk/include/llvm/CodeGen/LinkAllCodegenComponents.h Mon Oct 1 15:44:07 2007 @@ -35,9 +35,6 @@ (void) llvm::createSimpleRegisterCoalescer(); - (void) llvm::createBFS_DAGScheduler(NULL, NULL, NULL); - (void) llvm::createSimpleDAGScheduler(NULL, NULL, NULL); - (void) llvm::createNoItinsDAGScheduler(NULL, NULL, NULL); (void) llvm::createBURRListDAGScheduler(NULL, NULL, NULL); (void) llvm::createTDRRListDAGScheduler(NULL, NULL, NULL); (void) llvm::createTDListDAGScheduler(NULL, NULL, NULL); Modified: llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h?rev=42499&r1=42498&r2=42499&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h (original) +++ llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h Mon Oct 1 15:44:07 2007 @@ -354,24 +354,6 @@ DenseMap &VRBaseMap); }; - /// createBFS_DAGScheduler - This creates a simple breadth first instruction - /// scheduler. - ScheduleDAG *createBFS_DAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB); - - /// createSimpleDAGScheduler - This creates a simple two pass instruction - /// scheduler using instruction itinerary. - ScheduleDAG* createSimpleDAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB); - - /// createNoItinsDAGScheduler - This creates a simple two pass instruction - /// scheduler without using instruction itinerary. - ScheduleDAG* createNoItinsDAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB); - /// createBURRListDAGScheduler - This creates a bottom up register usage /// reduction list scheduler. ScheduleDAG* createBURRListDAGScheduler(SelectionDAGISel *IS, Removed: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp?rev=42498&view=auto ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGSimple.cpp (removed) @@ -1,1165 +0,0 @@ -//===-- ScheduleDAGSimple.cpp - Implement a trivial DAG scheduler ---------===// -// -// The LLVM Compiler Infrastructure -// -// This file was developed by James M. Laskey and is distributed under the -// University of Illinois Open Source License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This implements a simple two pass scheduler. The first pass attempts to push -// backward any lengthy instructions and critical paths. The second pass packs -// instructions into semi-optimal time slots. -// -//===----------------------------------------------------------------------===// - -#define DEBUG_TYPE "pre-RA-sched" -#include "llvm/CodeGen/MachineFunction.h" -#include "llvm/CodeGen/ScheduleDAG.h" -#include "llvm/CodeGen/SchedulerRegistry.h" -#include "llvm/CodeGen/SelectionDAG.h" -#include "llvm/CodeGen/SSARegMap.h" -#include "llvm/Target/TargetData.h" -#include "llvm/Target/TargetMachine.h" -#include "llvm/Target/TargetInstrInfo.h" -#include "llvm/Support/Debug.h" -#include "llvm/Support/Compiler.h" -#include -using namespace llvm; - -namespace { - -static RegisterScheduler - bfsDAGScheduler("none", " No scheduling: breadth first sequencing", - createBFS_DAGScheduler); -static RegisterScheduler - simpleDAGScheduler("simple", - " Simple two pass scheduling: minimize critical path " - "and maximize processor utilization", - createSimpleDAGScheduler); -static RegisterScheduler - noitinDAGScheduler("simple-noitin", - " Simple two pass scheduling: Same as simple " - "except using generic latency", - createNoItinsDAGScheduler); - -class NodeInfo; -typedef NodeInfo *NodeInfoPtr; -typedef std::vector NIVector; -typedef std::vector::iterator NIIterator; - -//===--------------------------------------------------------------------===// -/// -/// Node group - This struct is used to manage flagged node groups. -/// -class NodeGroup { -public: - NodeGroup *Next; -private: - NIVector Members; // Group member nodes - NodeInfo *Dominator; // Node with highest latency - unsigned Latency; // Total latency of the group - int Pending; // Number of visits pending before - // adding to order - -public: - // Ctor. - NodeGroup() : Next(NULL), Dominator(NULL), Pending(0) {} - - // Accessors - inline void setDominator(NodeInfo *D) { Dominator = D; } - inline NodeInfo *getTop() { return Members.front(); } - inline NodeInfo *getBottom() { return Members.back(); } - inline NodeInfo *getDominator() { return Dominator; } - inline void setLatency(unsigned L) { Latency = L; } - inline unsigned getLatency() { return Latency; } - inline int getPending() const { return Pending; } - inline void setPending(int P) { Pending = P; } - inline int addPending(int I) { return Pending += I; } - - // Pass thru - inline bool group_empty() { return Members.empty(); } - inline NIIterator group_begin() { return Members.begin(); } - inline NIIterator group_end() { return Members.end(); } - inline void group_push_back(const NodeInfoPtr &NI) { - Members.push_back(NI); - } - inline NIIterator group_insert(NIIterator Pos, const NodeInfoPtr &NI) { - return Members.insert(Pos, NI); - } - inline void group_insert(NIIterator Pos, NIIterator First, - NIIterator Last) { - Members.insert(Pos, First, Last); - } - - static void Add(NodeInfo *D, NodeInfo *U); -}; - -//===--------------------------------------------------------------------===// -/// -/// NodeInfo - This struct tracks information used to schedule the a node. -/// -class NodeInfo { -private: - int Pending; // Number of visits pending before - // adding to order -public: - SDNode *Node; // DAG node - InstrStage *StageBegin; // First stage in itinerary - InstrStage *StageEnd; // Last+1 stage in itinerary - unsigned Latency; // Total cycles to complete instr - bool IsCall : 1; // Is function call - bool IsLoad : 1; // Is memory load - bool IsStore : 1; // Is memory store - unsigned Slot; // Node's time slot - NodeGroup *Group; // Grouping information -#ifndef NDEBUG - unsigned Preorder; // Index before scheduling -#endif - - // Ctor. - NodeInfo(SDNode *N = NULL) - : Pending(0) - , Node(N) - , StageBegin(NULL) - , StageEnd(NULL) - , Latency(0) - , IsCall(false) - , IsLoad(false) - , IsStore(false) - , Slot(0) - , Group(NULL) -#ifndef NDEBUG - , Preorder(0) -#endif - {} - - // Accessors - inline bool isInGroup() const { - assert(!Group || !Group->group_empty() && "Group with no members"); - return Group != NULL; - } - inline bool isGroupDominator() const { - return isInGroup() && Group->getDominator() == this; - } - inline int getPending() const { - return Group ? Group->getPending() : Pending; - } - inline void setPending(int P) { - if (Group) Group->setPending(P); - else Pending = P; - } - inline int addPending(int I) { - if (Group) return Group->addPending(I); - else return Pending += I; - } -}; - -//===--------------------------------------------------------------------===// -/// -/// NodeGroupIterator - Iterates over all the nodes indicated by the node -/// info. If the node is in a group then iterate over the members of the -/// group, otherwise just the node info. -/// -class NodeGroupIterator { -private: - NodeInfo *NI; // Node info - NIIterator NGI; // Node group iterator - NIIterator NGE; // Node group iterator end - bool iter_valid; - -public: - // Ctor. - NodeGroupIterator(NodeInfo *N) : NI(N), iter_valid(false) { - // If the node is in a group then set up the group iterator. Otherwise - // the group iterators will trip first time out. - assert(N && "Bad node info"); - if (N->isInGroup()) { - // get Group - NodeGroup *Group = NI->Group; - NGI = Group->group_begin(); - NGE = Group->group_end(); - // Prevent this node from being used (will be in members list - NI = NULL; - iter_valid = true; - } - } - - /// next - Return the next node info, otherwise NULL. - /// - NodeInfo *next() { - if (iter_valid) { - // If members list - if (NGI != NGE) return *NGI++; - } - // Use node as the result (may be NULL) - NodeInfo *Result = NI; - // Only use once - NI = NULL; - // Return node or NULL - return Result; - } -}; -//===--------------------------------------------------------------------===// - - -//===--------------------------------------------------------------------===// -/// -/// NodeGroupOpIterator - Iterates over all the operands of a node. If the -/// node is a member of a group, this iterates over all the operands of all -/// the members of the group. -/// -class NodeGroupOpIterator { -private: - NodeInfo *NI; // Node containing operands - NodeGroupIterator GI; // Node group iterator - SDNode::op_iterator OI; // Operand iterator - SDNode::op_iterator OE; // Operand iterator end - - /// CheckNode - Test if node has more operands. If not get the next node - /// skipping over nodes that have no operands. - void CheckNode() { - // Only if operands are exhausted first - while (OI == OE) { - // Get next node info - NodeInfo *NI = GI.next(); - // Exit if nodes are exhausted - if (!NI) return; - // Get node itself - SDNode *Node = NI->Node; - // Set up the operand iterators - OI = Node->op_begin(); - OE = Node->op_end(); - } - } - -public: - // Ctor. - NodeGroupOpIterator(NodeInfo *N) - : NI(N), GI(N), OI(SDNode::op_iterator()), OE(SDNode::op_iterator()) {} - - /// isEnd - Returns true when not more operands are available. - /// - inline bool isEnd() { CheckNode(); return OI == OE; } - - /// next - Returns the next available operand. - /// - inline SDOperand next() { - assert(OI != OE && - "Not checking for end of NodeGroupOpIterator correctly"); - return *OI++; - } -}; - - -//===----------------------------------------------------------------------===// -/// -/// BitsIterator - Provides iteration through individual bits in a bit vector. -/// -template -class BitsIterator { -private: - T Bits; // Bits left to iterate through - -public: - /// Ctor. - BitsIterator(T Initial) : Bits(Initial) {} - - /// Next - Returns the next bit set or zero if exhausted. - inline T Next() { - // Get the rightmost bit set - T Result = Bits & -Bits; - // Remove from rest - Bits &= ~Result; - // Return single bit or zero - return Result; - } -}; - -//===----------------------------------------------------------------------===// - - -//===----------------------------------------------------------------------===// -/// -/// ResourceTally - Manages the use of resources over time intervals. Each -/// item (slot) in the tally vector represents the resources used at a given -/// moment. A bit set to 1 indicates that a resource is in use, otherwise -/// available. An assumption is made that the tally is large enough to schedule -/// all current instructions (asserts otherwise.) -/// -template -class ResourceTally { -private: - std::vector Tally; // Resources used per slot - typedef typename std::vector::iterator Iter; - // Tally iterator - - /// SlotsAvailable - Returns true if all units are available. - /// - bool SlotsAvailable(Iter Begin, unsigned N, unsigned ResourceSet, - unsigned &Resource) { - assert(N && "Must check availability with N != 0"); - // Determine end of interval - Iter End = Begin + N; - assert(End <= Tally.end() && "Tally is not large enough for schedule"); - - // Iterate thru each resource - BitsIterator Resources(ResourceSet & ~*Begin); - while (unsigned Res = Resources.Next()) { - // Check if resource is available for next N slots - Iter Interval = End; - do { - Interval--; - if (*Interval & Res) break; - } while (Interval != Begin); - - // If available for N - if (Interval == Begin) { - // Success - Resource = Res; - return true; - } - } - - // No luck - Resource = 0; - return false; - } - - /// RetrySlot - Finds a good candidate slot to retry search. - Iter RetrySlot(Iter Begin, unsigned N, unsigned ResourceSet) { - assert(N && "Must check availability with N != 0"); - // Determine end of interval - Iter End = Begin + N; - assert(End <= Tally.end() && "Tally is not large enough for schedule"); - - while (Begin != End--) { - // Clear units in use - ResourceSet &= ~*End; - // If no units left then we should go no further - if (!ResourceSet) return End + 1; - } - // Made it all the way through - return Begin; - } - - /// FindAndReserveStages - Return true if the stages can be completed. If - /// so mark as busy. - bool FindAndReserveStages(Iter Begin, - InstrStage *Stage, InstrStage *StageEnd) { - // If at last stage then we're done - if (Stage == StageEnd) return true; - // Get number of cycles for current stage - unsigned N = Stage->Cycles; - // Check to see if N slots are available, if not fail - unsigned Resource; - if (!SlotsAvailable(Begin, N, Stage->Units, Resource)) return false; - // Check to see if remaining stages are available, if not fail - if (!FindAndReserveStages(Begin + N, Stage + 1, StageEnd)) return false; - // Reserve resource - Reserve(Begin, N, Resource); - // Success - return true; - } - - /// Reserve - Mark busy (set) the specified N slots. - void Reserve(Iter Begin, unsigned N, unsigned Resource) { - // Determine end of interval - Iter End = Begin + N; - assert(End <= Tally.end() && "Tally is not large enough for schedule"); - - // Set resource bit in each slot - for (; Begin < End; Begin++) - *Begin |= Resource; - } - - /// FindSlots - Starting from Begin, locate consecutive slots where all stages - /// can be completed. Returns the address of first slot. - Iter FindSlots(Iter Begin, InstrStage *StageBegin, InstrStage *StageEnd) { - // Track position - Iter Cursor = Begin; - - // Try all possible slots forward - while (true) { - // Try at cursor, if successful return position. - if (FindAndReserveStages(Cursor, StageBegin, StageEnd)) return Cursor; - // Locate a better position - Cursor = RetrySlot(Cursor + 1, StageBegin->Cycles, StageBegin->Units); - } - } - -public: - /// Initialize - Resize and zero the tally to the specified number of time - /// slots. - inline void Initialize(unsigned N) { - Tally.assign(N, 0); // Initialize tally to all zeros. - } - - // FindAndReserve - Locate an ideal slot for the specified stages and mark - // as busy. - unsigned FindAndReserve(unsigned Slot, InstrStage *StageBegin, - InstrStage *StageEnd) { - // Where to begin - Iter Begin = Tally.begin() + Slot; - // Find a free slot - Iter Where = FindSlots(Begin, StageBegin, StageEnd); - // Distance is slot number - unsigned Final = Where - Tally.begin(); - return Final; - } - -}; - -//===----------------------------------------------------------------------===// -/// -/// ScheduleDAGSimple - Simple two pass scheduler. -/// -class VISIBILITY_HIDDEN ScheduleDAGSimple : public ScheduleDAG { -private: - bool NoSched; // Just do a BFS schedule, nothing fancy - bool NoItins; // Don't use itineraries? - ResourceTally Tally; // Resource usage tally - unsigned NSlots; // Total latency - static const unsigned NotFound = ~0U; // Search marker - - unsigned NodeCount; // Number of nodes in DAG - std::map Map; // Map nodes to info - bool HasGroups; // True if there are any groups - NodeInfo *Info; // Info for nodes being scheduled - NIVector Ordering; // Emit ordering of nodes - NodeGroup *HeadNG, *TailNG; // Keep track of allocated NodeGroups - -public: - - // Ctor. - ScheduleDAGSimple(bool noSched, bool noItins, SelectionDAG &dag, - MachineBasicBlock *bb, const TargetMachine &tm) - : ScheduleDAG(dag, bb, tm), NoSched(noSched), NoItins(noItins), NSlots(0), - NodeCount(0), HasGroups(false), Info(NULL), HeadNG(NULL), TailNG(NULL) { - assert(&TII && "Target doesn't provide instr info?"); - assert(&MRI && "Target doesn't provide register info?"); - } - - virtual ~ScheduleDAGSimple() { - if (Info) - delete[] Info; - - NodeGroup *NG = HeadNG; - while (NG) { - NodeGroup *NextSU = NG->Next; - delete NG; - NG = NextSU; - } - } - - void Schedule(); - - /// getNI - Returns the node info for the specified node. - /// - NodeInfo *getNI(SDNode *Node) { return Map[Node]; } - -private: - static bool isDefiner(NodeInfo *A, NodeInfo *B); - void IncludeNode(NodeInfo *NI); - void VisitAll(); - void GatherSchedulingInfo(); - void FakeGroupDominators(); - bool isStrongDependency(NodeInfo *A, NodeInfo *B); - bool isWeakDependency(NodeInfo *A, NodeInfo *B); - void ScheduleBackward(); - void ScheduleForward(); - - void AddToGroup(NodeInfo *D, NodeInfo *U); - /// PrepareNodeInfo - Set up the basic minimum node info for scheduling. - /// - void PrepareNodeInfo(); - - /// IdentifyGroups - Put flagged nodes into groups. - /// - void IdentifyGroups(); - - /// print - Print ordering to specified output stream. - /// - void print(std::ostream &O) const; - void print(std::ostream *O) const { if (O) print(*O); } - - void dump(const char *tag) const; - - virtual void dump() const; - - /// EmitAll - Emit all nodes in schedule sorted order. - /// - void EmitAll(); - - /// printNI - Print node info. - /// - void printNI(std::ostream &O, NodeInfo *NI) const; - void printNI(std::ostream *O, NodeInfo *NI) const { if (O) printNI(*O, NI); } - - /// printChanges - Hilight changes in order caused by scheduling. - /// - void printChanges(unsigned Index) const; -}; - -//===----------------------------------------------------------------------===// -/// Special case itineraries. -/// -enum { - CallLatency = 40, // To push calls back in time - - RSInteger = 0xC0000000, // Two integer units - RSFloat = 0x30000000, // Two float units - RSLoadStore = 0x0C000000, // Two load store units - RSBranch = 0x02000000 // One branch unit -}; -static InstrStage LoadStage = { 5, RSLoadStore }; -static InstrStage StoreStage = { 2, RSLoadStore }; -static InstrStage IntStage = { 2, RSInteger }; -static InstrStage FloatStage = { 3, RSFloat }; -//===----------------------------------------------------------------------===// - -} // namespace - -//===----------------------------------------------------------------------===// - -/// PrepareNodeInfo - Set up the basic minimum node info for scheduling. -/// -void ScheduleDAGSimple::PrepareNodeInfo() { - // Allocate node information - Info = new NodeInfo[NodeCount]; - - unsigned i = 0; - for (SelectionDAG::allnodes_iterator I = DAG.allnodes_begin(), - E = DAG.allnodes_end(); I != E; ++I, ++i) { - // Fast reference to node schedule info - NodeInfo* NI = &Info[i]; - // Set up map - Map[I] = NI; - // Set node - NI->Node = I; - // Set pending visit count - NI->setPending(I->use_size()); - } -} - -/// IdentifyGroups - Put flagged nodes into groups. -/// -void ScheduleDAGSimple::IdentifyGroups() { - for (unsigned i = 0, N = NodeCount; i < N; i++) { - NodeInfo* NI = &Info[i]; - SDNode *Node = NI->Node; - - // For each operand (in reverse to only look at flags) - for (unsigned N = Node->getNumOperands(); 0 < N--;) { - // Get operand - SDOperand Op = Node->getOperand(N); - // No more flags to walk - if (Op.getValueType() != MVT::Flag) break; - // Add to node group - AddToGroup(getNI(Op.Val), NI); - // Let everyone else know - HasGroups = true; - } - } -} - -/// CountInternalUses - Returns the number of edges between the two nodes. -/// -static unsigned CountInternalUses(NodeInfo *D, NodeInfo *U) { - unsigned N = 0; - for (unsigned M = U->Node->getNumOperands(); 0 < M--;) { - SDOperand Op = U->Node->getOperand(M); - if (Op.Val == D->Node) N++; - } - - return N; -} - -//===----------------------------------------------------------------------===// -/// Add - Adds a definer and user pair to a node group. -/// -void ScheduleDAGSimple::AddToGroup(NodeInfo *D, NodeInfo *U) { - // Get current groups - NodeGroup *DGroup = D->Group; - NodeGroup *UGroup = U->Group; - // If both are members of groups - if (DGroup && UGroup) { - // There may have been another edge connecting - if (DGroup == UGroup) return; - // Add the pending users count - DGroup->addPending(UGroup->getPending()); - // For each member of the users group - NodeGroupIterator UNGI(U); - while (NodeInfo *UNI = UNGI.next() ) { - // Change the group - UNI->Group = DGroup; - // For each member of the definers group - NodeGroupIterator DNGI(D); - while (NodeInfo *DNI = DNGI.next() ) { - // Remove internal edges - DGroup->addPending(-CountInternalUses(DNI, UNI)); - } - } - // Merge the two lists - DGroup->group_insert(DGroup->group_end(), - UGroup->group_begin(), UGroup->group_end()); - } else if (DGroup) { - // Make user member of definers group - U->Group = DGroup; - // Add users uses to definers group pending - DGroup->addPending(U->Node->use_size()); - // For each member of the definers group - NodeGroupIterator DNGI(D); - while (NodeInfo *DNI = DNGI.next() ) { - // Remove internal edges - DGroup->addPending(-CountInternalUses(DNI, U)); - } - DGroup->group_push_back(U); - } else if (UGroup) { - // Make definer member of users group - D->Group = UGroup; - // Add definers uses to users group pending - UGroup->addPending(D->Node->use_size()); - // For each member of the users group - NodeGroupIterator UNGI(U); - while (NodeInfo *UNI = UNGI.next() ) { - // Remove internal edges - UGroup->addPending(-CountInternalUses(D, UNI)); - } - UGroup->group_insert(UGroup->group_begin(), D); - } else { - D->Group = U->Group = DGroup = new NodeGroup(); - DGroup->addPending(D->Node->use_size() + U->Node->use_size() - - CountInternalUses(D, U)); - DGroup->group_push_back(D); - DGroup->group_push_back(U); - - if (HeadNG == NULL) - HeadNG = DGroup; - if (TailNG != NULL) - TailNG->Next = DGroup; - TailNG = DGroup; - } -} - - -/// print - Print ordering to specified output stream. -/// -void ScheduleDAGSimple::print(std::ostream &O) const { -#ifndef NDEBUG - O << "Ordering\n"; - for (unsigned i = 0, N = Ordering.size(); i < N; i++) { - NodeInfo *NI = Ordering[i]; - printNI(O, NI); - O << "\n"; - if (NI->isGroupDominator()) { - NodeGroup *Group = NI->Group; - for (NIIterator NII = Group->group_begin(), E = Group->group_end(); - NII != E; NII++) { - O << " "; - printNI(O, *NII); - O << "\n"; - } - } - } -#endif -} - -void ScheduleDAGSimple::dump(const char *tag) const { - cerr << tag; dump(); -} - -void ScheduleDAGSimple::dump() const { - print(cerr); -} - - -/// EmitAll - Emit all nodes in schedule sorted order. -/// -void ScheduleDAGSimple::EmitAll() { - // If this is the first basic block in the function, and if it has live ins - // that need to be copied into vregs, emit the copies into the top of the - // block before emitting the code for the block. - MachineFunction &MF = DAG.getMachineFunction(); - if (&MF.front() == BB && MF.livein_begin() != MF.livein_end()) { - for (MachineFunction::livein_iterator LI = MF.livein_begin(), - E = MF.livein_end(); LI != E; ++LI) - if (LI->second) { - const TargetRegisterClass *RC = RegMap->getRegClass(LI->second); - MRI->copyRegToReg(*MF.begin(), MF.begin()->end(), LI->second, - LI->first, RC, RC); - } - } - - DenseMap VRBaseMap; - - // For each node in the ordering - for (unsigned i = 0, N = Ordering.size(); i < N; i++) { - // Get the scheduling info - NodeInfo *NI = Ordering[i]; - if (NI->isInGroup()) { - NodeGroupIterator NGI(Ordering[i]); - while (NodeInfo *NI = NGI.next()) EmitNode(NI->Node, 0, VRBaseMap); - } else { - EmitNode(NI->Node, 0, VRBaseMap); - } - } -} - -/// isFlagDefiner - Returns true if the node defines a flag result. -static bool isFlagDefiner(SDNode *A) { - unsigned N = A->getNumValues(); - return N && A->getValueType(N - 1) == MVT::Flag; -} - -/// isFlagUser - Returns true if the node uses a flag result. -/// -static bool isFlagUser(SDNode *A) { - unsigned N = A->getNumOperands(); - return N && A->getOperand(N - 1).getValueType() == MVT::Flag; -} - -/// printNI - Print node info. -/// -void ScheduleDAGSimple::printNI(std::ostream &O, NodeInfo *NI) const { -#ifndef NDEBUG - SDNode *Node = NI->Node; - O << " " - << std::hex << Node << std::dec - << ", Lat=" << NI->Latency - << ", Slot=" << NI->Slot - << ", ARITY=(" << Node->getNumOperands() << "," - << Node->getNumValues() << ")" - << " " << Node->getOperationName(&DAG); - if (isFlagDefiner(Node)) O << "<#"; - if (isFlagUser(Node)) O << ">#"; -#endif -} - -/// printChanges - Hilight changes in order caused by scheduling. -/// -void ScheduleDAGSimple::printChanges(unsigned Index) const { -#ifndef NDEBUG - // Get the ordered node count - unsigned N = Ordering.size(); - // Determine if any changes - unsigned i = 0; - for (; i < N; i++) { - NodeInfo *NI = Ordering[i]; - if (NI->Preorder != i) break; - } - - if (i < N) { - cerr << Index << ". New Ordering\n"; - - for (i = 0; i < N; i++) { - NodeInfo *NI = Ordering[i]; - cerr << " " << NI->Preorder << ". "; - printNI(cerr, NI); - cerr << "\n"; - if (NI->isGroupDominator()) { - NodeGroup *Group = NI->Group; - for (NIIterator NII = Group->group_begin(), E = Group->group_end(); - NII != E; NII++) { - cerr << " "; - printNI(cerr, *NII); - cerr << "\n"; - } - } - } - } else { - cerr << Index << ". No Changes\n"; - } -#endif -} - -//===----------------------------------------------------------------------===// -/// isDefiner - Return true if node A is a definer for B. -/// -bool ScheduleDAGSimple::isDefiner(NodeInfo *A, NodeInfo *B) { - // While there are A nodes - NodeGroupIterator NII(A); - while (NodeInfo *NI = NII.next()) { - // Extract node - SDNode *Node = NI->Node; - // While there operands in nodes of B - NodeGroupOpIterator NGOI(B); - while (!NGOI.isEnd()) { - SDOperand Op = NGOI.next(); - // If node from A defines a node in B - if (Node == Op.Val) return true; - } - } - return false; -} - -/// IncludeNode - Add node to NodeInfo vector. -/// -void ScheduleDAGSimple::IncludeNode(NodeInfo *NI) { - // Get node - SDNode *Node = NI->Node; - // Ignore entry node - if (Node->getOpcode() == ISD::EntryToken) return; - // Check current count for node - int Count = NI->getPending(); - // If the node is already in list - if (Count < 0) return; - // Decrement count to indicate a visit - Count--; - // If count has gone to zero then add node to list - if (!Count) { - // Add node - if (NI->isInGroup()) { - Ordering.push_back(NI->Group->getDominator()); - } else { - Ordering.push_back(NI); - } - // indicate node has been added - Count--; - } - // Mark as visited with new count - NI->setPending(Count); -} - -/// GatherSchedulingInfo - Get latency and resource information about each node. -/// -void ScheduleDAGSimple::GatherSchedulingInfo() { - // Get instruction itineraries for the target - const InstrItineraryData &InstrItins = TM.getInstrItineraryData(); - - // For each node - for (unsigned i = 0, N = NodeCount; i < N; i++) { - // Get node info - NodeInfo* NI = &Info[i]; - SDNode *Node = NI->Node; - - // If there are itineraries and it is a machine instruction - if (InstrItins.isEmpty() || NoItins) { - // If machine opcode - if (Node->isTargetOpcode()) { - // Get return type to guess which processing unit - MVT::ValueType VT = Node->getValueType(0); - // Get machine opcode - MachineOpCode TOpc = Node->getTargetOpcode(); - NI->IsCall = TII->isCall(TOpc); - NI->IsLoad = TII->isLoad(TOpc); - NI->IsStore = TII->isStore(TOpc); - - if (TII->isLoad(TOpc)) NI->StageBegin = &LoadStage; - else if (TII->isStore(TOpc)) NI->StageBegin = &StoreStage; - else if (MVT::isInteger(VT)) NI->StageBegin = &IntStage; - else if (MVT::isFloatingPoint(VT)) NI->StageBegin = &FloatStage; - if (NI->StageBegin) NI->StageEnd = NI->StageBegin + 1; - } - } else if (Node->isTargetOpcode()) { - // get machine opcode - MachineOpCode TOpc = Node->getTargetOpcode(); - // Check to see if it is a call - NI->IsCall = TII->isCall(TOpc); - // Get itinerary stages for instruction - unsigned II = TII->getSchedClass(TOpc); - NI->StageBegin = InstrItins.begin(II); - NI->StageEnd = InstrItins.end(II); - } - - // One slot for the instruction itself - NI->Latency = 1; - - // Add long latency for a call to push it back in time - if (NI->IsCall) NI->Latency += CallLatency; - - // Sum up all the latencies - for (InstrStage *Stage = NI->StageBegin, *E = NI->StageEnd; - Stage != E; Stage++) { - NI->Latency += Stage->Cycles; - } - - // Sum up all the latencies for max tally size - NSlots += NI->Latency; - } - - // Unify metrics if in a group - if (HasGroups) { - for (unsigned i = 0, N = NodeCount; i < N; i++) { - NodeInfo* NI = &Info[i]; - - if (NI->isInGroup()) { - NodeGroup *Group = NI->Group; - - if (!Group->getDominator()) { - NIIterator NGI = Group->group_begin(), NGE = Group->group_end(); - NodeInfo *Dominator = *NGI; - unsigned Latency = 0; - - for (NGI++; NGI != NGE; NGI++) { - NodeInfo* NGNI = *NGI; - Latency += NGNI->Latency; - if (Dominator->Latency < NGNI->Latency) Dominator = NGNI; - } - - Dominator->Latency = Latency; - Group->setDominator(Dominator); - } - } - } - } -} - -/// VisitAll - Visit each node breadth-wise to produce an initial ordering. -/// Note that the ordering in the Nodes vector is reversed. -void ScheduleDAGSimple::VisitAll() { - // Add first element to list - NodeInfo *NI = getNI(DAG.getRoot().Val); - if (NI->isInGroup()) { - Ordering.push_back(NI->Group->getDominator()); - } else { - Ordering.push_back(NI); - } - - // Iterate through all nodes that have been added - for (unsigned i = 0; i < Ordering.size(); i++) { // note: size() varies - // Visit all operands - NodeGroupOpIterator NGI(Ordering[i]); - while (!NGI.isEnd()) { - // Get next operand - SDOperand Op = NGI.next(); - // Get node - SDNode *Node = Op.Val; - // Ignore passive nodes - if (isPassiveNode(Node)) continue; - // Check out node - IncludeNode(getNI(Node)); - } - } - - // Add entry node last (IncludeNode filters entry nodes) - if (DAG.getEntryNode().Val != DAG.getRoot().Val) - Ordering.push_back(getNI(DAG.getEntryNode().Val)); - - // Reverse the order - std::reverse(Ordering.begin(), Ordering.end()); -} - -/// FakeGroupDominators - Set dominators for non-scheduling. -/// -void ScheduleDAGSimple::FakeGroupDominators() { - for (unsigned i = 0, N = NodeCount; i < N; i++) { - NodeInfo* NI = &Info[i]; - - if (NI->isInGroup()) { - NodeGroup *Group = NI->Group; - - if (!Group->getDominator()) { - Group->setDominator(NI); - } - } - } -} - -/// isStrongDependency - Return true if node A has results used by node B. -/// I.E., B must wait for latency of A. -bool ScheduleDAGSimple::isStrongDependency(NodeInfo *A, NodeInfo *B) { - // If A defines for B then it's a strong dependency or - // if a load follows a store (may be dependent but why take a chance.) - return isDefiner(A, B) || (A->IsStore && B->IsLoad); -} - -/// isWeakDependency Return true if node A produces a result that will -/// conflict with operands of B. It is assumed that we have called -/// isStrongDependency prior. -bool ScheduleDAGSimple::isWeakDependency(NodeInfo *A, NodeInfo *B) { - // TODO check for conflicting real registers and aliases -#if 0 // FIXME - Since we are in SSA form and not checking register aliasing - return A->Node->getOpcode() == ISD::EntryToken || isStrongDependency(B, A); -#else - return A->Node->getOpcode() == ISD::EntryToken; -#endif -} - -/// ScheduleBackward - Schedule instructions so that any long latency -/// instructions and the critical path get pushed back in time. Time is run in -/// reverse to allow code reuse of the Tally and eliminate the overhead of -/// biasing every slot indices against NSlots. -void ScheduleDAGSimple::ScheduleBackward() { - // Size and clear the resource tally - Tally.Initialize(NSlots); - // Get number of nodes to schedule - unsigned N = Ordering.size(); - - // For each node being scheduled - for (unsigned i = N; 0 < i--;) { - NodeInfo *NI = Ordering[i]; - // Track insertion - unsigned Slot = NotFound; - - // Compare against those previously scheduled nodes - unsigned j = i + 1; - for (; j < N; j++) { - // Get following instruction - NodeInfo *Other = Ordering[j]; - - // Check dependency against previously inserted nodes - if (isStrongDependency(NI, Other)) { - Slot = Other->Slot + Other->Latency; - break; - } else if (isWeakDependency(NI, Other)) { - Slot = Other->Slot; - break; - } - } - - // If independent of others (or first entry) - if (Slot == NotFound) Slot = 0; - -#if 0 // FIXME - measure later - // Find a slot where the needed resources are available - if (NI->StageBegin != NI->StageEnd) - Slot = Tally.FindAndReserve(Slot, NI->StageBegin, NI->StageEnd); -#endif - - // Set node slot - NI->Slot = Slot; - - // Insert sort based on slot - j = i + 1; - for (; j < N; j++) { - // Get following instruction - NodeInfo *Other = Ordering[j]; - // Should we look further (remember slots are in reverse time) - if (Slot >= Other->Slot) break; - // Shuffle other into ordering - Ordering[j - 1] = Other; - } - // Insert node in proper slot - if (j != i + 1) Ordering[j - 1] = NI; - } -} - -/// ScheduleForward - Schedule instructions to maximize packing. -/// -void ScheduleDAGSimple::ScheduleForward() { - // Size and clear the resource tally - Tally.Initialize(NSlots); - // Get number of nodes to schedule - unsigned N = Ordering.size(); - - // For each node being scheduled - for (unsigned i = 0; i < N; i++) { - NodeInfo *NI = Ordering[i]; - // Track insertion - unsigned Slot = NotFound; - - // Compare against those previously scheduled nodes - unsigned j = i; - for (; 0 < j--;) { - // Get following instruction - NodeInfo *Other = Ordering[j]; - - // Check dependency against previously inserted nodes - if (isStrongDependency(Other, NI)) { - Slot = Other->Slot + Other->Latency; - break; - } else if (Other->IsCall || isWeakDependency(Other, NI)) { - Slot = Other->Slot; - break; - } - } - - // If independent of others (or first entry) - if (Slot == NotFound) Slot = 0; - - // Find a slot where the needed resources are available - if (NI->StageBegin != NI->StageEnd) - Slot = Tally.FindAndReserve(Slot, NI->StageBegin, NI->StageEnd); - - // Set node slot - NI->Slot = Slot; - - // Insert sort based on slot - j = i; - for (; 0 < j--;) { - // Get prior instruction - NodeInfo *Other = Ordering[j]; - // Should we look further - if (Slot >= Other->Slot) break; - // Shuffle other into ordering - Ordering[j + 1] = Other; - } - // Insert node in proper slot - if (j != i) Ordering[j + 1] = NI; - } -} - -/// Schedule - Order nodes according to selected style. -/// -void ScheduleDAGSimple::Schedule() { - // Number the nodes - NodeCount = std::distance(DAG.allnodes_begin(), DAG.allnodes_end()); - - // Set up minimum info for scheduling - PrepareNodeInfo(); - // Construct node groups for flagged nodes - IdentifyGroups(); - - // Test to see if scheduling should occur - bool ShouldSchedule = NodeCount > 3 && !NoSched; - // Don't waste time if is only entry and return - if (ShouldSchedule) { - // Get latency and resource requirements - GatherSchedulingInfo(); - } else if (HasGroups) { - // Make sure all the groups have dominators - FakeGroupDominators(); - } - - // Breadth first walk of DAG - VisitAll(); - -#ifndef NDEBUG - static unsigned Count = 0; - Count++; - for (unsigned i = 0, N = Ordering.size(); i < N; i++) { - NodeInfo *NI = Ordering[i]; - NI->Preorder = i; - } -#endif - - // Don't waste time if is only entry and return - if (ShouldSchedule) { - // Push back long instructions and critical path - ScheduleBackward(); - - // Pack instructions to maximize resource utilization - ScheduleForward(); - } - - DEBUG(printChanges(Count)); - - // Emit in scheduled order - EmitAll(); -} - - -/// createSimpleDAGScheduler - This creates a simple two pass instruction -/// scheduler using instruction itinerary. -llvm::ScheduleDAG* llvm::createSimpleDAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB) { - return new ScheduleDAGSimple(false, false, *DAG, BB, DAG->getTarget()); -} - -/// createNoItinsDAGScheduler - This creates a simple two pass instruction -/// scheduler without using instruction itinerary. -llvm::ScheduleDAG* llvm::createNoItinsDAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB) { - return new ScheduleDAGSimple(false, true, *DAG, BB, DAG->getTarget()); -} - -/// createBFS_DAGScheduler - This creates a simple breadth first instruction -/// scheduler. -llvm::ScheduleDAG* llvm::createBFS_DAGScheduler(SelectionDAGISel *IS, - SelectionDAG *DAG, - MachineBasicBlock *BB) { - return new ScheduleDAGSimple(true, false, *DAG, BB, DAG->getTarget()); -} From evan.cheng at apple.com Mon Oct 1 15:45:17 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 20:45:17 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42500 - /llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp Message-ID: <200710012045.l91KjH8x024456@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 15:45:17 2007 New Revision: 42500 URL: http://llvm.org/viewvc/llvm-project?rev=42500&view=rev Log: - Always use default scheduler even at -O0. - Use local register allocator at -O0. Modified: llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp?rev=42500&r1=42499&r2=42500&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp Mon Oct 1 15:45:17 2007 @@ -185,14 +185,13 @@ // optimizer use. TheModule->setDataLayout(TheTarget->getTargetData()-> getStringRepresentation()); + + RegisterScheduler::setDefault(createDefaultScheduler); - if (optimize) { - RegisterScheduler::setDefault(createDefaultScheduler); - } else { - RegisterScheduler::setDefault(createBFS_DAGScheduler); - } - - RegisterRegAlloc::setDefault(createLinearScanRegisterAllocator); + if (optimize) + RegisterRegAlloc::setDefault(createLinearScanRegisterAllocator); + else + RegisterRegAlloc::setDefault(createLocalRegisterAllocator); if (!optimize && debug_info_level > DINFO_LEVEL_NONE) TheDebugInfo = new DebugInfo(TheModule); From evan.cheng at apple.com Mon Oct 1 15:47:54 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 20:47:54 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42501 - /llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp Message-ID: <200710012047.l91Klscl024579@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 15:47:54 2007 New Revision: 42501 URL: http://llvm.org/viewvc/llvm-project?rev=42501&view=rev Log: Merge -r42499:42500 llvm-project/llvm-gcc-4.0/trunk Modified: llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp?rev=42501&r1=42500&r2=42501&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-backend.cpp Mon Oct 1 15:47:54 2007 @@ -185,14 +185,13 @@ // optimizer use. TheModule->setDataLayout(TheTarget->getTargetData()-> getStringRepresentation()); + + RegisterScheduler::setDefault(createDefaultScheduler); - if (optimize) { - RegisterScheduler::setDefault(createDefaultScheduler); - } else { - RegisterScheduler::setDefault(createBFS_DAGScheduler); - } - - RegisterRegAlloc::setDefault(createLinearScanRegisterAllocator); + if (optimize) + RegisterRegAlloc::setDefault(createLinearScanRegisterAllocator); + else + RegisterRegAlloc::setDefault(createLocalRegisterAllocator); if (!optimize && debug_info_level > DINFO_LEVEL_NONE) TheDebugInfo = new DebugInfo(TheModule); From dpatel at apple.com Mon Oct 1 16:22:57 2007 From: dpatel at apple.com (Devang Patel) Date: Mon, 01 Oct 2007 21:22:57 -0000 Subject: [llvm-commits] [llvm] r42502 - /llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c Message-ID: <200710012122.l91LMvEa025858@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 16:22:57 2007 New Revision: 42502 URL: http://llvm.org/viewvc/llvm-project?rev=42502&view=rev Log: PR 1603 test case. Added: llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c Added: llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c?rev=42502&view=auto ============================================================================== --- llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c (added) +++ llvm/trunk/test/CFrontend/2007-10-01-BuildArrayRef.c Mon Oct 1 16:22:57 2007 @@ -0,0 +1,8 @@ +// RUN: %llvmgcc -S %s -o /dev/null |& grep "error: assignment of read-only location" +// PR 1603 +int func() +{ + const int *arr; + arr[0] = 1; +} + From dpatel at apple.com Mon Oct 1 16:24:01 2007 From: dpatel at apple.com (Devang Patel) Date: Mon, 01 Oct 2007 21:24:01 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42503 - /llvm-gcc-4.0/trunk/gcc/c-typeck.c Message-ID: <200710012124.l91LO1IU025907@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 16:24:00 2007 New Revision: 42503 URL: http://llvm.org/viewvc/llvm-project?rev=42503&view=rev Log: Fix http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20071001/054127.html Modified: llvm-gcc-4.0/trunk/gcc/c-typeck.c Modified: llvm-gcc-4.0/trunk/gcc/c-typeck.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/c-typeck.c?rev=42503&r1=42502&r2=42503&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/c-typeck.c (original) +++ llvm-gcc-4.0/trunk/gcc/c-typeck.c Mon Oct 1 16:24:00 2007 @@ -1898,7 +1898,8 @@ */ { tree ty = TREE_TYPE(TREE_TYPE(ar)); - if (TREE_CODE(ty) != ARRAY_TYPE) + if (TREE_CODE(ty) == RECORD_TYPE || TREE_CODE(ty) == UNION_TYPE + || TREE_CODE(ty) == QUAL_UNION_TYPE) ty = TYPE_MAIN_VARIANT (ty); return build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); } From dpatel at apple.com Mon Oct 1 16:27:19 2007 From: dpatel at apple.com (Devang Patel) Date: Mon, 01 Oct 2007 21:27:19 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42504 - /llvm-gcc-4.2/trunk/gcc/c-typeck.c Message-ID: <200710012127.l91LRJGq026066@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 16:27:18 2007 New Revision: 42504 URL: http://llvm.org/viewvc/llvm-project?rev=42504&view=rev Log: Fix PR1603 Modified: llvm-gcc-4.2/trunk/gcc/c-typeck.c Modified: llvm-gcc-4.2/trunk/gcc/c-typeck.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/c-typeck.c?rev=42504&r1=42503&r2=42504&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/c-typeck.c (original) +++ llvm-gcc-4.2/trunk/gcc/c-typeck.c Mon Oct 1 16:27:18 2007 @@ -2082,7 +2082,8 @@ */ { tree ty = TREE_TYPE(TREE_TYPE(ar)); - if (TREE_CODE(ty) != ARRAY_TYPE) + if (TREE_CODE(ty) == RECORD_TYPE || TREE_CODE(ty) == UNION_TYPE + || TREE_CODE(ty) == QUAL_UNION_TYPE) ty = TYPE_MAIN_VARIANT (ty); return build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); } From dalej at apple.com Mon Oct 1 17:06:32 2007 From: dalej at apple.com (Dale Johannesen) Date: Mon, 1 Oct 2007 15:06:32 -0700 Subject: [llvm-commits] PR 1660 Message-ID: <53BF6421-4CB5-4981-9DF9-76C6AEAA2A5D@apple.com> points out that llvm-gcc-4.2 doesn't build on MacOSX PowerPC machines. I believe that the gcc-4.2 target-specific code for this environment assumes that gcc's XFloat mode (==128-bit long double) is supported, and makes no provision for the possibility that it isn't. The dependency was introduced in FSF between 4.0 and 4.2. That is, this isn't a recent regression (and the way to fix it, at least the best way if not the quickest, is to implement PPC long double.) Here is your chance to prove me wrong! Has anybody ever gotten this configuration to build? From evan.cheng at apple.com Mon Oct 1 17:17:20 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 22:17:20 -0000 Subject: [llvm-commits] [llvm] r42505 - in /llvm/trunk/test/CodeGen: Generic/2006-07-03-schedulers.ll X86/2007-01-13-StackPtrIndex.ll Message-ID: <200710012217.l91MHKH8028692@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 17:17:20 2007 New Revision: 42505 URL: http://llvm.org/viewvc/llvm-project?rev=42505&view=rev Log: -pre-RA-sched=none, simple, simple-noitin are gone. Modified: llvm/trunk/test/CodeGen/Generic/2006-07-03-schedulers.ll llvm/trunk/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll Modified: llvm/trunk/test/CodeGen/Generic/2006-07-03-schedulers.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Generic/2006-07-03-schedulers.ll?rev=42505&r1=42504&r2=42505&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/Generic/2006-07-03-schedulers.ll (original) +++ llvm/trunk/test/CodeGen/Generic/2006-07-03-schedulers.ll Mon Oct 1 17:17:20 2007 @@ -1,7 +1,4 @@ -; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=none ; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=default -; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=simple -; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=simple-noitin ; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=list-td ; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=list-tdrr ; RUN: llvm-upgrade %s | llvm-as | llc -pre-RA-sched=list-burr Modified: llvm/trunk/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll?rev=42505&r1=42504&r2=42505&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll (original) +++ llvm/trunk/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll Mon Oct 1 17:17:20 2007 @@ -1,5 +1,5 @@ -; RUN: llvm-as < %s | llc -march=x86-64 -pre-RA-sched=none | grep leaq -; RUN: llvm-as < %s | llc -march=x86-64 -pre-RA-sched=none | not grep {,%rsp)} +; RUN: llvm-as < %s | llc -march=x86-64 | grep leaq +; RUN: llvm-as < %s | llc -march=x86-64 | not grep {,%rsp)} ; PR1103 target datalayout = "e-p:64:64" From sabre at nondot.org Mon Oct 1 17:23:22 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 17:23:22 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/DemoInfo.html index.cgi Message-ID: <200710012223.l91MNMrG029180@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: DemoInfo.html added (r1.1) index.cgi updated: 1.78 -> 1.79 --- Log message: move information about options out to a separate page. --- Diffs of the changes: (+77 -21) DemoInfo.html | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ index.cgi | 32 +++++++++------------------- 2 files changed, 77 insertions(+), 21 deletions(-) Index: llvm-www/demo/DemoInfo.html diff -c /dev/null llvm-www/demo/DemoInfo.html:1.1 *** /dev/null Mon Oct 1 17:22:50 2007 --- llvm-www/demo/DemoInfo.html Mon Oct 1 17:22:39 2007 *************** *** 0 **** --- 1,66 ---- + + + + + Demo page information + + + + + +

Demo page information

+ +

Press "back" or click here to return to the demo + page.

+ +

Demangle C++ names with C++ filt

+ +

+ Select this option if you want to run the output LLVM IR through "c++filt", + which converts 'mangled' C++ names to their unmangled version. + Note that LLVM code produced will not be lexically valid, but it will + be easier to understand. +

+ +

Run link-time optimizer

+ +

+ Select this option to run the LLVM link-time optimizer, which is designed to + optimize across files in your application. Since the demo page doesn't allow + you to upload multiple files at once, and does not link in any libraries, we + configured the demo page optimizer to assume there are no calls + coming in from outside the source file, allowing it to optimize more + aggressively.

+ +

Note that you have to define 'main' in your program for this + to make much of a difference. +

+ +

Show detailed pass statistics

+ +

+ Select this option to enable compilation timings and statistics from various + optimizers.

+ + +

Analyze generated bytecode

+ +

+ Select this option to run the llvm-bcanalyzer tool + on the generated bytecode, which introspects into the format of the .bc file + itself.

+ + +

Show C++ API code

+ +

+ Select this option to run the llvm2cpp tool + on the generated bytecode, which auto generates the C++ API calls that could + be used to create the .bc file. +

+ + + + Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.78 llvm-www/demo/index.cgi:1.79 --- llvm-www/demo/index.cgi:1.78 Mon Oct 1 14:36:08 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 17:22:39 2007 @@ -10,7 +10,6 @@ $| = 1; my $ROOT = "/tmp/webcompile"; -#my $ROOT = "/home/vadve/lattner/webcompile"; open( STDERR, ">&STDOUT" ) or die "can't redirect stderr to stdout"; @@ -175,6 +174,8 @@ # Start the user out with something valid if no code. $source = $defaultsrc if (!defined($source)); +print '
'; + print $c->textarea( -name => "source", -rows => 16, @@ -194,46 +195,35 @@ -default => 'C' ); -print '

'; -print '
'; +print '\n

'; + +print "

Options

"; print $c->checkbox( -name => 'cxxdemangle', -label => 'Demangle C++ names with c++filt' - ),'		'; -print -"

Click here if you want Bitter Melon to run the resulting code through c++filt. Warning, the LLVM code produced will probably not be lexically valid, but it will be easier to understand.

"; + ),' ?

'; print $c->checkbox( -name => 'linkopt', -label => 'Run link-time optimizer', -checked => 'checked' - ),"

"; -print -"

Bitter Melon can also run the link-time optimizer for you, if you like (but she won't link your code with any libraries). Note that you have to define 'main' in your program for this to make much of a difference.

"; -print "
"; + ),' ?

'; print $c->checkbox( -name => 'showstats', -label => 'Show detailed pass statistics' - ), "

"; -print -"

Would you believe that BitterMelon even has a stopwatch? Click here to get detailed compilation statistic and timings from each pass.

"; -print "
"; + ), ' ?

'; print $c->checkbox( -name => 'showbcanalysis', -label => 'Show detailed bytecode analysis' - ),"

"; -print "

And further, the cat can even produce detailed information about the bytecode file generated! Click here to get a detailed description of the various artifacts in the bytecode file your program generates.

"; - -print "
"; + ),' ?

'; print $c->checkbox( -name => 'showllvm2cpp', - -label => 'Show corresponding LLVM C++ code' - ),"

"; -print "

And, BitterMelon, that fabulously talented cat can even turn your bytecode back into LLVM C++ that would produce the same bytecode! Click here to get a C++ program that generates the equivalent bytecode for your program by direct LLVM IR calls.

"; + -label => 'Show LLVM C++ API code' + ), ' ?

'; print "

"; From djg at cray.com Mon Oct 1 17:40:21 2007 From: djg at cray.com (Dan Gohman) Date: Mon, 01 Oct 2007 22:40:21 -0000 Subject: [llvm-commits] [llvm] r42506 - /llvm/trunk/lib/CodeGen/DwarfWriter.cpp Message-ID: <200710012240.l91MeLfI030394@zion.cs.uiuc.edu> Author: djg Date: Mon Oct 1 17:40:20 2007 New Revision: 42506 URL: http://llvm.org/viewvc/llvm-project?rev=42506&view=rev Log: Move the code that emits the .file directives so that it runs after the SourceFiles list is fully filled in so that it sees all of the files. Modified: llvm/trunk/lib/CodeGen/DwarfWriter.cpp Modified: llvm/trunk/lib/CodeGen/DwarfWriter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/DwarfWriter.cpp?rev=42506&r1=42505&r2=42506&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/DwarfWriter.cpp (original) +++ llvm/trunk/lib/CodeGen/DwarfWriter.cpp Mon Oct 1 17:40:20 2007 @@ -1971,19 +1971,6 @@ if (didInitial) return; didInitial = true; - // Print out .file directives to specify files for .loc directives. - if (TAI->hasDotLocAndDotFile()) { - const UniqueVector &SourceFiles = MMI->getSourceFiles(); - const UniqueVector &Directories = MMI->getDirectories(); - for (unsigned i = 1, e = SourceFiles.size(); i <= e; ++i) { - sys::Path FullPath(Directories[SourceFiles[i].getDirectoryID()]); - bool AppendOk = FullPath.appendComponent(SourceFiles[i].getName()); - assert(AppendOk && "Could not append filename to directory!"); - Asm->EmitFile(i, FullPath.toString()); - Asm->EOL(); - } - } - // Dwarf sections base addresses. if (TAI->doesDwarfRequireFrameSection()) { Asm->SwitchToDataSection(TAI->getDwarfFrameSection()); @@ -2626,9 +2613,6 @@ MMI = mmi; shouldEmit = true; - // Emit initial sections - EmitInitial(); - // Create all the compile unit DIEs. ConstructCompileUnitDIEs(); @@ -2640,6 +2624,23 @@ // Prime section data. SectionMap.insert(TAI->getTextSection()); + + // Print out .file directives to specify files for .loc directives. These + // are printed out early so that they precede any .loc directives. + if (TAI->hasDotLocAndDotFile()) { + const UniqueVector &SourceFiles = MMI->getSourceFiles(); + const UniqueVector &Directories = MMI->getDirectories(); + for (unsigned i = 1, e = SourceFiles.size(); i <= e; ++i) { + sys::Path FullPath(Directories[SourceFiles[i].getDirectoryID()]); + bool AppendOk = FullPath.appendComponent(SourceFiles[i].getName()); + assert(AppendOk && "Could not append filename to directory!"); + Asm->EmitFile(i, FullPath.toString()); + Asm->EOL(); + } + } + + // Emit initial sections + EmitInitial(); } } From sabre at nondot.org Mon Oct 1 17:44:49 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 17:44:49 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/DemoInfo.html index.cgi Message-ID: <200710012244.l91Min4R030730@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: DemoInfo.html updated: 1.1 -> 1.2 index.cgi updated: 1.79 -> 1.80 --- Log message: Further clean up --- Diffs of the changes: (+48 -39) DemoInfo.html | 17 ++++++++++++++ index.cgi | 70 +++++++++++++++++++++++++--------------------------------- 2 files changed, 48 insertions(+), 39 deletions(-) Index: llvm-www/demo/DemoInfo.html diff -u llvm-www/demo/DemoInfo.html:1.1 llvm-www/demo/DemoInfo.html:1.2 --- llvm-www/demo/DemoInfo.html:1.1 Mon Oct 1 17:22:39 2007 +++ llvm-www/demo/DemoInfo.html Mon Oct 1 17:44:25 2007 @@ -13,6 +13,23 @@

Press "back" or click here to return to the demo page.

+

Hints and Advice

+ +
    +
  • The generated LLVM code will be easier to read if +you use stdio (e.g., printf) than iostreams (e.g., std::cout).
    • + +
  • Unused inline functions and methods are not generated. Instead +of 'class foo { void bar() {}};', +try writing 'class foo { void bar(); }; void foo::bar() {}'.
    • + +
  • If you want to try out a file that uses non-standard header files, you should + preprocess it (e.g., with the -save-temps or -E options to + gcc) then upload the result.
    • + +
+ +

Demangle C++ names with C++ filt

Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.79 llvm-www/demo/index.cgi:1.80 --- llvm-www/demo/index.cgi:1.79 Mon Oct 1 17:22:39 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 17:44:25 2007 @@ -10,6 +10,7 @@ $| = 1; my $ROOT = "/tmp/webcompile"; +#my $ROOT = "/home/vadve/lattner/webcompile"; open( STDERR, ">&STDOUT" ) or die "can't redirect stderr to stdout"; @@ -127,10 +128,6 @@ -

- The LLVM Compiler Infrastructure -

-
Try out LLVM in your browser!
@@ -156,18 +153,12 @@ print <Bitter Melon the cat says, paste a C/C++ program in the text box or upload - one from your computer, and you can see the equivalent LLVM code, meow!!

- -

If you want to try out a file that uses non-standard header files, you should - preprocess it (e.g., with the -save-temps or -E options to - gcc) then upload the result.

+ one from your computer, and you can see LLVM compile it, meow!!

END print $c->start_multipart_form( 'POST', $FORM_URL ); -print "

Type your source code in below:

\n"; - my $source = $c->param('source'); @@ -176,70 +167,71 @@ print '
'; +print "Type your source code in below: (hints and +advice)
\n"; + print $c->textarea( -name => "source", -rows => 16, -columns => 60, -default => $source -); +), "
"; -print "

Or upload a file: "; +print "Or upload a file: "; print $c->filefield( -name => 'uploaded_file', -default => '' ); print "

\n"; -print "

By the way, what language is this code in?: ", + +print '

'; + +print "

General Options

"; + +print "Source language: ", $c->radio_group( -name => 'language', -values => [ 'C', 'C++' ], -default => 'C' - ); - -print '\n

'; - -print "

Options

"; - -print $c->checkbox( - -name => 'cxxdemangle', - -label => 'Demangle C++ names with c++filt' - ),' ?

'; + ), "

"; print $c->checkbox( -name => 'linkopt', -label => 'Run link-time optimizer', -checked => 'checked' - ),' ?

'; + ),' ?
'; print $c->checkbox( -name => 'showstats', -label => 'Show detailed pass statistics' - ), ' ?

'; + ), ' ?
'; + +print $c->checkbox( + -name => 'cxxdemangle', + -label => 'Demangle C++ names' + ),' ?

'; + + +print "

Output Options

"; print $c->checkbox( -name => 'showbcanalysis', -label => 'Show detailed bytecode analysis' - ),' ?

'; + ),' ?
'; print $c->checkbox( -name => 'showllvm2cpp', -label => 'Show LLVM C++ API code' - ), ' ?

'; + ), ' ?'; print "

"; -print "

", $c->submit, "\n", $c->endform; +print "

", $c->submit(-value=> 'Compile Source Code'), + "
\n", $c->endform; print "\n

If you have questions about the LLVM code generated by the -front-end, please check the FAQ. - If the FAQ does not address it, please -send an email to the LLVMdev mailing -list.

\n

Hints:

  • The generated LLVM code will be easier to read if -you use stdio (e.g., printf) than iostreams (e.g., std::cout).
    • -
  • Unused inline functions and methods are not generated. Instead -of 'class foo { void bar() {}};', -try writing 'class foo { void bar(); }; void foo::bar() {}'.
    • -
\n"; +front-end, please check the FAQ and +the demo page hints section. +

\n"; $ENV{'PATH'} = ( join ( ':', @PREPENDPATHDIRS ) ) . ":" . $ENV{'PATH'}; From sabre at nondot.org Mon Oct 1 17:52:27 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 1 Oct 2007 17:52:27 -0500 Subject: [llvm-commits] CVS: llvm-www/demo/index.cgi Message-ID: <200710012252.l91MqRL4031243@zion.cs.uiuc.edu> Changes in directory llvm-www/demo: index.cgi updated: 1.80 -> 1.81 --- Log message: minor reformating. --- Diffs of the changes: (+9 -10) index.cgi | 19 +++++++++---------- 1 files changed, 9 insertions(+), 10 deletions(-) Index: llvm-www/demo/index.cgi diff -u llvm-www/demo/index.cgi:1.80 llvm-www/demo/index.cgi:1.81 --- llvm-www/demo/index.cgi:1.80 Mon Oct 1 17:44:25 2007 +++ llvm-www/demo/index.cgi Mon Oct 1 17:52:08 2007 @@ -131,19 +131,19 @@
Try out LLVM in your browser!
-EOF -if ($LOGO_IMAGE_URL) { - print ""; -} +
+ + +EOF if ( -f "$ROOT/locked" ) { my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,$atime,$locktime) = stat("$ROOT/locked"); my $currtime = time(); if ($locktime + 60 > $currtime) { - print "

Bitter Melon the cat says, 'this page is already in use by someone else at this "; - print "time, try reloading in a second or two. Meow!'

\n"; + print "This page is already in use by someone else at this "; + print "time, try reloading in a second or two. Meow!
'\n"; exit 0; } } @@ -151,10 +151,9 @@ system("touch $ROOT/locked"); print <Bitter Melon the cat says, paste a C/C++ program in the text box or upload - one from your computer, and you can see LLVM compile it, meow!!

- +Bitter Melon the cat says, paste a C/C++ program in the text box or upload +one from your computer, and you can see LLVM compile it, meow!! +

END print $c->start_multipart_form( 'POST', $FORM_URL ); From dalej at apple.com Mon Oct 1 18:08:35 2007 From: dalej at apple.com (Dale Johannesen) Date: Mon, 01 Oct 2007 23:08:35 -0000 Subject: [llvm-commits] [llvm] r42508 - in /llvm/trunk/lib: CodeGen/AsmPrinter.cpp CodeGen/SelectionDAG/SelectionDAGISel.cpp Transforms/Scalar/LoopStrengthReduce.cpp Message-ID: <200710012308.l91N8ZKS032108@zion.cs.uiuc.edu> Author: johannes Date: Mon Oct 1 18:08:35 2007 New Revision: 42508 URL: http://llvm.org/viewvc/llvm-project?rev=42508&view=rev Log: Fix stride computations for long double arrays. Modified: llvm/trunk/lib/CodeGen/AsmPrinter.cpp llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp Modified: llvm/trunk/lib/CodeGen/AsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter.cpp?rev=42508&r1=42507&r2=42508&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/AsmPrinter.cpp (original) +++ llvm/trunk/lib/CodeGen/AsmPrinter.cpp Mon Oct 1 18:08:35 2007 @@ -816,8 +816,12 @@ if (CVA->isString()) { EmitString(CVA); } else { // Not a string. Print the values in successive locations - for (unsigned i = 0, e = CVA->getNumOperands(); i != e; ++i) + for (unsigned i = 0, e = CVA->getNumOperands(); i != e; ++i) { EmitGlobalConstant(CVA->getOperand(i)); + const Type* EltTy = CVA->getType()->getElementType(); + uint64_t padSize = TD->getABITypeSize(EltTy) - TD->getTypeSize(EltTy); + EmitZeros(padSize); + } } return; } else if (const ConstantStruct *CVS = dyn_cast(CV)) { Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp?rev=42508&r1=42507&r2=42508&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Mon Oct 1 18:08:35 2007 @@ -2279,13 +2279,13 @@ if (ConstantInt *CI = dyn_cast(Idx)) { if (CI->getZExtValue() == 0) continue; uint64_t Offs = - TD->getTypeSize(Ty)*cast(CI)->getSExtValue(); + TD->getABITypeSize(Ty)*cast(CI)->getSExtValue(); N = DAG.getNode(ISD::ADD, N.getValueType(), N, getIntPtrConstant(Offs)); continue; } // N = N + Idx * ElementSize; - uint64_t ElementSize = TD->getTypeSize(Ty); + uint64_t ElementSize = TD->getABITypeSize(Ty); SDOperand IdxN = getValue(Idx); // If the index is smaller or larger than intptr_t, truncate or extend Modified: llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp?rev=42508&r1=42507&r2=42508&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp Mon Oct 1 18:08:35 2007 @@ -286,7 +286,7 @@ Value *OpVal = getCastedVersionOf(opcode, GEP->getOperand(i)); SCEVHandle Idx = SE->getSCEV(OpVal); - uint64_t TypeSize = TD->getTypeSize(GTI.getIndexedType()); + uint64_t TypeSize = TD->getABITypeSize(GTI.getIndexedType()); if (TypeSize != 1) Idx = SCEVMulExpr::get(Idx, SCEVConstant::get(ConstantInt::get(UIntPtrTy, From evan.cheng at apple.com Mon Oct 1 18:44:33 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Mon, 01 Oct 2007 23:44:33 -0000 Subject: [llvm-commits] [llvm] r42509 - in /llvm/trunk/lib/Target/X86: X86RegisterInfo.cpp X86RegisterInfo.h Message-ID: <200710012344.l91NiYBe001006@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 18:44:33 2007 New Revision: 42509 URL: http://llvm.org/viewvc/llvm-project?rev=42509&view=rev Log: Refactor code to add load / store folded instructions -> register only instructions reverse map. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp llvm/trunk/lib/Target/X86/X86RegisterInfo.h Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42509&r1=42508&r2=42509&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Mon Oct 1 18:44:33 2007 @@ -63,6 +63,577 @@ StackPtr = X86::ESP; FramePtr = X86::EBP; } + + SmallVector AmbEntries; + static const unsigned OpTbl2Addr[][2] = { + { X86::ADC32ri, X86::ADC32mi }, + { X86::ADC32ri8, X86::ADC32mi8 }, + { X86::ADC32rr, X86::ADC32mr }, + { X86::ADC64ri32, X86::ADC64mi32 }, + { X86::ADC64ri8, X86::ADC64mi8 }, + { X86::ADC64rr, X86::ADC64mr }, + { X86::ADD16ri, X86::ADD16mi }, + { X86::ADD16ri8, X86::ADD16mi8 }, + { X86::ADD16rr, X86::ADD16mr }, + { X86::ADD32ri, X86::ADD32mi }, + { X86::ADD32ri8, X86::ADD32mi8 }, + { X86::ADD32rr, X86::ADD32mr }, + { X86::ADD64ri32, X86::ADD64mi32 }, + { X86::ADD64ri8, X86::ADD64mi8 }, + { X86::ADD64rr, X86::ADD64mr }, + { X86::ADD8ri, X86::ADD8mi }, + { X86::ADD8rr, X86::ADD8mr }, + { X86::AND16ri, X86::AND16mi }, + { X86::AND16ri8, X86::AND16mi8 }, + { X86::AND16rr, X86::AND16mr }, + { X86::AND32ri, X86::AND32mi }, + { X86::AND32ri8, X86::AND32mi8 }, + { X86::AND32rr, X86::AND32mr }, + { X86::AND64ri32, X86::AND64mi32 }, + { X86::AND64ri8, X86::AND64mi8 }, + { X86::AND64rr, X86::AND64mr }, + { X86::AND8ri, X86::AND8mi }, + { X86::AND8rr, X86::AND8mr }, + { X86::DEC16r, X86::DEC16m }, + { X86::DEC32r, X86::DEC32m }, + { X86::DEC64_16r, X86::DEC16m }, + { X86::DEC64_32r, X86::DEC32m }, + { X86::DEC64r, X86::DEC64m }, + { X86::DEC8r, X86::DEC8m }, + { X86::INC16r, X86::INC16m }, + { X86::INC32r, X86::INC32m }, + { X86::INC64_16r, X86::INC16m }, + { X86::INC64_32r, X86::INC32m }, + { X86::INC64r, X86::INC64m }, + { X86::INC8r, X86::INC8m }, + { X86::NEG16r, X86::NEG16m }, + { X86::NEG32r, X86::NEG32m }, + { X86::NEG64r, X86::NEG64m }, + { X86::NEG8r, X86::NEG8m }, + { X86::NOT16r, X86::NOT16m }, + { X86::NOT32r, X86::NOT32m }, + { X86::NOT64r, X86::NOT64m }, + { X86::NOT8r, X86::NOT8m }, + { X86::OR16ri, X86::OR16mi }, + { X86::OR16ri8, X86::OR16mi8 }, + { X86::OR16rr, X86::OR16mr }, + { X86::OR32ri, X86::OR32mi }, + { X86::OR32ri8, X86::OR32mi8 }, + { X86::OR32rr, X86::OR32mr }, + { X86::OR64ri32, X86::OR64mi32 }, + { X86::OR64ri8, X86::OR64mi8 }, + { X86::OR64rr, X86::OR64mr }, + { X86::OR8ri, X86::OR8mi }, + { X86::OR8rr, X86::OR8mr }, + { X86::ROL16r1, X86::ROL16m1 }, + { X86::ROL16rCL, X86::ROL16mCL }, + { X86::ROL16ri, X86::ROL16mi }, + { X86::ROL32r1, X86::ROL32m1 }, + { X86::ROL32rCL, X86::ROL32mCL }, + { X86::ROL32ri, X86::ROL32mi }, + { X86::ROL64r1, X86::ROL64m1 }, + { X86::ROL64rCL, X86::ROL64mCL }, + { X86::ROL64ri, X86::ROL64mi }, + { X86::ROL8r1, X86::ROL8m1 }, + { X86::ROL8rCL, X86::ROL8mCL }, + { X86::ROL8ri, X86::ROL8mi }, + { X86::ROR16r1, X86::ROR16m1 }, + { X86::ROR16rCL, X86::ROR16mCL }, + { X86::ROR16ri, X86::ROR16mi }, + { X86::ROR32r1, X86::ROR32m1 }, + { X86::ROR32rCL, X86::ROR32mCL }, + { X86::ROR32ri, X86::ROR32mi }, + { X86::ROR64r1, X86::ROR64m1 }, + { X86::ROR64rCL, X86::ROR64mCL }, + { X86::ROR64ri, X86::ROR64mi }, + { X86::ROR8r1, X86::ROR8m1 }, + { X86::ROR8rCL, X86::ROR8mCL }, + { X86::ROR8ri, X86::ROR8mi }, + { X86::SAR16r1, X86::SAR16m1 }, + { X86::SAR16rCL, X86::SAR16mCL }, + { X86::SAR16ri, X86::SAR16mi }, + { X86::SAR32r1, X86::SAR32m1 }, + { X86::SAR32rCL, X86::SAR32mCL }, + { X86::SAR32ri, X86::SAR32mi }, + { X86::SAR64r1, X86::SAR64m1 }, + { X86::SAR64rCL, X86::SAR64mCL }, + { X86::SAR64ri, X86::SAR64mi }, + { X86::SAR8r1, X86::SAR8m1 }, + { X86::SAR8rCL, X86::SAR8mCL }, + { X86::SAR8ri, X86::SAR8mi }, + { X86::SBB32ri, X86::SBB32mi }, + { X86::SBB32ri8, X86::SBB32mi8 }, + { X86::SBB32rr, X86::SBB32mr }, + { X86::SBB64ri32, X86::SBB64mi32 }, + { X86::SBB64ri8, X86::SBB64mi8 }, + { X86::SBB64rr, X86::SBB64mr }, + { X86::SHL16r1, X86::SHL16m1 }, + { X86::SHL16rCL, X86::SHL16mCL }, + { X86::SHL16ri, X86::SHL16mi }, + { X86::SHL32r1, X86::SHL32m1 }, + { X86::SHL32rCL, X86::SHL32mCL }, + { X86::SHL32ri, X86::SHL32mi }, + { X86::SHL64r1, X86::SHL64m1 }, + { X86::SHL64rCL, X86::SHL64mCL }, + { X86::SHL64ri, X86::SHL64mi }, + { X86::SHL8r1, X86::SHL8m1 }, + { X86::SHL8rCL, X86::SHL8mCL }, + { X86::SHL8ri, X86::SHL8mi }, + { X86::SHLD16rrCL, X86::SHLD16mrCL }, + { X86::SHLD16rri8, X86::SHLD16mri8 }, + { X86::SHLD32rrCL, X86::SHLD32mrCL }, + { X86::SHLD32rri8, X86::SHLD32mri8 }, + { X86::SHLD64rrCL, X86::SHLD64mrCL }, + { X86::SHLD64rri8, X86::SHLD64mri8 }, + { X86::SHR16r1, X86::SHR16m1 }, + { X86::SHR16rCL, X86::SHR16mCL }, + { X86::SHR16ri, X86::SHR16mi }, + { X86::SHR32r1, X86::SHR32m1 }, + { X86::SHR32rCL, X86::SHR32mCL }, + { X86::SHR32ri, X86::SHR32mi }, + { X86::SHR64r1, X86::SHR64m1 }, + { X86::SHR64rCL, X86::SHR64mCL }, + { X86::SHR64ri, X86::SHR64mi }, + { X86::SHR8r1, X86::SHR8m1 }, + { X86::SHR8rCL, X86::SHR8mCL }, + { X86::SHR8ri, X86::SHR8mi }, + { X86::SHRD16rrCL, X86::SHRD16mrCL }, + { X86::SHRD16rri8, X86::SHRD16mri8 }, + { X86::SHRD32rrCL, X86::SHRD32mrCL }, + { X86::SHRD32rri8, X86::SHRD32mri8 }, + { X86::SHRD64rrCL, X86::SHRD64mrCL }, + { X86::SHRD64rri8, X86::SHRD64mri8 }, + { X86::SUB16ri, X86::SUB16mi }, + { X86::SUB16ri8, X86::SUB16mi8 }, + { X86::SUB16rr, X86::SUB16mr }, + { X86::SUB32ri, X86::SUB32mi }, + { X86::SUB32ri8, X86::SUB32mi8 }, + { X86::SUB32rr, X86::SUB32mr }, + { X86::SUB64ri32, X86::SUB64mi32 }, + { X86::SUB64ri8, X86::SUB64mi8 }, + { X86::SUB64rr, X86::SUB64mr }, + { X86::SUB8ri, X86::SUB8mi }, + { X86::SUB8rr, X86::SUB8mr }, + { X86::XOR16ri, X86::XOR16mi }, + { X86::XOR16ri8, X86::XOR16mi8 }, + { X86::XOR16rr, X86::XOR16mr }, + { X86::XOR32ri, X86::XOR32mi }, + { X86::XOR32ri8, X86::XOR32mi8 }, + { X86::XOR32rr, X86::XOR32mr }, + { X86::XOR64ri32, X86::XOR64mi32 }, + { X86::XOR64ri8, X86::XOR64mi8 }, + { X86::XOR64rr, X86::XOR64mr }, + { X86::XOR8ri, X86::XOR8mi }, + { X86::XOR8rr, X86::XOR8mr } + }; + + for (unsigned i = 0, e = array_lengthof(OpTbl2Addr); i != e; ++i) { + unsigned RegOp = OpTbl2Addr[i][0]; + unsigned MemOp = OpTbl2Addr[i][1]; + if (!RegOp2MemOpTable2Addr.insert(std::make_pair((unsigned*)RegOp, MemOp))) + assert(false && "Duplicated entries?"); + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + AmbEntries.push_back(MemOp); + } + + static const unsigned OpTbl0[][2] = { + { X86::CALL32r, X86::CALL32m }, + { X86::CALL64r, X86::CALL64m }, + { X86::CMP16ri, X86::CMP16mi }, + { X86::CMP16ri8, X86::CMP16mi8 }, + { X86::CMP32ri, X86::CMP32mi }, + { X86::CMP32ri8, X86::CMP32mi8 }, + { X86::CMP64ri32, X86::CMP64mi32 }, + { X86::CMP64ri8, X86::CMP64mi8 }, + { X86::CMP8ri, X86::CMP8mi }, + { X86::DIV16r, X86::DIV16m }, + { X86::DIV32r, X86::DIV32m }, + { X86::DIV64r, X86::DIV64m }, + { X86::DIV8r, X86::DIV8m }, + { X86::FsMOVAPDrr, X86::MOVSDmr }, + { X86::FsMOVAPSrr, X86::MOVSSmr }, + { X86::IDIV16r, X86::IDIV16m }, + { X86::IDIV32r, X86::IDIV32m }, + { X86::IDIV64r, X86::IDIV64m }, + { X86::IDIV8r, X86::IDIV8m }, + { X86::IMUL16r, X86::IMUL16m }, + { X86::IMUL32r, X86::IMUL32m }, + { X86::IMUL64r, X86::IMUL64m }, + { X86::IMUL8r, X86::IMUL8m }, + { X86::JMP32r, X86::JMP32m }, + { X86::JMP64r, X86::JMP64m }, + { X86::MOV16ri, X86::MOV16mi }, + { X86::MOV16rr, X86::MOV16mr }, + { X86::MOV32ri, X86::MOV32mi }, + { X86::MOV32rr, X86::MOV32mr }, + { X86::MOV64ri32, X86::MOV64mi32 }, + { X86::MOV64rr, X86::MOV64mr }, + { X86::MOV8ri, X86::MOV8mi }, + { X86::MOV8rr, X86::MOV8mr }, + { X86::MOVAPDrr, X86::MOVAPDmr }, + { X86::MOVAPSrr, X86::MOVAPSmr }, + { X86::MOVPDI2DIrr, X86::MOVPDI2DImr }, + { X86::MOVPQIto64rr,X86::MOVPQIto64mr }, + { X86::MOVPS2SSrr, X86::MOVPS2SSmr }, + { X86::MOVSDrr, X86::MOVSDmr }, + { X86::MOVSDto64rr, X86::MOVSDto64mr }, + { X86::MOVSS2DIrr, X86::MOVSS2DImr }, + { X86::MOVSSrr, X86::MOVSSmr }, + { X86::MOVUPDrr, X86::MOVUPDmr }, + { X86::MOVUPSrr, X86::MOVUPSmr }, + { X86::MUL16r, X86::MUL16m }, + { X86::MUL32r, X86::MUL32m }, + { X86::MUL64r, X86::MUL64m }, + { X86::MUL8r, X86::MUL8m }, + { X86::SETAEr, X86::SETAEm }, + { X86::SETAr, X86::SETAm }, + { X86::SETBEr, X86::SETBEm }, + { X86::SETBr, X86::SETBm }, + { X86::SETEr, X86::SETEm }, + { X86::SETGEr, X86::SETGEm }, + { X86::SETGr, X86::SETGm }, + { X86::SETLEr, X86::SETLEm }, + { X86::SETLr, X86::SETLm }, + { X86::SETNEr, X86::SETNEm }, + { X86::SETNPr, X86::SETNPm }, + { X86::SETNSr, X86::SETNSm }, + { X86::SETPr, X86::SETPm }, + { X86::SETSr, X86::SETSm }, + { X86::TAILJMPr, X86::TAILJMPm }, + { X86::TEST16ri, X86::TEST16mi }, + { X86::TEST32ri, X86::TEST32mi }, + { X86::TEST64ri32, X86::TEST64mi32 }, + { X86::TEST8ri, X86::TEST8mi }, + { X86::XCHG16rr, X86::XCHG16mr }, + { X86::XCHG32rr, X86::XCHG32mr }, + { X86::XCHG64rr, X86::XCHG64mr }, + { X86::XCHG8rr, X86::XCHG8mr } + }; + + for (unsigned i = 0, e = array_lengthof(OpTbl0); i != e; ++i) { + unsigned RegOp = OpTbl0[i][0]; + unsigned MemOp = OpTbl0[i][1]; + if (!RegOp2MemOpTable0.insert(std::make_pair((unsigned*)RegOp, MemOp))) + assert(false && "Duplicated entries?"); + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + AmbEntries.push_back(MemOp); + } + + static const unsigned OpTbl1[][2] = { + { X86::CMP16rr, X86::CMP16rm }, + { X86::CMP32rr, X86::CMP32rm }, + { X86::CMP64rr, X86::CMP64rm }, + { X86::CMP8rr, X86::CMP8rm }, + { X86::CVTSD2SSrr, X86::CVTSD2SSrm }, + { X86::CVTSI2SD64rr, X86::CVTSI2SD64rm }, + { X86::CVTSI2SDrr, X86::CVTSI2SDrm }, + { X86::CVTSI2SS64rr, X86::CVTSI2SS64rm }, + { X86::CVTSI2SSrr, X86::CVTSI2SSrm }, + { X86::CVTSS2SDrr, X86::CVTSS2SDrm }, + { X86::CVTTSD2SI64rr, X86::CVTTSD2SI64rm }, + { X86::CVTTSD2SIrr, X86::CVTTSD2SIrm }, + { X86::CVTTSS2SI64rr, X86::CVTTSS2SI64rm }, + { X86::CVTTSS2SIrr, X86::CVTTSS2SIrm }, + { X86::FsMOVAPDrr, X86::MOVSDrm }, + { X86::FsMOVAPSrr, X86::MOVSSrm }, + { X86::IMUL16rri, X86::IMUL16rmi }, + { X86::IMUL16rri8, X86::IMUL16rmi8 }, + { X86::IMUL32rri, X86::IMUL32rmi }, + { X86::IMUL32rri8, X86::IMUL32rmi8 }, + { X86::IMUL64rri32, X86::IMUL64rmi32 }, + { X86::IMUL64rri8, X86::IMUL64rmi8 }, + { X86::Int_CMPSDrr, X86::Int_CMPSDrm }, + { X86::Int_CMPSSrr, X86::Int_CMPSSrm }, + { X86::Int_COMISDrr, X86::Int_COMISDrm }, + { X86::Int_COMISSrr, X86::Int_COMISSrm }, + { X86::Int_CVTDQ2PDrr, X86::Int_CVTDQ2PDrm }, + { X86::Int_CVTDQ2PSrr, X86::Int_CVTDQ2PSrm }, + { X86::Int_CVTPD2DQrr, X86::Int_CVTPD2DQrm }, + { X86::Int_CVTPD2PSrr, X86::Int_CVTPD2PSrm }, + { X86::Int_CVTPS2DQrr, X86::Int_CVTPS2DQrm }, + { X86::Int_CVTPS2PDrr, X86::Int_CVTPS2PDrm }, + { X86::Int_CVTSD2SI64rr,X86::Int_CVTSD2SI64rm }, + { X86::Int_CVTSD2SIrr, X86::Int_CVTSD2SIrm }, + { X86::Int_CVTSD2SSrr, X86::Int_CVTSD2SSrm }, + { X86::Int_CVTSI2SD64rr,X86::Int_CVTSI2SD64rm }, + { X86::Int_CVTSI2SDrr, X86::Int_CVTSI2SDrm }, + { X86::Int_CVTSI2SS64rr,X86::Int_CVTSI2SS64rm }, + { X86::Int_CVTSI2SSrr, X86::Int_CVTSI2SSrm }, + { X86::Int_CVTSS2SDrr, X86::Int_CVTSS2SDrm }, + { X86::Int_CVTSS2SI64rr,X86::Int_CVTSS2SI64rm }, + { X86::Int_CVTSS2SIrr, X86::Int_CVTSS2SIrm }, + { X86::Int_CVTTPD2DQrr, X86::Int_CVTTPD2DQrm }, + { X86::Int_CVTTPS2DQrr, X86::Int_CVTTPS2DQrm }, + { X86::Int_CVTTSD2SI64rr,X86::Int_CVTTSD2SI64rm }, + { X86::Int_CVTTSD2SIrr, X86::Int_CVTTSD2SIrm }, + { X86::Int_CVTTSS2SI64rr,X86::Int_CVTTSS2SI64rm }, + { X86::Int_CVTTSS2SIrr, X86::Int_CVTTSS2SIrm }, + { X86::Int_UCOMISDrr, X86::Int_UCOMISDrm }, + { X86::Int_UCOMISSrr, X86::Int_UCOMISSrm }, + { X86::MOV16rr, X86::MOV16rm }, + { X86::MOV32rr, X86::MOV32rm }, + { X86::MOV64rr, X86::MOV64rm }, + { X86::MOV64toPQIrr, X86::MOV64toPQIrm }, + { X86::MOV64toSDrr, X86::MOV64toSDrm }, + { X86::MOV8rr, X86::MOV8rm }, + { X86::MOVAPDrr, X86::MOVAPDrm }, + { X86::MOVAPSrr, X86::MOVAPSrm }, + { X86::MOVDDUPrr, X86::MOVDDUPrm }, + { X86::MOVDI2PDIrr, X86::MOVDI2PDIrm }, + { X86::MOVDI2SSrr, X86::MOVDI2SSrm }, + { X86::MOVSD2PDrr, X86::MOVSD2PDrm }, + { X86::MOVSDrr, X86::MOVSDrm }, + { X86::MOVSHDUPrr, X86::MOVSHDUPrm }, + { X86::MOVSLDUPrr, X86::MOVSLDUPrm }, + { X86::MOVSS2PSrr, X86::MOVSS2PSrm }, + { X86::MOVSSrr, X86::MOVSSrm }, + { X86::MOVSX16rr8, X86::MOVSX16rm8 }, + { X86::MOVSX32rr16, X86::MOVSX32rm16 }, + { X86::MOVSX32rr8, X86::MOVSX32rm8 }, + { X86::MOVSX64rr16, X86::MOVSX64rm16 }, + { X86::MOVSX64rr32, X86::MOVSX64rm32 }, + { X86::MOVSX64rr8, X86::MOVSX64rm8 }, + { X86::MOVUPDrr, X86::MOVUPDrm }, + { X86::MOVUPSrr, X86::MOVUPSrm }, + { X86::MOVZX16rr8, X86::MOVZX16rm8 }, + { X86::MOVZX32rr16, X86::MOVZX32rm16 }, + { X86::MOVZX32rr8, X86::MOVZX32rm8 }, + { X86::MOVZX64rr16, X86::MOVZX64rm16 }, + { X86::MOVZX64rr8, X86::MOVZX64rm8 }, + { X86::PSHUFDri, X86::PSHUFDmi }, + { X86::PSHUFHWri, X86::PSHUFHWmi }, + { X86::PSHUFLWri, X86::PSHUFLWmi }, + { X86::PsMOVZX64rr32, X86::PsMOVZX64rm32 }, + { X86::RCPPSr, X86::RCPPSm }, + { X86::RCPPSr_Int, X86::RCPPSm_Int }, + { X86::RSQRTPSr, X86::RSQRTPSm }, + { X86::RSQRTPSr_Int, X86::RSQRTPSm_Int }, + { X86::RSQRTSSr, X86::RSQRTSSm }, + { X86::RSQRTSSr_Int, X86::RSQRTSSm_Int }, + { X86::SQRTPDr, X86::SQRTPDm }, + { X86::SQRTPDr_Int, X86::SQRTPDm_Int }, + { X86::SQRTPSr, X86::SQRTPSm }, + { X86::SQRTPSr_Int, X86::SQRTPSm_Int }, + { X86::SQRTSDr, X86::SQRTSDm }, + { X86::SQRTSDr_Int, X86::SQRTSDm_Int }, + { X86::SQRTSSr, X86::SQRTSSm }, + { X86::SQRTSSr_Int, X86::SQRTSSm_Int }, + { X86::TEST16rr, X86::TEST16rm }, + { X86::TEST32rr, X86::TEST32rm }, + { X86::TEST64rr, X86::TEST64rm }, + { X86::TEST8rr, X86::TEST8rm }, + // FIXME: TEST*rr EAX,EAX ---> CMP [mem], 0 + { X86::UCOMISDrr, X86::UCOMISDrm }, + { X86::UCOMISSrr, X86::UCOMISSrm }, + { X86::XCHG16rr, X86::XCHG16rm }, + { X86::XCHG32rr, X86::XCHG32rm }, + { X86::XCHG64rr, X86::XCHG64rm }, + { X86::XCHG8rr, X86::XCHG8rm } + }; + + for (unsigned i = 0, e = array_lengthof(OpTbl1); i != e; ++i) { + unsigned RegOp = OpTbl1[i][0]; + unsigned MemOp = OpTbl1[i][1]; + if (!RegOp2MemOpTable1.insert(std::make_pair((unsigned*)RegOp, MemOp))) + assert(false && "Duplicated entries?"); + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + AmbEntries.push_back(MemOp); + } + + static const unsigned OpTbl2[][2] = { + { X86::ADC32rr, X86::ADC32rm }, + { X86::ADC64rr, X86::ADC64rm }, + { X86::ADD16rr, X86::ADD16rm }, + { X86::ADD32rr, X86::ADD32rm }, + { X86::ADD64rr, X86::ADD64rm }, + { X86::ADD8rr, X86::ADD8rm }, + { X86::ADDPDrr, X86::ADDPDrm }, + { X86::ADDPSrr, X86::ADDPSrm }, + { X86::ADDSDrr, X86::ADDSDrm }, + { X86::ADDSSrr, X86::ADDSSrm }, + { X86::ADDSUBPDrr, X86::ADDSUBPDrm }, + { X86::ADDSUBPSrr, X86::ADDSUBPSrm }, + { X86::AND16rr, X86::AND16rm }, + { X86::AND32rr, X86::AND32rm }, + { X86::AND64rr, X86::AND64rm }, + { X86::AND8rr, X86::AND8rm }, + { X86::ANDNPDrr, X86::ANDNPDrm }, + { X86::ANDNPSrr, X86::ANDNPSrm }, + { X86::ANDPDrr, X86::ANDPDrm }, + { X86::ANDPSrr, X86::ANDPSrm }, + { X86::CMOVA16rr, X86::CMOVA16rm }, + { X86::CMOVA32rr, X86::CMOVA32rm }, + { X86::CMOVA64rr, X86::CMOVA64rm }, + { X86::CMOVAE16rr, X86::CMOVAE16rm }, + { X86::CMOVAE32rr, X86::CMOVAE32rm }, + { X86::CMOVAE64rr, X86::CMOVAE64rm }, + { X86::CMOVB16rr, X86::CMOVB16rm }, + { X86::CMOVB32rr, X86::CMOVB32rm }, + { X86::CMOVB64rr, X86::CMOVB64rm }, + { X86::CMOVBE16rr, X86::CMOVBE16rm }, + { X86::CMOVBE32rr, X86::CMOVBE32rm }, + { X86::CMOVBE64rr, X86::CMOVBE64rm }, + { X86::CMOVE16rr, X86::CMOVE16rm }, + { X86::CMOVE32rr, X86::CMOVE32rm }, + { X86::CMOVE64rr, X86::CMOVE64rm }, + { X86::CMOVG16rr, X86::CMOVG16rm }, + { X86::CMOVG32rr, X86::CMOVG32rm }, + { X86::CMOVG64rr, X86::CMOVG64rm }, + { X86::CMOVGE16rr, X86::CMOVGE16rm }, + { X86::CMOVGE32rr, X86::CMOVGE32rm }, + { X86::CMOVGE64rr, X86::CMOVGE64rm }, + { X86::CMOVL16rr, X86::CMOVL16rm }, + { X86::CMOVL32rr, X86::CMOVL32rm }, + { X86::CMOVL64rr, X86::CMOVL64rm }, + { X86::CMOVLE16rr, X86::CMOVLE16rm }, + { X86::CMOVLE32rr, X86::CMOVLE32rm }, + { X86::CMOVLE64rr, X86::CMOVLE64rm }, + { X86::CMOVNE16rr, X86::CMOVNE16rm }, + { X86::CMOVNE32rr, X86::CMOVNE32rm }, + { X86::CMOVNE64rr, X86::CMOVNE64rm }, + { X86::CMOVNP16rr, X86::CMOVNP16rm }, + { X86::CMOVNP32rr, X86::CMOVNP32rm }, + { X86::CMOVNP64rr, X86::CMOVNP64rm }, + { X86::CMOVNS16rr, X86::CMOVNS16rm }, + { X86::CMOVNS32rr, X86::CMOVNS32rm }, + { X86::CMOVNS64rr, X86::CMOVNS64rm }, + { X86::CMOVP16rr, X86::CMOVP16rm }, + { X86::CMOVP32rr, X86::CMOVP32rm }, + { X86::CMOVP64rr, X86::CMOVP64rm }, + { X86::CMOVS16rr, X86::CMOVS16rm }, + { X86::CMOVS32rr, X86::CMOVS32rm }, + { X86::CMOVS64rr, X86::CMOVS64rm }, + { X86::CMPPDrri, X86::CMPPDrmi }, + { X86::CMPPSrri, X86::CMPPSrmi }, + { X86::CMPSDrr, X86::CMPSDrm }, + { X86::CMPSSrr, X86::CMPSSrm }, + { X86::DIVPDrr, X86::DIVPDrm }, + { X86::DIVPSrr, X86::DIVPSrm }, + { X86::DIVSDrr, X86::DIVSDrm }, + { X86::DIVSSrr, X86::DIVSSrm }, + { X86::HADDPDrr, X86::HADDPDrm }, + { X86::HADDPSrr, X86::HADDPSrm }, + { X86::HSUBPDrr, X86::HSUBPDrm }, + { X86::HSUBPSrr, X86::HSUBPSrm }, + { X86::IMUL16rr, X86::IMUL16rm }, + { X86::IMUL32rr, X86::IMUL32rm }, + { X86::IMUL64rr, X86::IMUL64rm }, + { X86::MAXPDrr, X86::MAXPDrm }, + { X86::MAXPDrr_Int, X86::MAXPDrm_Int }, + { X86::MAXPSrr, X86::MAXPSrm }, + { X86::MAXPSrr_Int, X86::MAXPSrm_Int }, + { X86::MAXSDrr, X86::MAXSDrm }, + { X86::MAXSDrr_Int, X86::MAXSDrm_Int }, + { X86::MAXSSrr, X86::MAXSSrm }, + { X86::MAXSSrr_Int, X86::MAXSSrm_Int }, + { X86::MINPDrr, X86::MINPDrm }, + { X86::MINPDrr_Int, X86::MINPDrm_Int }, + { X86::MINPSrr, X86::MINPSrm }, + { X86::MINPSrr_Int, X86::MINPSrm_Int }, + { X86::MINSDrr, X86::MINSDrm }, + { X86::MINSDrr_Int, X86::MINSDrm_Int }, + { X86::MINSSrr, X86::MINSSrm }, + { X86::MINSSrr_Int, X86::MINSSrm_Int }, + { X86::MULPDrr, X86::MULPDrm }, + { X86::MULPSrr, X86::MULPSrm }, + { X86::MULSDrr, X86::MULSDrm }, + { X86::MULSSrr, X86::MULSSrm }, + { X86::OR16rr, X86::OR16rm }, + { X86::OR32rr, X86::OR32rm }, + { X86::OR64rr, X86::OR64rm }, + { X86::OR8rr, X86::OR8rm }, + { X86::ORPDrr, X86::ORPDrm }, + { X86::ORPSrr, X86::ORPSrm }, + { X86::PACKSSDWrr, X86::PACKSSDWrm }, + { X86::PACKSSWBrr, X86::PACKSSWBrm }, + { X86::PACKUSWBrr, X86::PACKUSWBrm }, + { X86::PADDBrr, X86::PADDBrm }, + { X86::PADDDrr, X86::PADDDrm }, + { X86::PADDQrr, X86::PADDQrm }, + { X86::PADDSBrr, X86::PADDSBrm }, + { X86::PADDSWrr, X86::PADDSWrm }, + { X86::PADDWrr, X86::PADDWrm }, + { X86::PANDNrr, X86::PANDNrm }, + { X86::PANDrr, X86::PANDrm }, + { X86::PAVGBrr, X86::PAVGBrm }, + { X86::PAVGWrr, X86::PAVGWrm }, + { X86::PCMPEQBrr, X86::PCMPEQBrm }, + { X86::PCMPEQDrr, X86::PCMPEQDrm }, + { X86::PCMPEQWrr, X86::PCMPEQWrm }, + { X86::PCMPGTBrr, X86::PCMPGTBrm }, + { X86::PCMPGTDrr, X86::PCMPGTDrm }, + { X86::PCMPGTWrr, X86::PCMPGTWrm }, + { X86::PINSRWrri, X86::PINSRWrmi }, + { X86::PMADDWDrr, X86::PMADDWDrm }, + { X86::PMAXSWrr, X86::PMAXSWrm }, + { X86::PMAXUBrr, X86::PMAXUBrm }, + { X86::PMINSWrr, X86::PMINSWrm }, + { X86::PMINUBrr, X86::PMINUBrm }, + { X86::PMULHUWrr, X86::PMULHUWrm }, + { X86::PMULHWrr, X86::PMULHWrm }, + { X86::PMULLWrr, X86::PMULLWrm }, + { X86::PMULUDQrr, X86::PMULUDQrm }, + { X86::PORrr, X86::PORrm }, + { X86::PSADBWrr, X86::PSADBWrm }, + { X86::PSLLDrr, X86::PSLLDrm }, + { X86::PSLLQrr, X86::PSLLQrm }, + { X86::PSLLWrr, X86::PSLLWrm }, + { X86::PSRADrr, X86::PSRADrm }, + { X86::PSRAWrr, X86::PSRAWrm }, + { X86::PSRLDrr, X86::PSRLDrm }, + { X86::PSRLQrr, X86::PSRLQrm }, + { X86::PSRLWrr, X86::PSRLWrm }, + { X86::PSUBBrr, X86::PSUBBrm }, + { X86::PSUBDrr, X86::PSUBDrm }, + { X86::PSUBSBrr, X86::PSUBSBrm }, + { X86::PSUBSWrr, X86::PSUBSWrm }, + { X86::PSUBWrr, X86::PSUBWrm }, + { X86::PUNPCKHBWrr, X86::PUNPCKHBWrm }, + { X86::PUNPCKHDQrr, X86::PUNPCKHDQrm }, + { X86::PUNPCKHQDQrr, X86::PUNPCKHQDQrm }, + { X86::PUNPCKHWDrr, X86::PUNPCKHWDrm }, + { X86::PUNPCKLBWrr, X86::PUNPCKLBWrm }, + { X86::PUNPCKLDQrr, X86::PUNPCKLDQrm }, + { X86::PUNPCKLQDQrr, X86::PUNPCKLQDQrm }, + { X86::PUNPCKLWDrr, X86::PUNPCKLWDrm }, + { X86::PXORrr, X86::PXORrm }, + { X86::SBB32rr, X86::SBB32rm }, + { X86::SBB64rr, X86::SBB64rm }, + { X86::SHUFPDrri, X86::SHUFPDrmi }, + { X86::SHUFPSrri, X86::SHUFPSrmi }, + { X86::SUB16rr, X86::SUB16rm }, + { X86::SUB32rr, X86::SUB32rm }, + { X86::SUB64rr, X86::SUB64rm }, + { X86::SUB8rr, X86::SUB8rm }, + { X86::SUBPDrr, X86::SUBPDrm }, + { X86::SUBPSrr, X86::SUBPSrm }, + { X86::SUBSDrr, X86::SUBSDrm }, + { X86::SUBSSrr, X86::SUBSSrm }, + // FIXME: TEST*rr -> swapped operand of TEST*mr. + { X86::UNPCKHPDrr, X86::UNPCKHPDrm }, + { X86::UNPCKHPSrr, X86::UNPCKHPSrm }, + { X86::UNPCKLPDrr, X86::UNPCKLPDrm }, + { X86::UNPCKLPSrr, X86::UNPCKLPSrm }, + { X86::XOR16rr, X86::XOR16rm }, + { X86::XOR32rr, X86::XOR32rm }, + { X86::XOR64rr, X86::XOR64rm }, + { X86::XOR8rr, X86::XOR8rm }, + { X86::XORPDrr, X86::XORPDrm }, + { X86::XORPSrr, X86::XORPSrm } + }; + + for (unsigned i = 0, e = array_lengthof(OpTbl2); i != e; ++i) { + unsigned RegOp = OpTbl2[i][0]; + unsigned MemOp = OpTbl2[i][1]; + if (!RegOp2MemOpTable2.insert(std::make_pair((unsigned*)RegOp, MemOp))) + assert(false && "Duplicated entries?"); + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + AmbEntries.push_back(MemOp); + } + + // Remove ambiguous entries. + for (unsigned i = 0, e = AmbEntries.size(); i != e; ++i) + MemOp2RegOpTable.erase((unsigned*)AmbEntries[i]); } // getX86RegNum - This function maps LLVM register identifiers to their X86 @@ -411,70 +982,11 @@ return MIB.addImm(0); } - -//===----------------------------------------------------------------------===// -// Efficient Lookup Table Support -//===----------------------------------------------------------------------===// - -namespace { - /// TableEntry - Maps the 'from' opcode to a fused form of the 'to' opcode. - /// - struct TableEntry { - unsigned from; // Original opcode. - unsigned to; // New opcode. - - // less operators used by STL search. - bool operator<(const TableEntry &TE) const { return from < TE.from; } - friend bool operator<(const TableEntry &TE, unsigned V) { - return TE.from < V; - } - friend bool operator<(unsigned V, const TableEntry &TE) { - return V < TE.from; - } - }; -} - -/// TableIsSorted - Return true if the table is in 'from' opcode order. -/// -static bool TableIsSorted(const TableEntry *Table, unsigned NumEntries) { - for (unsigned i = 1; i != NumEntries; ++i) - if (!(Table[i-1] < Table[i])) { - cerr << "Entries out of order " << Table[i-1].from - << " " << Table[i].from << "\n"; - return false; - } - return true; -} - -/// TableLookup - Return the table entry matching the specified opcode. -/// Otherwise return NULL. -static const TableEntry *TableLookup(const TableEntry *Table, unsigned N, - unsigned Opcode) { - const TableEntry *I = std::lower_bound(Table, Table+N, Opcode); - if (I != Table+N && I->from == Opcode) - return I; - return NULL; -} - -#ifdef NDEBUG -#define ASSERT_SORTED(TABLE) -#else -#define ASSERT_SORTED(TABLE) \ - { static bool TABLE##Checked = false; \ - if (!TABLE##Checked) { \ - assert(TableIsSorted(TABLE, array_lengthof(TABLE)) && \ - "All lookup tables must be sorted for efficient access!"); \ - TABLE##Checked = true; \ - } \ - } -#endif - MachineInstr* X86RegisterInfo::foldMemoryOperand(MachineInstr *MI, unsigned i, SmallVector &MOs) const { // Table (and size) to search - const TableEntry *OpcodeTablePtr = NULL; - unsigned OpcodeTableSize = 0; + const DenseMap *OpcodeTablePtr = NULL; bool isTwoAddrFold = false; unsigned NumOps = TII.getNumOperands(MI->getOpcode()); bool isTwoAddr = NumOps > 1 && @@ -488,170 +1000,7 @@ MI->getOperand(0).isRegister() && MI->getOperand(1).isRegister() && MI->getOperand(0).getReg() == MI->getOperand(1).getReg()) { - static const TableEntry OpcodeTable[] = { - { X86::ADC32ri, X86::ADC32mi }, - { X86::ADC32ri8, X86::ADC32mi8 }, - { X86::ADC32rr, X86::ADC32mr }, - { X86::ADC64ri32, X86::ADC64mi32 }, - { X86::ADC64ri8, X86::ADC64mi8 }, - { X86::ADC64rr, X86::ADC64mr }, - { X86::ADD16ri, X86::ADD16mi }, - { X86::ADD16ri8, X86::ADD16mi8 }, - { X86::ADD16rr, X86::ADD16mr }, - { X86::ADD32ri, X86::ADD32mi }, - { X86::ADD32ri8, X86::ADD32mi8 }, - { X86::ADD32rr, X86::ADD32mr }, - { X86::ADD64ri32, X86::ADD64mi32 }, - { X86::ADD64ri8, X86::ADD64mi8 }, - { X86::ADD64rr, X86::ADD64mr }, - { X86::ADD8ri, X86::ADD8mi }, - { X86::ADD8rr, X86::ADD8mr }, - { X86::AND16ri, X86::AND16mi }, - { X86::AND16ri8, X86::AND16mi8 }, - { X86::AND16rr, X86::AND16mr }, - { X86::AND32ri, X86::AND32mi }, - { X86::AND32ri8, X86::AND32mi8 }, - { X86::AND32rr, X86::AND32mr }, - { X86::AND64ri32, X86::AND64mi32 }, - { X86::AND64ri8, X86::AND64mi8 }, - { X86::AND64rr, X86::AND64mr }, - { X86::AND8ri, X86::AND8mi }, - { X86::AND8rr, X86::AND8mr }, - { X86::DEC16r, X86::DEC16m }, - { X86::DEC32r, X86::DEC32m }, - { X86::DEC64_16r, X86::DEC16m }, - { X86::DEC64_32r, X86::DEC32m }, - { X86::DEC64r, X86::DEC64m }, - { X86::DEC8r, X86::DEC8m }, - { X86::INC16r, X86::INC16m }, - { X86::INC32r, X86::INC32m }, - { X86::INC64_16r, X86::INC16m }, - { X86::INC64_32r, X86::INC32m }, - { X86::INC64r, X86::INC64m }, - { X86::INC8r, X86::INC8m }, - { X86::NEG16r, X86::NEG16m }, - { X86::NEG32r, X86::NEG32m }, - { X86::NEG64r, X86::NEG64m }, - { X86::NEG8r, X86::NEG8m }, - { X86::NOT16r, X86::NOT16m }, - { X86::NOT32r, X86::NOT32m }, - { X86::NOT64r, X86::NOT64m }, - { X86::NOT8r, X86::NOT8m }, - { X86::OR16ri, X86::OR16mi }, - { X86::OR16ri8, X86::OR16mi8 }, - { X86::OR16rr, X86::OR16mr }, - { X86::OR32ri, X86::OR32mi }, - { X86::OR32ri8, X86::OR32mi8 }, - { X86::OR32rr, X86::OR32mr }, - { X86::OR64ri32, X86::OR64mi32 }, - { X86::OR64ri8, X86::OR64mi8 }, - { X86::OR64rr, X86::OR64mr }, - { X86::OR8ri, X86::OR8mi }, - { X86::OR8rr, X86::OR8mr }, - { X86::ROL16r1, X86::ROL16m1 }, - { X86::ROL16rCL, X86::ROL16mCL }, - { X86::ROL16ri, X86::ROL16mi }, - { X86::ROL32r1, X86::ROL32m1 }, - { X86::ROL32rCL, X86::ROL32mCL }, - { X86::ROL32ri, X86::ROL32mi }, - { X86::ROL64r1, X86::ROL64m1 }, - { X86::ROL64rCL, X86::ROL64mCL }, - { X86::ROL64ri, X86::ROL64mi }, - { X86::ROL8r1, X86::ROL8m1 }, - { X86::ROL8rCL, X86::ROL8mCL }, - { X86::ROL8ri, X86::ROL8mi }, - { X86::ROR16r1, X86::ROR16m1 }, - { X86::ROR16rCL, X86::ROR16mCL }, - { X86::ROR16ri, X86::ROR16mi }, - { X86::ROR32r1, X86::ROR32m1 }, - { X86::ROR32rCL, X86::ROR32mCL }, - { X86::ROR32ri, X86::ROR32mi }, - { X86::ROR64r1, X86::ROR64m1 }, - { X86::ROR64rCL, X86::ROR64mCL }, - { X86::ROR64ri, X86::ROR64mi }, - { X86::ROR8r1, X86::ROR8m1 }, - { X86::ROR8rCL, X86::ROR8mCL }, - { X86::ROR8ri, X86::ROR8mi }, - { X86::SAR16r1, X86::SAR16m1 }, - { X86::SAR16rCL, X86::SAR16mCL }, - { X86::SAR16ri, X86::SAR16mi }, - { X86::SAR32r1, X86::SAR32m1 }, - { X86::SAR32rCL, X86::SAR32mCL }, - { X86::SAR32ri, X86::SAR32mi }, - { X86::SAR64r1, X86::SAR64m1 }, - { X86::SAR64rCL, X86::SAR64mCL }, - { X86::SAR64ri, X86::SAR64mi }, - { X86::SAR8r1, X86::SAR8m1 }, - { X86::SAR8rCL, X86::SAR8mCL }, - { X86::SAR8ri, X86::SAR8mi }, - { X86::SBB32ri, X86::SBB32mi }, - { X86::SBB32ri8, X86::SBB32mi8 }, - { X86::SBB32rr, X86::SBB32mr }, - { X86::SBB64ri32, X86::SBB64mi32 }, - { X86::SBB64ri8, X86::SBB64mi8 }, - { X86::SBB64rr, X86::SBB64mr }, - { X86::SHL16r1, X86::SHL16m1 }, - { X86::SHL16rCL, X86::SHL16mCL }, - { X86::SHL16ri, X86::SHL16mi }, - { X86::SHL32r1, X86::SHL32m1 }, - { X86::SHL32rCL, X86::SHL32mCL }, - { X86::SHL32ri, X86::SHL32mi }, - { X86::SHL64r1, X86::SHL64m1 }, - { X86::SHL64rCL, X86::SHL64mCL }, - { X86::SHL64ri, X86::SHL64mi }, - { X86::SHL8r1, X86::SHL8m1 }, - { X86::SHL8rCL, X86::SHL8mCL }, - { X86::SHL8ri, X86::SHL8mi }, - { X86::SHLD16rrCL, X86::SHLD16mrCL }, - { X86::SHLD16rri8, X86::SHLD16mri8 }, - { X86::SHLD32rrCL, X86::SHLD32mrCL }, - { X86::SHLD32rri8, X86::SHLD32mri8 }, - { X86::SHLD64rrCL, X86::SHLD64mrCL }, - { X86::SHLD64rri8, X86::SHLD64mri8 }, - { X86::SHR16r1, X86::SHR16m1 }, - { X86::SHR16rCL, X86::SHR16mCL }, - { X86::SHR16ri, X86::SHR16mi }, - { X86::SHR32r1, X86::SHR32m1 }, - { X86::SHR32rCL, X86::SHR32mCL }, - { X86::SHR32ri, X86::SHR32mi }, - { X86::SHR64r1, X86::SHR64m1 }, - { X86::SHR64rCL, X86::SHR64mCL }, - { X86::SHR64ri, X86::SHR64mi }, - { X86::SHR8r1, X86::SHR8m1 }, - { X86::SHR8rCL, X86::SHR8mCL }, - { X86::SHR8ri, X86::SHR8mi }, - { X86::SHRD16rrCL, X86::SHRD16mrCL }, - { X86::SHRD16rri8, X86::SHRD16mri8 }, - { X86::SHRD32rrCL, X86::SHRD32mrCL }, - { X86::SHRD32rri8, X86::SHRD32mri8 }, - { X86::SHRD64rrCL, X86::SHRD64mrCL }, - { X86::SHRD64rri8, X86::SHRD64mri8 }, - { X86::SUB16ri, X86::SUB16mi }, - { X86::SUB16ri8, X86::SUB16mi8 }, - { X86::SUB16rr, X86::SUB16mr }, - { X86::SUB32ri, X86::SUB32mi }, - { X86::SUB32ri8, X86::SUB32mi8 }, - { X86::SUB32rr, X86::SUB32mr }, - { X86::SUB64ri32, X86::SUB64mi32 }, - { X86::SUB64ri8, X86::SUB64mi8 }, - { X86::SUB64rr, X86::SUB64mr }, - { X86::SUB8ri, X86::SUB8mi }, - { X86::SUB8rr, X86::SUB8mr }, - { X86::XOR16ri, X86::XOR16mi }, - { X86::XOR16ri8, X86::XOR16mi8 }, - { X86::XOR16rr, X86::XOR16mr }, - { X86::XOR32ri, X86::XOR32mi }, - { X86::XOR32ri8, X86::XOR32mi8 }, - { X86::XOR32rr, X86::XOR32mr }, - { X86::XOR64ri32, X86::XOR64mi32 }, - { X86::XOR64ri8, X86::XOR64mi8 }, - { X86::XOR64rr, X86::XOR64mr }, - { X86::XOR8ri, X86::XOR8mi }, - { X86::XOR8rr, X86::XOR8mr } - }; - ASSERT_SORTED(OpcodeTable); - OpcodeTablePtr = OpcodeTable; - OpcodeTableSize = array_lengthof(OpcodeTable); + OpcodeTablePtr = &RegOp2MemOpTable2Addr; isTwoAddrFold = true; } else if (i == 0) { // If operand 0 if (MI->getOpcode() == X86::MOV16r0) @@ -667,398 +1016,23 @@ return NewMI; } - static const TableEntry OpcodeTable[] = { - { X86::CALL32r, X86::CALL32m }, - { X86::CALL64r, X86::CALL64m }, - { X86::CMP16ri, X86::CMP16mi }, - { X86::CMP16ri8, X86::CMP16mi8 }, - { X86::CMP32ri, X86::CMP32mi }, - { X86::CMP32ri8, X86::CMP32mi8 }, - { X86::CMP64ri32, X86::CMP64mi32 }, - { X86::CMP64ri8, X86::CMP64mi8 }, - { X86::CMP8ri, X86::CMP8mi }, - { X86::DIV16r, X86::DIV16m }, - { X86::DIV32r, X86::DIV32m }, - { X86::DIV64r, X86::DIV64m }, - { X86::DIV8r, X86::DIV8m }, - { X86::FsMOVAPDrr, X86::MOVSDmr }, - { X86::FsMOVAPSrr, X86::MOVSSmr }, - { X86::IDIV16r, X86::IDIV16m }, - { X86::IDIV32r, X86::IDIV32m }, - { X86::IDIV64r, X86::IDIV64m }, - { X86::IDIV8r, X86::IDIV8m }, - { X86::IMUL16r, X86::IMUL16m }, - { X86::IMUL32r, X86::IMUL32m }, - { X86::IMUL64r, X86::IMUL64m }, - { X86::IMUL8r, X86::IMUL8m }, - { X86::JMP32r, X86::JMP32m }, - { X86::JMP64r, X86::JMP64m }, - { X86::MOV16ri, X86::MOV16mi }, - { X86::MOV16rr, X86::MOV16mr }, - { X86::MOV32ri, X86::MOV32mi }, - { X86::MOV32rr, X86::MOV32mr }, - { X86::MOV64ri32, X86::MOV64mi32 }, - { X86::MOV64rr, X86::MOV64mr }, - { X86::MOV8ri, X86::MOV8mi }, - { X86::MOV8rr, X86::MOV8mr }, - { X86::MOVAPDrr, X86::MOVAPDmr }, - { X86::MOVAPSrr, X86::MOVAPSmr }, - { X86::MOVPDI2DIrr, X86::MOVPDI2DImr }, - { X86::MOVPQIto64rr,X86::MOVPQIto64mr }, - { X86::MOVPS2SSrr, X86::MOVPS2SSmr }, - { X86::MOVSDrr, X86::MOVSDmr }, - { X86::MOVSDto64rr, X86::MOVSDto64mr }, - { X86::MOVSS2DIrr, X86::MOVSS2DImr }, - { X86::MOVSSrr, X86::MOVSSmr }, - { X86::MOVUPDrr, X86::MOVUPDmr }, - { X86::MOVUPSrr, X86::MOVUPSmr }, - { X86::MUL16r, X86::MUL16m }, - { X86::MUL32r, X86::MUL32m }, - { X86::MUL64r, X86::MUL64m }, - { X86::MUL8r, X86::MUL8m }, - { X86::SETAEr, X86::SETAEm }, - { X86::SETAr, X86::SETAm }, - { X86::SETBEr, X86::SETBEm }, - { X86::SETBr, X86::SETBm }, - { X86::SETEr, X86::SETEm }, - { X86::SETGEr, X86::SETGEm }, - { X86::SETGr, X86::SETGm }, - { X86::SETLEr, X86::SETLEm }, - { X86::SETLr, X86::SETLm }, - { X86::SETNEr, X86::SETNEm }, - { X86::SETNPr, X86::SETNPm }, - { X86::SETNSr, X86::SETNSm }, - { X86::SETPr, X86::SETPm }, - { X86::SETSr, X86::SETSm }, - { X86::TAILJMPr, X86::TAILJMPm }, - { X86::TEST16ri, X86::TEST16mi }, - { X86::TEST32ri, X86::TEST32mi }, - { X86::TEST64ri32, X86::TEST64mi32 }, - { X86::TEST8ri, X86::TEST8mi }, - { X86::XCHG16rr, X86::XCHG16mr }, - { X86::XCHG32rr, X86::XCHG32mr }, - { X86::XCHG64rr, X86::XCHG64mr }, - { X86::XCHG8rr, X86::XCHG8mr } - }; - - ASSERT_SORTED(OpcodeTable); - OpcodeTablePtr = OpcodeTable; - OpcodeTableSize = array_lengthof(OpcodeTable); + OpcodeTablePtr = &RegOp2MemOpTable0; } else if (i == 1) { - static const TableEntry OpcodeTable[] = { - { X86::CMP16rr, X86::CMP16rm }, - { X86::CMP32rr, X86::CMP32rm }, - { X86::CMP64rr, X86::CMP64rm }, - { X86::CMP8rr, X86::CMP8rm }, - { X86::CVTSD2SSrr, X86::CVTSD2SSrm }, - { X86::CVTSI2SD64rr, X86::CVTSI2SD64rm }, - { X86::CVTSI2SDrr, X86::CVTSI2SDrm }, - { X86::CVTSI2SS64rr, X86::CVTSI2SS64rm }, - { X86::CVTSI2SSrr, X86::CVTSI2SSrm }, - { X86::CVTSS2SDrr, X86::CVTSS2SDrm }, - { X86::CVTTSD2SI64rr, X86::CVTTSD2SI64rm }, - { X86::CVTTSD2SIrr, X86::CVTTSD2SIrm }, - { X86::CVTTSS2SI64rr, X86::CVTTSS2SI64rm }, - { X86::CVTTSS2SIrr, X86::CVTTSS2SIrm }, - { X86::FsMOVAPDrr, X86::MOVSDrm }, - { X86::FsMOVAPSrr, X86::MOVSSrm }, - { X86::IMUL16rri, X86::IMUL16rmi }, - { X86::IMUL16rri8, X86::IMUL16rmi8 }, - { X86::IMUL32rri, X86::IMUL32rmi }, - { X86::IMUL32rri8, X86::IMUL32rmi8 }, - { X86::IMUL64rri32, X86::IMUL64rmi32 }, - { X86::IMUL64rri8, X86::IMUL64rmi8 }, - { X86::Int_CMPSDrr, X86::Int_CMPSDrm }, - { X86::Int_CMPSSrr, X86::Int_CMPSSrm }, - { X86::Int_COMISDrr, X86::Int_COMISDrm }, - { X86::Int_COMISSrr, X86::Int_COMISSrm }, - { X86::Int_CVTDQ2PDrr, X86::Int_CVTDQ2PDrm }, - { X86::Int_CVTDQ2PSrr, X86::Int_CVTDQ2PSrm }, - { X86::Int_CVTPD2DQrr, X86::Int_CVTPD2DQrm }, - { X86::Int_CVTPD2PSrr, X86::Int_CVTPD2PSrm }, - { X86::Int_CVTPS2DQrr, X86::Int_CVTPS2DQrm }, - { X86::Int_CVTPS2PDrr, X86::Int_CVTPS2PDrm }, - { X86::Int_CVTSD2SI64rr,X86::Int_CVTSD2SI64rm }, - { X86::Int_CVTSD2SIrr, X86::Int_CVTSD2SIrm }, - { X86::Int_CVTSD2SSrr, X86::Int_CVTSD2SSrm }, - { X86::Int_CVTSI2SD64rr,X86::Int_CVTSI2SD64rm }, - { X86::Int_CVTSI2SDrr, X86::Int_CVTSI2SDrm }, - { X86::Int_CVTSI2SS64rr,X86::Int_CVTSI2SS64rm }, - { X86::Int_CVTSI2SSrr, X86::Int_CVTSI2SSrm }, - { X86::Int_CVTSS2SDrr, X86::Int_CVTSS2SDrm }, - { X86::Int_CVTSS2SI64rr,X86::Int_CVTSS2SI64rm }, - { X86::Int_CVTSS2SIrr, X86::Int_CVTSS2SIrm }, - { X86::Int_CVTTPD2DQrr, X86::Int_CVTTPD2DQrm }, - { X86::Int_CVTTPS2DQrr, X86::Int_CVTTPS2DQrm }, - { X86::Int_CVTTSD2SI64rr,X86::Int_CVTTSD2SI64rm }, - { X86::Int_CVTTSD2SIrr, X86::Int_CVTTSD2SIrm }, - { X86::Int_CVTTSS2SI64rr,X86::Int_CVTTSS2SI64rm }, - { X86::Int_CVTTSS2SIrr, X86::Int_CVTTSS2SIrm }, - { X86::Int_UCOMISDrr, X86::Int_UCOMISDrm }, - { X86::Int_UCOMISSrr, X86::Int_UCOMISSrm }, - { X86::MOV16rr, X86::MOV16rm }, - { X86::MOV32rr, X86::MOV32rm }, - { X86::MOV64rr, X86::MOV64rm }, - { X86::MOV64toPQIrr, X86::MOV64toPQIrm }, - { X86::MOV64toSDrr, X86::MOV64toSDrm }, - { X86::MOV8rr, X86::MOV8rm }, - { X86::MOVAPDrr, X86::MOVAPDrm }, - { X86::MOVAPSrr, X86::MOVAPSrm }, - { X86::MOVDDUPrr, X86::MOVDDUPrm }, - { X86::MOVDI2PDIrr, X86::MOVDI2PDIrm }, - { X86::MOVDI2SSrr, X86::MOVDI2SSrm }, - { X86::MOVSD2PDrr, X86::MOVSD2PDrm }, - { X86::MOVSDrr, X86::MOVSDrm }, - { X86::MOVSHDUPrr, X86::MOVSHDUPrm }, - { X86::MOVSLDUPrr, X86::MOVSLDUPrm }, - { X86::MOVSS2PSrr, X86::MOVSS2PSrm }, - { X86::MOVSSrr, X86::MOVSSrm }, - { X86::MOVSX16rr8, X86::MOVSX16rm8 }, - { X86::MOVSX32rr16, X86::MOVSX32rm16 }, - { X86::MOVSX32rr8, X86::MOVSX32rm8 }, - { X86::MOVSX64rr16, X86::MOVSX64rm16 }, - { X86::MOVSX64rr32, X86::MOVSX64rm32 }, - { X86::MOVSX64rr8, X86::MOVSX64rm8 }, - { X86::MOVUPDrr, X86::MOVUPDrm }, - { X86::MOVUPSrr, X86::MOVUPSrm }, - { X86::MOVZX16rr8, X86::MOVZX16rm8 }, - { X86::MOVZX32rr16, X86::MOVZX32rm16 }, - { X86::MOVZX32rr8, X86::MOVZX32rm8 }, - { X86::MOVZX64rr16, X86::MOVZX64rm16 }, - { X86::MOVZX64rr8, X86::MOVZX64rm8 }, - { X86::PSHUFDri, X86::PSHUFDmi }, - { X86::PSHUFHWri, X86::PSHUFHWmi }, - { X86::PSHUFLWri, X86::PSHUFLWmi }, - { X86::PsMOVZX64rr32, X86::PsMOVZX64rm32 }, - { X86::RCPPSr, X86::RCPPSm }, - { X86::RCPPSr_Int, X86::RCPPSm_Int }, - { X86::RSQRTPSr, X86::RSQRTPSm }, - { X86::RSQRTPSr_Int, X86::RSQRTPSm_Int }, - { X86::RSQRTSSr, X86::RSQRTSSm }, - { X86::RSQRTSSr_Int, X86::RSQRTSSm_Int }, - { X86::SQRTPDr, X86::SQRTPDm }, - { X86::SQRTPDr_Int, X86::SQRTPDm_Int }, - { X86::SQRTPSr, X86::SQRTPSm }, - { X86::SQRTPSr_Int, X86::SQRTPSm_Int }, - { X86::SQRTSDr, X86::SQRTSDm }, - { X86::SQRTSDr_Int, X86::SQRTSDm_Int }, - { X86::SQRTSSr, X86::SQRTSSm }, - { X86::SQRTSSr_Int, X86::SQRTSSm_Int }, - { X86::TEST16rr, X86::TEST16rm }, - { X86::TEST32rr, X86::TEST32rm }, - { X86::TEST64rr, X86::TEST64rm }, - { X86::TEST8rr, X86::TEST8rm }, - // FIXME: TEST*rr EAX,EAX ---> CMP [mem], 0 - { X86::UCOMISDrr, X86::UCOMISDrm }, - { X86::UCOMISSrr, X86::UCOMISSrm }, - { X86::XCHG16rr, X86::XCHG16rm }, - { X86::XCHG32rr, X86::XCHG32rm }, - { X86::XCHG64rr, X86::XCHG64rm }, - { X86::XCHG8rr, X86::XCHG8rm } - }; - - ASSERT_SORTED(OpcodeTable); - OpcodeTablePtr = OpcodeTable; - OpcodeTableSize = array_lengthof(OpcodeTable); + OpcodeTablePtr = &RegOp2MemOpTable1; } else if (i == 2) { - static const TableEntry OpcodeTable[] = { - { X86::ADC32rr, X86::ADC32rm }, - { X86::ADC64rr, X86::ADC64rm }, - { X86::ADD16rr, X86::ADD16rm }, - { X86::ADD32rr, X86::ADD32rm }, - { X86::ADD64rr, X86::ADD64rm }, - { X86::ADD8rr, X86::ADD8rm }, - { X86::ADDPDrr, X86::ADDPDrm }, - { X86::ADDPSrr, X86::ADDPSrm }, - { X86::ADDSDrr, X86::ADDSDrm }, - { X86::ADDSSrr, X86::ADDSSrm }, - { X86::ADDSUBPDrr, X86::ADDSUBPDrm }, - { X86::ADDSUBPSrr, X86::ADDSUBPSrm }, - { X86::AND16rr, X86::AND16rm }, - { X86::AND32rr, X86::AND32rm }, - { X86::AND64rr, X86::AND64rm }, - { X86::AND8rr, X86::AND8rm }, - { X86::ANDNPDrr, X86::ANDNPDrm }, - { X86::ANDNPSrr, X86::ANDNPSrm }, - { X86::ANDPDrr, X86::ANDPDrm }, - { X86::ANDPSrr, X86::ANDPSrm }, - { X86::CMOVA16rr, X86::CMOVA16rm }, - { X86::CMOVA32rr, X86::CMOVA32rm }, - { X86::CMOVA64rr, X86::CMOVA64rm }, - { X86::CMOVAE16rr, X86::CMOVAE16rm }, - { X86::CMOVAE32rr, X86::CMOVAE32rm }, - { X86::CMOVAE64rr, X86::CMOVAE64rm }, - { X86::CMOVB16rr, X86::CMOVB16rm }, - { X86::CMOVB32rr, X86::CMOVB32rm }, - { X86::CMOVB64rr, X86::CMOVB64rm }, - { X86::CMOVBE16rr, X86::CMOVBE16rm }, - { X86::CMOVBE32rr, X86::CMOVBE32rm }, - { X86::CMOVBE64rr, X86::CMOVBE64rm }, - { X86::CMOVE16rr, X86::CMOVE16rm }, - { X86::CMOVE32rr, X86::CMOVE32rm }, - { X86::CMOVE64rr, X86::CMOVE64rm }, - { X86::CMOVG16rr, X86::CMOVG16rm }, - { X86::CMOVG32rr, X86::CMOVG32rm }, - { X86::CMOVG64rr, X86::CMOVG64rm }, - { X86::CMOVGE16rr, X86::CMOVGE16rm }, - { X86::CMOVGE32rr, X86::CMOVGE32rm }, - { X86::CMOVGE64rr, X86::CMOVGE64rm }, - { X86::CMOVL16rr, X86::CMOVL16rm }, - { X86::CMOVL32rr, X86::CMOVL32rm }, - { X86::CMOVL64rr, X86::CMOVL64rm }, - { X86::CMOVLE16rr, X86::CMOVLE16rm }, - { X86::CMOVLE32rr, X86::CMOVLE32rm }, - { X86::CMOVLE64rr, X86::CMOVLE64rm }, - { X86::CMOVNE16rr, X86::CMOVNE16rm }, - { X86::CMOVNE32rr, X86::CMOVNE32rm }, - { X86::CMOVNE64rr, X86::CMOVNE64rm }, - { X86::CMOVNP16rr, X86::CMOVNP16rm }, - { X86::CMOVNP32rr, X86::CMOVNP32rm }, - { X86::CMOVNP64rr, X86::CMOVNP64rm }, - { X86::CMOVNS16rr, X86::CMOVNS16rm }, - { X86::CMOVNS32rr, X86::CMOVNS32rm }, - { X86::CMOVNS64rr, X86::CMOVNS64rm }, - { X86::CMOVP16rr, X86::CMOVP16rm }, - { X86::CMOVP32rr, X86::CMOVP32rm }, - { X86::CMOVP64rr, X86::CMOVP64rm }, - { X86::CMOVS16rr, X86::CMOVS16rm }, - { X86::CMOVS32rr, X86::CMOVS32rm }, - { X86::CMOVS64rr, X86::CMOVS64rm }, - { X86::CMPPDrri, X86::CMPPDrmi }, - { X86::CMPPSrri, X86::CMPPSrmi }, - { X86::CMPSDrr, X86::CMPSDrm }, - { X86::CMPSSrr, X86::CMPSSrm }, - { X86::DIVPDrr, X86::DIVPDrm }, - { X86::DIVPSrr, X86::DIVPSrm }, - { X86::DIVSDrr, X86::DIVSDrm }, - { X86::DIVSSrr, X86::DIVSSrm }, - { X86::HADDPDrr, X86::HADDPDrm }, - { X86::HADDPSrr, X86::HADDPSrm }, - { X86::HSUBPDrr, X86::HSUBPDrm }, - { X86::HSUBPSrr, X86::HSUBPSrm }, - { X86::IMUL16rr, X86::IMUL16rm }, - { X86::IMUL32rr, X86::IMUL32rm }, - { X86::IMUL64rr, X86::IMUL64rm }, - { X86::MAXPDrr, X86::MAXPDrm }, - { X86::MAXPDrr_Int, X86::MAXPDrm_Int }, - { X86::MAXPSrr, X86::MAXPSrm }, - { X86::MAXPSrr_Int, X86::MAXPSrm_Int }, - { X86::MAXSDrr, X86::MAXSDrm }, - { X86::MAXSDrr_Int, X86::MAXSDrm_Int }, - { X86::MAXSSrr, X86::MAXSSrm }, - { X86::MAXSSrr_Int, X86::MAXSSrm_Int }, - { X86::MINPDrr, X86::MINPDrm }, - { X86::MINPDrr_Int, X86::MINPDrm_Int }, - { X86::MINPSrr, X86::MINPSrm }, - { X86::MINPSrr_Int, X86::MINPSrm_Int }, - { X86::MINSDrr, X86::MINSDrm }, - { X86::MINSDrr_Int, X86::MINSDrm_Int }, - { X86::MINSSrr, X86::MINSSrm }, - { X86::MINSSrr_Int, X86::MINSSrm_Int }, - { X86::MULPDrr, X86::MULPDrm }, - { X86::MULPSrr, X86::MULPSrm }, - { X86::MULSDrr, X86::MULSDrm }, - { X86::MULSSrr, X86::MULSSrm }, - { X86::OR16rr, X86::OR16rm }, - { X86::OR32rr, X86::OR32rm }, - { X86::OR64rr, X86::OR64rm }, - { X86::OR8rr, X86::OR8rm }, - { X86::ORPDrr, X86::ORPDrm }, - { X86::ORPSrr, X86::ORPSrm }, - { X86::PACKSSDWrr, X86::PACKSSDWrm }, - { X86::PACKSSWBrr, X86::PACKSSWBrm }, - { X86::PACKUSWBrr, X86::PACKUSWBrm }, - { X86::PADDBrr, X86::PADDBrm }, - { X86::PADDDrr, X86::PADDDrm }, - { X86::PADDQrr, X86::PADDQrm }, - { X86::PADDSBrr, X86::PADDSBrm }, - { X86::PADDSWrr, X86::PADDSWrm }, - { X86::PADDWrr, X86::PADDWrm }, - { X86::PANDNrr, X86::PANDNrm }, - { X86::PANDrr, X86::PANDrm }, - { X86::PAVGBrr, X86::PAVGBrm }, - { X86::PAVGWrr, X86::PAVGWrm }, - { X86::PCMPEQBrr, X86::PCMPEQBrm }, - { X86::PCMPEQDrr, X86::PCMPEQDrm }, - { X86::PCMPEQWrr, X86::PCMPEQWrm }, - { X86::PCMPGTBrr, X86::PCMPGTBrm }, - { X86::PCMPGTDrr, X86::PCMPGTDrm }, - { X86::PCMPGTWrr, X86::PCMPGTWrm }, - { X86::PINSRWrri, X86::PINSRWrmi }, - { X86::PMADDWDrr, X86::PMADDWDrm }, - { X86::PMAXSWrr, X86::PMAXSWrm }, - { X86::PMAXUBrr, X86::PMAXUBrm }, - { X86::PMINSWrr, X86::PMINSWrm }, - { X86::PMINUBrr, X86::PMINUBrm }, - { X86::PMULHUWrr, X86::PMULHUWrm }, - { X86::PMULHWrr, X86::PMULHWrm }, - { X86::PMULLWrr, X86::PMULLWrm }, - { X86::PMULUDQrr, X86::PMULUDQrm }, - { X86::PORrr, X86::PORrm }, - { X86::PSADBWrr, X86::PSADBWrm }, - { X86::PSLLDrr, X86::PSLLDrm }, - { X86::PSLLQrr, X86::PSLLQrm }, - { X86::PSLLWrr, X86::PSLLWrm }, - { X86::PSRADrr, X86::PSRADrm }, - { X86::PSRAWrr, X86::PSRAWrm }, - { X86::PSRLDrr, X86::PSRLDrm }, - { X86::PSRLQrr, X86::PSRLQrm }, - { X86::PSRLWrr, X86::PSRLWrm }, - { X86::PSUBBrr, X86::PSUBBrm }, - { X86::PSUBDrr, X86::PSUBDrm }, - { X86::PSUBSBrr, X86::PSUBSBrm }, - { X86::PSUBSWrr, X86::PSUBSWrm }, - { X86::PSUBWrr, X86::PSUBWrm }, - { X86::PUNPCKHBWrr, X86::PUNPCKHBWrm }, - { X86::PUNPCKHDQrr, X86::PUNPCKHDQrm }, - { X86::PUNPCKHQDQrr, X86::PUNPCKHQDQrm }, - { X86::PUNPCKHWDrr, X86::PUNPCKHWDrm }, - { X86::PUNPCKLBWrr, X86::PUNPCKLBWrm }, - { X86::PUNPCKLDQrr, X86::PUNPCKLDQrm }, - { X86::PUNPCKLQDQrr, X86::PUNPCKLQDQrm }, - { X86::PUNPCKLWDrr, X86::PUNPCKLWDrm }, - { X86::PXORrr, X86::PXORrm }, - { X86::SBB32rr, X86::SBB32rm }, - { X86::SBB64rr, X86::SBB64rm }, - { X86::SHUFPDrri, X86::SHUFPDrmi }, - { X86::SHUFPSrri, X86::SHUFPSrmi }, - { X86::SUB16rr, X86::SUB16rm }, - { X86::SUB32rr, X86::SUB32rm }, - { X86::SUB64rr, X86::SUB64rm }, - { X86::SUB8rr, X86::SUB8rm }, - { X86::SUBPDrr, X86::SUBPDrm }, - { X86::SUBPSrr, X86::SUBPSrm }, - { X86::SUBSDrr, X86::SUBSDrm }, - { X86::SUBSSrr, X86::SUBSSrm }, - // FIXME: TEST*rr -> swapped operand of TEST*mr. - { X86::UNPCKHPDrr, X86::UNPCKHPDrm }, - { X86::UNPCKHPSrr, X86::UNPCKHPSrm }, - { X86::UNPCKLPDrr, X86::UNPCKLPDrm }, - { X86::UNPCKLPSrr, X86::UNPCKLPSrm }, - { X86::XOR16rr, X86::XOR16rm }, - { X86::XOR32rr, X86::XOR32rm }, - { X86::XOR64rr, X86::XOR64rm }, - { X86::XOR8rr, X86::XOR8rm }, - { X86::XORPDrr, X86::XORPDrm }, - { X86::XORPSrr, X86::XORPSrm } - }; - - ASSERT_SORTED(OpcodeTable); - OpcodeTablePtr = OpcodeTable; - OpcodeTableSize = array_lengthof(OpcodeTable); + OpcodeTablePtr = &RegOp2MemOpTable2; } // If table selected... if (OpcodeTablePtr) { // Find the Opcode to fuse - unsigned fromOpcode = MI->getOpcode(); - // Lookup fromOpcode in table - if (const TableEntry *Entry = TableLookup(OpcodeTablePtr, OpcodeTableSize, - fromOpcode)) { + DenseMap::iterator I = + OpcodeTablePtr->find((unsigned*)MI->getOpcode()); + if (I != OpcodeTablePtr->end()) { if (isTwoAddrFold) - NewMI = FuseTwoAddrInst(Entry->to, MOs, MI, TII); + NewMI = FuseTwoAddrInst(I->second, MOs, MI, TII); else - NewMI = FuseInst(Entry->to, i, MOs, MI, TII); + NewMI = FuseInst(I->second, i, MOs, MI, TII); NewMI->copyKillDeadInfo(MI); return NewMI; } Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.h?rev=42509&r1=42508&r2=42509&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.h (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.h Mon Oct 1 18:44:33 2007 @@ -14,6 +14,7 @@ #ifndef X86REGISTERINFO_H #define X86REGISTERINFO_H +#include "llvm/ADT/DenseMap.h" #include "llvm/ADT/SmallVector.h" #include "llvm/Target/MRegisterInfo.h" #include "X86GenRegisterInfo.h.inc" @@ -38,17 +39,33 @@ private: /// Is64Bit - Is the target 64-bits. + /// bool Is64Bit; /// SlotSize - Stack slot size in bytes. + /// unsigned SlotSize; /// StackPtr - X86 physical register used as stack ptr. + /// unsigned StackPtr; /// FramePtr - X86 physical register used as frame ptr. + /// unsigned FramePtr; + /// RegOp2MemOpTable2Addr, RegOp2MemOpTable0, RegOp2MemOpTable1, + /// RegOp2MemOpTable2 - Load / store folding opcode maps. + /// + DenseMap RegOp2MemOpTable2Addr; + DenseMap RegOp2MemOpTable0; + DenseMap RegOp2MemOpTable1; + DenseMap RegOp2MemOpTable2; + + /// MemOp2RegOpTable - Load / store unfolding opcode map. + /// + DenseMap MemOp2RegOpTable; + public: X86RegisterInfo(X86TargetMachine &tm, const TargetInstrInfo &tii); @@ -57,6 +74,7 @@ unsigned getX86RegNum(unsigned RegNo); /// Code Generation virtual methods... + /// bool spillCalleeSavedRegisters(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, const std::vector &CSI) const; From isanbard at gmail.com Mon Oct 1 18:48:07 2007 From: isanbard at gmail.com (Bill Wendling) Date: Mon, 01 Oct 2007 23:48:07 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42510 - /llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Message-ID: <200710012348.l91Nm78P001119@zion.cs.uiuc.edu> Author: void Date: Mon Oct 1 18:48:06 2007 New Revision: 42510 URL: http://llvm.org/viewvc/llvm-project?rev=42510&view=rev Log: createBFS_DAGScheduler doesn't exist anymore Modified: llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp?rev=42510&r1=42509&r2=42510&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Mon Oct 1 18:48:06 2007 @@ -51,7 +51,6 @@ llvm::createCFGSimplificationPass(); llvm::createDefaultScheduler(NULL, NULL, NULL); - llvm::createBFS_DAGScheduler(NULL, NULL, NULL); llvm::createLinearScanRegisterAllocator(); llvm::createGVNPass(); From isanbard at gmail.com Mon Oct 1 18:49:25 2007 From: isanbard at gmail.com (Bill Wendling) Date: Mon, 01 Oct 2007 23:49:25 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42511 - /llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Message-ID: <200710012349.l91NnPDF001180@zion.cs.uiuc.edu> Author: void Date: Mon Oct 1 18:49:21 2007 New Revision: 42511 URL: http://llvm.org/viewvc/llvm-project?rev=42511&view=rev Log: createBFS_DAGScheduler doesn't exist anymore Modified: llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp?rev=42511&r1=42510&r2=42511&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Mon Oct 1 18:49:21 2007 @@ -51,7 +51,6 @@ llvm::createCFGSimplificationPass(); llvm::createDefaultScheduler(NULL, NULL, NULL); - llvm::createBFS_DAGScheduler(NULL, NULL, NULL); llvm::createLinearScanRegisterAllocator(); llvm::createGVNPass(); From tonic at nondot.org Mon Oct 1 19:19:28 2007 From: tonic at nondot.org (Tanya Lattner) Date: Tue, 02 Oct 2007 00:19:28 -0000 Subject: [llvm-commits] [llvm] r42514 - /llvm/trunk/utils/NewNightlyTest.pl Message-ID: <200710020019.l920JSGK002668@zion.cs.uiuc.edu> Author: tbrethou Date: Mon Oct 1 19:19:27 2007 New Revision: 42514 URL: http://llvm.org/viewvc/llvm-project?rev=42514&view=rev Log: If the user did not check out LLVM and request it to be built, it should be a build error. This relies on the user having a successful build of LLVM, but the tests will fail if they dont. Modified: llvm/trunk/utils/NewNightlyTest.pl Modified: llvm/trunk/utils/NewNightlyTest.pl URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/NewNightlyTest.pl?rev=42514&r1=42513&r2=42514&view=diff ============================================================================== --- llvm/trunk/utils/NewNightlyTest.pl (original) +++ llvm/trunk/utils/NewNightlyTest.pl Mon Oct 1 19:19:27 2007 @@ -737,7 +737,6 @@ my $BuildError = 0, $BuildStatus = "OK"; if ($NOBUILD) { $BuildStatus = "Skipped by user"; - $BuildError = 1; } elsif (`grep '^$MAKECMD\[^:]*: .*Error' $BuildLog | wc -l` + 0 || `grep '^$MAKECMD: \*\*\*.*Stop.' $BuildLog | wc -l`+0) { From dalej at apple.com Mon Oct 1 19:27:23 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 00:27:23 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42515 - /llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Message-ID: <200710020027.l920RNVX003102@zion.cs.uiuc.edu> Author: johannes Date: Mon Oct 1 19:27:23 2007 New Revision: 42515 URL: http://llvm.org/viewvc/llvm-project?rev=42515&view=rev Log: Use getABITypeSizeInBits for superior checking. Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-types.cpp?rev=42515&r1=42514&r2=42515&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Mon Oct 1 19:27:23 2007 @@ -74,12 +74,11 @@ static const Type * llvm_set_type(tree Tr, const Type *Ty) { // For x86 long double, llvm records the size of the data (80) while - // gcc's TYPE_SIZE including alignment padding. Compensate. + // gcc's TYPE_SIZE including alignment padding. getABITypeSizeInBits + // is used to compensate for this. assert((!TYPE_SIZE(Tr) || !Ty->isSized() || !isInt64(TYPE_SIZE(Tr), true) || - getInt64(TYPE_SIZE(Tr), true) == getTargetData().getTypeSizeInBits(Ty) || - (getTargetData().getTypeSizeInBits(Ty) == 80 && - (getInt64(TYPE_SIZE(Tr), true) == 96 || - getInt64(TYPE_SIZE(Tr), true) == 128))) + getInt64(TYPE_SIZE(Tr), true) == + getTargetData().getABITypeSizeInBits(Ty)) && "LLVM type size doesn't match GCC type size!"); unsigned &TypeSlot = LTypesMap[Ty]; From dpatel at apple.com Mon Oct 1 19:39:20 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 00:39:20 -0000 Subject: [llvm-commits] [llvm] r42517 - /llvm/trunk/test/C++Frontend/2007-10-01-StructResize.ll Message-ID: <200710020039.l920dKsR003705@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 19:39:19 2007 New Revision: 42517 URL: http://llvm.org/viewvc/llvm-project?rev=42517&view=rev Log: New test. Added: llvm/trunk/test/C++Frontend/2007-10-01-StructResize.ll Added: llvm/trunk/test/C++Frontend/2007-10-01-StructResize.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/C%2B%2BFrontend/2007-10-01-StructResize.ll?rev=42517&view=auto ============================================================================== --- llvm/trunk/test/C++Frontend/2007-10-01-StructResize.ll (added) +++ llvm/trunk/test/C++Frontend/2007-10-01-StructResize.ll Mon Oct 1 19:39:19 2007 @@ -0,0 +1,14 @@ +// RUN: %llvmgxx -c %s -o /dev/null + +#pragma pack(4) + +struct Bork { + unsigned int f1 : 3; + unsigned int f2 : 30; +}; + +int Foo(Bork *hdr) { + hdr->f1 = 7; + hdr->f2 = 927; +} + From dpatel at apple.com Mon Oct 1 19:41:17 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 00:41:17 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42518 - /llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Message-ID: <200710020041.l920fHME003847@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 19:41:16 2007 New Revision: 42518 URL: http://llvm.org/viewvc/llvm-project?rev=42518&view=rev Log: If LLVM struct size matches GCC struct size then do not removing trailing 'extra' padding bytes. Fix http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20071001/054142.html Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-types.cpp?rev=42518&r1=42517&r2=42518&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Mon Oct 1 19:41:16 2007 @@ -1160,7 +1160,7 @@ // If this is a Packed struct and ExtraBitsAvailable is not zero then // remove Extra bytes if ExtraBitsAvailable > 8. - void RemoveExtraBytes () { + void RemoveExtraBytes (unsigned GCCTypeSize) { unsigned NoOfBytesToRemove = ExtraBitsAvailable/8; @@ -1170,6 +1170,10 @@ if (NoOfBytesToRemove == 0) return; + unsigned OriginalSize = ElementSizeInBytes.back(); + if (OriginalSize == GCCTypeSize) + return; + const Type *LastType = Elements.back(); unsigned PadBytes = 0; @@ -1191,7 +1195,6 @@ Elements.pop_back(); Elements.push_back(Pad); - unsigned OriginalSize = ElementSizeInBytes.back(); ElementSizeInBytes.pop_back(); ElementSizeInBytes.push_back(OriginalSize - NoOfBytesToRemove); } @@ -1667,12 +1670,12 @@ for (tree Field = TYPE_FIELDS(type); Field; Field = TREE_CHAIN(Field)) DecodeStructFields(Field, *Info); - Info->RemoveExtraBytes(); + uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; + Info->RemoveExtraBytes(GCCTypeSize); // If the LLVM struct requires explicit tail padding to be the same size as // the GCC struct, insert tail padding now. This handles, e.g., "{}" in C++. if (TYPE_SIZE(type) && TREE_CODE(TYPE_SIZE(type)) == INTEGER_CST) { uint64_t LLVMStructSize = Info->getSizeAsLLVMStruct(); - uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; if (LLVMStructSize != GCCTypeSize) { assert(LLVMStructSize < GCCTypeSize && From resistor at mac.com Mon Oct 1 19:43:25 2007 From: resistor at mac.com (Owen Anderson) Date: Tue, 02 Oct 2007 00:43:25 -0000 Subject: [llvm-commits] [llvm] r42519 - /llvm/trunk/docs/AliasAnalysis.html Message-ID: <200710020043.l920hQkk004003@zion.cs.uiuc.edu> Author: resistor Date: Mon Oct 1 19:43:25 2007 New Revision: 42519 URL: http://llvm.org/viewvc/llvm-project?rev=42519&view=rev Log: Mention memdep in the AA documentation. Modified: llvm/trunk/docs/AliasAnalysis.html Modified: llvm/trunk/docs/AliasAnalysis.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/AliasAnalysis.html?rev=42519&r1=42518&r2=42519&view=diff ============================================================================== --- llvm/trunk/docs/AliasAnalysis.html (original) +++ llvm/trunk/docs/AliasAnalysis.html Mon Oct 1 19:43:25 2007 @@ -50,6 +50,7 @@ implementations +

  • Memory Dependence Analysis
    • @@ -948,6 +949,24 @@
    +
    + Memory Dependence Analysis +
    + + +
    + +

    If you're just looking to be a client of alias analysis information, consider +using the Memory Dependence Analysis interface instead. MemDep is a lazy, +caching layer on top of alias analysis that is able to answer the question of +what preceding memory operations a given instruction depends on, either at an +intra- or inter-block level. Because of its laziness and caching +policy, using MemDep can be a significant performance win over accessing alias +analysis directly.

    + +
    + +
    From dpatel at apple.com Mon Oct 1 19:43:27 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 00:43:27 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42520 - /llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Message-ID: <200710020043.l920hRk1004032@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 19:43:27 2007 New Revision: 42520 URL: http://llvm.org/viewvc/llvm-project?rev=42520&view=rev Log: If LLVM struct size matches GCC struct size then do not removing trailing 'extra' padding bytes. Fix http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20071001/054142.html Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-types.cpp?rev=42520&r1=42519&r2=42520&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Mon Oct 1 19:43:27 2007 @@ -1185,7 +1185,7 @@ // If this is a Packed struct and ExtraBitsAvailable is not zero then // remove Extra bytes if ExtraBitsAvailable > 8. - void RemoveExtraBytes () { + void RemoveExtraBytes (unsigned GCCTypeSize) { unsigned NoOfBytesToRemove = ExtraBitsAvailable/8; @@ -1195,6 +1195,10 @@ if (NoOfBytesToRemove == 0) return; + unsigned OriginalSize = ElementSizeInBytes.back(); + if (OriginalSize == GCCTypeSize) + return; + const Type *LastType = Elements.back(); unsigned PadBytes = 0; @@ -1216,7 +1220,6 @@ Elements.pop_back(); Elements.push_back(Pad); - unsigned OriginalSize = ElementSizeInBytes.back(); ElementSizeInBytes.pop_back(); ElementSizeInBytes.push_back(OriginalSize - NoOfBytesToRemove); } @@ -1692,12 +1695,12 @@ for (tree Field = TYPE_FIELDS(type); Field; Field = TREE_CHAIN(Field)) DecodeStructFields(Field, *Info); - Info->RemoveExtraBytes(); + uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; + Info->RemoveExtraBytes(GCCTypeSize); // If the LLVM struct requires explicit tail padding to be the same size as // the GCC struct, insert tail padding now. This handles, e.g., "{}" in C++. if (TYPE_SIZE(type) && TREE_CODE(TYPE_SIZE(type)) == INTEGER_CST) { uint64_t LLVMStructSize = Info->getSizeAsLLVMStruct(); - uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; if (LLVMStructSize != GCCTypeSize) { assert(LLVMStructSize < GCCTypeSize && From resistor at mac.com Mon Oct 1 19:44:21 2007 From: resistor at mac.com (Owen Anderson) Date: Tue, 02 Oct 2007 00:44:21 -0000 Subject: [llvm-commits] [llvm] r42521 - /llvm/trunk/docs/AliasAnalysis.html Message-ID: <200710020044.l920iL2b004106@zion.cs.uiuc.edu> Author: resistor Date: Mon Oct 1 19:44:20 2007 New Revision: 42521 URL: http://llvm.org/viewvc/llvm-project?rev=42521&view=rev Log: Fix typo. Modified: llvm/trunk/docs/AliasAnalysis.html Modified: llvm/trunk/docs/AliasAnalysis.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/AliasAnalysis.html?rev=42521&r1=42520&r2=42521&view=diff ============================================================================== --- llvm/trunk/docs/AliasAnalysis.html (original) +++ llvm/trunk/docs/AliasAnalysis.html Mon Oct 1 19:44:20 2007 @@ -50,7 +50,7 @@ implementations -
  • Memory Dependence Analysis
    • +
  • Memory Dependence Analysis
    • From dpatel at apple.com Mon Oct 1 19:47:43 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 00:47:43 -0000 Subject: [llvm-commits] [llvm] r42522 - /llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Message-ID: <200710020047.l920lh0I004399@zion.cs.uiuc.edu> Author: dpatel Date: Mon Oct 1 19:47:43 2007 New Revision: 42522 URL: http://llvm.org/viewvc/llvm-project?rev=42522&view=rev Log: Fix test. Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/2007-09-25-EH.m?rev=42522&r1=42521&r2=42522&view=diff ============================================================================== --- llvm/trunk/test/FrontendObjC/2007-09-25-EH.m (original) +++ llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Mon Oct 1 19:47:43 2007 @@ -1,4 +1,4 @@ -// RUN: %llvmgcc -c -w -m64 %s -o /dev/null +// RUN: %llvmgcc -c -w -m64 -mmacosx-version-min=10.5 %s -o /dev/null @class NSDictionary, DSoBuffer, DSoDirectory, NSMutableArray; @interface NSException {} From evan.cheng at apple.com Mon Oct 1 20:50:08 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Tue, 02 Oct 2007 01:50:08 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42523 - /llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Message-ID: <200710020150.l921o8o0007378@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 20:50:08 2007 New Revision: 42523 URL: http://llvm.org/viewvc/llvm-project?rev=42523&view=rev Log: Link in llvm::createLocalRegisterAllocator. Modified: llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp?rev=42523&r1=42522&r2=42523&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-linker-hack.cpp Mon Oct 1 20:50:08 2007 @@ -52,6 +52,7 @@ llvm::createDefaultScheduler(NULL, NULL, NULL); llvm::createLinearScanRegisterAllocator(); + llvm::createLocalRegisterAllocator(); llvm::createGVNPass(); llvm::createGCSEPass(); From evan.cheng at apple.com Mon Oct 1 21:28:00 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Tue, 02 Oct 2007 02:28:00 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42525 - /llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Message-ID: <200710020228.l922S1Ms009090@zion.cs.uiuc.edu> Author: evancheng Date: Mon Oct 1 21:28:00 2007 New Revision: 42525 URL: http://llvm.org/viewvc/llvm-project?rev=42525&view=rev Log: Merge -r42522:42523 llvm-project/llvm-gcc-4.0/trunk Modified: llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp?rev=42525&r1=42524&r2=42525&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-linker-hack.cpp Mon Oct 1 21:28:00 2007 @@ -52,6 +52,7 @@ llvm::createDefaultScheduler(NULL, NULL, NULL); llvm::createLinearScanRegisterAllocator(); + llvm::createLocalRegisterAllocator(); llvm::createGVNPass(); llvm::createGCSEPass(); From baldrick at free.fr Tue Oct 2 01:33:12 2007 From: baldrick at free.fr (Duncan Sands) Date: Tue, 2 Oct 2007 08:33:12 +0200 Subject: [llvm-commits] [llvm-gcc-4.0] r42515 - /llvm-gcc-4.0/trunk/gcc/llvm-types.cpp In-Reply-To: <200710020027.l920RNVX003102@zion.cs.uiuc.edu> References: <200710020027.l920RNVX003102@zion.cs.uiuc.edu> Message-ID: <200710020833.12947.baldrick@free.fr> > Use getABITypeSizeInBits for superior checking. Thanks for doing this! Please also apply to 4.2. Thanks, Duncan. From gordonhenriksen at mac.com Tue Oct 2 04:50:18 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 09:50:18 -0000 Subject: [llvm-commits] [llvm] r42526 - in /llvm/trunk: Makefile.config.in autoconf/configure.ac bindings/Makefile test/Makefile Message-ID: <200710020950.l929oJpt008361@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 04:50:18 2007 New Revision: 42526 URL: http://llvm.org/viewvc/llvm-project?rev=42526&view=rev Log: Add explicit --enable-bindings option to configure. Modified: llvm/trunk/Makefile.config.in llvm/trunk/autoconf/configure.ac llvm/trunk/bindings/Makefile llvm/trunk/test/Makefile Modified: llvm/trunk/Makefile.config.in URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Makefile.config.in?rev=42526&r1=42525&r2=42526&view=diff ============================================================================== --- llvm/trunk/Makefile.config.in (original) +++ llvm/trunk/Makefile.config.in Tue Oct 2 04:50:18 2007 @@ -168,6 +168,9 @@ LIBS := @LIBS@ +# Targets that we should build +TARGETS_TO_BUILD=@TARGETS_TO_BUILD@ + # Path to location for LLVM C/C++ front-end. You can modify this if you # want to override the value set by configure. LLVMGCCDIR := @LLVMGCCDIR@ @@ -259,3 +262,8 @@ # Get the value of HUGE_VAL_SANITY which will be either "yes" or "no" depending # on the check. HUGE_VAL_SANITY = @HUGE_VAL_SANITY@ + +# Bindings that we should build +BINDINGS_TO_BUILD = @BINDINGS_TO_BUILD@ +ALL_BINDINGS = @ALL_BINDINGS@ + Modified: llvm/trunk/autoconf/configure.ac URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/autoconf/configure.ac?rev=42526&r1=42525&r2=42526&view=diff ============================================================================== --- llvm/trunk/autoconf/configure.ac (original) +++ llvm/trunk/autoconf/configure.ac Tue Oct 2 04:50:18 2007 @@ -432,6 +432,24 @@ esac AC_SUBST(EXTRA_OPTIONS,$EXTRA_OPTIONS) +dnl Allow specific bindings to be specified for building (or not) +AC_ARG_ENABLE([bindings],AS_HELP_STRING([--enable-bindings], + [Build specific language bindings: all,auto,none,{binding-name} (default=auto)]),, + enableval=default) +BINDINGS_TO_BUILD="" +case "$enableval" in + all | yes | default | auto) BINDINGS_TO_BUILD="auto" ;; + all ) BINDINGS_TO_BUILD="ocaml" ;; + none | no) BINDINGS_TO_BUILD="" ;; + *)for a_binding in `echo $enableval|sed -e 's/,/ /g' ` ; do + case "$a_binding" in + ocaml) BINDINGS_TO_BUILD="ocaml $BINDINGS_TO_BUILD" ;; + *) AC_MSG_ERROR([Unrecognized binding $a_binding]) ;; + esac + done + ;; +esac + dnl===-----------------------------------------------------------------------=== dnl=== dnl=== SECTION 4: Check for programs we need and that they are the right version @@ -855,6 +873,19 @@ AC_DEFINE_UNQUOTED(LLVM_CONFIGTIME, "$LLVM_CONFIGTIME", [Time at which LLVM was configured]) +# Determine which bindings to build. +if test "$BINDINGS_TO_BUILD" = auto ; then + BINDINGS_TO_BUILD="" + if test "x$OCAMLC" != x -a "x$OCAMLDEP" != x ; then + BINDINGS_TO_BUILD="ocaml $BINDINGS_TO_BUILD" + fi +fi +AC_SUBST(BINDINGS_TO_BUILD,$BINDINGS_TO_BUILD) + +# This isn't really configurey, but it avoids having to repeat the list in +# other files. +AC_SUBST(ALL_BINDINGS,ocaml) + dnl===-----------------------------------------------------------------------=== dnl=== dnl=== SECTION 10: Specify the output files and generate it Modified: llvm/trunk/bindings/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/Makefile?rev=42526&r1=42525&r2=42526&view=diff ============================================================================== --- llvm/trunk/bindings/Makefile (original) +++ llvm/trunk/bindings/Makefile Tue Oct 2 04:50:18 2007 @@ -11,8 +11,6 @@ include $(LEVEL)/Makefile.config -ifdef OCAMLC -PARALLEL_DIRS += ocaml -endif +PARALLEL_DIRS = $(BINDINGS_TO_BUILD) include $(LEVEL)/Makefile.common Modified: llvm/trunk/test/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Makefile?rev=42526&r1=42525&r2=42526&view=diff ============================================================================== --- llvm/trunk/test/Makefile (original) +++ llvm/trunk/test/Makefile Tue Oct 2 04:50:18 2007 @@ -34,9 +34,7 @@ IGNORE_TESTS += llvm2cpp.exp endif -ifndef OCAMLC -IGNORE_TESTS += ocaml.exp -endif +IGNORE_TESTS += $(filter-out $(BINDINGS_TO_BUILD:=.exp),$(ALL_BINDINGS:=.exp)) ifdef IGNORE_TESTS RUNTESTFLAGS += --ignore "$(strip $(IGNORE_TESTS))" From gordonhenriksen at mac.com Tue Oct 2 04:50:32 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 09:50:32 -0000 Subject: [llvm-commits] [llvm] r42527 - /llvm/trunk/configure Message-ID: <200710020950.l929oX43008383@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 04:50:32 2007 New Revision: 42527 URL: http://llvm.org/viewvc/llvm-project?rev=42527&view=rev Log: Regenerate. Modified: llvm/trunk/configure Modified: llvm/trunk/configure URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/configure?rev=42527&r1=42526&r2=42527&view=diff ============================================================================== --- llvm/trunk/configure (original) +++ llvm/trunk/configure Tue Oct 2 04:50:32 2007 @@ -924,6 +924,8 @@ LLVM_INFODIR LLVM_MANDIR LLVM_CONFIGTIME +BINDINGS_TO_BUILD +ALL_BINDINGS LIBOBJS LTLIBOBJS' ac_subst_files='' @@ -1542,6 +1544,8 @@ all,host-only,{target-name} (default=all) --enable-cbe-printf-a Enable C Backend output with hex floating point via %a (default is YES) + --enable-bindings Build specific language bindings: + all,auto,none,{binding-name} (default=auto) --enable-ltdl-install install libltdl --enable-shared[=PKGS] build shared libraries [default=yes] @@ -4823,6 +4827,29 @@ EXTRA_OPTIONS=$EXTRA_OPTIONS +# Check whether --enable-bindings was given. +if test "${enable_bindings+set}" = set; then + enableval=$enable_bindings; +else + enableval=default +fi + +BINDINGS_TO_BUILD="" +case "$enableval" in + all | yes | default | auto) BINDINGS_TO_BUILD="auto" ;; + all ) BINDINGS_TO_BUILD="ocaml" ;; + none | no) BINDINGS_TO_BUILD="" ;; + *)for a_binding in `echo $enableval|sed -e 's/,/ /g' ` ; do + case "$a_binding" in + ocaml) BINDINGS_TO_BUILD="ocaml $BINDINGS_TO_BUILD" ;; + *) { { echo "$as_me:$LINENO: error: Unrecognized binding $a_binding" >&5 +echo "$as_me: error: Unrecognized binding $a_binding" >&2;} + { (exit 1); exit 1; }; } ;; + esac + done + ;; +esac + ac_ext=c ac_cpp='$CPP $CPPFLAGS' @@ -10499,7 +10526,7 @@ lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext < conftest.$ac_ext + echo '#line 12673 "configure"' > conftest.$ac_ext if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5 (eval $ac_compile) 2>&5 ac_status=$? @@ -14361,11 +14388,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14364: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14391: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:14368: \$? = $ac_status" >&5 + echo "$as_me:14395: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -14629,11 +14656,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14632: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14659: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:14636: \$? = $ac_status" >&5 + echo "$as_me:14663: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -14733,11 +14760,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14736: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14763: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:14740: \$? = $ac_status" >&5 + echo "$as_me:14767: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -17185,7 +17212,7 @@ lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext < conftest.$ac_ext <&5) + (eval echo "\"\$as_me:19683: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:19660: \$? = $ac_status" >&5 + echo "$as_me:19687: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -19757,11 +19784,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:19760: $lt_compile\"" >&5) + (eval echo "\"\$as_me:19787: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:19764: \$? = $ac_status" >&5 + echo "$as_me:19791: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -21327,11 +21354,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:21330: $lt_compile\"" >&5) + (eval echo "\"\$as_me:21357: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:21334: \$? = $ac_status" >&5 + echo "$as_me:21361: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -21431,11 +21458,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:21434: $lt_compile\"" >&5) + (eval echo "\"\$as_me:21461: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:21438: \$? = $ac_status" >&5 + echo "$as_me:21465: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -23666,11 +23693,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:23669: $lt_compile\"" >&5) + (eval echo "\"\$as_me:23696: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:23673: \$? = $ac_status" >&5 + echo "$as_me:23700: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -23934,11 +23961,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:23937: $lt_compile\"" >&5) + (eval echo "\"\$as_me:23964: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:23941: \$? = $ac_status" >&5 + echo "$as_me:23968: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -24038,11 +24065,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:24041: $lt_compile\"" >&5) + (eval echo "\"\$as_me:24068: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:24045: \$? = $ac_status" >&5 + echo "$as_me:24072: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -34115,6 +34142,21 @@ _ACEOF +# Determine which bindings to build. +if test "$BINDINGS_TO_BUILD" = auto ; then + BINDINGS_TO_BUILD="" + if test "x$OCAMLC" != x -a "x$OCAMLDEP" != x ; then + BINDINGS_TO_BUILD="ocaml $BINDINGS_TO_BUILD" + fi +fi +BINDINGS_TO_BUILD=$BINDINGS_TO_BUILD + + +# This isn't really configurey, but it avoids having to repeat the list in +# other files. +ALL_BINDINGS=ocaml + + ac_config_headers="$ac_config_headers include/llvm/Config/config.h" @@ -35034,11 +35076,13 @@ LLVM_INFODIR!$LLVM_INFODIR$ac_delim LLVM_MANDIR!$LLVM_MANDIR$ac_delim LLVM_CONFIGTIME!$LLVM_CONFIGTIME$ac_delim +BINDINGS_TO_BUILD!$BINDINGS_TO_BUILD$ac_delim +ALL_BINDINGS!$ALL_BINDINGS$ac_delim LIBOBJS!$LIBOBJS$ac_delim LTLIBOBJS!$LTLIBOBJS$ac_delim _ACEOF - if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 73; then + if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 75; then break elif $ac_last_try; then { { echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5 From gordonhenriksen at mac.com Tue Oct 2 05:14:32 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 10:14:32 -0000 Subject: [llvm-commits] [llvm] r42528 - in /llvm/trunk: Makefile.config.in autoconf/configure.ac Message-ID: <200710021014.l92AEWHa009347@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 05:14:32 2007 New Revision: 42528 URL: http://llvm.org/viewvc/llvm-project?rev=42528&view=rev Log: Fixing (harmless) typos. Modified: llvm/trunk/Makefile.config.in llvm/trunk/autoconf/configure.ac Modified: llvm/trunk/Makefile.config.in URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Makefile.config.in?rev=42528&r1=42527&r2=42528&view=diff ============================================================================== --- llvm/trunk/Makefile.config.in (original) +++ llvm/trunk/Makefile.config.in Tue Oct 2 05:14:32 2007 @@ -105,9 +105,6 @@ # Target triple (cpu-vendor-os) for which we should generate code TARGET_TRIPLE=@target@ -# Targets that we should build -TARGETS_TO_BUILD=@TARGETS_TO_BUILD@ - # Extra options to compile LLVM with EXTRA_OPTIONS=@EXTRA_OPTIONS@ Modified: llvm/trunk/autoconf/configure.ac URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/autoconf/configure.ac?rev=42528&r1=42527&r2=42528&view=diff ============================================================================== --- llvm/trunk/autoconf/configure.ac (original) +++ llvm/trunk/autoconf/configure.ac Tue Oct 2 05:14:32 2007 @@ -438,7 +438,7 @@ enableval=default) BINDINGS_TO_BUILD="" case "$enableval" in - all | yes | default | auto) BINDINGS_TO_BUILD="auto" ;; + yes | default | auto) BINDINGS_TO_BUILD="auto" ;; all ) BINDINGS_TO_BUILD="ocaml" ;; none | no) BINDINGS_TO_BUILD="" ;; *)for a_binding in `echo $enableval|sed -e 's/,/ /g' ` ; do From gordonhenriksen at mac.com Tue Oct 2 05:14:42 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 10:14:42 -0000 Subject: [llvm-commits] [llvm] r42529 - /llvm/trunk/configure Message-ID: <200710021014.l92AEhTK009368@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 05:14:42 2007 New Revision: 42529 URL: http://llvm.org/viewvc/llvm-project?rev=42529&view=rev Log: Regenerate. Modified: llvm/trunk/configure Modified: llvm/trunk/configure URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/configure?rev=42529&r1=42528&r2=42529&view=diff ============================================================================== --- llvm/trunk/configure (original) +++ llvm/trunk/configure Tue Oct 2 05:14:42 2007 @@ -4836,7 +4836,7 @@ BINDINGS_TO_BUILD="" case "$enableval" in - all | yes | default | auto) BINDINGS_TO_BUILD="auto" ;; + yes | default | auto) BINDINGS_TO_BUILD="auto" ;; all ) BINDINGS_TO_BUILD="ocaml" ;; none | no) BINDINGS_TO_BUILD="" ;; *)for a_binding in `echo $enableval|sed -e 's/,/ /g' ` ; do From baldrick at free.fr Tue Oct 2 06:31:12 2007 From: baldrick at free.fr (Duncan Sands) Date: Tue, 2 Oct 2007 13:31:12 +0200 Subject: [llvm-commits] Size of long double (X86_FP80Ty) In-Reply-To: <4753238B-3DAB-454B-BE0D-491C1E815E23@apple.com> References: <200709282204.17074.baldrick@free.fr> <200709301023.44406.baldrick@free.fr> <4753238B-3DAB-454B-BE0D-491C1E815E23@apple.com> Message-ID: <200710021331.12297.baldrick@free.fr> Hi Dale, > The attached patch, plus the bits I've checked in, seems to work. > There are no problems in the llvm testsuite but there's something > funny going on in the gcc testsuite, although I'm not convinced it's > the fault of this patch, so I'm not checking it in until I get that > figured out. You could see if it works for you. it doesn't cause any additional Ada testsuite passes or failures. However I didn't check whether the tests fail differently (most of the tests are multipart, so it is possible that more or less parts failed). Best wishes, Duncan. From gordonhenriksen at mac.com Tue Oct 2 11:42:11 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 16:42:11 -0000 Subject: [llvm-commits] [llvm] r42532 - in /llvm/trunk: Makefile.config.in autoconf/configure.ac bindings/ocaml/Makefile.ocaml bindings/ocaml/bitwriter/llvm_bitwriter.mli Message-ID: <200710021642.l92GgBu1025978@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 11:42:10 2007 New Revision: 42532 URL: http://llvm.org/viewvc/llvm-project?rev=42532&view=rev Log: Providing --with-ocaml-libdir for ./configure. The default is the stdlib if it's beneath --prefix, and is libdir/ocaml otherwise. If someone has a better way than this to test whether $B is a path within $A, I'd love to hear it: if test "$A" \< "$B" -a "$B" \< "${A}~" Modified: llvm/trunk/Makefile.config.in llvm/trunk/autoconf/configure.ac llvm/trunk/bindings/ocaml/Makefile.ocaml llvm/trunk/bindings/ocaml/bitwriter/llvm_bitwriter.mli Modified: llvm/trunk/Makefile.config.in URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Makefile.config.in?rev=42532&r1=42531&r2=42532&view=diff ============================================================================== --- llvm/trunk/Makefile.config.in (original) +++ llvm/trunk/Makefile.config.in Tue Oct 2 11:42:10 2007 @@ -261,6 +261,7 @@ HUGE_VAL_SANITY = @HUGE_VAL_SANITY@ # Bindings that we should build -BINDINGS_TO_BUILD = @BINDINGS_TO_BUILD@ -ALL_BINDINGS = @ALL_BINDINGS@ +BINDINGS_TO_BUILD := @BINDINGS_TO_BUILD@ +ALL_BINDINGS := @ALL_BINDINGS@ +OCAML_LIBDIR := @OCAML_LIBDIR@ Modified: llvm/trunk/autoconf/configure.ac URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/autoconf/configure.ac?rev=42532&r1=42531&r2=42532&view=diff ============================================================================== --- llvm/trunk/autoconf/configure.ac (original) +++ llvm/trunk/autoconf/configure.ac Tue Oct 2 11:42:10 2007 @@ -450,6 +450,20 @@ ;; esac +dnl Allow the ocaml libdir to be overridden. This could go in a configure +dnl script for bindings/ocaml/configure, except that its auto value depends on +dnl OCAMLC, which is found here to support tests. +AC_ARG_WITH([ocaml-libdir], + [AS_HELP_STRING([--with-ocaml-libdir], + [Specify install location for ocaml bindings (default is stdlib)])], + [], + [withval=auto]) +case "$withval" in + auto) with_ocaml_libdir="$withval" ;; + /* | [[A-Za-z]]:[[\\/]]*) with_ocaml_libdir="$withval" ;; + *) AC_MSG_ERROR([Invalid path for --with-ocaml-libdir. Provide full path]) ;; +esac + dnl===-----------------------------------------------------------------------=== dnl=== dnl=== SECTION 4: Check for programs we need and that they are the right version @@ -886,6 +900,44 @@ # other files. AC_SUBST(ALL_BINDINGS,ocaml) +# Do any work necessary to ensure that bindings have what they need. +binding_prereqs_failed=0 +for a_binding in $BINDINGS_TO_BUILD ; do + case "$a_binding" in + ocaml) + if test "x$OCAMLC" = x ; then + AC_MSG_WARN([--enable-bindings=ocaml specified, but ocamlc not found. Try configure OCAMLC=/path/to/ocamlc]) + binding_prereqs_failed=1 + fi + if test "x$OCAMLDEP" = x ; then + AC_MSG_WARN([--enable-bindings=ocaml specified, but ocamldep not found. Try configure OCAMLDEP=/path/to/ocamldep]) + binding_prereqs_failed=1 + fi + if test "x$OCAMLOPT" = x ; then + AC_MSG_WARN([--enable-bindings=ocaml specified, but ocamlopt not found. Try configure OCAMLOPT=/path/to/ocamlopt]) + dnl ocamlopt is optional! + fi + if test "x$with_ocaml_libdir" != xauto ; then + AC_SUBST(OCAML_LIBDIR,$with_ocaml_libdir) + else + ocaml_stdlib="`"$OCAMLC" -where`" + if test "$LLVM_PREFIX" '<' "$ocaml_stdlib" -a "$ocaml_stdlib" '<' "$LLVM_PREFIX~" + then + # ocaml stdlib is beneath our prefix; use stdlib + AC_SUBST(OCAML_LIBDIR,$ocaml_stdlib) + else + # ocaml stdlib is outside our prefix; use libdir/ocaml + AC_SUBST(OCAML_LIBDIR,$LLVM_LIBDIR/ocaml) + fi + fi + ;; + esac +done +if test "$binding_prereqs_failed" = 1 ; then + AC_MSG_ERROR([Prequisites for bindings not satisfied. Fix them or use configure --disable-bindings.]) +fi + + dnl===-----------------------------------------------------------------------=== dnl=== dnl=== SECTION 10: Specify the output files and generate it Modified: llvm/trunk/bindings/ocaml/Makefile.ocaml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/Makefile.ocaml?rev=42532&r1=42531&r2=42532&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/Makefile.ocaml (original) +++ llvm/trunk/bindings/ocaml/Makefile.ocaml Tue Oct 2 11:42:10 2007 @@ -14,12 +14,6 @@ include $(LEVEL)/Makefile.config -# Find the ocaml stdlib root. /usr/local/lib/ocaml is the default when built -# from source; distros use something like /usr/lib/ocaml/3.10.0. -ifndef OCAML_LIBDIR -OCAML_LIBDIR := $(shell $(OCAMLC) -where) -endif - # CFLAGS needs to be set before Makefile.rules is included. Yes, ocaml puts its # includes under its libdir. CFLAGS += -I$(OCAML_LIBDIR) @@ -27,7 +21,7 @@ include $(LEVEL)/Makefile.common # Intentionally ignore PROJ_prefix here. We want the ocaml stdlib. However, the -# user can override this with OCAML_LIBDIR. +# user can override this with OCAML_LIBDIR or configure --with-ocaml-libdir=. PROJ_libocamldir := $(DESTDIR)$(OCAML_LIBDIR) OcamlDir := $(LibDir)/ocaml Modified: llvm/trunk/bindings/ocaml/bitwriter/llvm_bitwriter.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/bitwriter/llvm_bitwriter.mli?rev=42532&r1=42531&r2=42532&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/bitwriter/llvm_bitwriter.mli (original) +++ llvm/trunk/bindings/ocaml/bitwriter/llvm_bitwriter.mli Tue Oct 2 11:42:10 2007 @@ -8,7 +8,7 @@ *===----------------------------------------------------------------------=== * * This interface provides an ocaml API for the LLVM bitcode writer, the - * classes in the classes in the Bitwriter library. + * classes in the Bitwriter library. * *===----------------------------------------------------------------------===*) From gordonhenriksen at mac.com Tue Oct 2 11:42:22 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 16:42:22 -0000 Subject: [llvm-commits] [llvm] r42533 - /llvm/trunk/configure Message-ID: <200710021642.l92GgN8n025999@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 11:42:22 2007 New Revision: 42533 URL: http://llvm.org/viewvc/llvm-project?rev=42533&view=rev Log: Regenerate. Modified: llvm/trunk/configure Modified: llvm/trunk/configure URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/configure?rev=42533&r1=42532&r2=42533&view=diff ============================================================================== --- llvm/trunk/configure (original) +++ llvm/trunk/configure Tue Oct 2 11:42:22 2007 @@ -926,6 +926,7 @@ LLVM_CONFIGTIME BINDINGS_TO_BUILD ALL_BINDINGS +OCAML_LIBDIR LIBOBJS LTLIBOBJS' ac_subst_files='' @@ -1562,6 +1563,8 @@ --with-llvmgccdir Specify location of llvm-gcc install dir (default searches PATH) --with-extra-options Specify addtional options to compile LLVM with + --with-ocaml-libdir Specify install location for ocaml bindings (default + is stdlib) --with-tclinclude directory where tcl headers are --with-gnu-ld assume the C compiler uses GNU ld [default=no] --with-pic try to use only PIC/non-PIC objects [default=use @@ -4851,6 +4854,22 @@ esac +# Check whether --with-ocaml-libdir was given. +if test "${with_ocaml_libdir+set}" = set; then + withval=$with_ocaml_libdir; +else + withval=auto +fi + +case "$withval" in + auto) with_ocaml_libdir="$withval" ;; + /* | [A-Za-z]:[\\/]*) with_ocaml_libdir="$withval" ;; + *) { { echo "$as_me:$LINENO: error: Invalid path for --with-ocaml-libdir. Provide full path" >&5 +echo "$as_me: error: Invalid path for --with-ocaml-libdir. Provide full path" >&2;} + { (exit 1); exit 1; }; } ;; +esac + + ac_ext=c ac_cpp='$CPP $CPPFLAGS' ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' @@ -10526,7 +10545,7 @@ lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext < conftest.$ac_ext + echo '#line 12692 "configure"' > conftest.$ac_ext if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5 (eval $ac_compile) 2>&5 ac_status=$? @@ -14388,11 +14407,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14391: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14410: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:14395: \$? = $ac_status" >&5 + echo "$as_me:14414: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -14656,11 +14675,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14659: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14678: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:14663: \$? = $ac_status" >&5 + echo "$as_me:14682: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -14760,11 +14779,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14763: $lt_compile\"" >&5) + (eval echo "\"\$as_me:14782: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:14767: \$? = $ac_status" >&5 + echo "$as_me:14786: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -17212,7 +17231,7 @@ lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext < conftest.$ac_ext <&5) + (eval echo "\"\$as_me:19702: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:19687: \$? = $ac_status" >&5 + echo "$as_me:19706: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -19784,11 +19803,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:19787: $lt_compile\"" >&5) + (eval echo "\"\$as_me:19806: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:19791: \$? = $ac_status" >&5 + echo "$as_me:19810: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -21354,11 +21373,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:21357: $lt_compile\"" >&5) + (eval echo "\"\$as_me:21376: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:21361: \$? = $ac_status" >&5 + echo "$as_me:21380: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -21458,11 +21477,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:21461: $lt_compile\"" >&5) + (eval echo "\"\$as_me:21480: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:21465: \$? = $ac_status" >&5 + echo "$as_me:21484: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -23693,11 +23712,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:23696: $lt_compile\"" >&5) + (eval echo "\"\$as_me:23715: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:23700: \$? = $ac_status" >&5 + echo "$as_me:23719: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -23961,11 +23980,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:23964: $lt_compile\"" >&5) + (eval echo "\"\$as_me:23983: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:23968: \$? = $ac_status" >&5 + echo "$as_me:23987: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -24065,11 +24084,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:24068: $lt_compile\"" >&5) + (eval echo "\"\$as_me:24087: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:24072: \$? = $ac_status" >&5 + echo "$as_me:24091: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -34157,6 +34176,51 @@ ALL_BINDINGS=ocaml +# Do any work necessary to ensure that bindings have what they need. +binding_prereqs_failed=0 +for a_binding in $BINDINGS_TO_BUILD ; do + case "$a_binding" in + ocaml) + if test "x$OCAMLC" = x ; then + { echo "$as_me:$LINENO: WARNING: --enable-bindings=ocaml specified, but ocamlc not found. Try configure OCAMLC=/path/to/ocamlc" >&5 +echo "$as_me: WARNING: --enable-bindings=ocaml specified, but ocamlc not found. Try configure OCAMLC=/path/to/ocamlc" >&2;} + binding_prereqs_failed=1 + fi + if test "x$OCAMLDEP" = x ; then + { echo "$as_me:$LINENO: WARNING: --enable-bindings=ocaml specified, but ocamldep not found. Try configure OCAMLDEP=/path/to/ocamldep" >&5 +echo "$as_me: WARNING: --enable-bindings=ocaml specified, but ocamldep not found. Try configure OCAMLDEP=/path/to/ocamldep" >&2;} + binding_prereqs_failed=1 + fi + if test "x$OCAMLOPT" = x ; then + { echo "$as_me:$LINENO: WARNING: --enable-bindings=ocaml specified, but ocamlopt not found. Try configure OCAMLOPT=/path/to/ocamlopt" >&5 +echo "$as_me: WARNING: --enable-bindings=ocaml specified, but ocamlopt not found. Try configure OCAMLOPT=/path/to/ocamlopt" >&2;} + fi + if test "x$with_ocaml_libdir" != xauto ; then + OCAML_LIBDIR=$with_ocaml_libdir + + else + ocaml_stdlib="`"$OCAMLC" -where`" + if test "$LLVM_PREFIX" '<' "$ocaml_stdlib" -a "$ocaml_stdlib" '<' "$LLVM_PREFIX~" + then + # ocaml stdlib is beneath our prefix; use stdlib + OCAML_LIBDIR=$ocaml_stdlib + + else + # ocaml stdlib is outside our prefix; use libdir/ocaml + OCAML_LIBDIR=$LLVM_LIBDIR/ocaml + + fi + fi + ;; + esac +done +if test "$binding_prereqs_failed" = 1 ; then + { { echo "$as_me:$LINENO: error: Prequisites for bindings not satisfied. Fix them or use configure --disable-bindings." >&5 +echo "$as_me: error: Prequisites for bindings not satisfied. Fix them or use configure --disable-bindings." >&2;} + { (exit 1); exit 1; }; } +fi + + ac_config_headers="$ac_config_headers include/llvm/Config/config.h" @@ -35078,11 +35142,12 @@ LLVM_CONFIGTIME!$LLVM_CONFIGTIME$ac_delim BINDINGS_TO_BUILD!$BINDINGS_TO_BUILD$ac_delim ALL_BINDINGS!$ALL_BINDINGS$ac_delim +OCAML_LIBDIR!$OCAML_LIBDIR$ac_delim LIBOBJS!$LIBOBJS$ac_delim LTLIBOBJS!$LTLIBOBJS$ac_delim _ACEOF - if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 75; then + if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 76; then break elif $ac_last_try; then { { echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5 From dalej at apple.com Tue Oct 2 12:42:04 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 17:42:04 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42536 - /llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Message-ID: <200710021742.l92Hg5CZ028923@zion.cs.uiuc.edu> Author: johannes Date: Tue Oct 2 12:42:04 2007 New Revision: 42536 URL: http://llvm.org/viewvc/llvm-project?rev=42536&view=rev Log: Use getABITypeSizeInBits for improved type checking. Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-types.cpp?rev=42536&r1=42535&r2=42536&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Tue Oct 2 12:42:04 2007 @@ -74,12 +74,11 @@ static const Type * llvm_set_type(tree Tr, const Type *Ty) { // For x86 long double, llvm records the size of the data (80) while - // gcc's TYPE_SIZE including alignment padding. Compensate. + // gcc's TYPE_SIZE including alignment padding. getABITypeSizeInBits + // is used to compensate for this. assert((!TYPE_SIZE(Tr) || !Ty->isSized() || !isInt64(TYPE_SIZE(Tr), true) || - getInt64(TYPE_SIZE(Tr), true) == getTargetData().getTypeSizeInBits(Ty) || - (getTargetData().getTypeSizeInBits(Ty) == 80 && - (getInt64(TYPE_SIZE(Tr), true) == 96 || - getInt64(TYPE_SIZE(Tr), true) == 128))) + getInt64(TYPE_SIZE(Tr), true) == + getTargetData().getABITypeSizeInBits(Ty)) && "LLVM type size doesn't match GCC type size!"); unsigned &TypeSlot = LTypesMap[Ty]; From dalej at apple.com Tue Oct 2 12:44:00 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 17:44:00 -0000 Subject: [llvm-commits] [llvm] r42537 - in /llvm/trunk: include/llvm/Intrinsics.td lib/Analysis/ConstantFolding.cpp lib/CodeGen/IntrinsicLowering.cpp lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp lib/Target/CBackend/CBackend.cpp Message-ID: <200710021744.l92Hi0IJ029003@zion.cs.uiuc.edu> Author: johannes Date: Tue Oct 2 12:43:59 2007 New Revision: 42537 URL: http://llvm.org/viewvc/llvm-project?rev=42537&view=rev Log: Rewrite sqrt and powi to use anyfloat. By popular demand. Modified: llvm/trunk/include/llvm/Intrinsics.td llvm/trunk/lib/Analysis/ConstantFolding.cpp llvm/trunk/lib/CodeGen/IntrinsicLowering.cpp llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp llvm/trunk/lib/Target/CBackend/CBackend.cpp Modified: llvm/trunk/include/llvm/Intrinsics.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Intrinsics.td?rev=42537&r1=42536&r2=42537&view=diff ============================================================================== --- llvm/trunk/include/llvm/Intrinsics.td (original) +++ llvm/trunk/include/llvm/Intrinsics.td Tue Oct 2 12:43:59 2007 @@ -178,18 +178,8 @@ } let Properties = [IntrNoMem] in { - def int_sqrt_f32 : Intrinsic<[llvm_float_ty, llvm_float_ty]>; - def int_sqrt_f64 : Intrinsic<[llvm_double_ty, llvm_double_ty]>; - def int_sqrt_f80 : Intrinsic<[llvm_f80_ty, llvm_f80_ty]>; - def int_sqrt_f128 : Intrinsic<[llvm_f128_ty, llvm_f128_ty]>; - def int_sqrt_ppcf128 : Intrinsic<[llvm_ppcf128_ty, llvm_ppcf128_ty]>; - - def int_powi_f32 : Intrinsic<[llvm_float_ty, llvm_float_ty, llvm_i32_ty]>; - def int_powi_f64 : Intrinsic<[llvm_double_ty, llvm_double_ty, llvm_i32_ty]>; - def int_powi_f80 : Intrinsic<[llvm_f80_ty, llvm_f80_ty, llvm_i32_ty]>; - def int_powi_f128 : Intrinsic<[llvm_f128_ty, llvm_f128_ty, llvm_i32_ty]>; - def int_powi_ppcf128 : Intrinsic<[llvm_ppcf128_ty, llvm_ppcf128_ty, - llvm_i32_ty]>; + def int_sqrt : Intrinsic<[llvm_anyfloat_ty, LLVMMatchType<0>]>; + def int_powi : Intrinsic<[llvm_anyfloat_ty, LLVMMatchType<0>, llvm_i32_ty]>; } // NOTE: these are internal interfaces. Modified: llvm/trunk/lib/Analysis/ConstantFolding.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/ConstantFolding.cpp?rev=42537&r1=42536&r2=42537&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/ConstantFolding.cpp (original) +++ llvm/trunk/lib/Analysis/ConstantFolding.cpp Tue Oct 2 12:43:59 2007 @@ -329,16 +329,8 @@ bool llvm::canConstantFoldCallTo(Function *F) { switch (F->getIntrinsicID()) { - case Intrinsic::sqrt_f32: - case Intrinsic::sqrt_f64: - case Intrinsic::sqrt_f80: - case Intrinsic::sqrt_f128: - case Intrinsic::sqrt_ppcf128: - case Intrinsic::powi_f32: - case Intrinsic::powi_f64: - case Intrinsic::powi_f80: - case Intrinsic::powi_f128: - case Intrinsic::powi_ppcf128: + case Intrinsic::sqrt: + case Intrinsic::powi: case Intrinsic::bswap: case Intrinsic::ctpop: case Intrinsic::ctlz: @@ -539,12 +531,12 @@ } } else if (NumOperands == 2) { if (ConstantFP *Op1 = dyn_cast(Operands[0])) { + if (Ty!=Type::FloatTy && Ty!=Type::DoubleTy) + return 0; double Op1V = Ty==Type::FloatTy ? (double)Op1->getValueAPF().convertToFloat(): Op1->getValueAPF().convertToDouble(); if (ConstantFP *Op2 = dyn_cast(Operands[1])) { - if (Ty!=Type::FloatTy && Ty!=Type::DoubleTy) - return 0; double Op2V = Ty==Type::FloatTy ? (double)Op2->getValueAPF().convertToFloat(): Op2->getValueAPF().convertToDouble(); Modified: llvm/trunk/lib/CodeGen/IntrinsicLowering.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/IntrinsicLowering.cpp?rev=42537&r1=42536&r2=42537&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/IntrinsicLowering.cpp (original) +++ llvm/trunk/lib/CodeGen/IntrinsicLowering.cpp Tue Oct 2 12:43:59 2007 @@ -99,14 +99,20 @@ PointerType::get(Type::Int8Ty), Type::Int32Ty, TD.getIntPtrType(), (Type *)0); break; - case Intrinsic::sqrt_f32: - case Intrinsic::sqrt_f64: - if(I->arg_begin()->getType() == Type::FloatTy) + case Intrinsic::sqrt: + switch((int)I->arg_begin()->getType()->getTypeID()) { + case Type::FloatTyID: EnsureFunctionExists(M, "sqrtf", I->arg_begin(), I->arg_end(), Type::FloatTy); - else + case Type::DoubleTyID: EnsureFunctionExists(M, "sqrt", I->arg_begin(), I->arg_end(), Type::DoubleTy); + case Type::X86_FP80TyID: + case Type::FP128TyID: + case Type::PPC_FP128TyID: + EnsureFunctionExists(M, "sqrtl", I->arg_begin(), I->arg_end(), + I->arg_begin()->getType()); + } break; } } @@ -782,34 +788,27 @@ MemsetFCache); break; } - case Intrinsic::sqrt_f32: { + case Intrinsic::sqrt: { static Constant *sqrtfFCache = 0; - ReplaceCallWith("sqrtf", CI, CI->op_begin()+1, CI->op_end(), - Type::FloatTy, sqrtfFCache); - break; - } - case Intrinsic::sqrt_f64: { static Constant *sqrtFCache = 0; - ReplaceCallWith("sqrt", CI, CI->op_begin()+1, CI->op_end(), + static Constant *sqrtLDCache = 0; + switch (CI->getOperand(1)->getType()->getTypeID()) { + default: assert(0 && "Invalid type in sqrt"); abort(); + case Type::FloatTyID: + ReplaceCallWith("sqrtf", CI, CI->op_begin()+1, CI->op_end(), + Type::FloatTy, sqrtfFCache); + break; + case Type::DoubleTyID: + ReplaceCallWith("sqrt", CI, CI->op_begin()+1, CI->op_end(), Type::DoubleTy, sqrtFCache); - break; - } - case Intrinsic::sqrt_f80: { - static Constant *sqrtF80Cache = 0; - ReplaceCallWith("sqrtl", CI, CI->op_begin()+1, CI->op_end(), - Type::X86_FP80Ty, sqrtF80Cache); - break; - } - case Intrinsic::sqrt_f128: { - static Constant *sqrtF128Cache = 0; - ReplaceCallWith("sqrtl", CI, CI->op_begin()+1, CI->op_end(), - Type::FP128Ty, sqrtF128Cache); - break; - } - case Intrinsic::sqrt_ppcf128: { - static Constant *sqrtppcF128Cache = 0; - ReplaceCallWith("sqrtl", CI, CI->op_begin()+1, CI->op_end(), - Type::PPC_FP128Ty, sqrtppcF128Cache); + break; + case Type::X86_FP80TyID: + case Type::FP128TyID: + case Type::PPC_FP128TyID: + ReplaceCallWith("sqrtl", CI, CI->op_begin()+1, CI->op_end(), + CI->getOperand(1)->getType(), sqrtLDCache); + break; + } break; } } Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp?rev=42537&r1=42536&r2=42537&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Tue Oct 2 12:43:59 2007 @@ -2796,20 +2796,12 @@ return 0; } - case Intrinsic::sqrt_f32: - case Intrinsic::sqrt_f64: - case Intrinsic::sqrt_f80: - case Intrinsic::sqrt_f128: - case Intrinsic::sqrt_ppcf128: + case Intrinsic::sqrt: setValue(&I, DAG.getNode(ISD::FSQRT, getValue(I.getOperand(1)).getValueType(), getValue(I.getOperand(1)))); return 0; - case Intrinsic::powi_f32: - case Intrinsic::powi_f64: - case Intrinsic::powi_f80: - case Intrinsic::powi_f128: - case Intrinsic::powi_ppcf128: + case Intrinsic::powi: setValue(&I, DAG.getNode(ISD::FPOWI, getValue(I.getOperand(1)).getValueType(), getValue(I.getOperand(1)), Modified: llvm/trunk/lib/Target/CBackend/CBackend.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/CBackend/CBackend.cpp?rev=42537&r1=42536&r2=42537&view=diff ============================================================================== --- llvm/trunk/lib/Target/CBackend/CBackend.cpp (original) +++ llvm/trunk/lib/Target/CBackend/CBackend.cpp Tue Oct 2 12:43:59 2007 @@ -2417,8 +2417,7 @@ case Intrinsic::longjmp: case Intrinsic::prefetch: case Intrinsic::dbg_stoppoint: - case Intrinsic::powi_f32: - case Intrinsic::powi_f64: + case Intrinsic::powi: // We directly implement these intrinsics break; default: @@ -2537,8 +2536,7 @@ writeOperand(I.getOperand(1)); Out << ')'; return; - case Intrinsic::powi_f32: - case Intrinsic::powi_f64: + case Intrinsic::powi: Out << "__builtin_powi("; writeOperand(I.getOperand(1)); Out << ", "; From dalej at apple.com Tue Oct 2 12:44:34 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 17:44:34 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42538 - /llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp Message-ID: <200710021744.l92HiY2u029037@zion.cs.uiuc.edu> Author: johannes Date: Tue Oct 2 12:44:34 2007 New Revision: 42538 URL: http://llvm.org/viewvc/llvm-project?rev=42538&view=rev Log: Rewrite sqrt and powi to use anyfloat. Modified: llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp?rev=42538&r1=42537&r2=42538&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-convert.cpp Tue Oct 2 12:44:34 2007 @@ -4499,17 +4499,10 @@ Value *TreeToLLVM::EmitBuiltinSQRT(tree exp) { Value *Amt = Emit(TREE_VALUE(TREE_OPERAND(exp, 1)), 0); Intrinsic::ID Id = Intrinsic::not_intrinsic; + const Type* Ty = Amt->getType(); - switch (Amt->getType()->getTypeID()) { - default: assert(0 && "Unknown FP type!"); - case Type::FloatTyID: Id = Intrinsic::sqrt_f32; break; - case Type::DoubleTyID: Id = Intrinsic::sqrt_f64; break; - case Type::X86_FP80TyID: Id = Intrinsic::sqrt_f80; break; - case Type::FP128TyID: Id = Intrinsic::sqrt_f128; break; - case Type::PPC_FP128TyID: Id = Intrinsic::sqrt_ppcf128; break; - } - - return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, Id), + return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, + Intrinsic::sqrt, &Ty, 1), Amt, "tmp"); } @@ -4520,23 +4513,14 @@ Value *Val = Emit(TREE_VALUE(ArgList), 0); Value *Pow = Emit(TREE_VALUE(TREE_CHAIN(ArgList)), 0); + const Type *Ty = Val->getType(); Pow = CastToSIntType(Pow, Type::Int32Ty); - Intrinsic::ID Id = Intrinsic::not_intrinsic; - - switch (Val->getType()->getTypeID()) { - default: assert(0 && "Unknown FP type!"); - case Type::FloatTyID: Id = Intrinsic::powi_f32; break; - case Type::DoubleTyID: Id = Intrinsic::powi_f64; break; - case Type::X86_FP80TyID: Id = Intrinsic::powi_f80; break; - case Type::FP128TyID: Id = Intrinsic::powi_f128; break; - case Type::PPC_FP128TyID: Id = Intrinsic::powi_ppcf128; break; - } - SmallVector Args; Args.push_back(Val); Args.push_back(Pow); - return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, Id), + return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, + Intrinsic::powi, &Ty, 1), Args.begin(), Args.end(), "tmp"); } From dalej at apple.com Tue Oct 2 12:44:51 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 17:44:51 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42539 - /llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp Message-ID: <200710021744.l92Hipqm029060@zion.cs.uiuc.edu> Author: johannes Date: Tue Oct 2 12:44:51 2007 New Revision: 42539 URL: http://llvm.org/viewvc/llvm-project?rev=42539&view=rev Log: Rewrite sqrt and powi to use anyfloat. Modified: llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp?rev=42539&r1=42538&r2=42539&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp Tue Oct 2 12:44:51 2007 @@ -4052,17 +4052,10 @@ Value *TreeToLLVM::EmitBuiltinSQRT(tree exp) { Value *Amt = Emit(TREE_VALUE(TREE_OPERAND(exp, 1)), 0); Intrinsic::ID Id = Intrinsic::not_intrinsic; + const Type* Ty = Amt->getType(); - switch (Amt->getType()->getTypeID()) { - default: assert(0 && "Unknown FP type!"); - case Type::FloatTyID: Id = Intrinsic::sqrt_f32; break; - case Type::DoubleTyID: Id = Intrinsic::sqrt_f64; break; - case Type::X86_FP80TyID: Id = Intrinsic::sqrt_f80; break; - case Type::FP128TyID: Id = Intrinsic::sqrt_f128; break; - case Type::PPC_FP128TyID: Id = Intrinsic::sqrt_ppcf128; break; - } - - return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, Id), + return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, + Intrinsic::sqrt, &Ty, 1), Amt, "tmp"); } @@ -4073,23 +4066,14 @@ Value *Val = Emit(TREE_VALUE(ArgList), 0); Value *Pow = Emit(TREE_VALUE(TREE_CHAIN(ArgList)), 0); + const Type *Ty = Val->getType(); Pow = CastToSIntType(Pow, Type::Int32Ty); - Intrinsic::ID Id = Intrinsic::not_intrinsic; - - switch (Val->getType()->getTypeID()) { - default: assert(0 && "Unknown FP type!"); - case Type::FloatTyID: Id = Intrinsic::powi_f32; break; - case Type::DoubleTyID: Id = Intrinsic::powi_f64; break; - case Type::X86_FP80TyID: Id = Intrinsic::powi_f80; break; - case Type::FP128TyID: Id = Intrinsic::powi_f128; break; - case Type::PPC_FP128TyID: Id = Intrinsic::powi_ppcf128; break; - } - SmallVector Args; Args.push_back(Val); Args.push_back(Pow); - return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, Id), + return Builder.CreateCall(Intrinsic::getDeclaration(TheModule, + Intrinsic::powi, &Ty, 1), Args.begin(), Args.end(), "tmp"); } From dalej at apple.com Tue Oct 2 12:47:38 2007 From: dalej at apple.com (Dale Johannesen) Date: Tue, 02 Oct 2007 17:47:38 -0000 Subject: [llvm-commits] [llvm] r42540 - /llvm/trunk/docs/LangRef.html Message-ID: <200710021747.l92HlcZ7029155@zion.cs.uiuc.edu> Author: johannes Date: Tue Oct 2 12:47:38 2007 New Revision: 42540 URL: http://llvm.org/viewvc/llvm-project?rev=42540&view=rev Log: Document new sqrt and powi handling. Modified: llvm/trunk/docs/LangRef.html Modified: llvm/trunk/docs/LangRef.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.html?rev=42540&r1=42539&r2=42540&view=diff ============================================================================== --- llvm/trunk/docs/LangRef.html (original) +++ llvm/trunk/docs/LangRef.html Tue Oct 2 12:47:38 2007 @@ -4484,9 +4484,14 @@
    Syntax:
    +

    This is an overloaded intrinsic. You can use llvm.sqrt on any +floating point type. Not all targets support all types however. 

    -  declare float @llvm.sqrt.f32(float %Val)
-  declare double @llvm.sqrt.f64(double %Val)
+  declare float     @llvm.sqrt.f32(float %Val)
+  declare double    @llvm.sqrt.f64(double %Val)
+  declare x86_fp80  @llvm.sqrt.f80(x86_fp80 %Val)
+  declare fp128     @llvm.sqrt.f128(fp128 %Val)
+  declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
    Overview:
    @@ -4520,9 +4525,14 @@
    Syntax:
    +

    This is an overloaded intrinsic. You can use llvm.powi on any +floating point type. Not all targets support all types however. 

    -  declare float  @llvm.powi.f32(float  %Val, i32 %power)
-  declare double @llvm.powi.f64(double %Val, i32 %power)
+  declare float     @llvm.powi.f32(float  %Val, i32 %power)
+  declare double    @llvm.powi.f64(double %Val, i32 %power)
+  declare x86_fp80  @llvm.powi.f80(x86_fp80  %Val, i32 %power)
+  declare fp128     @llvm.powi.f128(fp128 %Val, i32 %power)
+  declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128  %Val, i32 %power)
    Overview:
    From gordonhenriksen at mac.com Tue Oct 2 13:04:26 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Tue, 02 Oct 2007 18:04:26 -0000 Subject: [llvm-commits] [llvm] r42541 - /llvm/trunk/CREDITS.TXT Message-ID: <200710021804.l92I4QDA029787@zion.cs.uiuc.edu> Author: gordon Date: Tue Oct 2 13:04:24 2007 New Revision: 42541 URL: http://llvm.org/viewvc/llvm-project?rev=42541&view=rev Log: Updating my credits entry. Modified: llvm/trunk/CREDITS.TXT Modified: llvm/trunk/CREDITS.TXT URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/CREDITS.TXT?rev=42541&r1=42540&r2=42541&view=diff ============================================================================== --- llvm/trunk/CREDITS.TXT (original) +++ llvm/trunk/CREDITS.TXT Tue Oct 2 13:04:24 2007 @@ -108,7 +108,9 @@ N: Gordon Henriksen E: gordonhenriksen at mac.com +D: Pluggable GC support D: C interface +D: Ocaml bindings N: Raul Fernandes Herbster E: raul at dsc.ufcg.edu.br From lattner at apple.com Tue Oct 2 13:51:31 2007 From: lattner at apple.com (Tanya Lattner) Date: Tue, 2 Oct 2007 11:51:31 -0700 Subject: [llvm-commits] make check errors Message-ID: <24EC4450-66E1-4E56-B5AB-867F4871F420@apple.com> I'm seeing several make check errors (this is x86 Darwin, Revision: 42541). Any ideas? FAIL: /Users/lattner/work/llvm/test/CFrontend/2004-11-27- VariableSizeInStructure.c for PR360 Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm /Users/lattner/work/llvm/test/CFrontend/2004-11-27- VariableSizeInStructure.c -S -o /dev/null /Users/lattner/work/llvm/test/CFrontend/2004-11-27- VariableSizeInStructure.c: In function 'sub1': /Users/lattner/work/llvm/test/CFrontend/2004-11-27- VariableSizeInStructure.c:7: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm- types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. FAIL: /Users/lattner/work/llvm/test/CFrontend/2005-10-18- VariableSizedElementCrash.c Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm /Users/lattner/work/llvm/test/CFrontend/2005-10-18- VariableSizedElementCrash.c -S -o - .abort /Users/lattner/work/llvm/test/CFrontend/2005-10-18- VariableSizedElementCrash.c: In function 'sub1': /Users/lattner/work/llvm/test/CFrontend/2005-10-18- VariableSizedElementCrash.c:5: warning: no semicolon at end of struct or union /Users/lattner/work/llvm/test/CFrontend/2005-10-18- VariableSizedElementCrash.c:3: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm- types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. FAIL: /Users/lattner/work/llvm/test/CFrontend/2007-02-04- WITH_SIZE_EXPR.c for PR1174 Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm /Users/lattner/work/llvm/test/CFrontend/2007-02-04- WITH_SIZE_EXPR.c -O3 -S -o - -emit-llvm .abort /Users/lattner/work/llvm/test/CFrontend/2007-02-04-WITH_SIZE_EXPR.c: In function 'zzz': /Users/lattner/work/llvm/test/CFrontend/2007-02-04-WITH_SIZE_EXPR.c: 5: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm-types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. FAIL: /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c for PR1170 Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm -S /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c -o - .abort /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c:3: warning: anonymous struct declared inside parameter list /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c:3: warning: its scope is only this definition or declaration, which is probably not what you want /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c: In function 'f': /Users/lattner/work/llvm/test/CFrontend/2007-02-16- VariableSizeStructArg.c:3: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm- types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. FAIL: /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct1.c Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct1.c -S -o - .abort /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct1.c: In function 'p': /Users/lattner/work/llvm/test/CFrontend/2007-03-06-VarSizeInStruct1.c: 7: warning: function returns address of local variable /Users/lattner/work/llvm/test/CFrontend/2007-03-06-VarSizeInStruct1.c: 2: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm-types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. FAIL: /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct2.c Failed with exit(1) at line 1 while running: /Users/lattner/work/llvm-gcc/install/bin/llvm-gcc - emit-llvm /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct2.c -S -o - .abort /Users/lattner/work/llvm/test/CFrontend/2007-03-06- VarSizeInStruct2.c: In function 'p': /Users/lattner/work/llvm/test/CFrontend/2007-03-06-VarSizeInStruct2.c: 2: internal compiler error: tree check: expected integer_cst, have var_decl in ConvertRECORD, at llvm-types.cpp:1673 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. -Tanya From dpatel at apple.com Tue Oct 2 13:56:40 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 2 Oct 2007 11:56:40 -0700 Subject: [llvm-commits] make check errors In-Reply-To: <24EC4450-66E1-4E56-B5AB-867F4871F420@apple.com> References: <24EC4450-66E1-4E56-B5AB-867F4871F420@apple.com> Message-ID: On Oct 2, 2007, at 11:51 AM, Tanya Lattner wrote: > I'm seeing several make check errors (this is x86 Darwin, Revision: > 42541). Any ideas? I'll take a look. - Devang From isanbard at gmail.com Tue Oct 2 14:55:05 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 19:55:05 -0000 Subject: [llvm-commits] [llvm] r42542 - /llvm/trunk/lib/Target/X86/README-SSE.txt Message-ID: <200710021955.l92Jt57O002086@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 14:55:05 2007 New Revision: 42542 URL: http://llvm.org/viewvc/llvm-project?rev=42542&view=rev Log: Micro-optimization -- missed LICM opportunity. Modified: llvm/trunk/lib/Target/X86/README-SSE.txt Modified: llvm/trunk/lib/Target/X86/README-SSE.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README-SSE.txt?rev=42542&r1=42541&r2=42542&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README-SSE.txt (original) +++ llvm/trunk/lib/Target/X86/README-SSE.txt Tue Oct 2 14:55:05 2007 @@ -671,3 +671,27 @@ //===---------------------------------------------------------------------===// +In this loop: + +LBB4_6: # bb47.preheader + shlw $2, %si + decw %si + movzwl %si, %eax + incl %eax + xorl %ecx, %ecx +LBB4_7: # bb47 + xorps %xmm0, %xmm0 # (1) + movaps %xmm0, (%edx) + movaps %xmm0, 16(%edx) + addl $32, %edx + incl %ecx + cmpl %eax, %ecx + jne LBB4_7 # bb47 +LBB4_8: # bb42.return_crit_edge + xorw %si, %si + jmp LBB4_12 # return + +The instruction at (1) can be moved out of the main body of the loop. + +//===---------------------------------------------------------------------===// + From isanbard at gmail.com Tue Oct 2 15:42:59 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 20:42:59 -0000 Subject: [llvm-commits] [llvm] r42546 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710022042.l92Kgx2O005184@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 15:42:59 2007 New Revision: 42546 URL: http://llvm.org/viewvc/llvm-project?rev=42546&view=rev Log: Another missed optimization. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42546&r1=42545&r2=42546&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 15:42:59 2007 @@ -1239,3 +1239,31 @@ //===---------------------------------------------------------------------===// +The following code: + +LBB3_5: # bb114.preheader + movswl -68(%ebp), %eax + movl $32, %ecx + movl %ecx, -80(%ebp) + subl %eax, -80(%ebp) + movswl -52(%ebp), %eax + movl %ecx, -84(%ebp) + subl %eax, -84(%ebp) + movswl -70(%ebp), %eax + movl %ecx, -88(%ebp) + subl %eax, -88(%ebp) + movswl -50(%ebp), %eax + subl %eax, %ecx + movl %ecx, -76(%ebp) + movswl -42(%ebp), %eax + movl %eax, -92(%ebp) + movswl -66(%ebp), %eax + movl %eax, -96(%ebp) + movw $0, -98(%ebp) + +has redundant subtractions of %eax from a stack slot. However, %ecx doesn't +change, so we could simply subtract %eax from %ecx first and then use %ecx (or +vice-versa). + +//===---------------------------------------------------------------------===// + From isanbard at gmail.com Tue Oct 2 15:54:32 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 20:54:32 -0000 Subject: [llvm-commits] [llvm] r42547 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710022054.l92KsWiX006088@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 15:54:32 2007 New Revision: 42547 URL: http://llvm.org/viewvc/llvm-project?rev=42547&view=rev Log: Now with LL code! Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42547&r1=42546&r2=42547&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 15:54:32 2007 @@ -1241,6 +1241,22 @@ The following code: +bb114.preheader: ; preds = %cond_next94 + %tmp231232 = sext i16 %tmp62 to i32 ; [#uses=1] + %tmp233 = sub i32 32, %tmp231232 ; [#uses=1] + %tmp245246 = sext i16 %tmp65 to i32 ; [#uses=1] + %tmp252253 = sext i16 %tmp68 to i32 ; [#uses=1] + %tmp254 = sub i32 32, %tmp252253 ; [#uses=1] + %tmp553554 = bitcast i16* %tmp37 to i8* ; [#uses=2] + %tmp583584 = sext i16 %tmp98 to i32 ; [#uses=1] + %tmp585 = sub i32 32, %tmp583584 ; [#uses=1] + %tmp614615 = sext i16 %tmp101 to i32 ; [#uses=1] + %tmp621622 = sext i16 %tmp104 to i32 ; [#uses=1] + %tmp623 = sub i32 32, %tmp621622 ; [#uses=1] + br label %bb114 + +produces: + LBB3_5: # bb114.preheader movswl -68(%ebp), %eax movl $32, %ecx @@ -1261,7 +1277,7 @@ movl %eax, -96(%ebp) movw $0, -98(%ebp) -has redundant subtractions of %eax from a stack slot. However, %ecx doesn't +This has redundant subtractions of %eax from a stack slot. However, %ecx doesn't change, so we could simply subtract %eax from %ecx first and then use %ecx (or vice-versa). From isanbard at gmail.com Tue Oct 2 16:01:16 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 21:01:16 -0000 Subject: [llvm-commits] [llvm] r42548 - /llvm/trunk/lib/Target/X86/README-SSE.txt Message-ID: <200710022101.l92L1GDe006547@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 16:01:16 2007 New Revision: 42548 URL: http://llvm.org/viewvc/llvm-project?rev=42548&view=rev Log: Now with source code. Modified: llvm/trunk/lib/Target/X86/README-SSE.txt Modified: llvm/trunk/lib/Target/X86/README-SSE.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README-SSE.txt?rev=42548&r1=42547&r2=42548&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README-SSE.txt (original) +++ llvm/trunk/lib/Target/X86/README-SSE.txt Tue Oct 2 16:01:16 2007 @@ -673,6 +673,22 @@ In this loop: +bb49: ; preds = %bb49, %bb49.preheader + %indvar = phi i32 [ 0, %bb49.preheader ], [ %indvar.next, %bb49 ] ; [#uses=2] + %dp.089.0.rec = shl i32 %indvar, 3 ; [#uses=2] + %dp.089.0 = getelementptr i32* %tmp89, i32 %dp.089.0.rec ; [#uses=1] + %tmp5051 = bitcast i32* %dp.089.0 to <2 x i64>* ; <<2 x i64>*> [#uses=1] + store <2 x i64> zeroinitializer, <2 x i64>* %tmp5051, align 16 + %dp.089.0.sum105 = or i32 %dp.089.0.rec, 4 ; [#uses=1] + %tmp56 = getelementptr i32* %tmp89, i32 %dp.089.0.sum105 ; [#uses=1] + %tmp5657 = bitcast i32* %tmp56 to <2 x i64>* ; <<2 x i64>*> [#uses=1] + store <2 x i64> zeroinitializer, <2 x i64>* %tmp5657, align 16 + %indvar.next = add i32 %indvar, 1 ; [#uses=2] + %exitcond = icmp eq i32 %indvar.next, %tmp98 ; [#uses=1] + br i1 %exitcond, label %bb72, label %bb49 + +we get: + LBB4_6: # bb47.preheader shlw $2, %si decw %si From isanbard at gmail.com Tue Oct 2 16:02:53 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 21:02:53 -0000 Subject: [llvm-commits] [llvm] r42549 - /llvm/trunk/lib/Target/X86/README-SSE.txt Message-ID: <200710022102.l92L2sTb006688@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 16:02:53 2007 New Revision: 42549 URL: http://llvm.org/viewvc/llvm-project?rev=42549&view=rev Log: Small label changes. Modified: llvm/trunk/lib/Target/X86/README-SSE.txt Modified: llvm/trunk/lib/Target/X86/README-SSE.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README-SSE.txt?rev=42549&r1=42548&r2=42549&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README-SSE.txt (original) +++ llvm/trunk/lib/Target/X86/README-SSE.txt Tue Oct 2 16:02:53 2007 @@ -689,13 +689,13 @@ we get: -LBB4_6: # bb47.preheader +LBB5_6: # bb49.preheader shlw $2, %si decw %si movzwl %si, %eax incl %eax xorl %ecx, %ecx -LBB4_7: # bb47 +LBB5_7: # bb49 xorps %xmm0, %xmm0 # (1) movaps %xmm0, (%edx) movaps %xmm0, 16(%edx) @@ -703,9 +703,6 @@ incl %ecx cmpl %eax, %ecx jne LBB4_7 # bb47 -LBB4_8: # bb42.return_crit_edge - xorw %si, %si - jmp LBB4_12 # return The instruction at (1) can be moved out of the main body of the loop. From dpatel at apple.com Tue Oct 2 16:28:45 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 21:28:45 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42550 - /llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Message-ID: <200710022128.l92LSjLQ008191@zion.cs.uiuc.edu> Author: dpatel Date: Tue Oct 2 16:28:44 2007 New Revision: 42550 URL: http://llvm.org/viewvc/llvm-project?rev=42550&view=rev Log: Remove extra bytes from LLVMStruct only if is bigger than GCC struct. This patch fixes gcc assertion failure caused by previous patch. Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.0/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-types.cpp?rev=42550&r1=42549&r2=42550&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-types.cpp Tue Oct 2 16:28:44 2007 @@ -1160,7 +1160,7 @@ // If this is a Packed struct and ExtraBitsAvailable is not zero then // remove Extra bytes if ExtraBitsAvailable > 8. - void RemoveExtraBytes (unsigned GCCTypeSize) { + void RemoveExtraBytes () { unsigned NoOfBytesToRemove = ExtraBitsAvailable/8; @@ -1170,10 +1170,6 @@ if (NoOfBytesToRemove == 0) return; - unsigned OriginalSize = ElementSizeInBytes.back(); - if (OriginalSize == GCCTypeSize) - return; - const Type *LastType = Elements.back(); unsigned PadBytes = 0; @@ -1192,6 +1188,7 @@ // Update last element type and size, element offset is unchanged. const Type *Pad = ArrayType::get(Type::Int8Ty, PadBytes); + unsigned OriginalSize = ElementSizeInBytes.back(); Elements.pop_back(); Elements.push_back(Pad); @@ -1670,13 +1667,17 @@ for (tree Field = TYPE_FIELDS(type); Field; Field = TREE_CHAIN(Field)) DecodeStructFields(Field, *Info); - uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; - Info->RemoveExtraBytes(GCCTypeSize); // If the LLVM struct requires explicit tail padding to be the same size as // the GCC struct, insert tail padding now. This handles, e.g., "{}" in C++. if (TYPE_SIZE(type) && TREE_CODE(TYPE_SIZE(type)) == INTEGER_CST) { + uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; uint64_t LLVMStructSize = Info->getSizeAsLLVMStruct(); - + + if (LLVMStructSize > GCCTypeSize) { + Info->RemoveExtraBytes(); + LLVMStructSize = Info->getSizeAsLLVMStruct(); + } + if (LLVMStructSize != GCCTypeSize) { assert(LLVMStructSize < GCCTypeSize && "LLVM type size doesn't match GCC type size!"); @@ -1702,7 +1703,9 @@ } } } - } + } else + Info->RemoveExtraBytes(); + // Now that the LLVM struct is finalized, figure out a safe place to index to // and set the DECL_LLVM values for each FieldDecl that doesn't start at a From dpatel at apple.com Tue Oct 2 16:33:23 2007 From: dpatel at apple.com (Devang Patel) Date: Tue, 02 Oct 2007 21:33:23 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42551 - /llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Message-ID: <200710022133.l92LXNfk008508@zion.cs.uiuc.edu> Author: dpatel Date: Tue Oct 2 16:33:22 2007 New Revision: 42551 URL: http://llvm.org/viewvc/llvm-project?rev=42551&view=rev Log: Remove extra bytes from LLVMStruct only if is bigger than GCC struct. This patch fixes gcc assertion failure caused by previous patch. Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Modified: llvm-gcc-4.2/trunk/gcc/llvm-types.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-types.cpp?rev=42551&r1=42550&r2=42551&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/llvm-types.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/llvm-types.cpp Tue Oct 2 16:33:22 2007 @@ -1184,7 +1184,7 @@ // If this is a Packed struct and ExtraBitsAvailable is not zero then // remove Extra bytes if ExtraBitsAvailable > 8. - void RemoveExtraBytes (unsigned GCCTypeSize) { + void RemoveExtraBytes () { unsigned NoOfBytesToRemove = ExtraBitsAvailable/8; @@ -1194,10 +1194,6 @@ if (NoOfBytesToRemove == 0) return; - unsigned OriginalSize = ElementSizeInBytes.back(); - if (OriginalSize == GCCTypeSize) - return; - const Type *LastType = Elements.back(); unsigned PadBytes = 0; @@ -1216,6 +1212,7 @@ // Update last element type and size, element offset is unchanged. const Type *Pad = ArrayType::get(Type::Int8Ty, PadBytes); + unsigned OriginalSize = ElementSizeInBytes.back(); Elements.pop_back(); Elements.push_back(Pad); @@ -1694,13 +1691,17 @@ for (tree Field = TYPE_FIELDS(type); Field; Field = TREE_CHAIN(Field)) DecodeStructFields(Field, *Info); - uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; - Info->RemoveExtraBytes(GCCTypeSize); // If the LLVM struct requires explicit tail padding to be the same size as // the GCC struct, insert tail padding now. This handles, e.g., "{}" in C++. if (TYPE_SIZE(type) && TREE_CODE(TYPE_SIZE(type)) == INTEGER_CST) { + uint64_t GCCTypeSize = ((uint64_t)TREE_INT_CST_LOW(TYPE_SIZE(type))+7)/8; uint64_t LLVMStructSize = Info->getSizeAsLLVMStruct(); - + + if (LLVMStructSize > GCCTypeSize) { + Info->RemoveExtraBytes(); + LLVMStructSize = Info->getSizeAsLLVMStruct(); + } + if (LLVMStructSize != GCCTypeSize) { assert(LLVMStructSize < GCCTypeSize && "LLVM type size doesn't match GCC type size!"); @@ -1726,7 +1727,9 @@ } } } - } + } else + Info->RemoveExtraBytes(); + // Now that the LLVM struct is finalized, figure out a safe place to index to // and set index values for each FieldDecl that doesn't start at a variable From isanbard at gmail.com Tue Oct 2 16:43:06 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 21:43:06 -0000 Subject: [llvm-commits] [llvm] r42552 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710022143.l92Lh6Ne009164@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 16:43:06 2007 New Revision: 42552 URL: http://llvm.org/viewvc/llvm-project?rev=42552&view=rev Log: Another missed optimization with LICM. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42552&r1=42551&r2=42552&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 16:43:06 2007 @@ -1283,3 +1283,26 @@ //===---------------------------------------------------------------------===// +For this code: + +cond_next603: ; preds = %bb493, %cond_true336, %cond_next599 + %v.21050.1 = phi i32 [ %v.21050.0, %cond_next599 ], [ %tmp344, %cond_true336 ], [ %v.2, %bb493 ] ; [#uses=1] + %maxz.21051.1 = phi i32 [ %maxz.21051.0, %cond_next599 ], [ 0, %cond_true336 ], [ %maxz.2, %bb493 ] ; [#uses=2] + %cnt.01055.1 = phi i32 [ %cnt.01055.0, %cond_next599 ], [ 0, %cond_true336 ], [ %cnt.0, %bb493 ] ; [#uses=2] + %byteptr.9 = phi i8* [ %byteptr.12, %cond_next599 ], [ %byteptr.0, %cond_true336 ], [ %byteptr.10, %bb493 ] ; [#uses=9] + %bitptr.6 = phi i32 [ %tmp5571104.1, %cond_next599 ], [ %tmp4921049, %cond_true336 ], [ %bitptr.7, %bb493 ] ; [#uses=4] + %source.5 = phi i32 [ %tmp602, %cond_next599 ], [ %source.0, %cond_true336 ], [ %source.6, %bb493 ] ; [#uses=7] + %tmp606 = getelementptr %struct.const_tables* @tables, i32 0, i32 0, i32 %cnt.01055.1 ; [#uses=1] + %tmp607 = load i8* %tmp606, align 1 ; [#uses=1] + +We produce this: + +LBB4_70: # cond_next603 + movl -20(%ebp), %esi + movl L_tables$non_lazy_ptr-"L4$pb"(%esi), %esi + +However, ICC caches this information before the loop and produces this: + + movl 88(%esp), %eax #481.12 + +//===---------------------------------------------------------------------===// From isanbard at gmail.com Tue Oct 2 16:49:31 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 21:49:31 -0000 Subject: [llvm-commits] [llvm] r42554 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710022149.l92LnVFU009573@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 16:49:31 2007 New Revision: 42554 URL: http://llvm.org/viewvc/llvm-project?rev=42554&view=rev Log: Another micro-opt. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42554&r1=42553&r2=42554&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 16:49:31 2007 @@ -1306,3 +1306,19 @@ movl 88(%esp), %eax #481.12 //===---------------------------------------------------------------------===// + +This code: + + %tmp659 = icmp slt i16 %tmp654, 0 ; [#uses=1] + br i1 %tmp659, label %cond_true662, label %cond_next715 + +produces this: + + testw %cx, %cx + movswl %cx, %esi + jns LBB4_109 # cond_next715 + +Shark tells us that using %cx in the testw instruction is sub-optimal. It +suggests using the 32-bit register (which is what ICC uses). + +//===---------------------------------------------------------------------===// From isanbard at gmail.com Tue Oct 2 17:26:38 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 22:26:38 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42557 - /llvm-gcc-4.0/tags/llvm-gcc-1200-16/ Message-ID: <200710022226.l92MQcox011349@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 17:26:38 2007 New Revision: 42557 URL: http://llvm.org/viewvc/llvm-project?rev=42557&view=rev Log: Creating llvm-gcc-1200-16 branch Added: llvm-gcc-4.0/tags/llvm-gcc-1200-16/ - copied from r42556, llvm-gcc-4.0/trunk/ From isanbard at gmail.com Tue Oct 2 17:26:53 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 22:26:53 -0000 Subject: [llvm-commits] [llvm] r42558 - /llvm/tags/Apple/llvm-1200-16/ Message-ID: <200710022226.l92MQrhe011374@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 17:26:53 2007 New Revision: 42558 URL: http://llvm.org/viewvc/llvm-project?rev=42558&view=rev Log: Creating llvm-1200-16 branch Added: llvm/tags/Apple/llvm-1200-16/ - copied from r42557, llvm/trunk/ From isanbard at gmail.com Tue Oct 2 18:56:53 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 23:56:53 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42560 - /llvm-gcc-4.0/trunk/build_gcc Message-ID: <200710022356.l92Nur8Q015907@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 18:56:53 2007 New Revision: 42560 URL: http://llvm.org/viewvc/llvm-project?rev=42560&view=rev Log: Port of rdar://5505456 to LLVM. Modified: llvm-gcc-4.0/trunk/build_gcc Modified: llvm-gcc-4.0/trunk/build_gcc URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/build_gcc?rev=42560&r1=42559&r2=42560&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/build_gcc (original) +++ llvm-gcc-4.0/trunk/build_gcc Tue Oct 2 18:56:53 2007 @@ -261,8 +261,10 @@ --host=$BUILD-apple-darwin$DARWIN_VERS --target=$BUILD-apple-darwin$DARWIN_VERS || exit 1 fi # Unset RC_DEBUG_OPTIONS because it causes the bootstrap to fail. -RC_DEBUG_OPTIONS= \ - make $MAKEFLAGS $BOOTSTRAP CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 +# Also keep unset for cross compilers so that the cross built libraries are +# comparable to the native built libraries. +unset RC_DEBUG_OPTIONS +make $MAKEFLAGS CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 make $MAKEFLAGS html CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 make $MAKEFLAGS DESTDIR=$DIR/dst-$BUILD-$BUILD install-gcc install-target \ CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 From isanbard at gmail.com Tue Oct 2 18:57:16 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 02 Oct 2007 23:57:16 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42561 - /llvm-gcc-4.2/trunk/build_gcc Message-ID: <200710022357.l92NvGWf015939@zion.cs.uiuc.edu> Author: void Date: Tue Oct 2 18:57:16 2007 New Revision: 42561 URL: http://llvm.org/viewvc/llvm-project?rev=42561&view=rev Log: Backport of radr://5505456 to LLVM Modified: llvm-gcc-4.2/trunk/build_gcc Modified: llvm-gcc-4.2/trunk/build_gcc URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/build_gcc?rev=42561&r1=42560&r2=42561&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/build_gcc (original) +++ llvm-gcc-4.2/trunk/build_gcc Tue Oct 2 18:57:16 2007 @@ -253,8 +253,10 @@ --host=$BUILD-apple-darwin$DARWIN_VERS --target=$BUILD-apple-darwin$DARWIN_VERS || exit 1 fi # Unset RC_DEBUG_OPTIONS because it causes the bootstrap to fail. -RC_DEBUG_OPTIONS= \ - make $MAKEFLAGS CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 +# Also keep unset for cross compilers so that the cross built libraries are +# comparable to the native built libraries. +unset RC_DEBUG_OPTIONS +make $MAKEFLAGS CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 make $MAKEFLAGS html CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 make $MAKEFLAGS DESTDIR=$DIR/dst-$BUILD-$BUILD install-gcc install-target \ CFLAGS="$CFLAGS" CXXFLAGS="$CFLAGS" || exit 1 From resistor at mac.com Tue Oct 2 22:20:17 2007 From: resistor at mac.com (Owen Anderson) Date: Wed, 03 Oct 2007 03:20:17 -0000 Subject: [llvm-commits] [llvm] r42563 - in /llvm/trunk: include/llvm/Analysis/PostDominators.h lib/Analysis/PostDominatorCalculation.h lib/Analysis/PostDominators.cpp lib/VMCore/DominatorCalculation.h lib/VMCore/Dominators.cpp Message-ID: <200710030320.l933KHBE027793@zion.cs.uiuc.edu> Author: resistor Date: Tue Oct 2 22:20:17 2007 New Revision: 42563 URL: http://llvm.org/viewvc/llvm-project?rev=42563&view=rev Log: Factor some code from the DomTree and PostDomTree calculate methods up into each one's runOnFunction method. Modified: llvm/trunk/include/llvm/Analysis/PostDominators.h llvm/trunk/lib/Analysis/PostDominatorCalculation.h llvm/trunk/lib/Analysis/PostDominators.cpp llvm/trunk/lib/VMCore/DominatorCalculation.h llvm/trunk/lib/VMCore/Dominators.cpp Modified: llvm/trunk/include/llvm/Analysis/PostDominators.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/PostDominators.h?rev=42563&r1=42562&r2=42563&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/PostDominators.h (original) +++ llvm/trunk/include/llvm/Analysis/PostDominators.h Tue Oct 2 22:20:17 2007 @@ -27,11 +27,7 @@ PostDominatorTree() : DominatorTreeBase((intptr_t)&ID, true) {} - virtual bool runOnFunction(Function &F) { - reset(); // Reset from the last time we were run... - PDTcalculate(*this, F); - return false; - } + virtual bool runOnFunction(Function &F); virtual void getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesAll(); Modified: llvm/trunk/lib/Analysis/PostDominatorCalculation.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/PostDominatorCalculation.h?rev=42563&r1=42562&r2=42563&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/PostDominatorCalculation.h (original) +++ llvm/trunk/lib/Analysis/PostDominatorCalculation.h Tue Oct 2 22:20:17 2007 @@ -19,24 +19,6 @@ namespace llvm { void PDTcalculate(PostDominatorTree& PDT, Function &F) { - // Step #0: Scan the function looking for the root nodes of the post-dominance - // relationships. These blocks, which have no successors, end with return and - // unwind instructions. - for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) { - TerminatorInst *Insn = I->getTerminator(); - if (Insn->getNumSuccessors() == 0) { - // Unreachable block is not a root node. - if (!isa(Insn)) - PDT.Roots.push_back(I); - } - - // Prepopulate maps so that we don't get iterator invalidation issues later. - PDT.IDoms[I] = 0; - PDT.DomTreeNodes[I] = 0; - } - - PDT.Vertex.push_back(0); - // Step #1: Number blocks in depth-first order and initialize variables used // in later stages of the algorithm. unsigned N = 0; Modified: llvm/trunk/lib/Analysis/PostDominators.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/PostDominators.cpp?rev=42563&r1=42562&r2=42563&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/PostDominators.cpp (original) +++ llvm/trunk/lib/Analysis/PostDominators.cpp Tue Oct 2 22:20:17 2007 @@ -28,6 +28,29 @@ static RegisterPass F("postdomtree", "Post-Dominator Tree Construction", true); +bool PostDominatorTree::runOnFunction(Function &F) { + reset(); // Reset from the last time we were run... + + // Initialize the roots list + for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) { + TerminatorInst *Insn = I->getTerminator(); + if (Insn->getNumSuccessors() == 0) { + // Unreachable block is not a root node. + if (!isa(Insn)) + Roots.push_back(I); + } + + // Prepopulate maps so that we don't get iterator invalidation issues later. + IDoms[I] = 0; + DomTreeNodes[I] = 0; + } + + Vertex.push_back(0); + + PDTcalculate(*this, F); + return false; +} + //===----------------------------------------------------------------------===// // PostDominanceFrontier Implementation //===----------------------------------------------------------------------===// Modified: llvm/trunk/lib/VMCore/DominatorCalculation.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/DominatorCalculation.h?rev=42563&r1=42562&r2=42563&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/DominatorCalculation.h (original) +++ llvm/trunk/lib/VMCore/DominatorCalculation.h Tue Oct 2 22:20:17 2007 @@ -40,8 +40,6 @@ // Add a node for the root... DT.DomTreeNodes[Root] = DT.RootNode = new DomTreeNode(Root, 0); - DT.Vertex.push_back(0); - // Step #1: Number blocks in depth-first order and initialize variables used // in later stages of the algorithm. unsigned N = DFSPass >(DT, Root, 0); Modified: llvm/trunk/lib/VMCore/Dominators.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Dominators.cpp?rev=42563&r1=42562&r2=42563&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Dominators.cpp (original) +++ llvm/trunk/lib/VMCore/Dominators.cpp Tue Oct 2 22:20:17 2007 @@ -350,7 +350,13 @@ bool DominatorTree::runOnFunction(Function &F) { reset(); // Reset from the last time we were run... + + // Initialize roots Roots.push_back(&F.getEntryBlock()); + IDoms[&F.getEntryBlock()] = 0; + DomTreeNodes[&F.getEntryBlock()] = 0; + Vertex.push_back(0); + DTcalculate(*this, F); return false; } From clattner at apple.com Tue Oct 2 22:29:21 2007 From: clattner at apple.com (Chris Lattner) Date: Tue, 2 Oct 2007 20:29:21 -0700 Subject: [llvm-commits] [llvm] r42546 - /llvm/trunk/lib/Target/X86/README.txt In-Reply-To: <200710022042.l92Kgx2O005184@zion.cs.uiuc.edu> References: <200710022042.l92Kgx2O005184@zion.cs.uiuc.edu> Message-ID: <9BFDF489-2B34-498D-8FAE-8790C4EEC1D1@apple.com> > URL: http://llvm.org/viewvc/llvm-project?rev=42546&view=rev > Log: > Another missed optimization. Hi Bill, Thank you for tracking these down. However, without .ll files or .c files to reproduce this, they are somewhat useless. In the example below, if I wanted to work on this, how would I demonstrate the issue? -Chris > Modified: > llvm/trunk/lib/Target/X86/README.txt > > Modified: llvm/trunk/lib/Target/X86/README.txt > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ > README.txt?rev=42546&r1=42545&r2=42546&view=diff > > ====================================================================== > ======== > --- llvm/trunk/lib/Target/X86/README.txt (original) > +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 15:42:59 2007 > @@ -1239,3 +1239,31 @@ > > // > ===------------------------------------------------------------------- > --===// > > +The following code: > + > +LBB3_5: # bb114.preheader > + movswl -68(%ebp), %eax > + movl $32, %ecx > + movl %ecx, -80(%ebp) > + subl %eax, -80(%ebp) > + movswl -52(%ebp), %eax > + movl %ecx, -84(%ebp) > + subl %eax, -84(%ebp) > + movswl -70(%ebp), %eax > + movl %ecx, -88(%ebp) > + subl %eax, -88(%ebp) > + movswl -50(%ebp), %eax > + subl %eax, %ecx > + movl %ecx, -76(%ebp) > + movswl -42(%ebp), %eax > + movl %eax, -92(%ebp) > + movswl -66(%ebp), %eax > + movl %eax, -96(%ebp) > + movw $0, -98(%ebp) > + > +has redundant subtractions of %eax from a stack slot. However, % > ecx doesn't > +change, so we could simply subtract %eax from %ecx first and then > use %ecx (or > +vice-versa). > + > +// > ===------------------------------------------------------------------- > --===// > + > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From clattner at apple.com Tue Oct 2 22:30:23 2007 From: clattner at apple.com (Chris Lattner) Date: Tue, 2 Oct 2007 20:30:23 -0700 Subject: [llvm-commits] [llvm] r42546 - /llvm/trunk/lib/Target/X86/README.txt In-Reply-To: <9BFDF489-2B34-498D-8FAE-8790C4EEC1D1@apple.com> References: <200710022042.l92Kgx2O005184@zion.cs.uiuc.edu> <9BFDF489-2B34-498D-8FAE-8790C4EEC1D1@apple.com> Message-ID: <5D2023C7-8B27-47B8-B541-22D49258D49D@apple.com> On Oct 2, 2007, at 8:29 PM, Chris Lattner wrote: >> URL: http://llvm.org/viewvc/llvm-project?rev=42546&view=rev >> Log: >> Another missed optimization. > > Hi Bill, > > Thank you for tracking these down. However, without .ll files or .c > files to reproduce this, they are somewhat useless. In the example > below, if I wanted to work on this, how would I demonstrate the issue? Ok, nevermind, you fixed this later, thanks! -Chris > -Chris > >> Modified: >> llvm/trunk/lib/Target/X86/README.txt >> >> Modified: llvm/trunk/lib/Target/X86/README.txt >> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ >> README.txt?rev=42546&r1=42545&r2=42546&view=diff >> >> ===================================================================== >> = >> ======== >> --- llvm/trunk/lib/Target/X86/README.txt (original) >> +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 15:42:59 2007 >> @@ -1239,3 +1239,31 @@ >> >> // >> ===------------------------------------------------------------------ >> - >> --===// >> >> +The following code: >> + >> +LBB3_5: # bb114.preheader >> + movswl -68(%ebp), %eax >> + movl $32, %ecx >> + movl %ecx, -80(%ebp) >> + subl %eax, -80(%ebp) >> + movswl -52(%ebp), %eax >> + movl %ecx, -84(%ebp) >> + subl %eax, -84(%ebp) >> + movswl -70(%ebp), %eax >> + movl %ecx, -88(%ebp) >> + subl %eax, -88(%ebp) >> + movswl -50(%ebp), %eax >> + subl %eax, %ecx >> + movl %ecx, -76(%ebp) >> + movswl -42(%ebp), %eax >> + movl %eax, -92(%ebp) >> + movswl -66(%ebp), %eax >> + movl %eax, -96(%ebp) >> + movw $0, -98(%ebp) >> + >> +has redundant subtractions of %eax from a stack slot. However, % >> ecx doesn't >> +change, so we could simply subtract %eax from %ecx first and then >> use %ecx (or >> +vice-versa). >> + >> +// >> ===------------------------------------------------------------------ >> - >> --===// >> + >> >> >> _______________________________________________ >> llvm-commits mailing list >> llvm-commits at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From clattner at apple.com Tue Oct 2 22:33:16 2007 From: clattner at apple.com (Chris Lattner) Date: Tue, 2 Oct 2007 20:33:16 -0700 Subject: [llvm-commits] [llvm] r42537 - in /llvm/trunk: include/llvm/Intrinsics.td lib/Analysis/ConstantFolding.cpp lib/CodeGen/IntrinsicLowering.cpp lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp lib/Target/CBackend/CBackend.cpp In-Reply-To: <200710021744.l92Hi0IJ029003@zion.cs.uiuc.edu> References: <200710021744.l92Hi0IJ029003@zion.cs.uiuc.edu> Message-ID: On Oct 2, 2007, at 10:44 AM, Dale Johannesen wrote: > Author: johannes > Date: Tue Oct 2 12:43:59 2007 > New Revision: 42537 > > URL: http://llvm.org/viewvc/llvm-project?rev=42537&view=rev > Log: > Rewrite sqrt and powi to use anyfloat. By popular demand. Nice shrinkification Dale! -Chris From clattner at apple.com Tue Oct 2 22:35:07 2007 From: clattner at apple.com (Chris Lattner) Date: Tue, 2 Oct 2007 20:35:07 -0700 Subject: [llvm-commits] [llvm] r42554 - /llvm/trunk/lib/Target/X86/README.txt In-Reply-To: <200710022149.l92LnVFU009573@zion.cs.uiuc.edu> References: <200710022149.l92LnVFU009573@zion.cs.uiuc.edu> Message-ID: Hi Bill, This is not enough context to reproduce this issue. Can you please include more in the .ll file? Specifically, the movswl wouldn't exist if "tmp654" weren't live out I hope. It would be best if the .ll fragments were actually compilable. -Chris On Oct 2, 2007, at 2:49 PM, Bill Wendling wrote: > Author: void > Date: Tue Oct 2 16:49:31 2007 > New Revision: 42554 > > URL: http://llvm.org/viewvc/llvm-project?rev=42554&view=rev > Log: > Another micro-opt. > > Modified: > llvm/trunk/lib/Target/X86/README.txt > > Modified: llvm/trunk/lib/Target/X86/README.txt > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ > README.txt?rev=42554&r1=42553&r2=42554&view=diff > > ====================================================================== > ======== > --- llvm/trunk/lib/Target/X86/README.txt (original) > +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 16:49:31 2007 > @@ -1306,3 +1306,19 @@ > movl 88(%esp), %eax > #481.12 > > // > ===------------------------------------------------------------------- > --===// > + > +This code: > + > + %tmp659 = icmp slt i16 %tmp654, 0 ; [#uses=1] > + br i1 %tmp659, label %cond_true662, label %cond_next715 > + > +produces this: > + > + testw %cx, %cx > + movswl %cx, %esi > + jns LBB4_109 # cond_next715 > + > +Shark tells us that using %cx in the testw instruction is sub- > optimal. It > +suggests using the 32-bit register (which is what ICC uses). > + > +// > ===------------------------------------------------------------------- > --===// > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From sabre at nondot.org Tue Oct 2 22:40:24 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 03:40:24 -0000 Subject: [llvm-commits] [llvm] r42564 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710030340.l933eOZe028977@zion.cs.uiuc.edu> Author: lattner Date: Tue Oct 2 22:40:24 2007 New Revision: 42564 URL: http://llvm.org/viewvc/llvm-project?rev=42564&view=rev Log: Bill's example is still not enough to repro this, but it has other issues that seem significant as well. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42564&r1=42563&r2=42564&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 22:40:24 2007 @@ -1277,6 +1277,14 @@ movl %eax, -96(%ebp) movw $0, -98(%ebp) +This appears to be bad because the RA is not folding the store to the stack +slot into the movl. The above instructions could be: + movl $32, -80(%ebp) +... + movl $32, -84(%ebp) +... +This seems like a cross between remat and spill folding. + This has redundant subtractions of %eax from a stack slot. However, %ecx doesn't change, so we could simply subtract %eax from %ecx first and then use %ecx (or vice-versa). From clattner at apple.com Tue Oct 2 22:49:04 2007 From: clattner at apple.com (Chris Lattner) Date: Tue, 2 Oct 2007 20:49:04 -0700 Subject: [llvm-commits] [llvm] r42519 - /llvm/trunk/docs/AliasAnalysis.html In-Reply-To: <200710020043.l920hQkk004003@zion.cs.uiuc.edu> References: <200710020043.l920hQkk004003@zion.cs.uiuc.edu> Message-ID: Hi Owen, Can you please expound a bit on what sort of clients should use memdep and why? The goal of this document is so that people can understand how the pieces fit together and have an idea of what to use when they have a particular problem to solve, thanks! -Chris On Oct 1, 2007, at 5:43 PM, Owen Anderson wrote: > Author: resistor > Date: Mon Oct 1 19:43:25 2007 > New Revision: 42519 > > URL: http://llvm.org/viewvc/llvm-project?rev=42519&view=rev > Log: > Mention memdep in the AA documentation. > > Modified: > llvm/trunk/docs/AliasAnalysis.html > > Modified: llvm/trunk/docs/AliasAnalysis.html > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/ > AliasAnalysis.html?rev=42519&r1=42518&r2=42519&view=diff > > ====================================================================== > ======== > --- llvm/trunk/docs/AliasAnalysis.html (original) > +++ llvm/trunk/docs/AliasAnalysis.html Mon Oct 1 19:43:25 2007 > @@ -50,6 +50,7 @@ > implementations > > > +
  • Memory Dependence Analysis
    • > > >
    > @@ -948,6 +949,24 @@ >
    > > > +
    > + Memory Dependence Analysis > +
    > + > + > +
    > + > +

    If you're just looking to be a client of alias analysis > information, consider > +using the Memory Dependence Analysis interface instead. MemDep is > a lazy, > +caching layer on top of alias analysis that is able to answer the > question of > +what preceding memory operations a given instruction depends on, > either at an > +intra- or inter-block level. Because of its laziness and caching > +policy, using MemDep can be a significant performance win over > accessing alias > +analysis directly.

    > + > +
    > + > + > >
    >
    > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From sabre at nondot.org Tue Oct 2 22:59:16 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 03:59:16 -0000 Subject: [llvm-commits] [llvm] r42565 - /llvm/trunk/lib/Transforms/IPO/Internalize.cpp Message-ID: <200710030359.l933xGRW029911@zion.cs.uiuc.edu> Author: lattner Date: Tue Oct 2 22:59:15 2007 New Revision: 42565 URL: http://llvm.org/viewvc/llvm-project?rev=42565&view=rev Log: Fix PR1719, by not marking llvm.noinline internal. Modified: llvm/trunk/lib/Transforms/IPO/Internalize.cpp Modified: llvm/trunk/lib/Transforms/IPO/Internalize.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/IPO/Internalize.cpp?rev=42565&r1=42564&r2=42565&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/IPO/Internalize.cpp (original) +++ llvm/trunk/lib/Transforms/IPO/Internalize.cpp Tue Oct 2 22:59:15 2007 @@ -130,6 +130,7 @@ ExternalNames.insert("llvm.dbg.subprograms"); ExternalNames.insert("llvm.global_ctors"); ExternalNames.insert("llvm.global_dtors"); + ExternalNames.insert("llvm.noinline"); // Mark all global variables with initializers as internal as well. for (Module::global_iterator I = M.global_begin(), E = M.global_end(); From resistor at mac.com Tue Oct 2 23:33:13 2007 From: resistor at mac.com (Owen Anderson) Date: Wed, 03 Oct 2007 04:33:13 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42566 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Message-ID: <200710030433.l934XDap031554@zion.cs.uiuc.edu> Author: resistor Date: Tue Oct 2 23:33:12 2007 New Revision: 42566 URL: http://llvm.org/viewvc/llvm-project?rev=42566&view=rev Log: Revert Bill's last patch to this file. It was causing a failure building Qt for Darwin. Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/objc/objc-act.c?rev=42566&r1=42565&r2=42566&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Tue Oct 2 23:33:12 2007 @@ -2566,8 +2566,7 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2577,8 +2576,7 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2588,8 +2586,7 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2670,8 +2667,7 @@ /* struct _prop_list_t *properties; */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -4978,8 +4974,7 @@ /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ if (flag_next_runtime || ! sel_ref_chain) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, @@ -9239,8 +9234,7 @@ (objc_protocol_template)), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - /* APPLE LOCAL LLVM - NUL pointer */ - refs_expr = null_pointer_node; + refs_expr = build_int_cst (NULL_TREE, 0); /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9310,8 +9304,7 @@ if (newabi) { /* 'isa' is NULL in the new ObjC abi */ - /* APPLE LOCAL LLVM - NUL pointer */ - expr = null_pointer_node; + expr = build_int_cst (NULL_TREE, 0); } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9319,8 +9312,7 @@ else { if (!objc_protocol_or_opt_ins_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - expr = null_pointer_node; + expr = build_int_cst (NULL_TREE, 0); else expr = convert (build_pointer_type (objc_protocol_extension_template), build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); @@ -9334,8 +9326,7 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -9346,8 +9337,7 @@ } if (!class_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -9362,8 +9352,7 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9371,8 +9360,7 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!opt_cls_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9381,8 +9369,7 @@ } /* APPLE LOCAL end radar 4695109 */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -10402,8 +10389,7 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - /* APPLE LOCAL LLVM - NUL pointer */ - ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); + ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); /* Set type */ encode_field_decl (field_decl, @@ -10460,8 +10446,7 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - /* APPLE LOCAL LLVM - NUL pointer */ - ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); + ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); /* Set type. */ encode_field_decl (field_decl, @@ -11157,8 +11142,7 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -11168,8 +11152,7 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -11181,8 +11164,7 @@ /* protocol_list = */ if (!protocol_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ @@ -11204,8 +11186,7 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -11272,8 +11253,7 @@ /* objc_ivar_list = */ if (!ivar_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -11285,8 +11265,7 @@ /* objc_method_list = */ if (!dispatch_table) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin mainline */ @@ -11298,27 +11277,22 @@ if (flag_next_runtime) /* method_cache = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* dtable = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* subclass_list = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* sibling_class = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); } /* protocol_list = */ if (! protocol_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (build_pointer_type @@ -11333,21 +11307,18 @@ { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ if (!objc_class_ext) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (build_pointer_type (objc_class_ext_template), @@ -11357,8 +11328,7 @@ } else /* gc_object_type = NULL */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11520,15 +11490,13 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11572,8 +11540,7 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11583,8 +11550,7 @@ /* baseMethods */ if (!baseMethods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, @@ -11594,34 +11560,31 @@ /* baseProtocols */ if (!baseProtocols) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { tree protocol_list_t_p = build_pointer_type ( xref_tag (RECORD_TYPE, get_identifier (UTAG_V2_PROTOCOL_LIST))); expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); + build_unary_op (ADDR_EXPR, baseProtocols, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* ivars */ if (!ivars) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11629,12 +11592,11 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11892,8 +11854,7 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - /* APPLE LOCAL LLVM - NUL pointer */ - class_superclass_expr = null_pointer_node; + class_superclass_expr = build_int_cst (NULL_TREE, 0); flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -12082,8 +12043,7 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - /* APPLE LOCAL LLVM - NUL pointer */ - super_expr = null_pointer_node; + super_expr = build_int_cst (NULL_TREE, 0); root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13343,8 +13303,7 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - /* APPLE LOCAL LLVM - NUL pointer */ - refs_expr = null_pointer_node; + refs_expr = build_int_cst (NULL_TREE, 0); /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From resistor at mac.com Tue Oct 2 23:49:04 2007 From: resistor at mac.com (Owen Anderson) Date: Wed, 03 Oct 2007 04:49:04 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42567 - /llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Message-ID: <200710030449.l934n4UX032323@zion.cs.uiuc.edu> Author: resistor Date: Tue Oct 2 23:49:04 2007 New Revision: 42567 URL: http://llvm.org/viewvc/llvm-project?rev=42567&view=rev Log: Revert Bill's last patch to this file. It was causing a failure building Qt for Darwin. Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c?rev=42567&r1=42566&r2=42567&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Tue Oct 2 23:49:04 2007 @@ -2576,8 +2576,7 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2587,8 +2586,7 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2598,8 +2596,7 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2680,8 +2677,7 @@ /* struct _prop_list_t *properties; */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -4875,8 +4871,7 @@ /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ if (flag_next_runtime || ! sel_ref_chain) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, @@ -9139,8 +9134,7 @@ (objc_protocol_template)), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - /* APPLE LOCAL LLVM - NUL pointer */ - refs_expr = null_pointer_node; + refs_expr = build_int_cst (NULL_TREE, 0); /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9210,8 +9204,7 @@ if (newabi) { /* 'isa' is NULL in the new ObjC abi */ - /* APPLE LOCAL LLVM - NUL pointer */ - expr = null_pointer_node; + expr = build_int_cst (NULL_TREE, 0); } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9219,8 +9212,7 @@ else { if (!objc_protocol_or_opt_ins_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - expr = null_pointer_node; + expr = build_int_cst (NULL_TREE, 0); else expr = convert (build_pointer_type (objc_protocol_extension_template), build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); @@ -9234,8 +9226,7 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9244,8 +9235,7 @@ } if (!class_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9258,8 +9248,7 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9267,8 +9256,7 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!opt_cls_meth) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_proto_list_ptr, @@ -9277,8 +9265,7 @@ } /* APPLE LOCAL end radar 4695109 */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -10304,8 +10291,7 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - /* APPLE LOCAL LLVM - NUL pointer */ - ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); + ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); /* Set type */ encode_field_decl (field_decl, @@ -10362,8 +10348,7 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - /* APPLE LOCAL LLVM - NUL pointer */ - ivar = tree_cons (NULL_TREE, null_pointer_node, ivar); + ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); /* Set type. */ encode_field_decl (field_decl, @@ -11057,8 +11042,7 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, @@ -11066,8 +11050,7 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, @@ -11077,8 +11060,7 @@ /* protocol_list = */ if (!protocol_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ @@ -11100,8 +11082,7 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, @@ -11168,8 +11149,7 @@ /* objc_ivar_list = */ if (!ivar_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_ivar_list_ptr, @@ -11179,8 +11159,7 @@ /* objc_method_list = */ if (!dispatch_table) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, @@ -11190,27 +11169,22 @@ if (flag_next_runtime) /* method_cache = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { /* dtable = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* subclass_list = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* sibling_class = */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); } /* protocol_list = */ if (! protocol_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (build_pointer_type @@ -11225,21 +11199,18 @@ { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ if (!objc_class_ext) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (build_pointer_type (objc_class_ext_template), @@ -11249,8 +11220,7 @@ } else /* gc_object_type = NULL */ - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11412,15 +11382,13 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11464,8 +11432,7 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11475,8 +11442,7 @@ /* baseMethods */ if (!baseMethods) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_method_list_ptr, @@ -11486,34 +11452,31 @@ /* baseProtocols */ if (!baseProtocols) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { tree protocol_list_t_p = build_pointer_type ( xref_tag (RECORD_TYPE, get_identifier (UTAG_V2_PROTOCOL_LIST))); expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); + build_unary_op (ADDR_EXPR, baseProtocols, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* ivars */ if (!ivars) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11521,12 +11484,11 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - /* APPLE LOCAL LLVM - NUL pointer */ - initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); + initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11773,8 +11735,7 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - /* APPLE LOCAL LLVM - NUL pointer */ - class_superclass_expr = null_pointer_node; + class_superclass_expr = build_int_cst (NULL_TREE, 0); flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -11966,8 +11927,7 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - /* APPLE LOCAL LLVM - NUL pointer */ - super_expr = null_pointer_node; + super_expr = build_int_cst (NULL_TREE, 0); root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13230,8 +13190,7 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - /* APPLE LOCAL LLVM - NUL pointer */ - refs_expr = null_pointer_node; + refs_expr = build_int_cst (NULL_TREE, 0); /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From isanbard at gmail.com Tue Oct 2 23:54:18 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 2 Oct 2007 21:54:18 -0700 Subject: [llvm-commits] [llvm-gcc-4.0] r42566 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c In-Reply-To: <200710030433.l934XDap031554@zion.cs.uiuc.edu> References: <200710030433.l934XDap031554@zion.cs.uiuc.edu> Message-ID: On Oct 2, 2007, at 9:33 PM, Owen Anderson wrote: > Author: resistor > Date: Tue Oct 2 23:33:12 2007 > New Revision: 42566 > > URL: http://llvm.org/viewvc/llvm-project?rev=42566&view=rev > Log: > Revert Bill's last patch to this file. It was causing a failure > building > Qt for Darwin. > Hi Owen, Could you file a radar or PR with a testcase? Thanks! -bw From isanbard at gmail.com Tue Oct 2 22:57:06 2007 From: isanbard at gmail.com (Bill Wendling) Date: Tue, 2 Oct 2007 20:57:06 -0700 Subject: [llvm-commits] [llvm] r42554 - /llvm/trunk/lib/Target/X86/README.txt In-Reply-To: References: <200710022149.l92LnVFU009573@zion.cs.uiuc.edu> Message-ID: Sure. I can try to whittle down the testcase that I'm working on. -bw On Oct 2, 2007, at 8:35 PM, Chris Lattner wrote: > Hi Bill, > > This is not enough context to reproduce this issue. Can you please > include more in the .ll file? Specifically, the movswl wouldn't > exist if "tmp654" weren't live out I hope. It would be best if > the .ll fragments were actually compilable. > > -Chris > > On Oct 2, 2007, at 2:49 PM, Bill Wendling wrote: > >> Author: void >> Date: Tue Oct 2 16:49:31 2007 >> New Revision: 42554 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=42554&view=rev >> Log: >> Another micro-opt. >> >> Modified: >> llvm/trunk/lib/Target/X86/README.txt >> >> Modified: llvm/trunk/lib/Target/X86/README.txt >> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ >> README.txt?rev=42554&r1=42553&r2=42554&view=diff >> >> ===================================================================== >> = >> ======== >> --- llvm/trunk/lib/Target/X86/README.txt (original) >> +++ llvm/trunk/lib/Target/X86/README.txt Tue Oct 2 16:49:31 2007 >> @@ -1306,3 +1306,19 @@ >> movl 88(%esp), %eax >> #481.12 >> >> // >> ===------------------------------------------------------------------ >> - >> --===// >> + >> +This code: >> + >> + %tmp659 = icmp slt i16 %tmp654, 0 ; [#uses=1] >> + br i1 %tmp659, label %cond_true662, label %cond_next715 >> + >> +produces this: >> + >> + testw %cx, %cx >> + movswl %cx, %esi >> + jns LBB4_109 # cond_next715 >> + >> +Shark tells us that using %cx in the testw instruction is sub- >> optimal. It >> +suggests using the 32-bit register (which is what ICC uses). >> + >> +// >> ===------------------------------------------------------------------ >> - >> --===// >> >> >> _______________________________________________ >> llvm-commits mailing list >> llvm-commits at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From isanbard at gmail.com Wed Oct 3 00:03:34 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 03 Oct 2007 05:03:34 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42568 - /llvm-gcc-4.0/tags/llvm-gcc-1200-17/ Message-ID: <200710030503.l9353YZS000826@zion.cs.uiuc.edu> Author: void Date: Wed Oct 3 00:03:34 2007 New Revision: 42568 URL: http://llvm.org/viewvc/llvm-project?rev=42568&view=rev Log: Creating llvm-gcc-1200-17 branch Added: llvm-gcc-4.0/tags/llvm-gcc-1200-17/ - copied from r42567, llvm-gcc-4.0/trunk/ From isanbard at gmail.com Wed Oct 3 00:03:40 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 03 Oct 2007 05:03:40 -0000 Subject: [llvm-commits] [llvm] r42569 - /llvm/tags/Apple/llvm-1200-17/ Message-ID: <200710030503.l9353eRh000841@zion.cs.uiuc.edu> Author: void Date: Wed Oct 3 00:03:39 2007 New Revision: 42569 URL: http://llvm.org/viewvc/llvm-project?rev=42569&view=rev Log: Creating llvm-1200-17 branch Added: llvm/tags/Apple/llvm-1200-17/ - copied from r42568, llvm/trunk/ From sabre at nondot.org Wed Oct 3 00:29:07 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 05:29:07 -0000 Subject: [llvm-commits] [llvm] r42570 - /llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c Message-ID: <200710030529.l935T7qZ001896@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 00:29:07 2007 New Revision: 42570 URL: http://llvm.org/viewvc/llvm-project?rev=42570&view=rev Log: testcase for PR1647 Added: llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c Added: llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c?rev=42570&view=auto ============================================================================== --- llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c (added) +++ llvm/trunk/test/CFrontend/2007-10-02-VolatileArray.c Wed Oct 3 00:29:07 2007 @@ -0,0 +1,7 @@ +// RUN: %llvmgcc -S %s -o - | grep volatile +// PR1647 + +void foo(volatile int *p) +{ +p[0] = 0; +} From sabre at nondot.org Wed Oct 3 00:31:52 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 05:31:52 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42571 - /llvm-gcc-4.0/trunk/gcc/c-typeck.c Message-ID: <200710030531.l935VqhF001981@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 00:31:52 2007 New Revision: 42571 URL: http://llvm.org/viewvc/llvm-project?rev=42571&view=rev Log: Fix PR1647 and test/CFrontend/2007-10-02-VolatileArray.c Modified: llvm-gcc-4.0/trunk/gcc/c-typeck.c Modified: llvm-gcc-4.0/trunk/gcc/c-typeck.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/c-typeck.c?rev=42571&r1=42570&r2=42571&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/c-typeck.c (original) +++ llvm-gcc-4.0/trunk/gcc/c-typeck.c Wed Oct 3 00:31:52 2007 @@ -1901,7 +1901,11 @@ if (TREE_CODE(ty) == RECORD_TYPE || TREE_CODE(ty) == UNION_TYPE || TREE_CODE(ty) == QUAL_UNION_TYPE) ty = TYPE_MAIN_VARIANT (ty); - return build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); + + ar = build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); + /* mirror logic from build_indirect_ref to set TREE_THIS_VOLATILE. */ + TREE_THIS_VOLATILE(ar) = TYPE_VOLATILE(TREE_TYPE(ar)); + return ar; } #endif /* APPLE LOCAL end LLVM */ From sabre at nondot.org Wed Oct 3 00:33:03 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 05:33:03 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42572 - /llvm-gcc-4.2/trunk/gcc/c-typeck.c Message-ID: <200710030533.l935X3hv002031@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 00:33:02 2007 New Revision: 42572 URL: http://llvm.org/viewvc/llvm-project?rev=42572&view=rev Log: Fix PR1647 and test/CFrontend/2007-10-02-VolatileArray.c Modified: llvm-gcc-4.2/trunk/gcc/c-typeck.c Modified: llvm-gcc-4.2/trunk/gcc/c-typeck.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/c-typeck.c?rev=42572&r1=42571&r2=42572&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/c-typeck.c (original) +++ llvm-gcc-4.2/trunk/gcc/c-typeck.c Wed Oct 3 00:33:02 2007 @@ -2085,7 +2085,11 @@ if (TREE_CODE(ty) == RECORD_TYPE || TREE_CODE(ty) == UNION_TYPE || TREE_CODE(ty) == QUAL_UNION_TYPE) ty = TYPE_MAIN_VARIANT (ty); - return build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); + + ar = build4 (ARRAY_REF, ty, ar, index, NULL_TREE, NULL_TREE); + /* mirror logic from build_indirect_ref to set TREE_THIS_VOLATILE. */ + TREE_THIS_VOLATILE(ar) = TYPE_VOLATILE(TREE_TYPE(ar)); + return ar; } #endif /* LLVM LOCAL end */ From sabre at nondot.org Wed Oct 3 01:10:59 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 06:10:59 -0000 Subject: [llvm-commits] [llvm] r42573 - /llvm/trunk/lib/Target/README.txt Message-ID: <200710030610.l936AxrD003783@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 01:10:59 2007 New Revision: 42573 URL: http://llvm.org/viewvc/llvm-project?rev=42573&view=rev Log: add a note Modified: llvm/trunk/lib/Target/README.txt Modified: llvm/trunk/lib/Target/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/README.txt?rev=42573&r1=42572&r2=42573&view=diff ============================================================================== --- llvm/trunk/lib/Target/README.txt (original) +++ llvm/trunk/lib/Target/README.txt Wed Oct 3 01:10:59 2007 @@ -427,6 +427,22 @@ //===---------------------------------------------------------------------===// +Tail recursion elimination is not transforming this function, because it is +returning n, which fails the isDynamicConstant check in the accumulator +recursion checks. + +long long fib(const long long n) { + switch(n) { + case 0: + case 1: + return n; + default: + return fib(n-1) + fib(n-2); + } +} + +//===---------------------------------------------------------------------===// + Argument promotion should promote arguments for recursive functions, like this: From gordonhenriksen at mac.com Wed Oct 3 07:06:42 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Wed, 03 Oct 2007 12:06:42 -0000 Subject: [llvm-commits] [llvm] r42574 - /llvm/trunk/autoconf/configure.ac Message-ID: <200710031206.l93C6hp1024959@zion.cs.uiuc.edu> Author: gordon Date: Wed Oct 3 07:06:41 2007 New Revision: 42574 URL: http://llvm.org/viewvc/llvm-project?rev=42574&view=rev Log: Fixing a typo in configure.ac. Patch by Adam Goode. Modified: llvm/trunk/autoconf/configure.ac Modified: llvm/trunk/autoconf/configure.ac URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/autoconf/configure.ac?rev=42574&r1=42573&r2=42574&view=diff ============================================================================== --- llvm/trunk/autoconf/configure.ac (original) +++ llvm/trunk/autoconf/configure.ac Wed Oct 3 07:06:41 2007 @@ -874,7 +874,7 @@ [Installation directory for libraries]) AC_DEFINE_UNQUOTED(LLVM_DATADIR, "$LLVM_DATADIR", [Installation directory for data files]) -AC_DEFINE_UNQUOTED(LLVM_DATADIR, "$LLVM_DOCSDIR", +AC_DEFINE_UNQUOTED(LLVM_DOCSDIR, "$LLVM_DOCSDIR", [Installation directory for documentation]) AC_DEFINE_UNQUOTED(LLVM_ETCDIR, "$LLVM_ETCDIR", [Installation directory for config files]) From gordonhenriksen at mac.com Wed Oct 3 07:07:14 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Wed, 03 Oct 2007 12:07:14 -0000 Subject: [llvm-commits] [llvm] r42575 - in /llvm/trunk: configure include/llvm/Config/config.h.in Message-ID: <200710031207.l93C7F6n024987@zion.cs.uiuc.edu> Author: gordon Date: Wed Oct 3 07:07:14 2007 New Revision: 42575 URL: http://llvm.org/viewvc/llvm-project?rev=42575&view=rev Log: Regenerate. Modified: llvm/trunk/configure llvm/trunk/include/llvm/Config/config.h.in Modified: llvm/trunk/configure URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/configure?rev=42575&r1=42574&r2=42575&view=diff ============================================================================== --- llvm/trunk/configure (original) +++ llvm/trunk/configure Wed Oct 3 07:07:14 2007 @@ -34132,7 +34132,7 @@ cat >>confdefs.h <<_ACEOF -#define LLVM_DATADIR "$LLVM_DOCSDIR" +#define LLVM_DOCSDIR "$LLVM_DOCSDIR" _ACEOF Modified: llvm/trunk/include/llvm/Config/config.h.in URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Config/config.h.in?rev=42575&r1=42574&r2=42575&view=diff ============================================================================== --- llvm/trunk/include/llvm/Config/config.h.in (original) +++ llvm/trunk/include/llvm/Config/config.h.in Wed Oct 3 07:07:14 2007 @@ -449,9 +449,12 @@ /* Time at which LLVM was configured */ #undef LLVM_CONFIGTIME -/* Installation directory for documentation */ +/* Installation directory for data files */ #undef LLVM_DATADIR +/* Installation directory for documentation */ +#undef LLVM_DOCSDIR + /* Installation directory for config files */ #undef LLVM_ETCDIR From neil at daikokuya.co.uk Wed Oct 3 10:16:42 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Wed, 03 Oct 2007 15:16:42 -0000 Subject: [llvm-commits] [llvm] r42576 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp Message-ID: <200710031516.l93FGg2Q031663@zion.cs.uiuc.edu> Author: neil Date: Wed Oct 3 10:16:41 2007 New Revision: 42576 URL: http://llvm.org/viewvc/llvm-project?rev=42576&view=rev Log: Tweak RoundAwayFromZero the bit number below which is truncated, and make it const. Preparation for APFloat -> hexadecimal string conversion. Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42576&r1=42575&r2=42576&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Wed Oct 3 10:16:41 2007 @@ -253,7 +253,7 @@ opStatus addOrSubtract(const APFloat &, roundingMode, bool subtract); cmpResult compareAbsoluteValue(const APFloat &) const; opStatus handleOverflow(roundingMode); - bool roundAwayFromZero(roundingMode, lostFraction); + bool roundAwayFromZero(roundingMode, lostFraction, unsigned int) const; opStatus convertFromUnsignedInteger(integerPart *, unsigned int, roundingMode); lostFraction combineLostFractions(lostFraction, lostFraction); Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42576&r1=42575&r2=42576&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Wed Oct 3 10:16:41 2007 @@ -184,7 +184,8 @@ return digitValue == 0 ? lfLessThanHalf: lfMoreThanHalf; } - /* Return the fraction lost were a bignum truncated. */ + /* Return the fraction lost were a bignum truncated losing the least + significant BITS bits. */ lostFraction lostFractionThroughTruncation(integerPart *parts, unsigned int partCount, @@ -694,16 +695,20 @@ return opInexact; } -/* This routine must work for fcZero of both signs, and fcNormal - numbers. */ +/* Returns TRUE if, when truncating the current number, with BIT the + new LSB, with the given lost fraction and rounding mode, the result + would need to be rounded away from zero (i.e., by increasing the + signficand). This routine must work for fcZero of both signs, and + fcNormal numbers. */ bool APFloat::roundAwayFromZero(roundingMode rounding_mode, - lostFraction lost_fraction) + lostFraction lost_fraction, + unsigned int bit) const { /* NaNs and infinities should not have lost fractions. */ assert(category == fcNormal || category == fcZero); - /* Our caller has already handled this case. */ + /* Current callers never pass this so we don't handle it. */ assert(lost_fraction != lfExactlyZero); switch(rounding_mode) { @@ -719,7 +724,7 @@ /* Our zeroes don't have a significand to test. */ if(lost_fraction == lfExactlyHalf && category != fcZero) - return significandParts()[0] & 1; + return APInt::tcExtractBit(significandParts(), bit); return false; @@ -802,7 +807,7 @@ } /* Increment the significand if we're rounding away from zero. */ - if(roundAwayFromZero(rounding_mode, lost_fraction)) { + if(roundAwayFromZero(rounding_mode, lost_fraction, 0)) { if(omsb == 0) exponent = semantics->minExponent; @@ -1451,7 +1456,7 @@ } if(lost_fraction != lfExactlyZero - && tmp.roundAwayFromZero(rounding_mode, lost_fraction)) + && tmp.roundAwayFromZero(rounding_mode, lost_fraction, 0)) tmp.incrementSignificand(); msb = tmp.significandMSB(); From lattner at apple.com Wed Oct 3 11:34:41 2007 From: lattner at apple.com (Tanya Lattner) Date: Wed, 3 Oct 2007 09:34:41 -0700 Subject: [llvm-commits] [llvm-gcc-4.0] r42566 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c In-Reply-To: References: <200710030433.l934XDap031554@zion.cs.uiuc.edu> Message-ID: Reverting this patch causes test/FrontendObjC/2007-09-25-EH.m to fail. It will most likely only fail on Tiger systems (<10.5). Please xfail this test and then re-enable it once this has been fixed. -Tanya On Oct 2, 2007, at 9:54 PM, Bill Wendling wrote: > On Oct 2, 2007, at 9:33 PM, Owen Anderson wrote: > >> Author: resistor >> Date: Tue Oct 2 23:33:12 2007 >> New Revision: 42566 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=42566&view=rev >> Log: >> Revert Bill's last patch to this file. It was causing a failure >> building >> Qt for Darwin. >> > Hi Owen, > > Could you file a radar or PR with a testcase? > > Thanks! > -bw > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20071003/4597be24/attachment.html From tonic at nondot.org Wed Oct 3 12:05:40 2007 From: tonic at nondot.org (Tanya Lattner) Date: Wed, 03 Oct 2007 17:05:40 -0000 Subject: [llvm-commits] [llvm] r42578 - /llvm/trunk/lib/Transforms/IPO/Internalize.cpp Message-ID: <200710031705.l93H5eb7004580@zion.cs.uiuc.edu> Author: tbrethou Date: Wed Oct 3 12:05:40 2007 New Revision: 42578 URL: http://llvm.org/viewvc/llvm-project?rev=42578&view=rev Log: Fix PR1719, by not marking llvm.global.annotations internal. Modified: llvm/trunk/lib/Transforms/IPO/Internalize.cpp Modified: llvm/trunk/lib/Transforms/IPO/Internalize.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/IPO/Internalize.cpp?rev=42578&r1=42577&r2=42578&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/IPO/Internalize.cpp (original) +++ llvm/trunk/lib/Transforms/IPO/Internalize.cpp Wed Oct 3 12:05:40 2007 @@ -131,6 +131,7 @@ ExternalNames.insert("llvm.global_ctors"); ExternalNames.insert("llvm.global_dtors"); ExternalNames.insert("llvm.noinline"); + ExternalNames.insert("llvm.global.annotations"); // Mark all global variables with initializers as internal as well. for (Module::global_iterator I = M.global_begin(), E = M.global_end(); From sabre at nondot.org Wed Oct 3 12:10:03 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 17:10:03 -0000 Subject: [llvm-commits] [llvm] r42579 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710031710.l93HA3L3004804@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 12:10:03 2007 New Revision: 42579 URL: http://llvm.org/viewvc/llvm-project?rev=42579&view=rev Log: add a note Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42579&r1=42578&r2=42579&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Wed Oct 3 12:10:03 2007 @@ -1330,3 +1330,35 @@ suggests using the 32-bit register (which is what ICC uses). //===---------------------------------------------------------------------===// + +rdar://5506677 - We compile this: + +define i32 @foo(double %x) { + %x14 = bitcast double %x to i64 ; [#uses=1] + %tmp713 = trunc i64 %x14 to i32 ; [#uses=1] + %tmp8 = and i32 %tmp713, 2147483647 ; [#uses=1] + ret i32 %tmp8 +} + +to: + +_foo: + subl $12, %esp + fldl 16(%esp) + fstpl (%esp) + movl $2147483647, %eax + andl (%esp), %eax + addl $12, %esp + #FP_REG_KILL + ret + +It would be much better to eliminate the fldl/fstpl by folding the bitcast +into the load SDNode. That would give us: + +_foo: + movl $2147483647, %eax + andl 4(%esp), %eax + ret + +//===---------------------------------------------------------------------===// + From sabre at nondot.org Wed Oct 3 12:33:52 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 17:33:52 -0000 Subject: [llvm-commits] [llvm] r42580 - /llvm/trunk/docs/LangRef.html Message-ID: <200710031733.l93HXq2e005894@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 12:33:51 2007 New Revision: 42580 URL: http://llvm.org/viewvc/llvm-project?rev=42580&view=rev Log: remove misleading documentation: LLVM has no atomic support yet. Modified: llvm/trunk/docs/LangRef.html Modified: llvm/trunk/docs/LangRef.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.html?rev=42580&r1=42579&r2=42580&view=diff ============================================================================== --- llvm/trunk/docs/LangRef.html (original) +++ llvm/trunk/docs/LangRef.html Wed Oct 3 12:33:51 2007 @@ -191,15 +191,6 @@
  • Debugger intrinsics
  • Exception Handling intrinsics
    • -
  • Atomic Operations and Synchronization Intrinsics -
      -
    1. 'llvm.atomic.lcs.*' Intrinsic
      2. -
    2. 'llvm.atomic.ls.*' Intrinsic
      3. -
    3. 'llvm.atomic.las.*' Intrinsic
      4. -
    4. 'llvm.atomic.lss.*' Intrinsic
      5. -
    5. 'llvm.memory.barrier' Intrinsic
      6. -
    -
  • Trampoline Intrinsic
    1. 'llvm.init.trampoline' Intrinsic
      2. @@ -291,7 +282,7 @@ the parser.

    • - +
    Identifiers
    @@ -4861,298 +4852,6 @@
    - Atomic Operations and Synchronization Intrinsics -
    - -
    -

    - These intrinsic functions expand the "universal IR" of LLVM to represent - hardware constructs for atomic operations and memory synchronization. This - provides an interface to the hardware, not an interface to the programmer. It - is aimed at a low enough level to allow any programming models or APIs which - need atomic behaviors to map cleanly onto it. It is also modeled primarily on - hardware behavior. Just as hardware provides a "universal IR" for source - languages, it also provides a starting point for developing a "universal" - atomic operation and synchronization IR. -

    -

    - These do not form an API such as high-level threading libraries, - software transaction memory systems, atomic primitives, and intrinsic - functions as found in BSD, GNU libc, atomic_ops, APR, and other system and - application libraries. The hardware interface provided by LLVM should allow - a clean implementation of all of these APIs and parallel programming models. - No one model or paradigm should be selected above others unless the hardware - itself ubiquitously does so. -

    -
    - - -
    - 'llvm.atomic.lcs.*' Intrinsic -
    -
    -
    Syntax:
    -

    - This is an overloaded intrinsic. You can use llvm.atomic.lcs on any - integer bit width. Not all targets support all bit widths however.

    -
    -declare i8 @llvm.atomic.lcs.i8.i8p.i8.i8( i8* <ptr>, i8 <cmp>, i8 <val> )
-declare i16 @llvm.atomic.lcs.i16.i16p.i16.i16( i16* <ptr>, i16 <cmp>, i16 <val> )
-declare i32 @llvm.atomic.lcs.i32.i32p.i32.i32( i32* <ptr>, i32 <cmp>, i32 <val> )
-declare i64 @llvm.atomic.lcs.i64.i64p.i64.i64( i64* <ptr>, i64 <cmp>, i64 <val> )
-
    -
    Overview:
    -

    - This loads a value in memory and compares it to a given value. If they are - equal, it stores a new value into the memory. -

    -
    Arguments:
    -

    - The llvm.atomic.lcs intrinsic takes three arguments. The result as - well as both cmp and val must be integer values with the - same bit width. The ptr argument must be a pointer to a value of - this integer type. While any bit width integer may be used, targets may only - lower representations they support in hardware. -

    -
    Semantics:
    -

    - This entire intrinsic must be executed atomically. It first loads the value - in memory pointed to by ptr and compares it with the value - cmp. If they are equal, val is stored into the memory. The - loaded value is yielded in all cases. This provides the equivalent of an - atomic compare-and-swap operation within the SSA framework. -

    -
    Examples:
    -
    -%ptr      = malloc i32
-            store i32 4, %ptr
-
-%val1     = add i32 4, 4
-%result1  = call i32 @llvm.atomic.lcs( i32* %ptr, i32 4, %val1 )
-                                          ; yields {i32}:result1 = 4
-%stored1  = icmp eq i32 %result1, 4       ; yields {i1}:stored1 = true
-%memval1  = load i32* %ptr                ; yields {i32}:memval1 = 8
-
-%val2     = add i32 1, 1
-%result2  = call i32 @llvm.atomic.lcs( i32* %ptr, i32 5, %val2 )
-                                          ; yields {i32}:result2 = 8
-%stored2  = icmp eq i32 %result2, 5       ; yields {i1}:stored2 = false
-%memval2  = load i32* %ptr                ; yields {i32}:memval2 = 8
-
    -
    - - -
    - 'llvm.atomic.ls.*' Intrinsic -
    -
    -
    Syntax:
    -

    - This is an overloaded intrinsic. You can use llvm.atomic.ls on any - integer bit width. Not all targets support all bit widths however.

    -
    -declare i8 @llvm.atomic.ls.i8.i8p.i8( i8* <ptr>, i8 <val> )
-declare i16 @llvm.atomic.ls.i16.i16p.i16( i16* <ptr>, i16 <val> )
-declare i32 @llvm.atomic.ls.i32.i32p.i32( i32* <ptr>, i32 <val> )
-declare i64 @llvm.atomic.ls.i64.i64p.i64( i64* <ptr>, i64 <val> )
-
    -
    Overview:
    -

    - This intrinsic loads the value stored in memory at ptr and yields - the value from memory. It then stores the value in val in the memory - at ptr. -

    -
    Arguments:
    -

    - The llvm.atomic.ls intrinsic takes two arguments. Both the - val argument and the result must be integers of the same bit width. - The first argument, ptr, must be a pointer to a value of this - integer type. The targets may only lower integer representations they - support. -

    -
    Semantics:
    -

    - This intrinsic loads the value pointed to by ptr, yields it, and - stores val back into ptr atomically. This provides the - equivalent of an atomic swap operation within the SSA framework. -

    -
    Examples:
    -
    -%ptr      = malloc i32
-            store i32 4, %ptr
-
-%val1     = add i32 4, 4
-%result1  = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val1 )
-                                        ; yields {i32}:result1 = 4
-%stored1  = icmp eq i32 %result1, 4     ; yields {i1}:stored1 = true
-%memval1  = load i32* %ptr              ; yields {i32}:memval1 = 8
-
-%val2     = add i32 1, 1
-%result2  = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val2 )
-                                        ; yields {i32}:result2 = 8
-%stored2  = icmp eq i32 %result2, 8     ; yields {i1}:stored2 = true
-%memval2  = load i32* %ptr              ; yields {i32}:memval2 = 2
-
    -
    - - -
    - 'llvm.atomic.las.*' Intrinsic -
    -
    -
    Syntax:
    -

    - This is an overloaded intrinsic. You can use llvm.atomic.las on any - integer bit width. Not all targets support all bit widths however.

    -
    -declare i8 @llvm.atomic.las.i8.i8p.i8( i8* <ptr>, i8 <delta> )
-declare i16 @llvm.atomic.las.i16.i16p.i16( i16* <ptr>, i16 <delta> )
-declare i32 @llvm.atomic.las.i32.i32p.i32( i32* <ptr>, i32 <delta> )
-declare i64 @llvm.atomic.las.i64.i64p.i64( i64* <ptr>, i64 <delta> )
-
    -
    Overview:
    -

    - This intrinsic adds delta to the value stored in memory at - ptr. It yields the original value at ptr. -

    -
    Arguments:
    -

    - The intrinsic takes two arguments, the first a pointer to an integer value - and the second an integer value. The result is also an integer value. These - integer types can have any bit width, but they must all have the same bit - width. The targets may only lower integer representations they support. -

    -
    Semantics:
    -

    - This intrinsic does a series of operations atomically. It first loads the - value stored at ptr. It then adds delta, stores the result - to ptr. It yields the original value stored at ptr. -

    -
    Examples:
    -
    -%ptr      = malloc i32
-        store i32 4, %ptr
-%result1  = call i32 @llvm.atomic.las( i32* %ptr, i32 4 )
-                                ; yields {i32}:result1 = 4
-%result2  = call i32 @llvm.atomic.las( i32* %ptr, i32 2 )
-                                ; yields {i32}:result2 = 8
-%result3  = call i32 @llvm.atomic.las( i32* %ptr, i32 5 )
-                                ; yields {i32}:result3 = 10
-%memval   = load i32* %ptr      ; yields {i32}:memval1 = 15
-
    -
    - - -
    - 'llvm.atomic.lss.*' Intrinsic -
    -
    -
    Syntax:
    -

    - This is an overloaded intrinsic. You can use llvm.atomic.lss on any - integer bit width. Not all targets support all bit widths however.

    -
    -declare i8 @llvm.atomic.lss.i8.i8.i8( i8* <ptr>, i8 <delta> )
-declare i16 @llvm.atomic.lss.i16.i16.i16( i16* <ptr>, i16 <delta> )
-declare i32 @llvm.atomic.lss.i32.i32.i32( i32* <ptr>, i32 <delta> )
-declare i64 @llvm.atomic.lss.i64.i64.i64( i64* <ptr>, i64 <delta> )
-
    -
    Overview:
    -

    - This intrinsic subtracts delta from the value stored in memory at - ptr. It yields the original value at ptr. -

    -
    Arguments:
    -

    - The intrinsic takes two arguments, the first a pointer to an integer value - and the second an integer value. The result is also an integer value. These - integer types can have any bit width, but they must all have the same bit - width. The targets may only lower integer representations they support. -

    -
    Semantics:
    -

    - This intrinsic does a series of operations atomically. It first loads the - value stored at ptr. It then subtracts delta, - stores the result to ptr. It yields the original value stored - at ptr. -

    -
    Examples:
    -
    -%ptr      = malloc i32
-        store i32 32, %ptr
-%result1  = call i32 @llvm.atomic.lss( i32* %ptr, i32 4 )
-                                    ; yields {i32}:result1 = 32
-%result2  = call i32 @llvm.atomic.lss( i32* %ptr, i32 2 )
-                                    ; yields {i32}:result2 = 28
-%result3  = call i32 @llvm.atomic.lss( i32* %ptr, i32 5 )
-                                    ; yields {i32}:result3 = 26
-%memval   = load i32* %ptr          ; yields {i32}:memval1 = 21
-
    -
    - - -
    - 'llvm.memory.barrier' Intrinsic -
    -
    -
    Syntax:
    -
    -declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss> )
-
    -
    Overview:
    -

    - The llvm.memory.barrier intrinsic guarantees ordering between - specific pairs of memory access types. -

    -
    Arguments:
    -

    - The llvm.memory.barrier intrinsic requires four boolean arguments. - Each argument enables a specific barrier as listed below. -

    -
      -
    • ll: load-load barrier
      • -
    • ls: load-store barrier
      • -
    • sl: store-load barrier
      • -
    • ss: store-store barrier
      • -
    -
    Semantics:
    -

    - This intrinsic causes the system to enforce some ordering constraints upon - the loads and stores of the program. This barrier does not indicate - when any events will occur, it only enforces an order in - which they occur. For any of the specified pairs of load and store operations - (f.ex. load-load, or store-load), all of the first operations preceding the - barrier will complete before any of the second operations succeeding the - barrier begin. Specifically the semantics for each pairing is as follows: -

    -
      -
    • ll: All loads before the barrier must complete before any load - after the barrier begins.
      • -
    • ls: All loads before the barrier must complete before any - store after the barrier begins.
      • -
    • ss: All stores before the barrier must complete before any - store after the barrier begins.
      • -
    • sl: All stores before the barrier must complete before any - load after the barrier begins.
      • -
    -

    - These semantics are applied with a logical "and" behavior when more than one - is enabled in a single memory barrier intrinsic. -

    -
    Example:
    -
    -%ptr      = malloc i32
-            store i32 4, %ptr
-
-%result1  = load i32* %ptr      ; yields {i32}:result1 = 4
-            call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
-                                ; guarantee the above finishes
-            store i32 8, %ptr   ; before this begins
-
    -
    - - -
    Trampoline Intrinsic
    From sabre at nondot.org Wed Oct 3 12:34:29 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 17:34:29 -0000 Subject: [llvm-commits] [llvm] r42581 - /llvm/trunk/docs/LangRef.html Message-ID: <200710031734.l93HYTlV005945@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 12:34:29 2007 New Revision: 42581 URL: http://llvm.org/viewvc/llvm-project?rev=42581&view=rev Log: remove extraneous tag Modified: llvm/trunk/docs/LangRef.html Modified: llvm/trunk/docs/LangRef.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.html?rev=42581&r1=42580&r2=42581&view=diff ============================================================================== --- llvm/trunk/docs/LangRef.html (original) +++ llvm/trunk/docs/LangRef.html Wed Oct 3 12:34:29 2007 @@ -282,7 +282,7 @@ the parser.

    - +
    Identifiers
    From dalej at apple.com Wed Oct 3 12:45:28 2007 From: dalej at apple.com (Dale Johannesen) Date: Wed, 03 Oct 2007 17:45:28 -0000 Subject: [llvm-commits] [llvm] r42582 - /llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp Message-ID: <200710031745.l93HjSJD006585@zion.cs.uiuc.edu> Author: johannes Date: Wed Oct 3 12:45:27 2007 New Revision: 42582 URL: http://llvm.org/viewvc/llvm-project?rev=42582&view=rev Log: Tone down an overzealous optimization. Modified: llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp Modified: llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp?rev=42582&r1=42581&r2=42582&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp Wed Oct 3 12:45:27 2007 @@ -7354,8 +7354,17 @@ if (FCmpInst *FCI = dyn_cast(CondVal)) { if (FCI->getOperand(0) == TrueVal && FCI->getOperand(1) == FalseVal) { // Transform (X == Y) ? X : Y -> Y - if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) + if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) { + // This is not safe in general for floating point: + // consider X== -0, Y== +0. + // It becomes safe if either operand is a nonzero constant. + ConstantFP *CFPt, *CFPf; + if (((CFPt = dyn_cast(TrueVal)) && + !CFPt->getValueAPF().isZero()) || + ((CFPf = dyn_cast(FalseVal)) && + !CFPf->getValueAPF().isZero())) return ReplaceInstUsesWith(SI, FalseVal); + } // Transform (X != Y) ? X : Y -> X if (FCI->getPredicate() == FCmpInst::FCMP_ONE) return ReplaceInstUsesWith(SI, TrueVal); @@ -7363,8 +7372,17 @@ } else if (FCI->getOperand(0) == FalseVal && FCI->getOperand(1) == TrueVal){ // Transform (X == Y) ? Y : X -> X - if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) - return ReplaceInstUsesWith(SI, FalseVal); + if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) { + // This is not safe in general for floating point: + // consider X== -0, Y== +0. + // It becomes safe if either operand is a nonzero constant. + ConstantFP *CFPt, *CFPf; + if (((CFPt = dyn_cast(TrueVal)) && + !CFPt->getValueAPF().isZero()) || + ((CFPf = dyn_cast(FalseVal)) && + !CFPf->getValueAPF().isZero())) + return ReplaceInstUsesWith(SI, FalseVal); + } // Transform (X != Y) ? Y : X -> Y if (FCI->getPredicate() == FCmpInst::FCMP_ONE) return ReplaceInstUsesWith(SI, TrueVal); From isanbard at gmail.com Wed Oct 3 12:46:54 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 03 Oct 2007 17:46:54 -0000 Subject: [llvm-commits] [llvm] r42583 - /llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Message-ID: <200710031746.l93Hks0u006702@zion.cs.uiuc.edu> Author: void Date: Wed Oct 3 12:46:54 2007 New Revision: 42583 URL: http://llvm.org/viewvc/llvm-project?rev=42583&view=rev Log: Temporarily XFAIL this test. Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/2007-09-25-EH.m?rev=42583&r1=42582&r2=42583&view=diff ============================================================================== --- llvm/trunk/test/FrontendObjC/2007-09-25-EH.m (original) +++ llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Wed Oct 3 12:46:54 2007 @@ -1,4 +1,6 @@ // RUN: %llvmgcc -c -w -m64 -mmacosx-version-min=10.5 %s -o /dev/null +// XFAIL: * +// Reenable when Obj-C FE problem is fixed. @class NSDictionary, DSoBuffer, DSoDirectory, NSMutableArray; @interface NSException {} From isanbard at gmail.com Wed Oct 3 12:47:27 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 3 Oct 2007 10:47:27 -0700 Subject: [llvm-commits] [llvm-gcc-4.0] r42566 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c In-Reply-To: References: <200710030433.l934XDap031554@zion.cs.uiuc.edu> Message-ID: <16e5fdf90710031047o2c042fbahee0ca7714cdd68a5@mail.gmail.com> On 10/3/07, Tanya Lattner wrote: > Reverting this patch causes test/FrontendObjC/2007-09-25-EH.m to fail. It > will most likely only fail on Tiger systems (<10.5). > > Please xfail this test and then re-enable it once this has been fixed. > Doh! I XFAILed it. This isn't a regression as it tested the patch I put in. :-) -bw From clattner at apple.com Wed Oct 3 13:14:35 2007 From: clattner at apple.com (Chris Lattner) Date: Wed, 3 Oct 2007 11:14:35 -0700 Subject: [llvm-commits] [llvm] r42582 - /llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp In-Reply-To: <200710031745.l93HjSJD006585@zion.cs.uiuc.edu> References: <200710031745.l93HjSJD006585@zion.cs.uiuc.edu> Message-ID: > URL: http://llvm.org/viewvc/llvm-project?rev=42582&view=rev > Log: > Tone down an overzealous optimization. Nice catch! -Chris > > Modified: > llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp > > Modified: llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/ > Scalar/InstructionCombining.cpp?rev=42582&r1=42581&r2=42582&view=diff > > ====================================================================== > ======== > --- llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp > (original) > +++ llvm/trunk/lib/Transforms/Scalar/InstructionCombining.cpp Wed > Oct 3 12:45:27 2007 > @@ -7354,8 +7354,17 @@ > if (FCmpInst *FCI = dyn_cast(CondVal)) { > if (FCI->getOperand(0) == TrueVal && FCI->getOperand(1) == > FalseVal) { > // Transform (X == Y) ? X : Y -> Y > - if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) > + if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) { > + // This is not safe in general for floating point: > + // consider X== -0, Y== +0. > + // It becomes safe if either operand is a nonzero constant. > + ConstantFP *CFPt, *CFPf; > + if (((CFPt = dyn_cast(TrueVal)) && > + !CFPt->getValueAPF().isZero()) || > + ((CFPf = dyn_cast(FalseVal)) && > + !CFPf->getValueAPF().isZero())) > return ReplaceInstUsesWith(SI, FalseVal); > + } > // Transform (X != Y) ? X : Y -> X > if (FCI->getPredicate() == FCmpInst::FCMP_ONE) > return ReplaceInstUsesWith(SI, TrueVal); > @@ -7363,8 +7372,17 @@ > > } else if (FCI->getOperand(0) == FalseVal && FCI->getOperand > (1) == TrueVal){ > // Transform (X == Y) ? Y : X -> X > - if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) > - return ReplaceInstUsesWith(SI, FalseVal); > + if (FCI->getPredicate() == FCmpInst::FCMP_OEQ) { > + // This is not safe in general for floating point: > + // consider X== -0, Y== +0. > + // It becomes safe if either operand is a nonzero constant. > + ConstantFP *CFPt, *CFPf; > + if (((CFPt = dyn_cast(TrueVal)) && > + !CFPt->getValueAPF().isZero()) || > + ((CFPf = dyn_cast(FalseVal)) && > + !CFPf->getValueAPF().isZero())) > + return ReplaceInstUsesWith(SI, FalseVal); > + } > // Transform (X != Y) ? Y : X -> Y > if (FCI->getPredicate() == FCmpInst::FCMP_ONE) > return ReplaceInstUsesWith(SI, TrueVal); > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From djg at cray.com Wed Oct 3 14:04:09 2007 From: djg at cray.com (Dan Gohman) Date: Wed, 03 Oct 2007 19:04:09 -0000 Subject: [llvm-commits] [llvm] r42584 - in /llvm/trunk: include/llvm/PassManagers.h lib/VMCore/PassManager.cpp Message-ID: <200710031904.l93J49KN010190@zion.cs.uiuc.edu> Author: djg Date: Wed Oct 3 14:04:09 2007 New Revision: 42584 URL: http://llvm.org/viewvc/llvm-project?rev=42584&view=rev Log: Fix a using namespace llvm; in a header file. Modified: llvm/trunk/include/llvm/PassManagers.h llvm/trunk/lib/VMCore/PassManager.cpp Modified: llvm/trunk/include/llvm/PassManagers.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/PassManagers.h?rev=42584&r1=42583&r2=42584&view=diff ============================================================================== --- llvm/trunk/include/llvm/PassManagers.h (original) +++ llvm/trunk/include/llvm/PassManagers.h Wed Oct 3 14:04:09 2007 @@ -13,7 +13,6 @@ #include "llvm/PassManager.h" #include "llvm/ADT/SmallVector.h" -using namespace llvm; class llvm::PMDataManager; class llvm::PMStack; @@ -381,8 +380,8 @@ } -extern void StartPassTimer(Pass *); -extern void StopPassTimer(Pass *); +extern void StartPassTimer(llvm::Pass *); +extern void StopPassTimer(llvm::Pass *); #endif Modified: llvm/trunk/lib/VMCore/PassManager.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/PassManager.cpp?rev=42584&r1=42583&r2=42584&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/PassManager.cpp (original) +++ llvm/trunk/lib/VMCore/PassManager.cpp Wed Oct 3 14:04:09 2007 @@ -22,6 +22,7 @@ #include #include #include +using namespace llvm; // See PassManagers.h for Pass Manager infrastructure overview. From djg at cray.com Wed Oct 3 14:26:29 2007 From: djg at cray.com (Dan Gohman) Date: Wed, 03 Oct 2007 19:26:29 -0000 Subject: [llvm-commits] [llvm] r42585 - in /llvm/trunk/lib: Analysis/ CodeGen/ CodeGen/SelectionDAG/ Target/ARM/ Target/IA64/ Target/PowerPC/ Target/X86/ Transforms/IPO/ Transforms/Scalar/ VMCore/ Message-ID: <200710031926.l93JQT5Z010983@zion.cs.uiuc.edu> Author: djg Date: Wed Oct 3 14:26:29 2007 New Revision: 42585 URL: http://llvm.org/viewvc/llvm-project?rev=42585&view=rev Log: Use empty() member functions when that's what's being tested for instead of comparing begin() and end(). Modified: llvm/trunk/lib/Analysis/AliasSetTracker.cpp llvm/trunk/lib/Analysis/LoopPass.cpp llvm/trunk/lib/CodeGen/LiveIntervalAnalysis.cpp llvm/trunk/lib/CodeGen/MachineBasicBlock.cpp llvm/trunk/lib/CodeGen/MachineFunction.cpp llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp llvm/trunk/lib/Target/IA64/IA64AsmPrinter.cpp llvm/trunk/lib/Target/PowerPC/PPCAsmPrinter.cpp llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp llvm/trunk/lib/Target/X86/X86IntelAsmPrinter.cpp llvm/trunk/lib/Transforms/IPO/Inliner.cpp llvm/trunk/lib/Transforms/IPO/RaiseAllocations.cpp llvm/trunk/lib/Transforms/Scalar/LICM.cpp llvm/trunk/lib/Transforms/Scalar/SCCP.cpp llvm/trunk/lib/VMCore/Value.cpp Modified: llvm/trunk/lib/Analysis/AliasSetTracker.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/AliasSetTracker.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/AliasSetTracker.cpp (original) +++ llvm/trunk/lib/Analysis/AliasSetTracker.cpp Wed Oct 3 14:26:29 2007 @@ -520,7 +520,7 @@ OS << " forwarding to " << (void*)Forward; - if (begin() != end()) { + if (!empty()) { OS << "Pointers: "; for (iterator I = begin(), E = end(); I != E; ++I) { if (I != begin()) OS << ", "; Modified: llvm/trunk/lib/Analysis/LoopPass.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/LoopPass.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/LoopPass.cpp (original) +++ llvm/trunk/lib/Analysis/LoopPass.cpp Wed Oct 3 14:26:29 2007 @@ -54,7 +54,7 @@ } // Move all subloops into the parent loop. - while (L->begin() != L->end()) + while (!L->empty()) ParentLoop->addChildLoop(L->removeChildLoop(L->end()-1)); } else { // Reparent all of the blocks in this loop. Since BBLoop had no parent, @@ -78,7 +78,7 @@ } // Move all of the subloops to the top-level. - while (L->begin() != L->end()) + while (!L->empty()) LI->addTopLevelLoop(L->removeChildLoop(L->end()-1)); } Modified: llvm/trunk/lib/CodeGen/LiveIntervalAnalysis.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveIntervalAnalysis.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/LiveIntervalAnalysis.cpp (original) +++ llvm/trunk/lib/CodeGen/LiveIntervalAnalysis.cpp Wed Oct 3 14:26:29 2007 @@ -817,17 +817,15 @@ MachineBasicBlock::iterator MI = MBB->begin(), miEnd = MBB->end(); - if (MBB->livein_begin() != MBB->livein_end()) { - // Create intervals for live-ins to this BB first. - for (MachineBasicBlock::const_livein_iterator LI = MBB->livein_begin(), - LE = MBB->livein_end(); LI != LE; ++LI) { - handleLiveInRegister(MBB, MIIndex, getOrCreateInterval(*LI)); - // Multiple live-ins can alias the same register. - for (const unsigned* AS = mri_->getSubRegisters(*LI); *AS; ++AS) - if (!hasInterval(*AS)) - handleLiveInRegister(MBB, MIIndex, getOrCreateInterval(*AS), - true); - } + // Create intervals for live-ins to this BB first. + for (MachineBasicBlock::const_livein_iterator LI = MBB->livein_begin(), + LE = MBB->livein_end(); LI != LE; ++LI) { + handleLiveInRegister(MBB, MIIndex, getOrCreateInterval(*LI)); + // Multiple live-ins can alias the same register. + for (const unsigned* AS = mri_->getSubRegisters(*LI); *AS; ++AS) + if (!hasInterval(*AS)) + handleLiveInRegister(MBB, MIIndex, getOrCreateInterval(*AS), + true); } for (; MI != miEnd; ++MI) { Modified: llvm/trunk/lib/CodeGen/MachineBasicBlock.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineBasicBlock.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/MachineBasicBlock.cpp (original) +++ llvm/trunk/lib/CodeGen/MachineBasicBlock.cpp Wed Oct 3 14:26:29 2007 @@ -119,7 +119,7 @@ OS << ":\n"; const MRegisterInfo *MRI = MF->getTarget().getRegisterInfo(); - if (livein_begin() != livein_end()) { + if (!livein_empty()) { OS << "Live Ins:"; for (const_livein_iterator I = livein_begin(),E = livein_end(); I != E; ++I) OutputReg(OS, *I, MRI); Modified: llvm/trunk/lib/CodeGen/MachineFunction.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineFunction.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/MachineFunction.cpp (original) +++ llvm/trunk/lib/CodeGen/MachineFunction.cpp Wed Oct 3 14:26:29 2007 @@ -208,7 +208,7 @@ const MRegisterInfo *MRI = getTarget().getRegisterInfo(); - if (livein_begin() != livein_end()) { + if (!livein_empty()) { OS << "Live Ins:"; for (livein_iterator I = livein_begin(), E = livein_end(); I != E; ++I) { if (MRI) @@ -221,7 +221,7 @@ } OS << "\n"; } - if (liveout_begin() != liveout_end()) { + if (!liveout_empty()) { OS << "Live Outs:"; for (liveout_iterator I = liveout_begin(), E = liveout_end(); I != E; ++I) if (MRI) Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp Wed Oct 3 14:26:29 2007 @@ -866,7 +866,7 @@ // that need to be copied into vregs, emit the copies into the top of the // block before emitting the code for the block. MachineFunction &MF = DAG.getMachineFunction(); - if (&MF.front() == BB && MF.livein_begin() != MF.livein_end()) { + if (&MF.front() == BB) { for (MachineFunction::livein_iterator LI = MF.livein_begin(), E = MF.livein_end(); LI != E; ++LI) if (LI->second) { Modified: llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -1001,7 +1001,7 @@ O << "\n"; // Output non-lazy-pointers for external and common global variables. - if (GVNonLazyPtrs.begin() != GVNonLazyPtrs.end()) + if (!GVNonLazyPtrs.empty()) SwitchToDataSection(".non_lazy_symbol_pointer", 0); for (std::set::iterator i = GVNonLazyPtrs.begin(), e = GVNonLazyPtrs.end(); i != e; ++i) { Modified: llvm/trunk/lib/Target/IA64/IA64AsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/IA64/IA64AsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/IA64/IA64AsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/IA64/IA64AsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -147,7 +147,7 @@ for (MachineFunction::const_iterator I = MF.begin(), E = MF.end(); I != E; ++I) { // Print a label for the basic block if there are any predecessors. - if (I->pred_begin() != I->pred_end()) { + if (!I->pred_empty()) { printBasicBlockLabel(I, true); O << '\n'; } Modified: llvm/trunk/lib/Target/PowerPC/PPCAsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCAsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCAsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/PowerPC/PPCAsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -1053,7 +1053,7 @@ O << "\n"; // Output stubs for external and common global variables. - if (GVStubs.begin() != GVStubs.end()) { + if (!GVStubs.empty()) { SwitchToDataSection(".non_lazy_symbol_pointer"); for (std::set::iterator I = GVStubs.begin(), E = GVStubs.end(); I != E; ++I) { Modified: llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -168,7 +168,7 @@ for (MachineFunction::const_iterator I = MF.begin(), E = MF.end(); I != E; ++I) { // Print a label for the basic block. - if (I->pred_begin() != I->pred_end()) { + if (!I->pred_empty()) { printBasicBlockLabel(I, true); O << '\n'; } Modified: llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -316,7 +316,7 @@ } // Output linker support code for dllexported globals - if (DLLExportedGVs.begin() != DLLExportedGVs.end()) { + if (!DLLExportedGVs.empty()) { SwitchToDataSection(".section .drectve"); } @@ -326,7 +326,7 @@ O << "\t.ascii \" -export:" << *i << ",data\"\n"; } - if (DLLExportedFns.begin() != DLLExportedFns.end()) { + if (!DLLExportedFns.empty()) { SwitchToDataSection(".section .drectve"); } @@ -362,7 +362,7 @@ } // Output stubs for external and common global variables. - if (GVStubs.begin() != GVStubs.end()) + if (!GVStubs.empty()) SwitchToDataSection( ".section __IMPORT,__pointers,non_lazy_symbol_pointers"); for (std::set::iterator i = GVStubs.begin(), e = GVStubs.end(); Modified: llvm/trunk/lib/Target/X86/X86IntelAsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86IntelAsmPrinter.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86IntelAsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/X86/X86IntelAsmPrinter.cpp Wed Oct 3 14:26:29 2007 @@ -77,7 +77,7 @@ for (MachineFunction::const_iterator I = MF.begin(), E = MF.end(); I != E; ++I) { // Print a label for the basic block if there are any predecessors. - if (I->pred_begin() != I->pred_end()) { + if (!I->pred_empty()) { printBasicBlockLabel(I, true); O << '\n'; } @@ -412,8 +412,8 @@ } // Output linker support code for dllexported globals - if ((DLLExportedGVs.begin() != DLLExportedGVs.end()) || - (DLLExportedFns.begin() != DLLExportedFns.end())) { + if (!DLLExportedGVs.empty() || + !DLLExportedFns.empty()) { SwitchToDataSection(""); O << "; WARNING: The following code is valid only with MASM v8.x and (possible) higher\n" << "; This version of MASM is usually shipped with Microsoft Visual Studio 2005\n" @@ -434,8 +434,8 @@ O << "\t db ' /EXPORT:" << *i << "'\n"; } - if ((DLLExportedGVs.begin() != DLLExportedGVs.end()) || - (DLLExportedFns.begin() != DLLExportedFns.end())) { + if (!DLLExportedGVs.empty() || + !DLLExportedFns.empty()) { O << "_drectve\t ends\n"; } Modified: llvm/trunk/lib/Transforms/IPO/Inliner.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/IPO/Inliner.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/IPO/Inliner.cpp (original) +++ llvm/trunk/lib/Transforms/IPO/Inliner.cpp Wed Oct 3 14:26:29 2007 @@ -63,7 +63,7 @@ // Remove any call graph edges from the callee to its callees. CallGraphNode *CalleeNode = CG[Callee]; - while (CalleeNode->begin() != CalleeNode->end()) + while (!CalleeNode->empty()) CalleeNode->removeCallEdgeTo((CalleeNode->end()-1)->second); // Removing the node for callee from the call graph and delete it. @@ -188,7 +188,7 @@ F->use_empty()) { // Remove any call graph edges from the function to its callees. - while (CGN->begin() != CGN->end()) + while (!CGN->empty()) CGN->removeCallEdgeTo((CGN->end()-1)->second); // Remove any edges from the external node to the function's call graph Modified: llvm/trunk/lib/Transforms/IPO/RaiseAllocations.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/IPO/RaiseAllocations.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/IPO/RaiseAllocations.cpp (original) +++ llvm/trunk/lib/Transforms/IPO/RaiseAllocations.cpp Wed Oct 3 14:26:29 2007 @@ -154,7 +154,7 @@ if (Instruction *I = dyn_cast(U)) { CallSite CS = CallSite::get(I); - if (CS.getInstruction() && CS.arg_begin() != CS.arg_end() && + if (CS.getInstruction() && !CS.arg_empty() && (CS.getCalledFunction() == MallocFunc || std::find(EqPointers.begin(), EqPointers.end(), CS.getCalledValue()) != EqPointers.end())) { @@ -205,7 +205,7 @@ if (Instruction *I = dyn_cast(U)) { CallSite CS = CallSite::get(I); - if (CS.getInstruction() && CS.arg_begin() != CS.arg_end() && + if (CS.getInstruction() && !CS.arg_empty() && (CS.getCalledFunction() == FreeFunc || std::find(EqPointers.begin(), EqPointers.end(), CS.getCalledValue()) != EqPointers.end())) { Modified: llvm/trunk/lib/Transforms/Scalar/LICM.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LICM.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LICM.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LICM.cpp Wed Oct 3 14:26:29 2007 @@ -786,7 +786,7 @@ // volatile loads or stores. if (!AS.isForwardingAliasSet() && AS.isMod() && AS.isMustAlias() && !AS.isVolatile() && CurLoop->isLoopInvariant(AS.begin()->first)) { - assert(AS.begin() != AS.end() && + assert(!AS.empty() && "Must alias set should have at least one pointer element in it!"); Value *V = AS.begin()->first; Modified: llvm/trunk/lib/Transforms/Scalar/SCCP.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/SCCP.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/SCCP.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/SCCP.cpp Wed Oct 3 14:26:29 2007 @@ -1574,7 +1574,7 @@ for (unsigned i = 0, e = TI->getNumSuccessors(); i != e; ++i) { BasicBlock *Succ = TI->getSuccessor(i); - if (Succ->begin() != Succ->end() && isa(Succ->begin())) + if (!Succ->empty() && isa(Succ->begin())) TI->getSuccessor(i)->removePredecessor(BB); } if (!TI->use_empty()) Modified: llvm/trunk/lib/VMCore/Value.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Value.cpp?rev=42585&r1=42584&r2=42585&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Value.cpp (original) +++ llvm/trunk/lib/VMCore/Value.cpp Wed Oct 3 14:26:29 2007 @@ -47,14 +47,14 @@ // still being referenced. The value in question should be printed as // a // - if (use_begin() != use_end()) { + if (!use_empty()) { DOUT << "While deleting: " << *Ty << " %" << Name << "\n"; for (use_iterator I = use_begin(), E = use_end(); I != E; ++I) DOUT << "Use still stuck around after Def is destroyed:" << **I << "\n"; } #endif - assert(use_begin() == use_end() && "Uses remain when a value is destroyed!"); + assert(use_empty() && "Uses remain when a value is destroyed!"); // If this value is named, destroy the name. This should not be in a symtab // at this point. From sabre at nondot.org Wed Oct 3 15:45:43 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 20:45:43 -0000 Subject: [llvm-commits] [llvm] r42586 - /llvm/trunk/lib/Support/FoldingSet.cpp Message-ID: <200710032045.l93Kjh51014033@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 15:45:43 2007 New Revision: 42586 URL: http://llvm.org/viewvc/llvm-project?rev=42586&view=rev Log: Simplify implementation of the FoldingSet circular list, a necessary step to giving it iterators. Modified: llvm/trunk/lib/Support/FoldingSet.cpp Modified: llvm/trunk/lib/Support/FoldingSet.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/FoldingSet.cpp?rev=42586&r1=42585&r2=42586&view=diff ============================================================================== --- llvm/trunk/lib/Support/FoldingSet.cpp (original) +++ llvm/trunk/lib/Support/FoldingSet.cpp Wed Oct 3 15:45:43 2007 @@ -139,17 +139,19 @@ /// The problem with this is that the start of the hash buckets are not /// Nodes. If NextInBucketPtr is a bucket pointer, this method returns null: /// use GetBucketPtr when this happens. -static FoldingSetImpl::Node *GetNextPtr(void *NextInBucketPtr, - void **Buckets, unsigned NumBuckets) { - if (NextInBucketPtr >= Buckets && NextInBucketPtr < Buckets + NumBuckets) +static FoldingSetImpl::Node *GetNextPtr(void *NextInBucketPtr) { + // The low bit is set if this is the pointer back to the bucket. + if (reinterpret_cast(NextInBucketPtr) & 1) return 0; + return static_cast(NextInBucketPtr); } /// GetBucketPtr - Provides a casting of a bucket pointer for isNode /// testing. static void **GetBucketPtr(void *NextInBucketPtr) { - return static_cast(NextInBucketPtr); + intptr_t Ptr = reinterpret_cast(NextInBucketPtr); + return reinterpret_cast(Ptr & ~intptr_t(1)); } /// GetBucketFor - Hash the specified node ID and return the hash bucket for @@ -168,8 +170,11 @@ assert(5 < Log2InitSize && Log2InitSize < 32 && "Initial hash table size out of range"); NumBuckets = 1 << Log2InitSize; - Buckets = new void*[NumBuckets]; + Buckets = new void*[NumBuckets+1]; memset(Buckets, 0, NumBuckets*sizeof(void*)); + + // Set the very last bucket to be a non-null "pointer". + Buckets[NumBuckets] = reinterpret_cast(-2); } FoldingSetImpl::~FoldingSetImpl() { delete [] Buckets; @@ -186,14 +191,17 @@ NumNodes = 0; // Clear out new buckets. - Buckets = new void*[NumBuckets]; + Buckets = new void*[NumBuckets+1]; memset(Buckets, 0, NumBuckets*sizeof(void*)); + // Set the very last bucket to be a non-null "pointer". + Buckets[NumBuckets] = reinterpret_cast(-1); + // Walk the old buckets, rehashing nodes into their new place. for (unsigned i = 0; i != OldNumBuckets; ++i) { void *Probe = OldBuckets[i]; if (!Probe) continue; - while (Node *NodeInBucket = GetNextPtr(Probe, OldBuckets, OldNumBuckets)) { + while (Node *NodeInBucket = GetNextPtr(Probe)) { // Figure out the next link, remove NodeInBucket from the old link. Probe = NodeInBucket->getNextInBucket(); NodeInBucket->SetNextInBucket(0); @@ -218,7 +226,7 @@ InsertPos = 0; - while (Node *NodeInBucket = GetNextPtr(Probe, Buckets, NumBuckets)) { + while (Node *NodeInBucket = GetNextPtr(Probe)) { NodeID OtherID; GetNodeProfile(OtherID, NodeInBucket); if (OtherID == ID) @@ -253,9 +261,10 @@ void *Next = *Bucket; // If this is the first insertion into this bucket, its next pointer will be - // null. Pretend as if it pointed to itself. + // null. Pretend as if it pointed to itself, setting the low bit to indicate + // that it is a pointer to the bucket. if (Next == 0) - Next = Bucket; + Next = reinterpret_cast(reinterpret_cast(Bucket)|1); // Set the node's next pointer, and make the bucket point to the node. N->SetNextInBucket(Next); @@ -278,7 +287,7 @@ // Chase around the list until we find the node (or bucket) which points to N. while (true) { - if (Node *NodeInBucket = GetNextPtr(Ptr, Buckets, NumBuckets)) { + if (Node *NodeInBucket = GetNextPtr(Ptr)) { // Advance pointer. Ptr = NodeInBucket->getNextInBucket(); From sabre at nondot.org Wed Oct 3 16:01:14 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 21:01:14 -0000 Subject: [llvm-commits] [llvm] r42588 - /llvm/trunk/docs/LangRef.html Message-ID: <200710032101.l93L1Ef0014768@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 16:01:14 2007 New Revision: 42588 URL: http://llvm.org/viewvc/llvm-project?rev=42588&view=rev Log: Clarify that shifts that are too large are undefined. Modified: llvm/trunk/docs/LangRef.html Modified: llvm/trunk/docs/LangRef.html URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.html?rev=42588&r1=42587&r2=42588&view=diff ============================================================================== --- llvm/trunk/docs/LangRef.html (original) +++ llvm/trunk/docs/LangRef.html Wed Oct 3 16:01:14 2007 @@ -2168,18 +2168,28 @@
    Syntax:
      <result> = shl <ty> <var1>, <var2>   ; yields {ty}:result
    +
    Overview:
    +

    The 'shl' instruction returns the first operand shifted to the left a specified number of bits.

    +
    Arguments:
    +

    Both arguments to the 'shl' instruction must be the same integer type.

    +
    Semantics:
    -

    The value produced is var1 * 2var2.

    + +

    The value produced is var1 * 2var2. If +var2 is (statically or dynamically) equal to or larger than the number +of bits in var1, the result is undefined.

    +
    Example:
       <result> = shl i32 4, %var   ; yields {i32}: 4 << %var
   <result> = shl i32 4, 2      ; yields {i32}: 16
   <result> = shl i32 1, 10     ; yields {i32}: 1024
+  <result> = shl i32 1, 32     ; undefined
    @@ -2199,9 +2209,11 @@ integer type.

    Semantics:
    +

    This instruction always performs a logical shift right operation. The most significant bits of the result will be filled with zero bits after the -shift.

    +shift. If var2 is (statically or dynamically) equal to or larger than +the number of bits in var1, the result is undefined.

    Example:
    @@ -2209,6 +2221,7 @@
   <result> = lshr i32 4, 2   ; yields {i32}:result = 1
   <result> = lshr i8  4, 3   ; yields {i8}:result = 0
   <result> = lshr i8 -2, 1   ; yields {i8}:result = 0x7FFFFFFF 
+  <result> = lshr i32 1, 32  ; undefined
    @@ -2232,7 +2245,9 @@
    Semantics:

    This instruction always performs an arithmetic shift right operation, The most significant bits of the result will be filled with the sign bit -of var1.

    +of var1. If var2 is (statically or dynamically) equal to or +larger than the number of bits in var1, the result is undefined. +

    Example:
    @@ -2240,6 +2255,7 @@
   <result> = ashr i32 4, 2   ; yields {i32}:result = 1
   <result> = ashr i8  4, 3   ; yields {i8}:result = 0
   <result> = ashr i8 -2, 1   ; yields {i8}:result = -1
+  <result> = ashr i32 1, 32  ; undefined
    From sabre at nondot.org Wed Oct 3 16:12:10 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 03 Oct 2007 21:12:10 -0000 Subject: [llvm-commits] [llvm] r42589 - in /llvm/trunk: include/llvm/ADT/FoldingSet.h lib/Support/FoldingSet.cpp Message-ID: <200710032112.l93LCAMa015236@zion.cs.uiuc.edu> Author: lattner Date: Wed Oct 3 16:12:09 2007 New Revision: 42589 URL: http://llvm.org/viewvc/llvm-project?rev=42589&view=rev Log: Add initial iterator support for folding set. Modified: llvm/trunk/include/llvm/ADT/FoldingSet.h llvm/trunk/lib/Support/FoldingSet.cpp Modified: llvm/trunk/include/llvm/ADT/FoldingSet.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/FoldingSet.h?rev=42589&r1=42588&r2=42589&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/FoldingSet.h (original) +++ llvm/trunk/include/llvm/ADT/FoldingSet.h Wed Oct 3 16:12:09 2007 @@ -220,6 +220,8 @@ typedef FoldingSetImpl::Node FoldingSetNode; typedef FoldingSetImpl::NodeID FoldingSetNodeID; +template class FoldingSetIterator; + //===----------------------------------------------------------------------===// /// FoldingSet - This template class is used to instantiate a specialized /// implementation of the folding set to the node class T. T must be a @@ -238,6 +240,14 @@ explicit FoldingSet(unsigned Log2InitSize = 6) : FoldingSetImpl(Log2InitSize) {} + + typedef FoldingSetIterator iterator; + iterator begin() { return iterator(Buckets); } + iterator end() { return iterator(Buckets+NumBuckets); } + + typedef FoldingSetIterator const_iterator; + const_iterator begin() const { return const_iterator(Buckets); } + const_iterator end() const { return const_iterator(Buckets+NumBuckets); } /// GetOrInsertNode - If there is an existing simple Node exactly /// equal to the specified node, return it. Otherwise, insert 'N' and @@ -254,6 +264,47 @@ } }; +//===----------------------------------------------------------------------===// +/// FoldingSetIteratorImpl - This is the common iterator support shared by all +/// folding sets, which knows how to walk the folding set hash table. +class FoldingSetIteratorImpl { +protected: + FoldingSetNode *NodePtr; + FoldingSetIteratorImpl(void **Bucket); + void advance(); + +public: + bool operator==(const FoldingSetIteratorImpl &RHS) const { + return NodePtr == RHS.NodePtr; + } + bool operator!=(const FoldingSetIteratorImpl &RHS) const { + return NodePtr != RHS.NodePtr; + } +}; + + +template +class FoldingSetIterator : public FoldingSetIteratorImpl { +public: + FoldingSetIterator(void **Bucket) : FoldingSetIteratorImpl(Bucket) {} + + T &operator*() const { + return *static_cast(NodePtr); + } + + T *operator->() const { + return static_cast(NodePtr); + } + + inline FoldingSetIterator& operator++() { // Preincrement + advance(); + return *this; + } + FoldingSetIterator operator++(int) { // Postincrement + FoldingSetIterator tmp = *this; ++*this; return tmp; + } +}; + } // End of namespace llvm. Modified: llvm/trunk/lib/Support/FoldingSet.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/FoldingSet.cpp?rev=42589&r1=42588&r2=42589&view=diff ============================================================================== --- llvm/trunk/lib/Support/FoldingSet.cpp (original) +++ llvm/trunk/lib/Support/FoldingSet.cpp Wed Oct 3 16:12:09 2007 @@ -151,6 +151,7 @@ /// testing. static void **GetBucketPtr(void *NextInBucketPtr) { intptr_t Ptr = reinterpret_cast(NextInBucketPtr); + assert((Ptr & 1) && "Not a bucket pointer"); return reinterpret_cast(Ptr & ~intptr_t(1)); } @@ -323,3 +324,34 @@ InsertNode(N, IP); return N; } + +//===----------------------------------------------------------------------===// +// FoldingSetIteratorImpl Implementation + +FoldingSetIteratorImpl::FoldingSetIteratorImpl(void **Bucket) { + // Skip to the first non-null non-self-cycle bucket. + while (*Bucket == 0 || GetNextPtr(*Bucket) == 0) + ++Bucket; + + NodePtr = static_cast(*Bucket); +} + +void FoldingSetIteratorImpl::advance() { + // If there is another link within this bucket, go to it. + void *Probe = NodePtr->getNextInBucket(); + + if (FoldingSetNode *NextNodeInBucket = GetNextPtr(Probe)) + NodePtr = NextNodeInBucket; + else { + // Otherwise, this is the last link in this bucket. + void **Bucket = GetBucketPtr(Probe); + + // Skip to the next non-null non-self-cycle bucket. + do { + ++Bucket; + } while (*Bucket == 0 || GetNextPtr(*Bucket) == 0); + + NodePtr = static_cast(*Bucket); + } +} + From dpatel at apple.com Wed Oct 3 16:16:08 2007 From: dpatel at apple.com (Devang Patel) Date: Wed, 03 Oct 2007 21:16:08 -0000 Subject: [llvm-commits] [llvm] r42590 - /llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Message-ID: <200710032116.l93LG8nm015463@zion.cs.uiuc.edu> Author: dpatel Date: Wed Oct 3 16:16:08 2007 New Revision: 42590 URL: http://llvm.org/viewvc/llvm-project?rev=42590&view=rev Log: Refactor code in a separate method. Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp?rev=42590&r1=42589&r2=42590&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Wed Oct 3 16:16:08 2007 @@ -103,6 +103,11 @@ if (I != LoopProcessWorklist.end()) LoopProcessWorklist.erase(I); } + + /// Split all of the edges from inside the loop to their exit blocks. Update + /// the appropriate Phi nodes as we do so. + void SplitExitEdges(const SmallVector &ExitBlocks, + SmallVector &MiddleBlocks); bool UnswitchIfProfitable(Value *LoopCond, Constant *Val,Loop *L); unsigned getLoopUnswitchCost(Loop *L, Value *LIC); @@ -596,38 +601,11 @@ ++NumTrivial; } -/// VersionLoop - We determined that the loop is profitable to unswitch when LIC -/// equal Val. Split it into loop versions and test the condition outside of -/// either loop. Return the loops created as Out1/Out2. -void LoopUnswitch::UnswitchNontrivialCondition(Value *LIC, Constant *Val, - Loop *L) { - Function *F = L->getHeader()->getParent(); - DOUT << "loop-unswitch: Unswitching loop %" - << L->getHeader()->getName() << " [" << L->getBlocks().size() - << " blocks] in Function " << F->getName() - << " when '" << *Val << "' == " << *LIC << "\n"; - - // LoopBlocks contains all of the basic blocks of the loop, including the - // preheader of the loop, the body of the loop, and the exit blocks of the - // loop, in that order. - std::vector LoopBlocks; - - // First step, split the preheader and exit blocks, and add these blocks to - // the LoopBlocks list. - BasicBlock *OrigHeader = L->getHeader(); - BasicBlock *OrigPreheader = L->getLoopPreheader(); - BasicBlock *NewPreheader = SplitEdge(OrigPreheader, L->getHeader(), this); - LoopBlocks.push_back(NewPreheader); - - // We want the loop to come after the preheader, but before the exit blocks. - LoopBlocks.insert(LoopBlocks.end(), L->block_begin(), L->block_end()); - - SmallVector ExitBlocks; - L->getUniqueExitBlocks(ExitBlocks); - - // Split all of the edges from inside the loop to their exit blocks. Update - // the appropriate Phi nodes as we do so. - SmallVector MiddleBlocks; +/// SplitExitEdges - +/// Split all of the edges from inside the loop to their exit blocks. Update +/// the appropriate Phi nodes as we do so. +void LoopUnswitch::SplitExitEdges(const SmallVector &ExitBlocks, + SmallVector &MiddleBlocks) { for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i) { BasicBlock *ExitBlock = ExitBlocks[i]; std::vector Preds(pred_begin(ExitBlock), pred_end(ExitBlock)); @@ -671,7 +649,42 @@ } } } - +} + +/// UnswitchNontrivialCondition - We determined that the loop is profitable to unswitch when LIC +/// equal Val. Split it into loop versions and test the condition outside of +/// either loop. Return the loops created as Out1/Out2. +void LoopUnswitch::UnswitchNontrivialCondition(Value *LIC, Constant *Val, + Loop *L) { + Function *F = L->getHeader()->getParent(); + DOUT << "loop-unswitch: Unswitching loop %" + << L->getHeader()->getName() << " [" << L->getBlocks().size() + << " blocks] in Function " << F->getName() + << " when '" << *Val << "' == " << *LIC << "\n"; + + // LoopBlocks contains all of the basic blocks of the loop, including the + // preheader of the loop, the body of the loop, and the exit blocks of the + // loop, in that order. + std::vector LoopBlocks; + + // First step, split the preheader and exit blocks, and add these blocks to + // the LoopBlocks list. + BasicBlock *OrigHeader = L->getHeader(); + BasicBlock *OrigPreheader = L->getLoopPreheader(); + BasicBlock *NewPreheader = SplitEdge(OrigPreheader, L->getHeader(), this); + LoopBlocks.push_back(NewPreheader); + + // We want the loop to come after the preheader, but before the exit blocks. + LoopBlocks.insert(LoopBlocks.end(), L->block_begin(), L->block_end()); + + SmallVector ExitBlocks; + L->getUniqueExitBlocks(ExitBlocks); + + // Split all of the edges from inside the loop to their exit blocks. Update + // the appropriate Phi nodes as we do so. + SmallVector MiddleBlocks; + SplitExitEdges(ExitBlocks, MiddleBlocks); + // The exit blocks may have been changed due to edge splitting, recompute. ExitBlocks.clear(); L->getUniqueExitBlocks(ExitBlocks); From dpatel at apple.com Wed Oct 3 16:17:43 2007 From: dpatel at apple.com (Devang Patel) Date: Wed, 03 Oct 2007 21:17:43 -0000 Subject: [llvm-commits] [llvm] r42591 - /llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Message-ID: <200710032117.l93LHh1d015537@zion.cs.uiuc.edu> Author: dpatel Date: Wed Oct 3 16:17:43 2007 New Revision: 42591 URL: http://llvm.org/viewvc/llvm-project?rev=42591&view=rev Log: Fix 80 col violation. Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp?rev=42591&r1=42590&r2=42591&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Wed Oct 3 16:17:43 2007 @@ -651,9 +651,9 @@ } } -/// UnswitchNontrivialCondition - We determined that the loop is profitable to unswitch when LIC -/// equal Val. Split it into loop versions and test the condition outside of -/// either loop. Return the loops created as Out1/Out2. +/// UnswitchNontrivialCondition - We determined that the loop is profitable +/// to unswitch when LIC equal Val. Split it into loop versions and test the +/// condition outside of either loop. Return the loops created as Out1/Out2. void LoopUnswitch::UnswitchNontrivialCondition(Value *LIC, Constant *Val, Loop *L) { Function *F = L->getHeader()->getParent(); From resistor at mac.com Wed Oct 3 16:24:38 2007 From: resistor at mac.com (Owen Anderson) Date: Wed, 03 Oct 2007 21:24:38 -0000 Subject: [llvm-commits] [llvm] r42592 - /llvm/trunk/include/llvm/ADT/GraphTraits.h Message-ID: <200710032124.l93LOck3015993@zion.cs.uiuc.edu> Author: resistor Date: Wed Oct 3 16:24:38 2007 New Revision: 42592 URL: http://llvm.org/viewvc/llvm-project?rev=42592&view=rev Log: Add a GraphTraits partial specialization to make the inverse of an inverse be the same as the underlying graph. Modified: llvm/trunk/include/llvm/ADT/GraphTraits.h Modified: llvm/trunk/include/llvm/ADT/GraphTraits.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/GraphTraits.h?rev=42592&r1=42591&r2=42592&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/GraphTraits.h (original) +++ llvm/trunk/include/llvm/ADT/GraphTraits.h Wed Oct 3 16:24:38 2007 @@ -78,6 +78,26 @@ inline Inverse(GraphType &G) : Graph(G) {} }; +// Provide a partial specialization of GraphTraits so that the inverse of an inverse +// falls back to the original graph. +template +struct GraphTraits > > { + typedef typename GraphTraits::NodeType NodeType; + typedef typename GraphTraits::ChildIteratorType ChildIteratorType; + + static NodeType *getEntryNode(Inverse > *G) { + return GraphTraits::getEntryNode(G.Graph.Graph); + } + + static ChildIteratorType child_begin(NodeType* N) { + return GraphTraits::child_begin(N); + } + + static ChildIteratorType child_end(NodeType* N) { + return GraphTraits::child_end(N); + } +}; + } // End llvm namespace #endif From resistor at mac.com Wed Oct 3 16:25:45 2007 From: resistor at mac.com (Owen Anderson) Date: Wed, 03 Oct 2007 21:25:45 -0000 Subject: [llvm-commits] [llvm] r42593 - in /llvm/trunk: include/llvm/Analysis/DominatorInternals.h include/llvm/Analysis/Dominators.h include/llvm/Analysis/PostDominators.h lib/Analysis/PostDominatorCalculation.h lib/Analysis/PostDominators.cpp lib/VMCore/DominatorCalculation.h lib/VMCore/Dominators.cpp Message-ID: <200710032125.l93LPkPH016053@zion.cs.uiuc.edu> Author: resistor Date: Wed Oct 3 16:25:45 2007 New Revision: 42593 URL: http://llvm.org/viewvc/llvm-project?rev=42593&view=rev Log: Completely merge the implementation details of DomTree and PostDomTree. Also, add a FIXME for a bug in PostDomTree calculation I noticed while writing this, Removed: llvm/trunk/lib/Analysis/PostDominatorCalculation.h llvm/trunk/lib/VMCore/DominatorCalculation.h Modified: llvm/trunk/include/llvm/Analysis/DominatorInternals.h llvm/trunk/include/llvm/Analysis/Dominators.h llvm/trunk/include/llvm/Analysis/PostDominators.h llvm/trunk/lib/Analysis/PostDominators.cpp llvm/trunk/lib/VMCore/Dominators.cpp Modified: llvm/trunk/include/llvm/Analysis/DominatorInternals.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/DominatorInternals.h?rev=42593&r1=42592&r2=42593&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/DominatorInternals.h (original) +++ llvm/trunk/include/llvm/Analysis/DominatorInternals.h Wed Oct 3 16:25:45 2007 @@ -208,6 +208,93 @@ #endif } +template +void Calculate(DominatorTreeBase& DT, Function& F) { + // Step #1: Number blocks in depth-first order and initialize variables used + // in later stages of the algorithm. + unsigned N = 0; + for (unsigned i = 0, e = DT.Roots.size(); i != e; ++i) + N = DFSPass >(DT, DT.Roots[i], N); + + for (unsigned i = N; i >= 2; --i) { + typename GraphTraits::NodeType* W = DT.Vertex[i]; + DominatorTree::InfoRec &WInfo = DT.Info[W]; + + // Step #2: Calculate the semidominators of all vertices + for (typename GraphTraits >::ChildIteratorType CI = + GraphTraits >::child_begin(W), + E = GraphTraits >::child_end(W); CI != E; ++CI) + if (DT.Info.count(*CI)) { // Only if this predecessor is reachable! + unsigned SemiU = DT.Info[Eval >(DT, *CI)].Semi; + if (SemiU < WInfo.Semi) + WInfo.Semi = SemiU; + } + + DT.Info[DT.Vertex[WInfo.Semi]].Bucket.push_back(W); + + typename GraphTraits::NodeType* WParent = WInfo.Parent; + Link >(DT, WParent, W, WInfo); + + // Step #3: Implicitly define the immediate dominator of vertices + std::vector::NodeType*> &WParentBucket = + DT.Info[WParent].Bucket; + while (!WParentBucket.empty()) { + typename GraphTraits::NodeType* V = WParentBucket.back(); + WParentBucket.pop_back(); + typename GraphTraits::NodeType* U = + Eval >(DT, V); + DT.IDoms[V] = DT.Info[U].Semi < DT.Info[V].Semi ? U : WParent; + } + } + + // Step #4: Explicitly define the immediate dominator of each vertex + for (unsigned i = 2; i <= N; ++i) { + typename GraphTraits::NodeType* W = DT.Vertex[i]; + typename GraphTraits::NodeType*& WIDom = DT.IDoms[W]; + if (WIDom != DT.Vertex[DT.Info[W].Semi]) + WIDom = DT.IDoms[WIDom]; + } + + if (DT.Roots.empty()) return; + + // Add a node for the root. This node might be the actual root, if there is + // one exit block, or it may be the virtual exit (denoted by (BasicBlock *)0) + // which postdominates all real exits if there are multiple exit blocks. + typename GraphTraits::NodeType* Root = DT.Roots.size() == 1 ? DT.Roots[0] + : 0; + DT.DomTreeNodes[Root] = DT.RootNode = new DomTreeNode(Root, 0); + + // Loop over all of the reachable blocks in the function... + for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) + if (typename GraphTraits::NodeType* ImmDom = DT.getIDom(I)) { + // Reachable block. + DomTreeNode *BBNode = DT.DomTreeNodes[I]; + if (BBNode) continue; // Haven't calculated this node yet? + + // Get or calculate the node for the immediate dominator + DomTreeNode *IDomNode = DT.getNodeForBlock(ImmDom); + + // Add a new tree node for this BasicBlock, and link it as a child of + // IDomNode + DomTreeNode *C = new DomTreeNode(I, IDomNode); + DT.DomTreeNodes[I] = IDomNode->addChild(C); + } + + // Free temporary memory used to construct idom's + DT.IDoms.clear(); + DT.Info.clear(); + std::vector::NodeType*>().swap(DT.Vertex); + + // FIXME: This does not work on PostDomTrees. It seems likely that this is + // due to an error in the algorithm for post-dominators. This really should + // be investigated and fixed at some point. + // DT.updateDFSNumbers(); + + // Start out with the DFS numbers being invalid. Let them be computed if + // demanded. + DT.DFSInfoValid = false; +} + } #endif Modified: llvm/trunk/include/llvm/Analysis/Dominators.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/Dominators.h?rev=42593&r1=42592&r2=42593&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/Dominators.h (original) +++ llvm/trunk/include/llvm/Analysis/Dominators.h Wed Oct 3 16:25:45 2007 @@ -289,6 +289,9 @@ typename GraphT::NodeType* V, unsigned N); + template friend void Calculate(DominatorTreeBase& DT, + Function& F); + /// updateDFSNumbers - Assign In and Out numbers to the nodes while walking /// dominator tree in dfs order. void updateDFSNumbers(); @@ -325,9 +328,6 @@ /// BB is split and now it has one successor. Update dominator tree to /// reflect this change. void splitBlock(BasicBlock *BB); - -private: - friend void DTcalculate(DominatorTree& DT, Function& F); }; //===------------------------------------- Modified: llvm/trunk/include/llvm/Analysis/PostDominators.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/PostDominators.h?rev=42593&r1=42592&r2=42593&view=diff ============================================================================== --- llvm/trunk/include/llvm/Analysis/PostDominators.h (original) +++ llvm/trunk/include/llvm/Analysis/PostDominators.h Wed Oct 3 16:25:45 2007 @@ -32,8 +32,6 @@ virtual void getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesAll(); } -private: - friend void PDTcalculate(PostDominatorTree& PDT, Function &F); }; Removed: llvm/trunk/lib/Analysis/PostDominatorCalculation.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/PostDominatorCalculation.h?rev=42592&view=auto ============================================================================== --- llvm/trunk/lib/Analysis/PostDominatorCalculation.h (original) +++ llvm/trunk/lib/Analysis/PostDominatorCalculation.h (removed) @@ -1,99 +0,0 @@ -//==- PostDominatorCalculation.h - Post-Dominator Calculation ----*- C++ -*-==// -// -// The LLVM Compiler Infrastructure -// -// This file was developed by Owen Anderson and is distributed under -// the University of Illinois Open Source License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// PostDominatorTree calculation implementation. -//===----------------------------------------------------------------------===// - -#ifndef LLVM_ANALYSIS_POST_DOMINATOR_CALCULATION_H -#define LLVM_ANALYSIS_POST_DOMINATOR_CALCULATION_H - -#include "llvm/Analysis/PostDominators.h" -#include "llvm/Analysis/DominatorInternals.h" - -namespace llvm { - -void PDTcalculate(PostDominatorTree& PDT, Function &F) { - // Step #1: Number blocks in depth-first order and initialize variables used - // in later stages of the algorithm. - unsigned N = 0; - for (unsigned i = 0, e = PDT.Roots.size(); i != e; ++i) - N = DFSPass > >(PDT, PDT.Roots[i], N); - - for (unsigned i = N; i >= 2; --i) { - BasicBlock *W = PDT.Vertex[i]; - PostDominatorTree::InfoRec &WInfo = PDT.Info[W]; - - // Step #2: Calculate the semidominators of all vertices - for (succ_iterator SI = succ_begin(W), SE = succ_end(W); SI != SE; ++SI) - if (PDT.Info.count(*SI)) { // Only if this predecessor is reachable! - unsigned SemiU = - PDT.Info[Eval > >(PDT, *SI)].Semi; - if (SemiU < WInfo.Semi) - WInfo.Semi = SemiU; - } - - PDT.Info[PDT.Vertex[WInfo.Semi]].Bucket.push_back(W); - - BasicBlock *WParent = WInfo.Parent; - Link > >(PDT, WParent, W, WInfo); - - // Step #3: Implicitly define the immediate dominator of vertices - std::vector &WParentBucket = PDT.Info[WParent].Bucket; - while (!WParentBucket.empty()) { - BasicBlock *V = WParentBucket.back(); - WParentBucket.pop_back(); - BasicBlock *U = Eval > >(PDT, V); - PDT.IDoms[V] = PDT.Info[U].Semi < PDT.Info[V].Semi ? U : WParent; - } - } - - // Step #4: Explicitly define the immediate dominator of each vertex - for (unsigned i = 2; i <= N; ++i) { - BasicBlock *W = PDT.Vertex[i]; - BasicBlock *&WIDom = PDT.IDoms[W]; - if (WIDom != PDT.Vertex[PDT.Info[W].Semi]) - WIDom = PDT.IDoms[WIDom]; - } - - if (PDT.Roots.empty()) return; - - // Add a node for the root. This node might be the actual root, if there is - // one exit block, or it may be the virtual exit (denoted by (BasicBlock *)0) - // which postdominates all real exits if there are multiple exit blocks. - BasicBlock *Root = PDT.Roots.size() == 1 ? PDT.Roots[0] : 0; - PDT.DomTreeNodes[Root] = PDT.RootNode = new DomTreeNode(Root, 0); - - // Loop over all of the reachable blocks in the function... - for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) - if (BasicBlock *ImmPostDom = PDT.getIDom(I)) { // Reachable block. - DomTreeNode *&BBNode = PDT.DomTreeNodes[I]; - if (!BBNode) { // Haven't calculated this node yet? - // Get or calculate the node for the immediate dominator - DomTreeNode *IPDomNode = PDT.getNodeForBlock(ImmPostDom); - - // Add a new tree node for this BasicBlock, and link it as a child of - // IDomNode - DomTreeNode *C = new DomTreeNode(I, IPDomNode); - PDT.DomTreeNodes[I] = C; - BBNode = IPDomNode->addChild(C); - } - } - - // Free temporary memory used to construct idom's - PDT.IDoms.clear(); - PDT.Info.clear(); - std::vector().swap(PDT.Vertex); - - // Start out with the DFS numbers being invalid. Let them be computed if - // demanded. - PDT.DFSInfoValid = false; -} - -} -#endif Modified: llvm/trunk/lib/Analysis/PostDominators.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/PostDominators.cpp?rev=42593&r1=42592&r2=42593&view=diff ============================================================================== --- llvm/trunk/lib/Analysis/PostDominators.cpp (original) +++ llvm/trunk/lib/Analysis/PostDominators.cpp Wed Oct 3 16:25:45 2007 @@ -16,7 +16,7 @@ #include "llvm/Support/CFG.h" #include "llvm/ADT/DepthFirstIterator.h" #include "llvm/ADT/SetOperations.h" -#include "PostDominatorCalculation.h" +#include "llvm/Analysis/DominatorInternals.h" using namespace llvm; //===----------------------------------------------------------------------===// @@ -47,7 +47,7 @@ Vertex.push_back(0); - PDTcalculate(*this, F); + Calculate >(*this, F); return false; } Removed: llvm/trunk/lib/VMCore/DominatorCalculation.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/DominatorCalculation.h?rev=42592&view=auto ============================================================================== --- llvm/trunk/lib/VMCore/DominatorCalculation.h (original) +++ llvm/trunk/lib/VMCore/DominatorCalculation.h (removed) @@ -1,106 +0,0 @@ -//==- DominatorCalculation.h - Dominator Calculation -------------*- C++ -*-==// -// -// The LLVM Compiler Infrastructure -// -// This file was developed by Owen Anderson and is distributed under -// the University of Illinois Open Source License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_VMCORE_DOMINATOR_CALCULATION_H -#define LLVM_VMCORE_DOMINATOR_CALCULATION_H - -#include "llvm/Analysis/Dominators.h" -#include "llvm/Analysis/DominatorInternals.h" - -//===----------------------------------------------------------------------===// -// -// DominatorTree construction - This pass constructs immediate dominator -// information for a flow-graph based on the algorithm described in this -// document: -// -// A Fast Algorithm for Finding Dominators in a Flowgraph -// T. Lengauer & R. Tarjan, ACM TOPLAS July 1979, pgs 121-141. -// -// This implements both the O(n*ack(n)) and the O(n*log(n)) versions of EVAL and -// LINK, but it turns out that the theoretically slower O(n*log(n)) -// implementation is actually faster than the "efficient" algorithm (even for -// large CFGs) because the constant overheads are substantially smaller. The -// lower-complexity version can be enabled with the following #define: -// -#define BALANCE_IDOM_TREE 0 -// -//===----------------------------------------------------------------------===// - -namespace llvm { - -void DTcalculate(DominatorTree& DT, Function &F) { - BasicBlock* Root = DT.Roots[0]; - - // Add a node for the root... - DT.DomTreeNodes[Root] = DT.RootNode = new DomTreeNode(Root, 0); - - // Step #1: Number blocks in depth-first order and initialize variables used - // in later stages of the algorithm. - unsigned N = DFSPass >(DT, Root, 0); - - for (unsigned i = N; i >= 2; --i) { - BasicBlock *W = DT.Vertex[i]; - DominatorTree::InfoRec &WInfo = DT.Info[W]; - - // Step #2: Calculate the semidominators of all vertices - for (pred_iterator PI = pred_begin(W), E = pred_end(W); PI != E; ++PI) - if (DT.Info.count(*PI)) { // Only if this predecessor is reachable! - unsigned SemiU = DT.Info[Eval >(DT, *PI)].Semi; - if (SemiU < WInfo.Semi) - WInfo.Semi = SemiU; - } - - DT.Info[DT.Vertex[WInfo.Semi]].Bucket.push_back(W); - - BasicBlock *WParent = WInfo.Parent; - Link >(DT, WParent, W, WInfo); - - // Step #3: Implicitly define the immediate dominator of vertices - std::vector &WParentBucket = DT.Info[WParent].Bucket; - while (!WParentBucket.empty()) { - BasicBlock *V = WParentBucket.back(); - WParentBucket.pop_back(); - BasicBlock *U = Eval >(DT, V); - DT.IDoms[V] = DT.Info[U].Semi < DT.Info[V].Semi ? U : WParent; - } - } - - // Step #4: Explicitly define the immediate dominator of each vertex - for (unsigned i = 2; i <= N; ++i) { - BasicBlock *W = DT.Vertex[i]; - BasicBlock *&WIDom = DT.IDoms[W]; - if (WIDom != DT.Vertex[DT.Info[W].Semi]) - WIDom = DT.IDoms[WIDom]; - } - - // Loop over all of the reachable blocks in the function... - for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) - if (BasicBlock *ImmDom = DT.getIDom(I)) { // Reachable block. - DomTreeNode *BBNode = DT.DomTreeNodes[I]; - if (BBNode) continue; // Haven't calculated this node yet? - - // Get or calculate the node for the immediate dominator - DomTreeNode *IDomNode = DT.getNodeForBlock(ImmDom); - - // Add a new tree node for this BasicBlock, and link it as a child of - // IDomNode - DomTreeNode *C = new DomTreeNode(I, IDomNode); - DT.DomTreeNodes[I] = IDomNode->addChild(C); - } - - // Free temporary memory used to construct idom's - DT.Info.clear(); - DT.IDoms.clear(); - std::vector().swap(DT.Vertex); - - DT.updateDFSNumbers(); -} - -} -#endif Modified: llvm/trunk/lib/VMCore/Dominators.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Dominators.cpp?rev=42593&r1=42592&r2=42593&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Dominators.cpp (original) +++ llvm/trunk/lib/VMCore/Dominators.cpp Wed Oct 3 16:25:45 2007 @@ -21,9 +21,9 @@ #include "llvm/ADT/SetOperations.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallVector.h" +#include "llvm/Analysis/DominatorInternals.h" #include "llvm/Instructions.h" #include "llvm/Support/Streams.h" -#include "DominatorCalculation.h" #include using namespace llvm; @@ -357,7 +357,10 @@ DomTreeNodes[&F.getEntryBlock()] = 0; Vertex.push_back(0); - DTcalculate(*this, F); + Calculate(*this, F); + + updateDFSNumbers(); + return false; } From isanbard at gmail.com Wed Oct 3 17:02:01 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 03 Oct 2007 22:02:01 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42595 - /llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Message-ID: <200710032202.l93M21LR017742@zion.cs.uiuc.edu> Author: void Date: Wed Oct 3 17:02:00 2007 New Revision: 42595 URL: http://llvm.org/viewvc/llvm-project?rev=42595&view=rev Log: LLVM needs a pointer type instead of "i32" for a null pointer. Also, Objective-C++ is picky when it comes to converting a "void*" to another pointer type. Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/objc/objc-act.c?rev=42595&r1=42594&r2=42595&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Wed Oct 3 17:02:00 2007 @@ -2566,7 +2566,12 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2576,7 +2581,12 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2586,7 +2596,12 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2667,7 +2682,12 @@ /* struct _prop_list_t *properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, @@ -4972,9 +4992,13 @@ build_int_cst (long_integer_type_node, 0)); /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ - if (flag_next_runtime || ! sel_ref_chain) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (objc_selector_type), + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, @@ -9229,13 +9253,16 @@ /* static struct objc_protocol _OBJC_PROTOCOL_; */ protocol_name_expr = add_objc_string (PROTOCOL_NAME (p), class_names); + /* APPLE LOCAL LLVM - begin NUL pointer */ if (refs_decl) - refs_expr = convert (build_pointer_type (build_pointer_type - (objc_protocol_template)), - build_unary_op (ADDR_EXPR, refs_decl, 0)); + refs_expr = + convert (build_pointer_type (build_pointer_type (objc_protocol_template)), + build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); - + refs_expr = + convert (build_pointer_type (build_pointer_type (objc_protocol_template)), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ save_objc_implementation_context = objc_implementation_context; @@ -9301,10 +9328,16 @@ /* APPLE LOCAL begin ObjC new abi */ /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ /* APPLE LOCAL begin radar 4533974 - ObjC newprotocol - radar 4695109 */ + /* APPLE LOCAL LLVM - begin NUL pointer */ if (newabi) { + if (!objc_protocol_extension_template) + build_objc_protocol_extension_template (); + /* 'isa' is NULL in the new ObjC abi */ - expr = build_int_cst (NULL_TREE, 0); + expr = + convert (build_pointer_type (objc_protocol_extension_template), + build_int_cst (NULL_TREE, 0)); } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9312,11 +9345,20 @@ else { if (!objc_protocol_or_opt_ins_meth) - expr = build_int_cst (NULL_TREE, 0); + { + if (!objc_protocol_extension_template) + build_objc_protocol_extension_template (); + + expr = + convert (build_pointer_type (objc_protocol_extension_template), + build_int_cst (NULL_TREE, 0)); + } else - expr = convert (build_pointer_type (objc_protocol_extension_template), - build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); + expr = convert (build_pointer_type (objc_protocol_extension_template), + build_unary_op (ADDR_EXPR, + objc_protocol_or_opt_ins_meth, 0)); } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ initlist = tree_cons (NULL_TREE, expr, initlist); @@ -9326,23 +9368,33 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, instance_methods, 0)); + build_unary_op (ADDR_EXPR, instance_methods, 0)); /* APPLE LOCAL end mainline */ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, class_methods, 0)); + build_unary_op (ADDR_EXPR, class_methods, 0)); /* APPLE LOCAL end mainline */ initlist = tree_cons (NULL_TREE, expr, initlist); } @@ -9352,31 +9404,49 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_method_proto_list_ptr, + build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + if (!opt_cls_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, opt_cls_meth, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_method_proto_list_ptr, + build_unary_op (ADDR_EXPR, opt_cls_meth, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + /* APPLE LOCAL end radar 4695109 */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_prop_list_ptr, + build_unary_op (ADDR_EXPR, property_list, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } } + /* APPLE LOCAL end C* property metadata (Radar 4498373) */ return objc_build_constructor (type, nreverse (initlist)); } @@ -10389,7 +10459,12 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - begin NUL pointer */ + ivar = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + ivar); + /* APPLE LOCAL LLVM - end NUL pointer */ /* Set type */ encode_field_decl (field_decl, @@ -10446,7 +10521,12 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - begin NUL pointer */ + ivar = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + ivar); + /* APPLE LOCAL LLVM - end NUL pointer */ /* Set type. */ encode_field_decl (field_decl, @@ -11142,7 +11222,12 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ @@ -11152,7 +11237,12 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ @@ -11163,18 +11253,26 @@ } /* protocol_list = */ - if (!protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ - tree protocol_list_ptr = - (abi_v2) ? build_pointer_type (objc_v2_protocol_template) - : build_pointer_type (build_pointer_type (objc_protocol_template)); - expr = convert (protocol_list_ptr, build_unary_op (ADDR_EXPR, protocol_list, 0)); - /* APPLE LOCAL end radar 4533974 - ObjC new protocol */ - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree protocol_list_ptr = + (abi_v2) ? build_pointer_type (objc_v2_protocol_template) + : build_pointer_type (build_pointer_type (objc_protocol_template)); + + if (!protocol_list) + initlist = tree_cons (NULL_TREE, + convert (protocol_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ + expr = convert (protocol_list_ptr, build_unary_op (ADDR_EXPR, protocol_list, 0)); + /* APPLE LOCAL end radar 4533974 - ObjC new protocol */ + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ if (!abi_v2) { @@ -11186,7 +11284,12 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, @@ -11253,7 +11356,12 @@ /* objc_ivar_list = */ if (!ivar_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_ivar_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ @@ -11265,7 +11373,12 @@ /* objc_method_list = */ if (!dispatch_table) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { /* APPLE LOCAL begin mainline */ @@ -11275,60 +11388,109 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } + /* APPLE LOCAL LLVM - begin NUL pointer */ if (flag_next_runtime) /* method_cache = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type + (xref_tag (RECORD_TYPE, + get_identifier + ("objc_cache"))), + build_int_cst (NULL_TREE, 0)), + initlist); else { + tree null_objc_class_ptr = + convert (build_pointer_type (objc_class_template), + build_int_cst (NULL_TREE, 0)); + /* dtable = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (xref_tag (RECORD_TYPE, + get_identifier + ("sarray"))), + build_int_cst (NULL_TREE, 0)), + initlist); /* subclass_list = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, null_objc_class_ptr, initlist); /* sibling_class = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, null_objc_class_ptr, initlist); } + /* APPLE LOCAL LLVM - end NUL pointer */ /* protocol_list = */ - if (! protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - expr = convert (build_pointer_type - (build_pointer_type - (objc_protocol_template)), - build_unary_op (ADDR_EXPR, protocol_list, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree objc_protocol_ptr_ptr = + build_pointer_type (build_pointer_type (objc_protocol_template)); + + if (!protocol_list) + initlist = tree_cons (NULL_TREE, + convert (objc_protocol_ptr_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (objc_protocol_ptr_ptr, + build_unary_op (ADDR_EXPR, protocol_list, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ if (flag_next_runtime) { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ - if (!objc_class_ext) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - expr = convert (build_pointer_type (objc_class_ext_template), - build_unary_op (ADDR_EXPR, objc_class_ext, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + if (!objc_class_ext_template) + build_objc_class_ext_template (); + + { + tree objc_class_ext_ptr = + build_pointer_type (objc_class_ext_template); + + if (!objc_class_ext) + initlist = tree_cons (NULL_TREE, + convert (objc_class_ext_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (objc_class_ext_ptr, + build_unary_op (ADDR_EXPR, objc_class_ext, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ } else /* gc_object_type = NULL */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11490,13 +11652,19 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (objc_imp_type), + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11540,7 +11708,12 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11550,7 +11723,12 @@ /* baseMethods */ if (!baseMethods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, @@ -11559,32 +11737,50 @@ } /* baseProtocols */ - if (!baseProtocols) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - tree protocol_list_t_p = build_pointer_type ( - xref_tag (RECORD_TYPE, - get_identifier (UTAG_V2_PROTOCOL_LIST))); - expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree protocol_list_t_p = + build_pointer_type (xref_tag (RECORD_TYPE, + get_identifier (UTAG_V2_PROTOCOL_LIST))); + + if (!baseProtocols) + initlist = tree_cons (NULL_TREE, + convert (protocol_list_t_p, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (protocol_list_t_p, + build_unary_op (ADDR_EXPR, baseProtocols, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* ivars */ if (!ivars) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_v2_ivar_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11592,11 +11788,16 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11854,7 +12055,10 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - class_superclass_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + class_superclass_expr = convert (build_pointer_type (objc_v2_class_template), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -12043,7 +12247,10 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - super_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + super_expr = convert (string_type_node, + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13303,7 +13510,11 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + refs_expr = + convert (build_pointer_type (objc_v2_protocol_template), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From isanbard at gmail.com Wed Oct 3 17:20:58 2007 From: isanbard at gmail.com (Bill Wendling) Date: Wed, 03 Oct 2007 22:20:58 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42597 - /llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Message-ID: <200710032220.l93MKwXc018734@zion.cs.uiuc.edu> Author: void Date: Wed Oct 3 17:20:58 2007 New Revision: 42597 URL: http://llvm.org/viewvc/llvm-project?rev=42597&view=rev Log: LLVM needs a pointer type instead of "i32" for a null pointer. Also, Objective-C++ is picky when it comes to converting a "void*" to another pointer type. Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c?rev=42597&r1=42596&r2=42597&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Wed Oct 3 17:20:58 2007 @@ -2576,7 +2576,12 @@ /* struct objc_method_list *optional_instance_methods; */ if (!optional_instance_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2586,7 +2591,12 @@ /* struct objc_method_list *optional_class_methods; */ if (!optional_class_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, build_unary_op (ADDR_EXPR, @@ -2596,7 +2606,12 @@ /* struct objc_prop_list *instance_properties; */ if (!instance_prop) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, build_unary_op (ADDR_EXPR, @@ -2677,7 +2692,12 @@ /* struct _prop_list_t *properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, @@ -4869,9 +4889,13 @@ build_int_cst (long_integer_type_node, 0)); /* refs = { ..., _OBJC_SELECTOR_TABLE, ... } */ - if (flag_next_runtime || ! sel_ref_chain) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (objc_selector_type), + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, @@ -9129,13 +9153,16 @@ /* static struct objc_protocol _OBJC_PROTOCOL_; */ protocol_name_expr = add_objc_string (PROTOCOL_NAME (p), class_names); + /* APPLE LOCAL LLVM - begin NUL pointer */ if (refs_decl) - refs_expr = convert (build_pointer_type (build_pointer_type - (objc_protocol_template)), - build_unary_op (ADDR_EXPR, refs_decl, 0)); + refs_expr = + convert (build_pointer_type (build_pointer_type (objc_protocol_template)), + build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); - + refs_expr = + convert (build_pointer_type (build_pointer_type (objc_protocol_template)), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ save_objc_implementation_context = objc_implementation_context; @@ -9201,10 +9228,16 @@ /* APPLE LOCAL begin ObjC new abi */ /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ /* APPLE LOCAL begin radar 4533974 - ObjC newprotocol - radar 4695109 */ + /* APPLE LOCAL LLVM - begin NUL pointer */ if (newabi) { + if (!objc_protocol_extension_template) + build_objc_protocol_extension_template (); + /* 'isa' is NULL in the new ObjC abi */ - expr = build_int_cst (NULL_TREE, 0); + expr = + convert (build_pointer_type (objc_protocol_extension_template), + build_int_cst (NULL_TREE, 0)); } /* APPLE LOCAL end radar 4533974 - ObjC newprotocol - radar 4695109 */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ @@ -9212,11 +9245,20 @@ else { if (!objc_protocol_or_opt_ins_meth) - expr = build_int_cst (NULL_TREE, 0); + { + if (!objc_protocol_extension_template) + build_objc_protocol_extension_template (); + + expr = + convert (build_pointer_type (objc_protocol_extension_template), + build_int_cst (NULL_TREE, 0)); + } else - expr = convert (build_pointer_type (objc_protocol_extension_template), - build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); + expr = convert (build_pointer_type (objc_protocol_extension_template), + build_unary_op (ADDR_EXPR, + objc_protocol_or_opt_ins_meth, 0)); } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ initlist = tree_cons (NULL_TREE, expr, initlist); @@ -9226,7 +9268,12 @@ initlist = tree_cons (NULL_TREE, protocol_list, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_proto_list_ptr, @@ -9235,7 +9282,12 @@ } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_proto_list_ptr, @@ -9248,30 +9300,45 @@ { /* APPLE LOCAL begin radar 4695109 */ if (!objc_protocol_or_opt_ins_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_method_proto_list_ptr, + build_unary_op (ADDR_EXPR, objc_protocol_or_opt_ins_meth, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } if (!opt_cls_meth) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_proto_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_method_proto_list_ptr, - build_unary_op (ADDR_EXPR, opt_cls_meth, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_method_proto_list_ptr, + build_unary_op (ADDR_EXPR, opt_cls_meth, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } /* APPLE LOCAL end radar 4695109 */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else - { - expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + { + expr = convert (objc_prop_list_ptr, + build_unary_op (ADDR_EXPR, property_list, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } /* APPLE LOCAL begin radar 5192466 */ /* const uint32_t size; = sizeof(struct protocol_t) */ expr = build_int_cst ( @@ -10291,7 +10358,12 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - begin NUL pointer */ + ivar = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + ivar); + /* APPLE LOCAL LLVM - end NUL pointer */ /* Set type */ encode_field_decl (field_decl, @@ -10348,7 +10420,12 @@ ivar); else /* Unnamed bit-field ivar (yuck). */ - ivar = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), ivar); + /* APPLE LOCAL LLVM - begin NUL pointer */ + ivar = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + ivar); + /* APPLE LOCAL LLVM - end NUL pointer */ /* Set type. */ encode_field_decl (field_decl, @@ -11042,7 +11119,12 @@ initlist = tree_cons (NULL_TREE, class_name, initlist); if (!instance_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, @@ -11050,7 +11132,12 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } if (!class_methods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, @@ -11059,18 +11146,26 @@ } /* protocol_list = */ - if (!protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ - tree protocol_list_ptr = - (abi_v2) ? build_pointer_type (objc_v2_protocol_template) - : build_pointer_type (build_pointer_type (objc_protocol_template)); - expr = convert (protocol_list_ptr, build_unary_op (ADDR_EXPR, protocol_list, 0)); - /* APPLE LOCAL end radar 4533974 - ObjC new protocol */ - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree protocol_list_ptr = + (abi_v2) ? build_pointer_type (objc_v2_protocol_template) + : build_pointer_type (build_pointer_type (objc_protocol_template)); + + if (!protocol_list) + initlist = tree_cons (NULL_TREE, + convert (protocol_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + /* APPLE LOCAL begin radar 4533974 - ObjC new protocol */ + expr = convert (protocol_list_ptr, build_unary_op (ADDR_EXPR, protocol_list, 0)); + /* APPLE LOCAL end radar 4533974 - ObjC new protocol */ + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ if (!abi_v2) { @@ -11082,7 +11177,12 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* struct _objc_property_list *instance_properties; */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, @@ -11149,7 +11249,12 @@ /* objc_ivar_list = */ if (!ivar_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_ivar_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_ivar_list_ptr, @@ -11159,7 +11264,12 @@ /* objc_method_list = */ if (!dispatch_table) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, @@ -11167,60 +11277,109 @@ initlist = tree_cons (NULL_TREE, expr, initlist); } + /* APPLE LOCAL LLVM - begin NUL pointer */ if (flag_next_runtime) /* method_cache = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type + (xref_tag (RECORD_TYPE, + get_identifier + ("objc_cache"))), + build_int_cst (NULL_TREE, 0)), + initlist); else { + tree null_objc_class_ptr = + convert (build_pointer_type (objc_class_template), + build_int_cst (NULL_TREE, 0)); + /* dtable = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (xref_tag (RECORD_TYPE, + get_identifier + ("sarray"))), + build_int_cst (NULL_TREE, 0)), + initlist); + /* subclass_list = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - + initlist = tree_cons (NULL_TREE, null_objc_class_ptr, initlist); + /* sibling_class = */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + initlist = tree_cons (NULL_TREE, null_objc_class_ptr, initlist); } - + /* APPLE LOCAL LLVM - end NUL pointer */ + /* protocol_list = */ - if (! protocol_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - expr = convert (build_pointer_type - (build_pointer_type - (objc_protocol_template)), - build_unary_op (ADDR_EXPR, protocol_list, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree objc_protocol_ptr_ptr = + build_pointer_type (build_pointer_type (objc_protocol_template)); + + if (!protocol_list) + initlist = tree_cons (NULL_TREE, + convert (objc_protocol_ptr_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (objc_protocol_ptr_ptr, + build_unary_op (ADDR_EXPR, protocol_list, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4585769 - Objective-C 1.0 extensions */ if (flag_next_runtime) { /* const char *ivar_layout; */ if (IS_CLS_META (status)) /* Meta Class ? */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { tree ivar_layout = objc_build_ivar_layout (true); if (!ivar_layout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, ivar_layout, initlist); } /* struct _objc_class_ext *ext; */ - if (!objc_class_ext) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - expr = convert (build_pointer_type (objc_class_ext_template), - build_unary_op (ADDR_EXPR, objc_class_ext, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + if (!objc_class_ext_template) + build_objc_class_ext_template (); + + { + tree objc_class_ext_ptr = + build_pointer_type (objc_class_ext_template); + + if (!objc_class_ext) + initlist = tree_cons (NULL_TREE, + convert (objc_class_ext_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (objc_class_ext_ptr, + build_unary_op (ADDR_EXPR, objc_class_ext, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ } else /* gc_object_type = NULL */ - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* APPLE LOCAL end radar 4585769 - Objective-C 1.0 extensions */ return objc_build_constructor (type, nreverse (initlist)); @@ -11382,13 +11541,19 @@ if (cache) initlist = tree_cons (NULL_TREE, cache, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - NUL pointer */ + initlist = tree_cons (NULL_TREE, null_pointer_node, initlist); /* vtable */ if (vtable) initlist = tree_cons (NULL_TREE, vtable, initlist); else - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (build_pointer_type (objc_imp_type), + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ /* ro */ initlist = tree_cons (NULL_TREE, ro, initlist); @@ -11432,7 +11597,12 @@ /* APPLE LOCAL begin radar 4695101 */ /* ivarLayout */ if (!ivarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, ivarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11442,7 +11612,12 @@ /* baseMethods */ if (!baseMethods) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_method_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_method_list_ptr, @@ -11451,32 +11626,50 @@ } /* baseProtocols */ - if (!baseProtocols) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); - else - { - tree protocol_list_t_p = build_pointer_type ( - xref_tag (RECORD_TYPE, - get_identifier (UTAG_V2_PROTOCOL_LIST))); - expr = convert (protocol_list_t_p, - build_unary_op (ADDR_EXPR, baseProtocols, 0)); - initlist = tree_cons (NULL_TREE, expr, initlist); - } + /* APPLE LOCAL LLVM - begin NUL pointer */ + { + tree protocol_list_t_p = + build_pointer_type (xref_tag (RECORD_TYPE, + get_identifier (UTAG_V2_PROTOCOL_LIST))); + + if (!baseProtocols) + initlist = tree_cons (NULL_TREE, + convert (protocol_list_t_p, + build_int_cst (NULL_TREE, 0)), + initlist); + else + { + expr = convert (protocol_list_t_p, + build_unary_op (ADDR_EXPR, baseProtocols, 0)); + initlist = tree_cons (NULL_TREE, expr, initlist); + } + } + /* APPLE LOCAL LLVM - end NUL pointer */ /* ivars */ if (!ivars) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_v2_ivar_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_v2_ivar_list_ptr, - build_unary_op (ADDR_EXPR, ivars, 0)); + build_unary_op (ADDR_EXPR, ivars, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL begin radar 4695101 */ /* weakIvarLayout */ if (!weakIvarLayout) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (string_type_node, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else initlist = tree_cons (NULL_TREE, weakIvarLayout, initlist); /* APPLE LOCAL end radar 4695101 */ @@ -11484,11 +11677,16 @@ /* APPLE LOCAL begin C* property metadata (Radar 4498373) */ /* property list */ if (!property_list) - initlist = tree_cons (NULL_TREE, build_int_cst (NULL_TREE, 0), initlist); + /* APPLE LOCAL LLVM - begin NUL pointer */ + initlist = tree_cons (NULL_TREE, + convert (objc_prop_list_ptr, + build_int_cst (NULL_TREE, 0)), + initlist); + /* APPLE LOCAL LLVM - end NUL pointer */ else { expr = convert (objc_prop_list_ptr, - build_unary_op (ADDR_EXPR, property_list, 0)); + build_unary_op (ADDR_EXPR, property_list, 0)); initlist = tree_cons (NULL_TREE, expr, initlist); } /* APPLE LOCAL end C* property metadata (Radar 4498373) */ @@ -11735,7 +11933,10 @@ /* root class. */ root_expr = build_unary_op (ADDR_EXPR, metaclass_decl, 0); metaclass_superclass_expr = build_unary_op (ADDR_EXPR, class_decl, 0); - class_superclass_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + class_superclass_expr = convert (build_pointer_type (objc_v2_class_template), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ flags |= 0x2; /* CLS_ROOT: it is also a root meta class */ } @@ -11927,7 +12128,10 @@ super_expr = build_c_cast (cast_type, super_expr); /* cast! */ } else - super_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + super_expr = convert (string_type_node, + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ root_expr = add_objc_string (my_root_id, class_names); root_expr = build_c_cast (cast_type, root_expr); /* cast! */ @@ -13190,7 +13394,11 @@ refs_expr = convert (build_pointer_type (objc_v2_protocol_template), build_unary_op (ADDR_EXPR, refs_decl, 0)); else - refs_expr = build_int_cst (NULL_TREE, 0); + /* APPLE LOCAL LLVM - begin NUL pointer */ + refs_expr = + convert (build_pointer_type (objc_v2_protocol_template), + build_int_cst (NULL_TREE, 0)); + /* APPLE LOCAL LLVM - end NUL pointer */ /* APPLE LOCAL begin radar 4695101 */ /* Build table of list of properties for this protocol. */ From neil at daikokuya.co.uk Wed Oct 3 17:26:02 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Wed, 03 Oct 2007 22:26:02 -0000 Subject: [llvm-commits] [llvm] r42598 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp Message-ID: <200710032226.l93MQ25v019027@zion.cs.uiuc.edu> Author: neil Date: Wed Oct 3 17:26:02 2007 New Revision: 42598 URL: http://llvm.org/viewvc/llvm-project?rev=42598&view=rev Log: Add APFloat -> hexadecimal string conversion, as per %a and %A in C99. Useful for diagnostics and debugging. Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42598&r1=42597&r2=42598&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Wed Oct 3 17:26:02 2007 @@ -13,10 +13,10 @@ //===----------------------------------------------------------------------===// /* A self-contained host- and target-independent arbitrary-precision - floating-point software implementation using bignum integer - arithmetic, as provided by static functions in the APInt class. + floating-point software implementation. It uses bignum integer + arithmetic as provided by static functions in the APInt class. The library will work with bignum integers whose parts are any - unsigned type at least 16 bits wide. 64 bits is recommended. + unsigned type at least 16 bits wide, but 64 bits is recommended. Written for clarity rather than speed, in particular with a view to use in the front-end of a cross compiler so that target @@ -30,10 +30,7 @@ are add, subtract, multiply, divide, fused-multiply-add, conversion-to-float, conversion-to-integer and conversion-from-integer. New rounding modes (e.g. away from zero) - can be added with three or four lines of code. The library reads - and correctly rounds hexadecimal floating point numbers as per - C99; syntax is required to have been validated by the caller. - Conversion from decimal is not currently implemented. + can be added with three or four lines of code. Four formats are built-in: IEEE single precision, double precision, quadruple precision, and x87 80-bit extended double @@ -54,6 +51,17 @@ should be straight forward to add support for the before-rounding case too. + The library reads hexadecimal floating point numbers as per C99, + and correctly rounds if necessary according to the specified + rounding mode. Syntax is required to have been validated by the + caller. It also converts floating point numbers to hexadecimal + text as per the C99 %a and %A conversions. The output precision + (or alternatively the natural minimal precision) can be specified; + if the requested precision is less than the natural precision the + output is correctly rounded for the specified rounding mode. + + Conversion to and from decimal text is not currently implemented. + Non-zero finite numbers are represented internally as a sign bit, a 16-bit signed exponent, and the significand as an array of integer parts. After normalization of a number of precision P the @@ -77,17 +85,14 @@ Conversions to and from decimal strings (hard). - Conversions to hexadecimal string. - - Read and write IEEE-format in-memory representations. - Optional ability to detect underflow tininess before rounding. New formats: x87 in single and double precision mode (IEEE apart from extended exponent range) and IBM two-double extended precision (hard). - New operations: sqrt, nextafter, nexttoward. + New operations: sqrt, IEEE remainder, C90 fmod, nextafter, + nexttoward. */ #ifndef LLVM_FLOAT_H @@ -205,6 +210,13 @@ compare unordered, 0==-0). */ cmpResult compare(const APFloat &) const; + /* Write out a hexadecimal representation of the floating point + value to DST, which must be of sufficient size, in the C99 form + [-]0xh.hhhhp[+-]d. Return the number of characters written, + excluding the terminating NUL. */ + unsigned int convertToHexString(char *dst, unsigned int hexDigits, + bool upperCase, roundingMode) const; + /* Bitwise comparison for equality (QNaNs compare equal, 0!=-0). */ bool bitwiseIsEqual(const APFloat &) const; @@ -255,9 +267,11 @@ opStatus handleOverflow(roundingMode); bool roundAwayFromZero(roundingMode, lostFraction, unsigned int) const; opStatus convertFromUnsignedInteger(integerPart *, unsigned int, - roundingMode); + roundingMode); lostFraction combineLostFractions(lostFraction, lostFraction); opStatus convertFromHexadecimalString(const char *, roundingMode); + char *convertNormalToHexString(char *, unsigned int, bool, + roundingMode) const; APInt convertFloatAPFloatToAPInt() const; APInt convertDoubleAPFloatToAPInt() const; APInt convertF80LongDoubleAPFloatToAPInt() const; Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42598&r1=42597&r2=42598&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Wed Oct 3 17:26:02 2007 @@ -13,6 +13,7 @@ //===----------------------------------------------------------------------===// #include +#include #include "llvm/ADT/APFloat.h" #include "llvm/Support/MathExtras.h" @@ -20,7 +21,8 @@ #define convolve(lhs, rhs) ((lhs) * 4 + (rhs)) -/* Assumed in hexadecimal significand parsing. */ +/* Assumed in hexadecimal significand parsing, and conversion to + hexadecimal strings. */ COMPILE_TIME_ASSERT(integerPartWidth % 4 == 0); namespace llvm { @@ -187,7 +189,7 @@ /* Return the fraction lost were a bignum truncated losing the least significant BITS bits. */ lostFraction - lostFractionThroughTruncation(integerPart *parts, + lostFractionThroughTruncation(const integerPart *parts, unsigned int partCount, unsigned int bits) { @@ -219,6 +221,66 @@ return lost_fraction; } + + + /* Zero at the end to avoid modular arithmetic when adding one; used + when rounding up during hexadecimal output. */ + static const char hexDigitsLower[] = "0123456789abcdef0"; + static const char hexDigitsUpper[] = "0123456789ABCDEF0"; + static const char infinityL[] = "infinity"; + static const char infinityU[] = "INFINITY"; + static const char NaNL[] = "nan"; + static const char NaNU[] = "NAN"; + + /* Write out an integerPart in hexadecimal, starting with the most + significant nibble. Write out exactly COUNT hexdigits, return + COUNT. */ + static unsigned int + partAsHex (char *dst, integerPart part, unsigned int count, + const char *hexDigitChars) + { + unsigned int result = count; + + assert (count != 0 && count <= integerPartWidth / 4); + + part >>= (integerPartWidth - 4 * count); + while (count--) { + dst[count] = hexDigitChars[part & 0xf]; + part >>= 4; + } + + return result; + } + + /* Write out a decimal exponent. */ + static char * + writeDecimalExponent (char *dst, int exponent) + { + assert (exponent >= -65536 && exponent <= 65535); + + if (exponent < 0) { + *dst++ = '-'; + exponent = -exponent; + } + + if (exponent == 0) { + *dst++ = '0'; + } else { + char buff[12], *p; + + p = buff; + while (exponent) { + *p++ = '0' + exponent % 10; + exponent /= 10; + } + + do + *dst++ = *--p; + while (p != buff); + } + + return dst; + } } /* Constructors. */ @@ -1167,7 +1229,7 @@ return fs; } -/* Normalized remainder. */ +/* Normalized remainder. This is not currently doing TRT. */ APFloat::opStatus APFloat::mod(const APFloat &rhs, roundingMode rounding_mode) { @@ -1630,6 +1692,181 @@ abort(); } +/* Write out a hexadecimal representation of the floating point value + to DST, which must be of sufficient size, in the C99 form + [-]0xh.hhhhp[+-]d. Return the number of characters written, + excluding the terminating NUL. + + If UPPERCASE, the output is in upper case, otherwise in lower case. + + HEXDIGITS digits appear altogether, rounding the value if + necessary. If HEXDIGITS is 0, the minimal precision to display the + number precisely is used instead. If nothing would appear after + the decimal point it is suppressed. + + The decimal exponent is always printed and has at least one digit. + Zero values display an exponent of zero. Infinities and NaNs + appear as "infinity" or "nan" respectively. + + The above rules are as specified by C99. There is ambiguity about + what the leading hexadecimal digit should be. This implementation + uses whatever is necessary so that the exponent is displayed as + stored. This implies the exponent will fall within the IEEE format + range, and the leading hexadecimal digit will be 0 (for denormals), + 1 (normal numbers) or 2 (normal numbers rounded-away-from-zero with + any other digits zero). +*/ +unsigned int +APFloat::convertToHexString(char *dst, unsigned int hexDigits, + bool upperCase, roundingMode rounding_mode) const +{ + char *p; + + p = dst; + if (sign) + *dst++ = '-'; + + switch (category) { + case fcInfinity: + memcpy (dst, upperCase ? infinityU: infinityL, sizeof infinityU - 1); + dst += sizeof infinityL - 1; + break; + + case fcNaN: + memcpy (dst, upperCase ? NaNU: NaNL, sizeof NaNU - 1); + dst += sizeof NaNU - 1; + break; + + case fcZero: + *dst++ = '0'; + *dst++ = upperCase ? 'X': 'x'; + *dst++ = '0'; + if (hexDigits > 1) { + *dst++ = '.'; + memset (dst, '0', hexDigits - 1); + dst += hexDigits - 1; + } + *dst++ = upperCase ? 'P': 'p'; + *dst++ = '0'; + break; + + case fcNormal: + dst = convertNormalToHexString (dst, hexDigits, upperCase, rounding_mode); + break; + } + + *dst = 0; + + return dst - p; +} + +/* Does the hard work of outputting the correctly rounded hexadecimal + form of a normal floating point number with the specified number of + hexadecimal digits. If HEXDIGITS is zero the minimum number of + digits necessary to print the value precisely is output. */ +char * +APFloat::convertNormalToHexString(char *dst, unsigned int hexDigits, + bool upperCase, + roundingMode rounding_mode) const +{ + unsigned int count, valueBits, shift, partsCount, outputDigits; + const char *hexDigitChars; + const integerPart *significand; + char *p; + bool roundUp; + + *dst++ = '0'; + *dst++ = upperCase ? 'X': 'x'; + + roundUp = false; + hexDigitChars = upperCase ? hexDigitsUpper: hexDigitsLower; + + significand = significandParts(); + partsCount = partCount(); + + /* +3 because the first digit only uses the single integer bit, so + we have 3 virtual zero most-significant-bits. */ + valueBits = semantics->precision + 3; + shift = integerPartWidth - valueBits % integerPartWidth; + + /* The natural number of digits required ignoring trailing + insignificant zeroes. */ + outputDigits = (valueBits - significandLSB () + 3) / 4; + + /* hexDigits of zero means use the required number for the + precision. Otherwise, see if we are truncating. If we are, + found out if we need to round away from zero. */ + if (hexDigits) { + if (hexDigits < outputDigits) { + /* We are dropping non-zero bits, so need to check how to round. + "bits" is the number of dropped bits. */ + unsigned int bits; + lostFraction fraction; + + bits = valueBits - hexDigits * 4; + fraction = lostFractionThroughTruncation (significand, partsCount, bits); + roundUp = roundAwayFromZero(rounding_mode, fraction, bits); + } + outputDigits = hexDigits; + } + + /* Write the digits consecutively, and start writing in the location + of the hexadecimal point. We move the most significant digit + left and add the hexadecimal point later. */ + p = ++dst; + + count = (valueBits + integerPartWidth - 1) / integerPartWidth; + + while (outputDigits && count) { + integerPart part; + + /* Put the most significant integerPartWidth bits in "part". */ + if (--count == partsCount) + part = 0; /* An imaginary higher zero part. */ + else + part = significand[count] << shift; + + if (count && shift) + part |= significand[count - 1] >> (integerPartWidth - shift); + + /* Convert as much of "part" to hexdigits as we can. */ + unsigned int curDigits = integerPartWidth / 4; + + if (curDigits > outputDigits) + curDigits = outputDigits; + dst += partAsHex (dst, part, curDigits, hexDigitChars); + outputDigits -= curDigits; + } + + if (roundUp) { + char *q = dst; + + /* Note that hexDigitChars has a trailing '0'. */ + do { + q--; + *q = hexDigitChars[hexDigitValue (*q) + 1]; + } while (*q == '0' && q > p); + } else { + /* Add trailing zeroes. */ + memset (dst, '0', outputDigits); + dst += outputDigits; + } + + /* Move the most significant digit to before the point, and if there + is something after the decimal point add it. This must come + after rounding above. */ + p[-1] = p[0]; + if (dst -1 == p) + dst--; + else + p[0] = '.'; + + /* Finally output the exponent. */ + *dst++ = upperCase ? 'P': 'p'; + + return writeDecimalExponent (dst, exponent); +} + // For good performance it is desirable for different APFloats // to produce different integers. uint32_t From gordonhenriksen at mac.com Wed Oct 3 19:07:51 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Thu, 04 Oct 2007 00:07:51 -0000 Subject: [llvm-commits] [llvm] r42599 - /llvm/trunk/bindings/ocaml/Makefile.ocaml Message-ID: <200710040007.l9407pTB022789@zion.cs.uiuc.edu> Author: gordon Date: Wed Oct 3 19:07:50 2007 New Revision: 42599 URL: http://llvm.org/viewvc/llvm-project?rev=42599&view=rev Log: Do use the actual ocaml stdlib (not the install dir) to find the caml/*.h headers. Modified: llvm/trunk/bindings/ocaml/Makefile.ocaml Modified: llvm/trunk/bindings/ocaml/Makefile.ocaml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/Makefile.ocaml?rev=42599&r1=42598&r2=42599&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/Makefile.ocaml (original) +++ llvm/trunk/bindings/ocaml/Makefile.ocaml Wed Oct 3 19:07:50 2007 @@ -14,9 +14,8 @@ include $(LEVEL)/Makefile.config -# CFLAGS needs to be set before Makefile.rules is included. Yes, ocaml puts its -# includes under its libdir. -CFLAGS += -I$(OCAML_LIBDIR) +# CFLAGS needs to be set before Makefile.rules is included. +CFLAGS += -I$(shell $(OCAMLC) -where) include $(LEVEL)/Makefile.common From sabre at nondot.org Wed Oct 3 23:40:37 2007 From: sabre at nondot.org (Chris Lattner) Date: Wed, 3 Oct 2007 23:40:37 -0500 Subject: [llvm-commits] CVS: llvm-www/Users.html Message-ID: <200710040440.l944eb27032169@zion.cs.uiuc.edu> Changes in directory llvm-www: Users.html updated: 1.21 -> 1.22 --- Log message: add an entry. --- Diffs of the changes: (+7 -1) Users.html | 8 +++++++- 1 files changed, 7 insertions(+), 1 deletion(-) Index: llvm-www/Users.html diff -u llvm-www/Users.html:1.21 llvm-www/Users.html:1.22 --- llvm-www/Users.html:1.21 Wed Jul 4 15:10:08 2007 +++ llvm-www/Users.html Wed Oct 3 23:39:57 2007 @@ -242,6 +242,12 @@ Faust Signal Processing Language Signal processing language, uses the LLVM JIT for runtime codegen. + + + + + iPhone tool chain + llvm-gcc Compiler for iPhone Dev Wiki toolchain. @@ -257,6 +263,6 @@ src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!">
    LLVM Development List
    - Last modified: $Date: 2007/07/04 20:10:08 $ + Last modified: $Date: 2007/10/04 04:39:57 $
    From dberlin at dberlin.org Thu Oct 4 00:25:51 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Thu, 04 Oct 2007 05:25:51 -0000 Subject: [llvm-commits] [llvm] r42604 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710040525.l945PpKi000954@zion.cs.uiuc.edu> Author: dannyb Date: Thu Oct 4 00:25:51 2007 New Revision: 42604 URL: http://llvm.org/viewvc/llvm-project?rev=42604&view=rev Log: Make SparseBitVector::set actually work properly when sets are not in ascending or descending order. Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42604&r1=42603&r2=42604&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Thu Oct 4 00:25:51 2007 @@ -334,9 +334,8 @@ --ElementIter; } else { while (ElementIter != Elements.end() && - ElementIter->index() <= ElementIndex) + ElementIter->index() < ElementIndex) ++ElementIter; - --ElementIter; } CurrElementIter = ElementIter; return ElementIter; @@ -536,10 +535,17 @@ if (ElementIter == Elements.end() || ElementIter->index() != ElementIndex) { Element = new SparseBitVectorElement(ElementIndex); - // Insert does insert before, and lower bound gives the one before. - ElementIter = Elements.insert(++ElementIter, Element); + // We may have hit the beginning of our SparseBitVector, in which case, + // we may need to insert right after this element, which requires moving + // the current iterator forward one, because insert does insert before. + if (ElementIter->index() < ElementIndex) + ElementIter = Elements.insert(++ElementIter, Element); + else + ElementIter = Elements.insert(ElementIter, Element); } } + CurrElementIter = ElementIter; + ElementIter->set(Idx % ElementSize); } From dberlin at dberlin.org Thu Oct 4 00:29:36 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Thu, 04 Oct 2007 05:29:36 -0000 Subject: [llvm-commits] [llvm] r42605 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710040529.l945TavG001073@zion.cs.uiuc.edu> Author: dannyb Date: Thu Oct 4 00:29:36 2007 New Revision: 42605 URL: http://llvm.org/viewvc/llvm-project?rev=42605&view=rev Log: It helps a lot when you check for the end of your list before randomly dereferencing things. Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42605&r1=42604&r2=42605&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Thu Oct 4 00:29:36 2007 @@ -538,7 +538,8 @@ // We may have hit the beginning of our SparseBitVector, in which case, // we may need to insert right after this element, which requires moving // the current iterator forward one, because insert does insert before. - if (ElementIter->index() < ElementIndex) + if (ElementIter != Elements.end() && + ElementIter->index() < ElementIndex) ElementIter = Elements.insert(++ElementIter, Element); else ElementIter = Elements.insert(ElementIter, Element); From isanbard at gmail.com Thu Oct 4 04:33:10 2007 From: isanbard at gmail.com (Bill Wendling) Date: Thu, 04 Oct 2007 09:33:10 -0000 Subject: [llvm-commits] [llvm] r42606 - /llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Message-ID: <200710040933.l949XB8X015933@zion.cs.uiuc.edu> Author: void Date: Thu Oct 4 04:33:08 2007 New Revision: 42606 URL: http://llvm.org/viewvc/llvm-project?rev=42606&view=rev Log: un-XFAIL this test. Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Modified: llvm/trunk/test/FrontendObjC/2007-09-25-EH.m URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/2007-09-25-EH.m?rev=42606&r1=42605&r2=42606&view=diff ============================================================================== --- llvm/trunk/test/FrontendObjC/2007-09-25-EH.m (original) +++ llvm/trunk/test/FrontendObjC/2007-09-25-EH.m Thu Oct 4 04:33:08 2007 @@ -1,6 +1,4 @@ // RUN: %llvmgcc -c -w -m64 -mmacosx-version-min=10.5 %s -o /dev/null -// XFAIL: * -// Reenable when Obj-C FE problem is fixed. @class NSDictionary, DSoBuffer, DSoDirectory, NSMutableArray; @interface NSException {} From sabre at nondot.org Thu Oct 4 10:47:28 2007 From: sabre at nondot.org (Chris Lattner) Date: Thu, 04 Oct 2007 15:47:28 -0000 Subject: [llvm-commits] [llvm] r42607 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710041547.l94FlSFa028162@zion.cs.uiuc.edu> Author: lattner Date: Thu Oct 4 10:47:27 2007 New Revision: 42607 URL: http://llvm.org/viewvc/llvm-project?rev=42607&view=rev Log: add a note. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42607&r1=42606&r2=42607&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Thu Oct 4 10:47:27 2007 @@ -1362,3 +1362,43 @@ //===---------------------------------------------------------------------===// +We compile this: + +void compare (long long foo) { + if (foo < 4294967297LL) + abort(); +} + +to: + +_compare: + subl $12, %esp + cmpl $0, 16(%esp) + setne %al + movzbw %al, %ax + cmpl $1, 20(%esp) + setg %cl + movzbw %cl, %cx + cmove %ax, %cx + movw %cx, %ax + testb $1, %al + je LBB1_2 # cond_true + +(also really horrible code on ppc). This is due to the expand code for 64-bit +compares. GCC produces multiple branches, which is much nicer: + +_compare: + pushl %ebp + movl %esp, %ebp + subl $8, %esp + movl 8(%ebp), %eax + movl 12(%ebp), %edx + subl $1, %edx + jg L5 +L7: + jl L4 + cmpl $0, %eax + jbe L4 +L5: + +//===---------------------------------------------------------------------===// From isanbard at gmail.com Thu Oct 4 13:46:07 2007 From: isanbard at gmail.com (Bill Wendling) Date: Thu, 04 Oct 2007 18:46:07 -0000 Subject: [llvm-commits] [llvm] r42609 - /llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm Message-ID: <200710041846.l94Ik7QL002895@zion.cs.uiuc.edu> Author: void Date: Thu Oct 4 13:46:06 2007 New Revision: 42609 URL: http://llvm.org/viewvc/llvm-project?rev=42609&view=rev Log: Added testcase for creation of metadata with null pointers. Added: llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm Added: llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm?rev=42609&view=auto ============================================================================== --- llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm (added) +++ llvm/trunk/test/FrontendObjC/2007-10-03-MetadataPointers.mm Thu Oct 4 13:46:06 2007 @@ -0,0 +1,7 @@ +// RUN: %llvmgcc -x objective-c++ -c %s -o /dev/null + + at class NSImage; +void bork() { + NSImage *nsimage; + [nsimage release]; +} From isanbard at gmail.com Thu Oct 4 13:47:25 2007 From: isanbard at gmail.com (Bill Wendling) Date: Thu, 04 Oct 2007 18:47:25 -0000 Subject: [llvm-commits] [llvm] r42610 - /llvm/trunk/test/FrontendObjC/dg.exp Message-ID: <200710041847.l94IlPli002964@zion.cs.uiuc.edu> Author: void Date: Thu Oct 4 13:47:25 2007 New Revision: 42610 URL: http://llvm.org/viewvc/llvm-project?rev=42610&view=rev Log: Support Objective C++ too. Modified: llvm/trunk/test/FrontendObjC/dg.exp Modified: llvm/trunk/test/FrontendObjC/dg.exp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/dg.exp?rev=42610&r1=42609&r2=42610&view=diff ============================================================================== --- llvm/trunk/test/FrontendObjC/dg.exp (original) +++ llvm/trunk/test/FrontendObjC/dg.exp Thu Oct 4 13:47:25 2007 @@ -2,5 +2,6 @@ if [ llvm_gcc_supports objc ] then { RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{m}]] + RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{mm}]] } From dberlin at dberlin.org Thu Oct 4 16:18:20 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Thu, 04 Oct 2007 21:18:20 -0000 Subject: [llvm-commits] [llvm] r42612 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710042118.l94LIKgG009660@zion.cs.uiuc.edu> Author: dannyb Date: Thu Oct 4 16:18:20 2007 New Revision: 42612 URL: http://llvm.org/viewvc/llvm-project?rev=42612&view=rev Log: Fix off by one error in iterator Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42612&r1=42611&r2=42612&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Thu Oct 4 16:18:20 2007 @@ -390,7 +390,7 @@ // See if we ran out of Bits in this word. if (!Bits) { - int NextSetBitNumber = Iter->find_next(BitNumber % ElementSize) ; + int NextSetBitNumber = Iter->find_next((BitNumber - 1) % ElementSize) ; // If we ran out of set bits in this element, move to next element. if (NextSetBitNumber == -1 || (BitNumber % ElementSize == 0)) { ++Iter; From dberlin at dberlin.org Thu Oct 4 16:27:17 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Thu, 04 Oct 2007 21:27:17 -0000 Subject: [llvm-commits] [llvm] r42613 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710042127.l94LRHcg010191@zion.cs.uiuc.edu> Author: dannyb Date: Thu Oct 4 16:27:17 2007 New Revision: 42613 URL: http://llvm.org/viewvc/llvm-project?rev=42613&view=rev Log: Fix the previous bug a slightly different way (by modifying how find_next works) Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42613&r1=42612&r2=42613&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Thu Oct 4 16:27:17 2007 @@ -168,15 +168,14 @@ assert(0 && "Illegal empty element"); } - /// find_next - Returns the index of the next set bit following the - /// "Prev" bit. Returns -1 if the next set bit is not found. - int find_next(unsigned Prev) const { - ++Prev; - if (Prev >= BITS_PER_ELEMENT) + /// find_next - Returns the index of the next set bit starting from the + /// "Curr" bit. Returns -1 if the next set bit is not found. + int find_next(unsigned Curr) const { + if (Curr >= BITS_PER_ELEMENT) return -1; - unsigned WordPos = Prev / BITWORD_SIZE; - unsigned BitPos = Prev % BITWORD_SIZE; + unsigned WordPos = Curr / BITWORD_SIZE; + unsigned BitPos = Curr % BITWORD_SIZE; BitWord Copy = Bits[WordPos]; assert (WordPos <= BITWORDS_PER_ELEMENT && "Word Position outside of element"); @@ -390,7 +389,7 @@ // See if we ran out of Bits in this word. if (!Bits) { - int NextSetBitNumber = Iter->find_next((BitNumber - 1) % ElementSize) ; + int NextSetBitNumber = Iter->find_next(BitNumber % ElementSize) ; // If we ran out of set bits in this element, move to next element. if (NextSetBitNumber == -1 || (BitNumber % ElementSize == 0)) { ++Iter; @@ -546,7 +545,7 @@ } } CurrElementIter = ElementIter; - + ElementIter->set(Idx % ElementSize); } From isanbard at gmail.com Thu Oct 4 17:15:58 2007 From: isanbard at gmail.com (Bill Wendling) Date: Thu, 04 Oct 2007 22:15:58 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42614 - in /llvm-gcc-4.0/trunk/gcc: c-common.h llvm-backend.cpp objc/objc-act.c stub-objc.c Message-ID: <200710042215.l94MFwL8012295@zion.cs.uiuc.edu> Author: void Date: Thu Oct 4 17:15:58 2007 New Revision: 42614 URL: http://llvm.org/viewvc/llvm-project?rev=42614&view=rev Log: rdar://5476262 - Try to lookup the name of the global decl before creating a new one. Modified: llvm-gcc-4.0/trunk/gcc/c-common.h llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp llvm-gcc-4.0/trunk/gcc/objc/objc-act.c llvm-gcc-4.0/trunk/gcc/stub-objc.c Modified: llvm-gcc-4.0/trunk/gcc/c-common.h URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/c-common.h?rev=42614&r1=42613&r2=42614&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/c-common.h (original) +++ llvm-gcc-4.0/trunk/gcc/c-common.h Thu Oct 4 17:15:58 2007 @@ -1139,12 +1139,6 @@ extern void objc_remove_weak_read (tree*); /* APPLE LOCAL end radar 4426814 */ -/* APPLE LOCAL begin - LLVM radar 5476262 */ -#ifdef ENABLE_LLVM -extern bool objc_is_protocol_reference (const char *name); -#endif -/* APPLE LOCAL end - LLVM radar 5476262 */ - /* APPLE LOCAL begin C* language */ extern void objc_set_method_opt (int); void objc_finish_foreach_loop (location_t, tree, tree, tree, tree); Modified: llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp?rev=42614&r1=42613&r2=42614&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/llvm-backend.cpp Thu Oct 4 17:15:58 2007 @@ -66,7 +66,6 @@ #include "function.h" #include "tree-inline.h" #include "langhooks.h" -#include "c-common.h" } // Non-zero if bytecode from PCH is successfully read. @@ -1061,14 +1060,6 @@ // being created. GlobalVariable *GVE = TheModule->getGlobalVariable(Name); - // And Objective-C "@protocol" will create a decl for the - // protocol metadata and then when the protocol is - // referenced. However, protocols have file-scope, so they - // aren't found in the GlobalVariable list unless we look at - // non-extern globals as well. - if (!GVE && c_dialect_objc() && objc_is_protocol_reference(Name)) - GVE = TheModule->getGlobalVariable(Name, true); - if (GVE == 0) { GV = new GlobalVariable(Ty, false, GlobalValue::ExternalLinkage,0, Name, TheModule); Modified: llvm-gcc-4.0/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/objc/objc-act.c?rev=42614&r1=42613&r2=42614&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.0/trunk/gcc/objc/objc-act.c Thu Oct 4 17:15:58 2007 @@ -11867,29 +11867,21 @@ tree decl = TREE_PURPOSE (chain); gcc_assert (TREE_CODE (expr) == PROTOCOL_INTERFACE_TYPE); /* APPLE LOCAL begin radar 4695109 */ - sprintf (string, "_OBJC_PROTOCOL_$_%s", + /* APPLE LOCAL begin - LLVM radar 5476262 */ + sprintf (string, "L_OBJC_PROTOCOL_$_%s", IDENTIFIER_POINTER (PROTOCOL_NAME (expr))); - expr = start_var_decl (objc_v2_protocol_template, string); +#ifdef ENABLE_LLVM + expr = lookup_name (get_identifier(string)); + if (expr == NULL_TREE) +#endif + expr = start_var_decl (objc_v2_protocol_template, &string[1]); + /* APPLE LOCAL end - LLVM radar 5476262 */ /* APPLE LOCAL end radar 4695109 */ expr = convert (objc_protocol_type, build_fold_addr_expr (expr)); finish_var_decl (decl, expr); } } -/* APPLE LOCAL begin - LLVM radar 5476262 */ -#ifdef ENABLE_LLVM -/* This routine returns true if the name is the same as a protocol - reference name. */ - -bool -objc_is_protocol_reference (const char *name) -{ - return flag_objc_abi == 2 && strlen (name) > 19 && - strncmp (name, "\01L_OBJC_PROTOCOL_$_", 19) == 0; -} -#endif -/* APPLE LOCAL end - LLVM radar 5476262 */ - /* This routine builds the protocol_reference_chain for each protocol name used @protocol(MyProtocol) expression. IDENT is current protocol name. */ @@ -13402,7 +13394,7 @@ /* Declare variable which holds 'struct protocol_t' meta data. */ /* APPLE LOCAL begin radar 4695109 */ -/* poniter to protocol metadata in new ABI protocol is now generated +/* pointer to protocol metadata in new ABI protocol is now generated like the old abi. */ static void build_v2_protocol_reference (tree p) @@ -13415,6 +13407,9 @@ proto_name = synth_id_with_class_suffix ("_OBJC_PROTOCOL_$", p); decl = start_var_decl (objc_v2_protocol_template, proto_name); PROTOCOL_V2_FORWARD_DECL (p) = decl; + /* APPLE LOCAL begin - LLVM radar 5476262 */ + pushdecl_top_level(decl); + /* APPLE LOCAL end - LLVM radar 5476262 */ } /* APPLE LOCAL end radar 4695109 */ Modified: llvm-gcc-4.0/trunk/gcc/stub-objc.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/stub-objc.c?rev=42614&r1=42613&r2=42614&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/stub-objc.c (original) +++ llvm-gcc-4.0/trunk/gcc/stub-objc.c Thu Oct 4 17:15:58 2007 @@ -548,13 +548,3 @@ return false; } /* APPLE LOCAL end radar 4985544 */ - -/* APPLE LOCAL begin - LLVM radar 5476262 */ -#ifdef ENABLE_LLVM -bool -objc_is_protocol_reference (const char * ARG_UNUSED(name)) -{ - return false; -} -#endif -/* APPLE LOCAL end - LLVM radar 5476262 */ From isanbard at gmail.com Thu Oct 4 17:18:01 2007 From: isanbard at gmail.com (Bill Wendling) Date: Thu, 04 Oct 2007 22:18:01 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42615 - /llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Message-ID: <200710042218.l94MI1jY012403@zion.cs.uiuc.edu> Author: void Date: Thu Oct 4 17:18:01 2007 New Revision: 42615 URL: http://llvm.org/viewvc/llvm-project?rev=42615&view=rev Log: rdar://5476262 - Try to lookup the name of the global decl before creating a new one. Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c?rev=42615&r1=42614&r2=42615&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/objc/objc-act.c (original) +++ llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Thu Oct 4 17:18:01 2007 @@ -11756,9 +11756,15 @@ tree decl = TREE_PURPOSE (chain); gcc_assert (TREE_CODE (expr) == PROTOCOL_INTERFACE_TYPE); /* APPLE LOCAL begin radar 4695109 */ - sprintf (string, "_OBJC_PROTOCOL_$_%s", + /* APPLE LOCAL begin - LLVM radar 5476262 */ + sprintf (string, "L_OBJC_PROTOCOL_$_%s", IDENTIFIER_POINTER (PROTOCOL_NAME (expr))); - expr = start_var_decl (objc_v2_protocol_template, string); +#ifdef ENABLE_LLVM + expr = lookup_name (get_identifier(string)); + if (expr == NULL_TREE) +#endif + expr = start_var_decl (objc_v2_protocol_template, &string[1]); + /* APPLE LOCAL end - LLVM radar 5476262 */ /* APPLE LOCAL end radar 4695109 */ expr = convert (objc_protocol_type, build_fold_addr_expr (expr)); finish_var_decl (decl, expr); @@ -13299,6 +13305,9 @@ proto_name = synth_id_with_class_suffix ("_OBJC_PROTOCOL_$", p); decl = start_var_decl (objc_v2_protocol_template, proto_name); PROTOCOL_V2_FORWARD_DECL (p) = decl; + /* APPLE LOCAL begin - LLVM radar 5476262 */ + pushdecl_top_level(decl); + /* APPLE LOCAL end - LLVM radar 5476262 */ } /* APPLE LOCAL end radar 4695109 */ From asl at math.spbu.ru Thu Oct 4 18:54:25 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Thu, 04 Oct 2007 23:54:25 -0000 Subject: [llvm-commits] [llvm-gcc-4.2] r42617 - /llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp Message-ID: <200710042354.l94NsPAG016999@zion.cs.uiuc.edu> Author: asl Date: Thu Oct 4 18:54:25 2007 New Revision: 42617 URL: http://llvm.org/viewvc/llvm-project?rev=42617&view=rev Log: Don't zero high 24 bit of %mxcsr by emitting 32 bit load instead of 8 bit one. Modified: llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp Modified: llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp?rev=42617&r1=42616&r2=42617&view=diff ============================================================================== --- llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp (original) +++ llvm-gcc-4.2/trunk/gcc/config/i386/llvm-i386.cpp Thu Oct 4 18:54:25 2007 @@ -610,9 +610,11 @@ case IX86_BUILTIN_STMXCSR: { Function *stmxcsr = Intrinsic::getDeclaration(TheModule, Intrinsic::x86_sse_stmxcsr); - Value *Ptr = CreateTemporary(Type::Int32Ty); - Ptr = Builder.CreateBitCast(Ptr, PointerType::get(Type::Int8Ty), "tmp"); - Builder.CreateCall(stmxcsr, Ptr); + Value *Ptr = CreateTemporary(Type::Int32Ty); + Value *BPtr = Builder.CreateBitCast(Ptr, PointerType::get(Type::Int8Ty), + "tmp"); + Builder.CreateCall(stmxcsr, BPtr); + Result = Builder.CreateLoad(Ptr, "tmp"); return true; } From asl at math.spbu.ru Thu Oct 4 18:55:59 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Thu, 04 Oct 2007 23:55:59 -0000 Subject: [llvm-commits] [llvm-gcc-4.0] r42618 - /llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp Message-ID: <200710042355.l94NtxQQ017055@zion.cs.uiuc.edu> Author: asl Date: Thu Oct 4 18:55:59 2007 New Revision: 42618 URL: http://llvm.org/viewvc/llvm-project?rev=42618&view=rev Log: Don't zero high 24 bit of %mxcsr by emitting 32 bit load instead of 8 bit one. Modified: llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp Modified: llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp?rev=42618&r1=42617&r2=42618&view=diff ============================================================================== --- llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp (original) +++ llvm-gcc-4.0/trunk/gcc/config/i386/llvm-i386.cpp Thu Oct 4 18:55:59 2007 @@ -610,9 +610,11 @@ case IX86_BUILTIN_STMXCSR: { Function *stmxcsr = Intrinsic::getDeclaration(TheModule, Intrinsic::x86_sse_stmxcsr); - Value *Ptr = CreateTemporary(Type::Int32Ty); - Ptr = Builder.CreateBitCast(Ptr, PointerType::get(Type::Int8Ty), "tmp"); - Builder.CreateCall(stmxcsr, Ptr); + Value *Ptr = CreateTemporary(Type::Int32Ty); + Value *BPtr = Builder.CreateBitCast(Ptr, PointerType::get(Type::Int8Ty), + "tmp"); + Builder.CreateCall(stmxcsr, BPtr); + Result = Builder.CreateLoad(Ptr, "tmp"); return true; } From evan.cheng at apple.com Thu Oct 4 20:09:32 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:09:32 -0000 Subject: [llvm-commits] [llvm] r42619 - /llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Message-ID: <200710050109.l9519W4i020221@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:09:32 2007 New Revision: 42619 URL: http://llvm.org/viewvc/llvm-project?rev=42619&view=rev Log: Silence a warning. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=42619&r1=42618&r2=42619&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Thu Oct 4 20:09:32 2007 @@ -5395,7 +5395,7 @@ } } - RTLIB::Libcall LC; + RTLIB::Libcall LC = RTLIB::UNKNOWN_LIBCALL; if (Node->getOperand(0).getValueType() == MVT::f32) LC = RTLIB::FPTOUINT_F32_I64; else if (Node->getOperand(0).getValueType() == MVT::f64) From evan.cheng at apple.com Thu Oct 4 20:10:50 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:10:50 -0000 Subject: [llvm-commits] [llvm] r42620 - in /llvm/trunk: include/llvm/CodeGen/SelectionDAG.h lib/CodeGen/SelectionDAG/SelectionDAG.cpp Message-ID: <200710050110.l951Ao9x020309@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:10:49 2007 New Revision: 42620 URL: http://llvm.org/viewvc/llvm-project?rev=42620&view=rev Log: Add a variant of getTargetNode() that takes a vector of MVT::ValueType. Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAG.h llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAG.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SelectionDAG.h?rev=42620&r1=42619&r2=42620&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/SelectionDAG.h (original) +++ llvm/trunk/include/llvm/CodeGen/SelectionDAG.h Thu Oct 4 20:10:49 2007 @@ -414,6 +414,8 @@ MVT::ValueType VT2, MVT::ValueType VT3, MVT::ValueType VT4, const SDOperand *Ops, unsigned NumOps); + SDNode *getTargetNode(unsigned Opcode, std::vector &ResultTys, + const SDOperand *Ops, unsigned NumOps); /// ReplaceAllUsesWith - Modify anything using 'From' to use 'To' instead. /// This can cause recursive merging of nodes in the DAG. Use the first Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=42620&r1=42619&r2=42620&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Thu Oct 4 20:10:49 2007 @@ -3054,6 +3054,13 @@ const MVT::ValueType *VTs = getNodeValueTypes(VTList); return getNode(ISD::BUILTIN_OP_END+Opcode, VTs, 4, Ops, NumOps).Val; } +SDNode *SelectionDAG::getTargetNode(unsigned Opcode, + std::vector &ResultTys, + const SDOperand *Ops, unsigned NumOps) { + const MVT::ValueType *VTs = getNodeValueTypes(ResultTys); + return getNode(ISD::BUILTIN_OP_END+Opcode, VTs, ResultTys.size(), + Ops, NumOps).Val; +} /// ReplaceAllUsesWith - Modify anything using 'From' to use 'To' instead. /// This can cause recursive merging of nodes in the DAG. From evan.cheng at apple.com Thu Oct 4 20:32:41 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:32:41 -0000 Subject: [llvm-commits] [llvm] r42621 - in /llvm/trunk: include/llvm/Target/MRegisterInfo.h lib/Target/ARM/ARMRegisterInfo.cpp lib/Target/ARM/ARMRegisterInfo.h lib/Target/Alpha/AlphaRegisterInfo.cpp lib/Target/Alpha/AlphaRegisterInfo.h lib/Target/Mips/MipsRegisterInfo.cpp lib/Target/Mips/MipsRegisterInfo.h lib/Target/PowerPC/PPCRegisterInfo.cpp lib/Target/PowerPC/PPCRegisterInfo.h lib/Target/Sparc/SparcRegisterInfo.cpp lib/Target/Sparc/SparcRegisterInfo.h Message-ID: <200710050132.l951Wf8x021244@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:32:41 2007 New Revision: 42621 URL: http://llvm.org/viewvc/llvm-project?rev=42621&view=rev Log: - Added a few target hooks to generate load / store instructions from / to any address (not just from / to frameindexes). - Added target hooks to unfold load / store instructions / SDNodes into separate load, data processing, store instructions / SDNodes. Modified: llvm/trunk/include/llvm/Target/MRegisterInfo.h llvm/trunk/lib/Target/ARM/ARMRegisterInfo.cpp llvm/trunk/lib/Target/ARM/ARMRegisterInfo.h llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.cpp llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.h llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp llvm/trunk/lib/Target/Mips/MipsRegisterInfo.h llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.cpp llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.h llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.cpp llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.h Modified: llvm/trunk/include/llvm/Target/MRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/MRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/include/llvm/Target/MRegisterInfo.h (original) +++ llvm/trunk/include/llvm/Target/MRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -16,6 +16,7 @@ #ifndef LLVM_TARGET_MREGISTERINFO_H #define LLVM_TARGET_MREGISTERINFO_H +#include "llvm/ADT/SmallVector.h" #include "llvm/CodeGen/MachineBasicBlock.h" #include "llvm/CodeGen/ValueTypes.h" #include @@ -30,6 +31,9 @@ class MachineLocation; class MachineMove; class RegScavenger; +class SDNode; +class SelectionDAG; +class SSARegMap; class TargetRegisterClass; class Type; @@ -503,11 +507,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const = 0; + virtual void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const = 0; + virtual void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const = 0; + virtual void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const = 0; + virtual void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, unsigned SrcReg, @@ -550,6 +564,20 @@ return 0; } + /// unfoldMemoryOperand - Separate a single instruction which folded a load or a + /// a store or a load and a store into two or more instruction. If this is + /// possible, returns true as well as the new instructions by reference. + virtual bool unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI, + SSARegMap *RegMap, + SmallVector &NewMIs) const { + return false; + } + + virtual bool unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N, + SmallVector &NewNodes) const { + return false; + } + /// targetHandlesStackFrameRounding - Returns true if the target is responsible /// for rounding up the stack frame (probably at emitPrologue time). virtual bool targetHandlesStackFrameRounding() const { Modified: llvm/trunk/lib/Target/ARM/ARMRegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMRegisterInfo.cpp?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/ARM/ARMRegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/ARM/ARMRegisterInfo.cpp Thu Oct 4 20:32:41 2007 @@ -132,6 +132,30 @@ return true; } +static inline +const MachineInstrBuilder &AddDefaultPred(const MachineInstrBuilder &MIB) { + return MIB.addImm((int64_t)ARMCC::AL).addReg(0); +} + +static inline +const MachineInstrBuilder &AddDefaultCC(const MachineInstrBuilder &MIB) { + return MIB.addReg(0); +} + +static const MachineInstrBuilder &ARMInstrAddOperand(MachineInstrBuilder &MIB, + MachineOperand &MO) { + if (MO.isRegister()) + MIB = MIB.addReg(MO.getReg(), MO.isDef(), MO.isImplicit()); + else if (MO.isImmediate()) + MIB = MIB.addImm(MO.getImm()); + else if (MO.isFrameIndex()) + MIB = MIB.addFrameIndex(MO.getFrameIndex()); + else + assert(0 && "Unknown operand for ARMInstrAddOperand!"); + + return MIB; +} + void ARMRegisterInfo:: storeRegToStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned SrcReg, int FI, @@ -143,17 +167,52 @@ BuildMI(MBB, I, TII.get(ARM::tSpill)).addReg(SrcReg, false, false, true) .addFrameIndex(FI).addImm(0); else - BuildMI(MBB, I, TII.get(ARM::STR)).addReg(SrcReg, false, false, true) - .addFrameIndex(FI).addReg(0).addImm(0).addImm((int64_t)ARMCC::AL) - .addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::STR)) + .addReg(SrcReg, false, false, true) + .addFrameIndex(FI).addReg(0).addImm(0)); + } else if (RC == ARM::DPRRegisterClass) { + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FSTD)) + .addReg(SrcReg, false, false, true) + .addFrameIndex(FI).addImm(0)); + } else { + assert(RC == ARM::SPRRegisterClass && "Unknown regclass!"); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FSTS)) + .addReg(SrcReg, false, false, true) + .addFrameIndex(FI).addImm(0)); + } +} + +void ARMRegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == ARM::GPRRegisterClass) { + ARMFunctionInfo *AFI = MF.getInfo(); + if (AFI->isThumbFunction()) { + Opc = Addr[0].isFrameIndex() ? ARM::tSpill : ARM::tSTR; + MachineInstrBuilder MIB = + BuildMI(TII.get(Opc)).addReg(SrcReg, false, false, true); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = ARMInstrAddOperand(MIB, Addr[i]); + NewMIs.push_back(MIB); + return; + } + Opc = ARM::STR; } else if (RC == ARM::DPRRegisterClass) { - BuildMI(MBB, I, TII.get(ARM::FSTD)).addReg(SrcReg, false, false, true) - .addFrameIndex(FI).addImm(0).addImm((int64_t)ARMCC::AL).addReg(0); + Opc = ARM::FSTD; } else { assert(RC == ARM::SPRRegisterClass && "Unknown regclass!"); - BuildMI(MBB, I, TII.get(ARM::FSTS)).addReg(SrcReg, false, false, true) - .addFrameIndex(FI).addImm(0).addImm((int64_t)ARMCC::AL).addReg(0); + Opc = ARM::FSTS; } + + MachineInstrBuilder MIB = + BuildMI(TII.get(Opc)).addReg(SrcReg, false, false, true); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = ARMInstrAddOperand(MIB, Addr[i]); + AddDefaultPred(MIB); + NewMIs.push_back(MIB); + return; } void ARMRegisterInfo:: @@ -167,19 +226,49 @@ BuildMI(MBB, I, TII.get(ARM::tRestore), DestReg) .addFrameIndex(FI).addImm(0); else - BuildMI(MBB, I, TII.get(ARM::LDR), DestReg) - .addFrameIndex(FI).addReg(0).addImm(0).addImm((int64_t)ARMCC::AL) - .addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::LDR), DestReg) + .addFrameIndex(FI).addReg(0).addImm(0)); } else if (RC == ARM::DPRRegisterClass) { - BuildMI(MBB, I, TII.get(ARM::FLDD), DestReg) - .addFrameIndex(FI).addImm(0).addImm((int64_t)ARMCC::AL).addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FLDD), DestReg) + .addFrameIndex(FI).addImm(0)); } else { assert(RC == ARM::SPRRegisterClass && "Unknown regclass!"); - BuildMI(MBB, I, TII.get(ARM::FLDS), DestReg) - .addFrameIndex(FI).addImm(0).addImm((int64_t)ARMCC::AL).addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FLDS), DestReg) + .addFrameIndex(FI).addImm(0)); } } +void ARMRegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == ARM::GPRRegisterClass) { + ARMFunctionInfo *AFI = MF.getInfo(); + if (AFI->isThumbFunction()) { + Opc = Addr[0].isFrameIndex() ? ARM::tRestore : ARM::tLDR; + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = ARMInstrAddOperand(MIB, Addr[i]); + NewMIs.push_back(MIB); + return; + } + Opc = ARM::LDR; + } else if (RC == ARM::DPRRegisterClass) { + Opc = ARM::FLDD; + } else { + assert(RC == ARM::SPRRegisterClass && "Unknown regclass!"); + Opc = ARM::FLDS; + } + + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = ARMInstrAddOperand(MIB, Addr[i]); + AddDefaultPred(MIB); + NewMIs.push_back(MIB); + return; +} + void ARMRegisterInfo::copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg, unsigned SrcReg, @@ -196,14 +285,14 @@ if (AFI->isThumbFunction()) BuildMI(MBB, I, TII.get(ARM::tMOVr), DestReg).addReg(SrcReg); else - BuildMI(MBB, I, TII.get(ARM::MOVr), DestReg).addReg(SrcReg) - .addImm((int64_t)ARMCC::AL).addReg(0).addReg(0); + AddDefaultCC(AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::MOVr), DestReg) + .addReg(SrcReg))); } else if (DestRC == ARM::SPRRegisterClass) - BuildMI(MBB, I, TII.get(ARM::FCPYS), DestReg).addReg(SrcReg) - .addImm((int64_t)ARMCC::AL).addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FCPYS), DestReg) + .addReg(SrcReg)); else if (DestRC == ARM::DPRRegisterClass) - BuildMI(MBB, I, TII.get(ARM::FCPYD), DestReg).addReg(SrcReg) - .addImm((int64_t)ARMCC::AL).addReg(0); + AddDefaultPred(BuildMI(MBB, I, TII.get(ARM::FCPYD), DestReg) + .addReg(SrcReg)); else abort(); } @@ -1390,7 +1479,7 @@ MachineInstrBuilder MIB = BuildMI(MBB, MBBI, TII.get(isThumb ? ARM::tADDrSPi : ARM::ADDri),FramePtr) .addFrameIndex(FramePtrSpillFI).addImm(0); - if (!isThumb) MIB.addImm(ARMCC::AL).addReg(0).addReg(0); + if (!isThumb) AddDefaultCC(AddDefaultPred(MIB)); } if (!isThumb) { Modified: llvm/trunk/lib/Target/ARM/ARMRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/ARM/ARMRegisterInfo.h (original) +++ llvm/trunk/lib/Target/ARM/ARMRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -51,11 +51,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, unsigned SrcReg, const TargetRegisterClass *DestRC, Modified: llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.cpp?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.cpp Thu Oct 4 20:32:41 2007 @@ -82,6 +82,31 @@ abort(); } +void AlphaRegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == Alpha::F4RCRegisterClass) + Opc = Alpha::STS; + else if (RC == Alpha::F8RCRegisterClass) + Opc = Alpha::STT; + else if (RC == Alpha::GPRCRegisterClass) + Opc = Alpha::STQ; + else + abort(); + MachineInstrBuilder MIB = + BuildMI(TII.get(Opc)).addReg(SrcReg, false, false, true); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg(), MO.isDef(), MO.isImplicit()); + else + MIB.addImm(MO.getImm()); + } + NewMIs.push_back(MIB); +} + void AlphaRegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, @@ -102,6 +127,31 @@ abort(); } +void AlphaRegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == Alpha::F4RCRegisterClass) + Opc = Alpha::LDS; + else if (RC == Alpha::F8RCRegisterClass) + Opc = Alpha::LDT; + else if (RC == Alpha::GPRCRegisterClass) + Opc = Alpha::LDQ; + else + abort(); + MachineInstrBuilder MIB = + BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg(), MO.isDef(), MO.isImplicit()); + else + MIB.addImm(MO.getImm()); + } + NewMIs.push_back(MIB); +} + MachineInstr *AlphaRegisterInfo::foldMemoryOperand(MachineInstr *MI, unsigned OpNum, int FrameIndex) const { Modified: llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.h (original) +++ llvm/trunk/lib/Target/Alpha/AlphaRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -33,11 +33,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + MachineInstr* foldMemoryOperand(MachineInstr *MI, unsigned OpNum, int FrameIndex) const; Modified: llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp Thu Oct 4 20:32:41 2007 @@ -95,6 +95,27 @@ assert(0 && "Can't store this register to stack slot"); } +void MipsRegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + if (RC != Mips::CPURegsRegisterClass) + assert(0 && "Can't store this register"); + MachineInstrBuilder MIB = BuildMI(TII.get(Mips::SW)) + .addReg(SrcReg, false, false, true); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; +} + void MipsRegisterInfo:: loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg, int FI, @@ -106,6 +127,26 @@ assert(0 && "Can't load this register from stack slot"); } +void MipsRegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + if (RC != Mips::CPURegsRegisterClass) + assert(0 && "Can't load this register"); + MachineInstrBuilder MIB = BuildMI(TII.get(Mips::LW), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; +} + void MipsRegisterInfo:: copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg, unsigned SrcReg, Modified: llvm/trunk/lib/Target/Mips/MipsRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Mips/MipsRegisterInfo.h (original) +++ llvm/trunk/lib/Target/Mips/MipsRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -37,11 +37,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void reMaterialize(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, const MachineInstr *Orig) const; Modified: llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.cpp?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.cpp Thu Oct 4 20:32:41 2007 @@ -103,68 +103,67 @@ ImmToIdxMap[PPC::ADDI8] = PPC::ADD8; ImmToIdxMap[PPC::STD_32] = PPC::STDX_32; } -void -PPCRegisterInfo::storeRegToStackSlot(MachineBasicBlock &MBB, - MachineBasicBlock::iterator MI, - unsigned SrcReg, int FrameIdx, - const TargetRegisterClass *RC) const { +static void StoreRegToStackSlot(const TargetInstrInfo &TII, + unsigned SrcReg, int FrameIdx, + const TargetRegisterClass *RC, + SmallVector &NewMIs) { if (RC == PPC::GPRCRegisterClass) { if (SrcReg != PPC::LR) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STW)) - .addReg(SrcReg, false, false, true), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STW)) + .addReg(SrcReg, false, false, true), FrameIdx)); } else { // FIXME: this spills LR immediately to memory in one step. To do this, // we use R11, which we know cannot be used in the prolog/epilog. This is // a hack. - BuildMI(MBB, MI, TII.get(PPC::MFLR), PPC::R11); - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STW)) - .addReg(PPC::R11, false, false, true), FrameIdx); + NewMIs.push_back(BuildMI(TII.get(PPC::MFLR), PPC::R11)); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STW)) + .addReg(PPC::R11, false, false, true), FrameIdx)); } } else if (RC == PPC::G8RCRegisterClass) { if (SrcReg != PPC::LR8) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STD)) - .addReg(SrcReg, false, false, true), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STD)) + .addReg(SrcReg, false, false, true), FrameIdx)); } else { // FIXME: this spills LR immediately to memory in one step. To do this, // we use R11, which we know cannot be used in the prolog/epilog. This is // a hack. - BuildMI(MBB, MI, TII.get(PPC::MFLR8), PPC::X11); - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STD)) - .addReg(PPC::X11, false, false, true), FrameIdx); + NewMIs.push_back(BuildMI(TII.get(PPC::MFLR8), PPC::X11)); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STD)) + .addReg(PPC::X11, false, false, true), FrameIdx)); } } else if (RC == PPC::F8RCRegisterClass) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STFD)) - .addReg(SrcReg, false, false, true), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STFD)) + .addReg(SrcReg, false, false, true), FrameIdx)); } else if (RC == PPC::F4RCRegisterClass) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STFS)) - .addReg(SrcReg, false, false, true), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STFS)) + .addReg(SrcReg, false, false, true), FrameIdx)); } else if (RC == PPC::CRRCRegisterClass) { // FIXME: We use R0 here, because it isn't available for RA. // We need to store the CR in the low 4-bits of the saved value. First, // issue a MFCR to save all of the CRBits. - BuildMI(MBB, MI, TII.get(PPC::MFCR), PPC::R0); + NewMIs.push_back(BuildMI(TII.get(PPC::MFCR), PPC::R0)); // If the saved register wasn't CR0, shift the bits left so that they are in // CR0's slot. if (SrcReg != PPC::CR0) { unsigned ShiftBits = PPCRegisterInfo::getRegisterNumbering(SrcReg)*4; // rlwinm r0, r0, ShiftBits, 0, 31. - BuildMI(MBB, MI, TII.get(PPC::RLWINM), PPC::R0) - .addReg(PPC::R0).addImm(ShiftBits).addImm(0).addImm(31); + NewMIs.push_back(BuildMI(TII.get(PPC::RLWINM), PPC::R0) + .addReg(PPC::R0).addImm(ShiftBits).addImm(0).addImm(31)); } - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::STW)) - .addReg(PPC::R0, false, false, true), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::STW)) + .addReg(PPC::R0, false, false, true), FrameIdx)); } else if (RC == PPC::VRRCRegisterClass) { // We don't have indexed addressing for vector loads. Emit: // R0 = ADDI FI# // STVX VAL, 0, R0 // // FIXME: We use R0 here, because it isn't available for RA. - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::ADDI), PPC::R0), - FrameIdx, 0, 0); - BuildMI(MBB, MI, TII.get(PPC::STVX)) - .addReg(SrcReg, false, false, true).addReg(PPC::R0).addReg(PPC::R0); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::ADDI), PPC::R0), + FrameIdx, 0, 0)); + NewMIs.push_back(BuildMI(TII.get(PPC::STVX)) + .addReg(SrcReg, false, false, true).addReg(PPC::R0).addReg(PPC::R0)); } else { assert(0 && "Unknown regclass!"); abort(); @@ -172,55 +171,163 @@ } void -PPCRegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, - MachineBasicBlock::iterator MI, - unsigned DestReg, int FrameIdx, - const TargetRegisterClass *RC) const { +PPCRegisterInfo::storeRegToStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MI, + unsigned SrcReg, int FrameIdx, + const TargetRegisterClass *RC) const { + SmallVector NewMIs; + StoreRegToStackSlot(TII, SrcReg, FrameIdx, RC, NewMIs); + for (unsigned i = 0, e = NewMIs.size(); i != e; ++i) + MBB.insert(MI, NewMIs[i]); +} + +void PPCRegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + if (Addr[0].isFrameIndex()) { + StoreRegToStackSlot(TII, SrcReg, Addr[0].getFrameIndex(), RC, NewMIs); + return; + } + + unsigned Opc = 0; + if (RC == PPC::GPRCRegisterClass) { + Opc = PPC::STW; + } else if (RC == PPC::G8RCRegisterClass) { + Opc = PPC::STD; + } else if (RC == PPC::F8RCRegisterClass) { + Opc = PPC::STFD; + } else if (RC == PPC::F4RCRegisterClass) { + Opc = PPC::STFS; + } else if (RC == PPC::VRRCRegisterClass) { + Opc = PPC::STVX; + } else { + assert(0 && "Unknown regclass!"); + abort(); + } + MachineInstrBuilder MIB = BuildMI(TII.get(Opc)) + .addReg(SrcReg, false, false, true); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; +} + +static void LoadRegFromStackSlot(const TargetInstrInfo &TII, + unsigned DestReg, int FrameIdx, + const TargetRegisterClass *RC, + SmallVector &NewMIs) { if (RC == PPC::GPRCRegisterClass) { if (DestReg != PPC::LR) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LWZ), DestReg), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LWZ), DestReg), + FrameIdx)); } else { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LWZ), PPC::R11),FrameIdx); - BuildMI(MBB, MI, TII.get(PPC::MTLR)).addReg(PPC::R11); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LWZ), PPC::R11), + FrameIdx)); + NewMIs.push_back(BuildMI(TII.get(PPC::MTLR)).addReg(PPC::R11)); } } else if (RC == PPC::G8RCRegisterClass) { if (DestReg != PPC::LR8) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LD), DestReg), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LD), DestReg), + FrameIdx)); } else { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LD), PPC::R11), FrameIdx); - BuildMI(MBB, MI, TII.get(PPC::MTLR8)).addReg(PPC::R11); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LD), PPC::R11), + FrameIdx)); + NewMIs.push_back(BuildMI(TII.get(PPC::MTLR8)).addReg(PPC::R11)); } } else if (RC == PPC::F8RCRegisterClass) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LFD), DestReg), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LFD), DestReg), + FrameIdx)); } else if (RC == PPC::F4RCRegisterClass) { - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LFS), DestReg), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LFS), DestReg), + FrameIdx)); } else if (RC == PPC::CRRCRegisterClass) { // FIXME: We use R0 here, because it isn't available for RA. - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::LWZ), PPC::R0), FrameIdx); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::LWZ), PPC::R0), + FrameIdx)); // If the reloaded register isn't CR0, shift the bits right so that they are // in the right CR's slot. if (DestReg != PPC::CR0) { unsigned ShiftBits = PPCRegisterInfo::getRegisterNumbering(DestReg)*4; // rlwinm r11, r11, 32-ShiftBits, 0, 31. - BuildMI(MBB, MI, TII.get(PPC::RLWINM), PPC::R0) - .addReg(PPC::R0).addImm(32-ShiftBits).addImm(0).addImm(31); + NewMIs.push_back(BuildMI(TII.get(PPC::RLWINM), PPC::R0) + .addReg(PPC::R0).addImm(32-ShiftBits).addImm(0).addImm(31)); } - BuildMI(MBB, MI, TII.get(PPC::MTCRF), DestReg).addReg(PPC::R0); + NewMIs.push_back(BuildMI(TII.get(PPC::MTCRF), DestReg).addReg(PPC::R0)); } else if (RC == PPC::VRRCRegisterClass) { // We don't have indexed addressing for vector loads. Emit: // R0 = ADDI FI# // Dest = LVX 0, R0 // // FIXME: We use R0 here, because it isn't available for RA. - addFrameReference(BuildMI(MBB, MI, TII.get(PPC::ADDI), PPC::R0), - FrameIdx, 0, 0); - BuildMI(MBB, MI, TII.get(PPC::LVX),DestReg).addReg(PPC::R0).addReg(PPC::R0); + NewMIs.push_back(addFrameReference(BuildMI(TII.get(PPC::ADDI), PPC::R0), + FrameIdx, 0, 0)); + NewMIs.push_back(BuildMI(TII.get(PPC::LVX),DestReg).addReg(PPC::R0) + .addReg(PPC::R0)); + } else { + assert(0 && "Unknown regclass!"); + abort(); + } +} + +void +PPCRegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MI, + unsigned DestReg, int FrameIdx, + const TargetRegisterClass *RC) const { + SmallVector NewMIs; + LoadRegFromStackSlot(TII, DestReg, FrameIdx, RC, NewMIs); + for (unsigned i = 0, e = NewMIs.size(); i != e; ++i) + MBB.insert(MI, NewMIs[i]); +} + +void PPCRegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + if (Addr[0].isFrameIndex()) { + LoadRegFromStackSlot(TII, DestReg, Addr[0].getFrameIndex(), RC, NewMIs); + return; + } + + unsigned Opc = 0; + if (RC == PPC::GPRCRegisterClass) { + assert(DestReg != PPC::LR && "Can't handle this yet!"); + Opc = PPC::LWZ; + } else if (RC == PPC::G8RCRegisterClass) { + assert(DestReg != PPC::LR8 && "Can't handle this yet!"); + Opc = PPC::LD; + } else if (RC == PPC::F8RCRegisterClass) { + Opc = PPC::LFD; + } else if (RC == PPC::F4RCRegisterClass) { + Opc = PPC::LFS; + } else if (RC == PPC::VRRCRegisterClass) { + Opc = PPC::LVX; } else { assert(0 && "Unknown regclass!"); abort(); } + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; } void PPCRegisterInfo::copyRegToReg(MachineBasicBlock &MBB, Modified: llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.h (original) +++ llvm/trunk/lib/Target/PowerPC/PPCRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -40,11 +40,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, unsigned SrcReg, const TargetRegisterClass *DestRC, Modified: llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.cpp?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.cpp Thu Oct 4 20:32:41 2007 @@ -48,6 +48,34 @@ assert(0 && "Can't store this register to stack slot"); } +void SparcRegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == SP::IntRegsRegisterClass) + Opc = SP::STri; + else if (RC == SP::FPRegsRegisterClass) + Opc = SP::STFri; + else if (RC == SP::DFPRegsRegisterClass) + Opc = SP::STDFri; + else + assert(0 && "Can't load this register"); + MachineInstrBuilder MIB = BuildMI(TII.get(Opc)); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + MIB.addReg(SrcReg, false, false, true); + NewMIs.push_back(MIB); + return; +} + void SparcRegisterInfo:: loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg, int FI, @@ -62,6 +90,33 @@ assert(0 && "Can't load this register from stack slot"); } +void SparcRegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == SP::IntRegsRegisterClass) + Opc = SP::LDri; + else if (RC == SP::FPRegsRegisterClass) + Opc = SP::LDFri; + else if (RC == SP::DFPRegsRegisterClass) + Opc = SP::LDDFri; + else + assert(0 && "Can't load this register"); + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; +} + void SparcRegisterInfo::copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, unsigned DestReg, unsigned SrcReg, Modified: llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.h?rev=42621&r1=42620&r2=42621&view=diff ============================================================================== --- llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.h (original) +++ llvm/trunk/lib/Target/Sparc/SparcRegisterInfo.h Thu Oct 4 20:32:41 2007 @@ -35,11 +35,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, unsigned DestReg, unsigned SrcReg, const TargetRegisterClass *DestRC, From evan.cheng at apple.com Thu Oct 4 20:33:45 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:33:45 -0000 Subject: [llvm-commits] [llvm] r42622 - in /llvm/trunk/lib/Target/IA64: IA64RegisterInfo.cpp IA64RegisterInfo.h Message-ID: <200710050133.l951XjEn021299@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:33:45 2007 New Revision: 42622 URL: http://llvm.org/viewvc/llvm-project?rev=42622&view=rev Log: Forgot these. Modified: llvm/trunk/lib/Target/IA64/IA64RegisterInfo.cpp llvm/trunk/lib/Target/IA64/IA64RegisterInfo.h Modified: llvm/trunk/lib/Target/IA64/IA64RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/IA64/IA64RegisterInfo.cpp?rev=42622&r1=42621&r2=42622&view=diff ============================================================================== --- llvm/trunk/lib/Target/IA64/IA64RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/IA64/IA64RegisterInfo.cpp Thu Oct 4 20:33:45 2007 @@ -60,6 +60,38 @@ "sorry, I don't know how to store this sort of reg in the stack\n"); } +void IA64RegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == IA64::FPRegisterClass) { + Opc = IA64::STF8; + } else if (RC == IA64::GRRegisterClass) { + Opc = IA64::ST8; + } else if (RC == IA64::PRRegisterClass) { + Opc = IA64::ST1; + } else { + assert(0 && + "sorry, I don't know how to store this sort of reg\n"); + } + + MachineInstrBuilder MIB = BuildMI(TII.get(Opc)); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + MIB.addReg(SrcReg, false, false, true); + NewMIs.push_back(MIB); + return; + +} + void IA64RegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, int FrameIdx, @@ -80,6 +112,36 @@ "sorry, I don't know how to load this sort of reg from the stack\n"); } +void IA64RegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = 0; + if (RC == IA64::FPRegisterClass) { + Opc = IA64::LDF8; + } else if (RC == IA64::GRRegisterClass) { + Opc = IA64::LD8; + } else if (RC == IA64::PRRegisterClass) { + Opc = IA64::LD1; + } else { + assert(0 && + "sorry, I don't know how to store this sort of reg\n"); + } + + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) { + MachineOperand &MO = Addr[i]; + if (MO.isRegister()) + MIB.addReg(MO.getReg()); + else if (MO.isImmediate()) + MIB.addImm(MO.getImmedValue()); + else + MIB.addFrameIndex(MO.getFrameIndex()); + } + NewMIs.push_back(MIB); + return; +} + void IA64RegisterInfo::copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, unsigned SrcReg, Modified: llvm/trunk/lib/Target/IA64/IA64RegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/IA64/IA64RegisterInfo.h?rev=42622&r1=42621&r2=42622&view=diff ============================================================================== --- llvm/trunk/lib/Target/IA64/IA64RegisterInfo.h (original) +++ llvm/trunk/lib/Target/IA64/IA64RegisterInfo.h Thu Oct 4 20:33:45 2007 @@ -34,11 +34,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, unsigned SrcReg, From evan.cheng at apple.com Thu Oct 4 20:34:14 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:34:14 -0000 Subject: [llvm-commits] [llvm] r42623 - /llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Message-ID: <200710050134.l951YEof021335@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:34:14 2007 New Revision: 42623 URL: http://llvm.org/viewvc/llvm-project?rev=42623&view=rev Log: Not needed any more. Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=42623&r1=42622&r2=42623&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original) +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Thu Oct 4 20:34:14 2007 @@ -31,7 +31,6 @@ #include "llvm/CodeGen/MachineInstrBuilder.h" #include "llvm/CodeGen/SelectionDAG.h" #include "llvm/CodeGen/SSARegMap.h" -#include "llvm/Support/CommandLine.h" #include "llvm/Support/MathExtras.h" #include "llvm/Target/TargetOptions.h" #include "llvm/ADT/StringExtras.h" From evan.cheng at apple.com Thu Oct 4 20:34:55 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:34:55 -0000 Subject: [llvm-commits] [llvm] r42624 - in /llvm/trunk/lib/Target/X86: X86RegisterInfo.cpp X86RegisterInfo.h Message-ID: <200710050134.l951YtmT021371@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:34:55 2007 New Revision: 42624 URL: http://llvm.org/viewvc/llvm-project?rev=42624&view=rev Log: Added storeRegToAddr, loadRegFromAddr, and unfoldMemoryOperand's. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp llvm/trunk/lib/Target/X86/X86RegisterInfo.h Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42624&r1=42623&r2=42624&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Thu Oct 4 20:34:55 2007 @@ -26,6 +26,7 @@ #include "llvm/CodeGen/MachineFunction.h" #include "llvm/CodeGen/MachineFrameInfo.h" #include "llvm/CodeGen/MachineLocation.h" +#include "llvm/CodeGen/SSARegMap.h" #include "llvm/Target/TargetAsmInfo.h" #include "llvm/Target/TargetFrameInfo.h" #include "llvm/Target/TargetInstrInfo.h" @@ -232,82 +233,85 @@ unsigned MemOp = OpTbl2Addr[i][1]; if (!RegOp2MemOpTable2Addr.insert(std::make_pair((unsigned*)RegOp, MemOp))) assert(false && "Duplicated entries?"); - if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + unsigned AuxInfo = 0 | (1 << 4) | (1 << 5); // Index 0,folded load and store + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, + std::make_pair(RegOp, AuxInfo)))) AmbEntries.push_back(MemOp); } - static const unsigned OpTbl0[][2] = { - { X86::CALL32r, X86::CALL32m }, - { X86::CALL64r, X86::CALL64m }, - { X86::CMP16ri, X86::CMP16mi }, - { X86::CMP16ri8, X86::CMP16mi8 }, - { X86::CMP32ri, X86::CMP32mi }, - { X86::CMP32ri8, X86::CMP32mi8 }, - { X86::CMP64ri32, X86::CMP64mi32 }, - { X86::CMP64ri8, X86::CMP64mi8 }, - { X86::CMP8ri, X86::CMP8mi }, - { X86::DIV16r, X86::DIV16m }, - { X86::DIV32r, X86::DIV32m }, - { X86::DIV64r, X86::DIV64m }, - { X86::DIV8r, X86::DIV8m }, - { X86::FsMOVAPDrr, X86::MOVSDmr }, - { X86::FsMOVAPSrr, X86::MOVSSmr }, - { X86::IDIV16r, X86::IDIV16m }, - { X86::IDIV32r, X86::IDIV32m }, - { X86::IDIV64r, X86::IDIV64m }, - { X86::IDIV8r, X86::IDIV8m }, - { X86::IMUL16r, X86::IMUL16m }, - { X86::IMUL32r, X86::IMUL32m }, - { X86::IMUL64r, X86::IMUL64m }, - { X86::IMUL8r, X86::IMUL8m }, - { X86::JMP32r, X86::JMP32m }, - { X86::JMP64r, X86::JMP64m }, - { X86::MOV16ri, X86::MOV16mi }, - { X86::MOV16rr, X86::MOV16mr }, - { X86::MOV32ri, X86::MOV32mi }, - { X86::MOV32rr, X86::MOV32mr }, - { X86::MOV64ri32, X86::MOV64mi32 }, - { X86::MOV64rr, X86::MOV64mr }, - { X86::MOV8ri, X86::MOV8mi }, - { X86::MOV8rr, X86::MOV8mr }, - { X86::MOVAPDrr, X86::MOVAPDmr }, - { X86::MOVAPSrr, X86::MOVAPSmr }, - { X86::MOVPDI2DIrr, X86::MOVPDI2DImr }, - { X86::MOVPQIto64rr,X86::MOVPQIto64mr }, - { X86::MOVPS2SSrr, X86::MOVPS2SSmr }, - { X86::MOVSDrr, X86::MOVSDmr }, - { X86::MOVSDto64rr, X86::MOVSDto64mr }, - { X86::MOVSS2DIrr, X86::MOVSS2DImr }, - { X86::MOVSSrr, X86::MOVSSmr }, - { X86::MOVUPDrr, X86::MOVUPDmr }, - { X86::MOVUPSrr, X86::MOVUPSmr }, - { X86::MUL16r, X86::MUL16m }, - { X86::MUL32r, X86::MUL32m }, - { X86::MUL64r, X86::MUL64m }, - { X86::MUL8r, X86::MUL8m }, - { X86::SETAEr, X86::SETAEm }, - { X86::SETAr, X86::SETAm }, - { X86::SETBEr, X86::SETBEm }, - { X86::SETBr, X86::SETBm }, - { X86::SETEr, X86::SETEm }, - { X86::SETGEr, X86::SETGEm }, - { X86::SETGr, X86::SETGm }, - { X86::SETLEr, X86::SETLEm }, - { X86::SETLr, X86::SETLm }, - { X86::SETNEr, X86::SETNEm }, - { X86::SETNPr, X86::SETNPm }, - { X86::SETNSr, X86::SETNSm }, - { X86::SETPr, X86::SETPm }, - { X86::SETSr, X86::SETSm }, - { X86::TAILJMPr, X86::TAILJMPm }, - { X86::TEST16ri, X86::TEST16mi }, - { X86::TEST32ri, X86::TEST32mi }, - { X86::TEST64ri32, X86::TEST64mi32 }, - { X86::TEST8ri, X86::TEST8mi }, - { X86::XCHG16rr, X86::XCHG16mr }, - { X86::XCHG32rr, X86::XCHG32mr }, - { X86::XCHG64rr, X86::XCHG64mr }, - { X86::XCHG8rr, X86::XCHG8mr } + // If the third value is 1, then it's folding either a load or a store. + static const unsigned OpTbl0[][3] = { + { X86::CALL32r, X86::CALL32m, 1 }, + { X86::CALL64r, X86::CALL64m, 1 }, + { X86::CMP16ri, X86::CMP16mi, 1 }, + { X86::CMP16ri8, X86::CMP16mi8, 1 }, + { X86::CMP32ri, X86::CMP32mi, 1 }, + { X86::CMP32ri8, X86::CMP32mi8, 1 }, + { X86::CMP64ri32, X86::CMP64mi32, 1 }, + { X86::CMP64ri8, X86::CMP64mi8, 1 }, + { X86::CMP8ri, X86::CMP8mi, 1 }, + { X86::DIV16r, X86::DIV16m, 1 }, + { X86::DIV32r, X86::DIV32m, 1 }, + { X86::DIV64r, X86::DIV64m, 1 }, + { X86::DIV8r, X86::DIV8m, 1 }, + { X86::FsMOVAPDrr, X86::MOVSDmr, 0 }, + { X86::FsMOVAPSrr, X86::MOVSSmr, 0 }, + { X86::IDIV16r, X86::IDIV16m, 1 }, + { X86::IDIV32r, X86::IDIV32m, 1 }, + { X86::IDIV64r, X86::IDIV64m, 1 }, + { X86::IDIV8r, X86::IDIV8m, 1 }, + { X86::IMUL16r, X86::IMUL16m, 1 }, + { X86::IMUL32r, X86::IMUL32m, 1 }, + { X86::IMUL64r, X86::IMUL64m, 1 }, + { X86::IMUL8r, X86::IMUL8m, 1 }, + { X86::JMP32r, X86::JMP32m, 1 }, + { X86::JMP64r, X86::JMP64m, 1 }, + { X86::MOV16ri, X86::MOV16mi, 0 }, + { X86::MOV16rr, X86::MOV16mr, 0 }, + { X86::MOV32ri, X86::MOV32mi, 0 }, + { X86::MOV32rr, X86::MOV32mr, 0 }, + { X86::MOV64ri32, X86::MOV64mi32, 0 }, + { X86::MOV64rr, X86::MOV64mr, 0 }, + { X86::MOV8ri, X86::MOV8mi, 0 }, + { X86::MOV8rr, X86::MOV8mr, 0 }, + { X86::MOVAPDrr, X86::MOVAPDmr, 0 }, + { X86::MOVAPSrr, X86::MOVAPSmr, 0 }, + { X86::MOVPDI2DIrr, X86::MOVPDI2DImr, 0 }, + { X86::MOVPQIto64rr,X86::MOVPQIto64mr, 0 }, + { X86::MOVPS2SSrr, X86::MOVPS2SSmr, 0 }, + { X86::MOVSDrr, X86::MOVSDmr, 0 }, + { X86::MOVSDto64rr, X86::MOVSDto64mr, 0 }, + { X86::MOVSS2DIrr, X86::MOVSS2DImr, 0 }, + { X86::MOVSSrr, X86::MOVSSmr, 0 }, + { X86::MOVUPDrr, X86::MOVUPDmr, 0 }, + { X86::MOVUPSrr, X86::MOVUPSmr, 0 }, + { X86::MUL16r, X86::MUL16m, 1 }, + { X86::MUL32r, X86::MUL32m, 1 }, + { X86::MUL64r, X86::MUL64m, 1 }, + { X86::MUL8r, X86::MUL8m, 1 }, + { X86::SETAEr, X86::SETAEm, 0 }, + { X86::SETAr, X86::SETAm, 0 }, + { X86::SETBEr, X86::SETBEm, 0 }, + { X86::SETBr, X86::SETBm, 0 }, + { X86::SETEr, X86::SETEm, 0 }, + { X86::SETGEr, X86::SETGEm, 0 }, + { X86::SETGr, X86::SETGm, 0 }, + { X86::SETLEr, X86::SETLEm, 0 }, + { X86::SETLr, X86::SETLm, 0 }, + { X86::SETNEr, X86::SETNEm, 0 }, + { X86::SETNPr, X86::SETNPm, 0 }, + { X86::SETNSr, X86::SETNSm, 0 }, + { X86::SETPr, X86::SETPm, 0 }, + { X86::SETSr, X86::SETSm, 0 }, + { X86::TAILJMPr, X86::TAILJMPm, 1 }, + { X86::TEST16ri, X86::TEST16mi, 1 }, + { X86::TEST32ri, X86::TEST32mi, 1 }, + { X86::TEST64ri32, X86::TEST64mi32, 1 }, + { X86::TEST8ri, X86::TEST8mi, 1 }, + { X86::XCHG16rr, X86::XCHG16mr, 0 }, + { X86::XCHG32rr, X86::XCHG32mr, 0 }, + { X86::XCHG64rr, X86::XCHG64mr, 0 }, + { X86::XCHG8rr, X86::XCHG8mr, 0 } }; for (unsigned i = 0, e = array_lengthof(OpTbl0); i != e; ++i) { @@ -315,7 +319,11 @@ unsigned MemOp = OpTbl0[i][1]; if (!RegOp2MemOpTable0.insert(std::make_pair((unsigned*)RegOp, MemOp))) assert(false && "Duplicated entries?"); - if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + unsigned FoldedLoad = OpTbl0[i][2]; + // Index 0, folded load or store. + unsigned AuxInfo = 0 | (FoldedLoad << 4) | ((FoldedLoad^1) << 5); + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, + std::make_pair(RegOp, AuxInfo)))) AmbEntries.push_back(MemOp); } @@ -436,7 +444,9 @@ unsigned MemOp = OpTbl1[i][1]; if (!RegOp2MemOpTable1.insert(std::make_pair((unsigned*)RegOp, MemOp))) assert(false && "Duplicated entries?"); - if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + unsigned AuxInfo = 1 | (1 << 4); // Index 1, folded load + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, + std::make_pair(RegOp, AuxInfo)))) AmbEntries.push_back(MemOp); } @@ -627,7 +637,9 @@ unsigned MemOp = OpTbl2[i][1]; if (!RegOp2MemOpTable2.insert(std::make_pair((unsigned*)RegOp, MemOp))) assert(false && "Duplicated entries?"); - if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, RegOp))) + unsigned AuxInfo = 2 | (1 << 4); // Index 1, folded load + if (!MemOp2RegOpTable.insert(std::make_pair((unsigned*)MemOp, + std::make_pair(RegOp, AuxInfo)))) AmbEntries.push_back(MemOp); } @@ -722,11 +734,30 @@ return true; } -void X86RegisterInfo::storeRegToStackSlot(MachineBasicBlock &MBB, - MachineBasicBlock::iterator MI, - unsigned SrcReg, int FrameIdx, - const TargetRegisterClass *RC) const { - unsigned Opc; +static const MachineInstrBuilder &X86InstrAddOperand(MachineInstrBuilder &MIB, + MachineOperand &MO) { + if (MO.isRegister()) + MIB = MIB.addReg(MO.getReg(), MO.isDef(), MO.isImplicit()); + else if (MO.isImmediate()) + MIB = MIB.addImm(MO.getImm()); + else if (MO.isFrameIndex()) + MIB = MIB.addFrameIndex(MO.getFrameIndex()); + else if (MO.isGlobalAddress()) + MIB = MIB.addGlobalAddress(MO.getGlobal(), MO.getOffset()); + else if (MO.isConstantPoolIndex()) + MIB = MIB.addConstantPoolIndex(MO.getConstantPoolIndex(), MO.getOffset()); + else if (MO.isJumpTableIndex()) + MIB = MIB.addJumpTableIndex(MO.getJumpTableIndex()); + else if (MO.isExternalSymbol()) + MIB = MIB.addExternalSymbol(MO.getSymbolName()); + else + assert(0 && "Unknown operand for X86InstrAddOperand!"); + + return MIB; +} + +static unsigned getStoreRegOpcode(const TargetRegisterClass *RC) { + unsigned Opc = 0; if (RC == &X86::GR64RegClass) { Opc = X86::MOV64mr; } else if (RC == &X86::GR32RegClass) { @@ -757,15 +788,33 @@ assert(0 && "Unknown regclass"); abort(); } + + return Opc; +} + +void X86RegisterInfo::storeRegToStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MI, + unsigned SrcReg, int FrameIdx, + const TargetRegisterClass *RC) const { + unsigned Opc = getStoreRegOpcode(RC); addFrameReference(BuildMI(MBB, MI, TII.get(Opc)), FrameIdx) .addReg(SrcReg, false, false, true); } -void X86RegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, - MachineBasicBlock::iterator MI, - unsigned DestReg, int FrameIdx, - const TargetRegisterClass *RC) const{ - unsigned Opc; +void X86RegisterInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = getStoreRegOpcode(RC); + MachineInstrBuilder MIB = BuildMI(TII.get(Opc)); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = X86InstrAddOperand(MIB, Addr[i]); + MIB.addReg(SrcReg, false, false, true); + NewMIs.push_back(MIB); +} + +static unsigned getLoadRegOpcode(const TargetRegisterClass *RC) { + unsigned Opc = 0; if (RC == &X86::GR64RegClass) { Opc = X86::MOV64rm; } else if (RC == &X86::GR32RegClass) { @@ -796,9 +845,29 @@ assert(0 && "Unknown regclass"); abort(); } + + return Opc; +} + +void X86RegisterInfo::loadRegFromStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MI, + unsigned DestReg, int FrameIdx, + const TargetRegisterClass *RC) const{ + unsigned Opc = getLoadRegOpcode(RC); addFrameReference(BuildMI(MBB, MI, TII.get(Opc), DestReg), FrameIdx); } +void X86RegisterInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const { + unsigned Opc = getLoadRegOpcode(RC); + MachineInstrBuilder MIB = BuildMI(TII.get(Opc), DestReg); + for (unsigned i = 0, e = Addr.size(); i != e; ++i) + MIB = X86InstrAddOperand(MIB, Addr[i]); + NewMIs.push_back(MIB); +} + void X86RegisterInfo::copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, unsigned SrcReg, @@ -905,28 +974,6 @@ } } -static const MachineInstrBuilder &FuseInstrAddOperand(MachineInstrBuilder &MIB, - MachineOperand &MO) { - if (MO.isRegister()) - MIB = MIB.addReg(MO.getReg(), MO.isDef(), MO.isImplicit()); - else if (MO.isImmediate()) - MIB = MIB.addImm(MO.getImm()); - else if (MO.isFrameIndex()) - MIB = MIB.addFrameIndex(MO.getFrameIndex()); - else if (MO.isGlobalAddress()) - MIB = MIB.addGlobalAddress(MO.getGlobal(), MO.getOffset()); - else if (MO.isConstantPoolIndex()) - MIB = MIB.addConstantPoolIndex(MO.getConstantPoolIndex(), MO.getOffset()); - else if (MO.isJumpTableIndex()) - MIB = MIB.addJumpTableIndex(MO.getJumpTableIndex()); - else if (MO.isExternalSymbol()) - MIB = MIB.addExternalSymbol(MO.getSymbolName()); - else - assert(0 && "Unknown operand for FuseInst!"); - - return MIB; -} - static MachineInstr *FuseTwoAddrInst(unsigned Opcode, SmallVector &MOs, MachineInstr *MI, const TargetInstrInfo &TII) { @@ -936,14 +983,14 @@ MachineInstrBuilder MIB = BuildMI(TII.get(Opcode)); unsigned NumAddrOps = MOs.size(); for (unsigned i = 0; i != NumAddrOps; ++i) - MIB = FuseInstrAddOperand(MIB, MOs[i]); + MIB = X86InstrAddOperand(MIB, MOs[i]); if (NumAddrOps < 4) // FrameIndex only MIB.addImm(1).addReg(0).addImm(0); // Loop over the rest of the ri operands, converting them over. for (unsigned i = 0; i != NumOps; ++i) { MachineOperand &MO = MI->getOperand(i+2); - MIB = FuseInstrAddOperand(MIB, MO); + MIB = X86InstrAddOperand(MIB, MO); } return MIB; } @@ -959,11 +1006,11 @@ assert(MO.isRegister() && "Expected to fold into reg operand!"); unsigned NumAddrOps = MOs.size(); for (unsigned i = 0; i != NumAddrOps; ++i) - MIB = FuseInstrAddOperand(MIB, MOs[i]); + MIB = X86InstrAddOperand(MIB, MOs[i]); if (NumAddrOps < 4) // FrameIndex only MIB.addImm(1).addReg(0).addImm(0); } else { - MIB = FuseInstrAddOperand(MIB, MO); + MIB = X86InstrAddOperand(MIB, MO); } } return MIB; @@ -976,7 +1023,7 @@ unsigned NumAddrOps = MOs.size(); for (unsigned i = 0; i != NumAddrOps; ++i) - MIB = FuseInstrAddOperand(MIB, MOs[i]); + MIB = X86InstrAddOperand(MIB, MOs[i]); if (NumAddrOps < 4) // FrameIndex only MIB.addImm(1).addReg(0).addImm(0); return MIB.addImm(0); @@ -1066,6 +1113,156 @@ return foldMemoryOperand(MI, OpNum, MOs); } +bool X86RegisterInfo::unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI, + SSARegMap *RegMap, + SmallVector &NewMIs) const { + DenseMap >::iterator I = + MemOp2RegOpTable.find((unsigned*)MI->getOpcode()); + if (I == MemOp2RegOpTable.end()) + return false; + unsigned Opc = I->second.first; + unsigned Index = I->second.second & 0xf; + bool HasLoad = I->second.second & (1 << 4); + bool HasStore = I->second.second & (1 << 5); + const TargetInstrDescriptor &TID = TII.get(Opc); + const TargetOperandInfo &TOI = TID.OpInfo[Index]; + const TargetRegisterClass *RC = (TOI.Flags & M_LOOK_UP_PTR_REG_CLASS) + ? TII.getPointerRegClass() : getRegClass(TOI.RegClass); + SmallVector AddrOps; + SmallVector BeforeOps; + SmallVector AfterOps; + SmallVector ImpOps; + for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { + MachineOperand &Op = MI->getOperand(i); + if (i >= Index && i < Index+4) + AddrOps.push_back(Op); + else if (Op.isRegister() && Op.isImplicit()) + ImpOps.push_back(Op); + else if (i < Index) + BeforeOps.push_back(Op); + else if (i > Index) + AfterOps.push_back(Op); + } + + // Emit the load instruction. + unsigned LoadReg = 0; + if (HasLoad) { + LoadReg = RegMap->createVirtualRegister(RC); + loadRegFromAddr(MF, LoadReg, AddrOps, RC, NewMIs); + if (HasStore) { + // Address operands cannot be marked isKill. + for (unsigned i = 1; i != 5; ++i) { + MachineOperand &MO = NewMIs[0]->getOperand(i); + if (MO.isRegister()) + MO.unsetIsKill(); + } + } + } + + // Emit the data processing instruction. + MachineInstrBuilder MIB = BuildMI(TII.get(Opc)); + unsigned StoreReg = 0; + const TargetRegisterClass *DstRC = 0; + if (HasStore) { + const TargetOperandInfo &DstTOI = TID.OpInfo[0]; + DstRC = (DstTOI.Flags & M_LOOK_UP_PTR_REG_CLASS) + ? TII.getPointerRegClass() : getRegClass(DstTOI.RegClass); + StoreReg = RegMap->createVirtualRegister(RC); + MIB.addReg(StoreReg, true); + } + for (unsigned i = 0, e = BeforeOps.size(); i != e; ++i) + MIB = X86InstrAddOperand(MIB, BeforeOps[i]); + if (LoadReg) + MIB.addReg(LoadReg); + for (unsigned i = 0, e = AfterOps.size(); i != e; ++i) + MIB = X86InstrAddOperand(MIB, AfterOps[i]); + NewMIs.push_back(MIB); + + // Emit the store instruction. + if (HasStore) + storeRegToAddr(MF, StoreReg, AddrOps, DstRC, NewMIs); + + return true; +} + + +bool +X86RegisterInfo::unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N, + SmallVector &NewNodes) const { + if (!N->isTargetOpcode()) + return false; + + DenseMap >::iterator I = + MemOp2RegOpTable.find((unsigned*)N->getTargetOpcode()); + if (I == MemOp2RegOpTable.end()) + return false; + unsigned Opc = I->second.first; + unsigned Index = I->second.second & 0xf; + bool HasLoad = I->second.second & (1 << 4); + bool HasStore = I->second.second & (1 << 5); + const TargetInstrDescriptor &TID = TII.get(Opc); + const TargetOperandInfo &TOI = TID.OpInfo[Index]; + const TargetRegisterClass *RC = (TOI.Flags & M_LOOK_UP_PTR_REG_CLASS) + ? TII.getPointerRegClass() : getRegClass(TOI.RegClass); + std::vector AddrOps; + std::vector BeforeOps; + std::vector AfterOps; + unsigned NumOps = N->getNumOperands(); + for (unsigned i = 0; i != NumOps-1; ++i) { + SDOperand Op = N->getOperand(i); + if (i >= Index && i < Index+4) + AddrOps.push_back(Op); + else if (i < Index) + BeforeOps.push_back(Op); + else if (i > Index) + AfterOps.push_back(Op); + } + SDOperand Chain = N->getOperand(NumOps-1); + AddrOps.push_back(Chain); + + // Emit the load instruction. + SDNode *Load = 0; + if (HasLoad) { + MVT::ValueType VT = *RC->vt_begin(); + Load = DAG.getTargetNode(getLoadRegOpcode(RC), VT, MVT::Other, + &AddrOps[0], AddrOps.size()); + NewNodes.push_back(Load); + } + + // Emit the data processing instruction. + std::vector VTs; + const TargetRegisterClass *DstRC = 0; + if (TID.numDefs > 0) { + const TargetOperandInfo &DstTOI = TID.OpInfo[0]; + DstRC = (DstTOI.Flags & M_LOOK_UP_PTR_REG_CLASS) + ? TII.getPointerRegClass() : getRegClass(DstTOI.RegClass); + VTs.push_back(*DstRC->vt_begin()); + } + for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) { + MVT::ValueType VT = N->getValueType(i); + if (VT != MVT::Other && i >= TID.numDefs) + VTs.push_back(VT); + } + if (Load) + BeforeOps.push_back(SDOperand(Load, 0)); + std::copy(AfterOps.begin(), AfterOps.end(), std::back_inserter(BeforeOps)); + SDNode *NewNode= DAG.getTargetNode(Opc, VTs, &BeforeOps[0], BeforeOps.size()); + NewNodes.push_back(NewNode); + + // Emit the store instruction. + if (HasStore) { + AddrOps.pop_back(); + AddrOps.push_back(SDOperand(NewNode, 0)); + AddrOps.push_back(Chain); + SDNode *Store = DAG.getTargetNode(getStoreRegOpcode(DstRC), + MVT::Other, &AddrOps[0], AddrOps.size()); + NewNodes.push_back(Store); + } + + return true; +} + + const unsigned * X86RegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const { static const unsigned CalleeSavedRegs32Bit[] = { Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.h?rev=42624&r1=42623&r2=42624&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.h (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.h Thu Oct 4 20:34:55 2007 @@ -64,7 +64,7 @@ /// MemOp2RegOpTable - Load / store unfolding opcode map. /// - DenseMap MemOp2RegOpTable; + DenseMap > MemOp2RegOpTable; public: X86RegisterInfo(X86TargetMachine &tm, const TargetInstrInfo &tii); @@ -88,11 +88,21 @@ unsigned SrcReg, int FrameIndex, const TargetRegisterClass *RC) const; + void storeRegToAddr(MachineFunction &MF, unsigned SrcReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, int FrameIndex, const TargetRegisterClass *RC) const; + void loadRegFromAddr(MachineFunction &MF, unsigned DestReg, + SmallVector Addr, + const TargetRegisterClass *RC, + SmallVector &NewMIs) const; + void copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, unsigned DestReg, unsigned SrcReg, @@ -122,6 +132,16 @@ unsigned OpNum, MachineInstr* LoadMI) const; + /// unfoldMemoryOperand - Separate a single instruction which folded a load or a + /// a store or a load and a store into two or more instruction. If this is + /// possible, returns true as well as the new instructions by reference. + bool unfoldMemoryOperand(MachineFunction &MF, MachineInstr *MI, + SSARegMap *RegMap, + SmallVector &NewMIs) const; + + bool unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N, + SmallVector &NewNodes) const; + /// getCalleeSavedRegs - Return a null-terminated list of all of the /// callee-save registers on this target. const unsigned *getCalleeSavedRegs(const MachineFunction* MF = 0) const; From evan.cheng at apple.com Thu Oct 4 20:39:18 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:39:18 -0000 Subject: [llvm-commits] [llvm] r42625 - in /llvm/trunk: include/llvm/CodeGen/ScheduleDAG.h lib/CodeGen/SelectionDAG/ScheduleDAG.cpp lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Message-ID: <200710050139.l951dI1P021546@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:39:18 2007 New Revision: 42625 URL: http://llvm.org/viewvc/llvm-project?rev=42625&view=rev Log: If a node that defines a physical register that is expensive to copy. The scheduler will try a number of tricks in order to avoid generating the copies. This may not be possible in case the node produces a chain value that prevent movement. Try unfolding the load from the node before to allow it to be moved / cloned. Modified: llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Modified: llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h?rev=42625&r1=42624&r2=42625&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h (original) +++ llvm/trunk/include/llvm/CodeGen/ScheduleDAG.h Thu Oct 4 20:39:18 2007 @@ -115,7 +115,7 @@ short NumSuccsLeft; // # of succs not scheduled. bool isTwoAddress : 1; // Is a two-address instruction. bool isCommutable : 1; // Is a commutable instruction. - bool hasPhysRegDefs : 1; // Has physreg defs that are being used. + bool hasPhysRegDefs : 1; // Has physreg defs that are being used. bool isPending : 1; // True once pending. bool isAvailable : 1; // True once available. bool isScheduled : 1; // True once scheduled. @@ -297,6 +297,10 @@ /// together nodes with a single SUnit. void BuildSchedUnits(); + /// ComputeLatency - Compute node latency. + /// + void ComputeLatency(SUnit *SU); + /// CalculateDepths, CalculateHeights - Calculate node depth / height. /// void CalculateDepths(); Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp?rev=42625&r1=42624&r2=42625&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp Thu Oct 4 20:39:18 2007 @@ -68,6 +68,7 @@ return SU; } + /// BuildSchedUnits - Build SUnits from the selection dag that we are input. /// This SUnit graph is similar to the SelectionDAG, but represents flagged /// together nodes with a single SUnit. @@ -77,8 +78,6 @@ // invalidated. SUnits.reserve(std::distance(DAG.allnodes_begin(), DAG.allnodes_end())); - const InstrItineraryData &InstrItins = TM.getInstrItineraryData(); - for (SelectionDAG::allnodes_iterator NI = DAG.allnodes_begin(), E = DAG.allnodes_end(); NI != E; ++NI) { if (isPassiveNode(NI)) // Leaf node, e.g. a TargetImmediate. @@ -131,32 +130,8 @@ // Update the SUnit NodeSUnit->Node = N; SUnitMap[N].push_back(NodeSUnit); - - // Compute the latency for the node. We use the sum of the latencies for - // all nodes flagged together into this SUnit. - if (InstrItins.isEmpty()) { - // No latency information. - NodeSUnit->Latency = 1; - } else { - NodeSUnit->Latency = 0; - if (N->isTargetOpcode()) { - unsigned SchedClass = TII->getSchedClass(N->getTargetOpcode()); - InstrStage *S = InstrItins.begin(SchedClass); - InstrStage *E = InstrItins.end(SchedClass); - for (; S != E; ++S) - NodeSUnit->Latency += S->Cycles; - } - for (unsigned i = 0, e = NodeSUnit->FlaggedNodes.size(); i != e; ++i) { - SDNode *FNode = NodeSUnit->FlaggedNodes[i]; - if (FNode->isTargetOpcode()) { - unsigned SchedClass = TII->getSchedClass(FNode->getTargetOpcode()); - InstrStage *S = InstrItins.begin(SchedClass); - InstrStage *E = InstrItins.end(SchedClass); - for (; S != E; ++S) - NodeSUnit->Latency += S->Cycles; - } - } - } + + ComputeLatency(NodeSUnit); } // Pass 2: add the preds, succs, etc. @@ -214,6 +189,36 @@ return; } +void ScheduleDAG::ComputeLatency(SUnit *SU) { + const InstrItineraryData &InstrItins = TM.getInstrItineraryData(); + + // Compute the latency for the node. We use the sum of the latencies for + // all nodes flagged together into this SUnit. + if (InstrItins.isEmpty()) { + // No latency information. + SU->Latency = 1; + } else { + SU->Latency = 0; + if (SU->Node->isTargetOpcode()) { + unsigned SchedClass = TII->getSchedClass(SU->Node->getTargetOpcode()); + InstrStage *S = InstrItins.begin(SchedClass); + InstrStage *E = InstrItins.end(SchedClass); + for (; S != E; ++S) + SU->Latency += S->Cycles; + } + for (unsigned i = 0, e = SU->FlaggedNodes.size(); i != e; ++i) { + SDNode *FNode = SU->FlaggedNodes[i]; + if (FNode->isTargetOpcode()) { + unsigned SchedClass = TII->getSchedClass(FNode->getTargetOpcode()); + InstrStage *S = InstrItins.begin(SchedClass); + InstrStage *E = InstrItins.end(SchedClass); + for (; S != E; ++S) + SU->Latency += S->Cycles; + } + } + } +} + void ScheduleDAG::CalculateDepths() { std::vector > WorkList; for (unsigned i = 0, e = SUnits.size(); i != e; ++i) Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp?rev=42625&r1=42624&r2=42625&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Thu Oct 4 20:39:18 2007 @@ -34,6 +34,7 @@ using namespace llvm; STATISTIC(NumBacktracks, "Number of times scheduler backtraced"); +STATISTIC(NumUnfolds, "Number of nodes unfolded"); STATISTIC(NumDups, "Number of duplicated nodes"); STATISTIC(NumCCCopies, "Number of cross class copies"); @@ -385,32 +386,145 @@ ++NumBacktracks; } -/// isSafeToCopy - True if the SUnit for the given SDNode can safely cloned, -/// i.e. the node does not produce a flag, it does not read a flag and it does -/// not have an incoming chain. -static bool isSafeToCopy(SDNode *N) { +/// CopyAndMoveSuccessors - Clone the specified node and move its scheduled +/// successors to the newly created node. +SUnit *ScheduleDAGRRList::CopyAndMoveSuccessors(SUnit *SU) { + if (SU->FlaggedNodes.size()) + return NULL; + + SDNode *N = SU->Node; if (!N) - return true; + return NULL; + SUnit *NewSU; for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) if (N->getValueType(i) == MVT::Flag) - return false; + return NULL; + bool TryUnfold = false; for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) { const SDOperand &Op = N->getOperand(i); MVT::ValueType VT = Op.Val->getValueType(Op.ResNo); - if (VT == MVT::Other || VT == MVT::Flag) - return false; - } + if (VT == MVT::Flag) + return NULL; + else if (VT == MVT::Other) + TryUnfold = true; + } + + if (TryUnfold) { + SmallVector NewNodes; + if (!MRI->unfoldMemoryOperand(DAG, N, NewNodes)) + return NULL; + + DOUT << "Unfolding SU # " << SU->NodeNum << "\n"; + assert(NewNodes.size() == 2 && "Expected a load folding node!"); + + N = NewNodes[1]; + SDNode *LoadNode = NewNodes[0]; + std::vector Deleted; + unsigned NumVals = N->getNumValues(); + unsigned OldNumVals = SU->Node->getNumValues(); + for (unsigned i = 0; i != NumVals; ++i) + DAG.ReplaceAllUsesOfValueWith(SDOperand(SU->Node, i), + SDOperand(N, i), Deleted); + DAG.ReplaceAllUsesOfValueWith(SDOperand(SU->Node, OldNumVals-1), + SDOperand(LoadNode, 1), Deleted); + + SUnit *LoadSU = NewSUnit(LoadNode); + SUnit *NewSU = NewSUnit(N); + SUnitMap[LoadNode].push_back(LoadSU); + SUnitMap[N].push_back(NewSU); + const TargetInstrDescriptor *TID = &TII->get(LoadNode->getTargetOpcode()); + for (unsigned i = 0; i != TID->numOperands; ++i) { + if (TID->getOperandConstraint(i, TOI::TIED_TO) != -1) { + LoadSU->isTwoAddress = true; + break; + } + } + if (TID->Flags & M_COMMUTABLE) + LoadSU->isCommutable = true; - return true; -} + TID = &TII->get(N->getTargetOpcode()); + for (unsigned i = 0; i != TID->numOperands; ++i) { + if (TID->getOperandConstraint(i, TOI::TIED_TO) != -1) { + NewSU->isTwoAddress = true; + break; + } + } + if (TID->Flags & M_COMMUTABLE) + NewSU->isCommutable = true; -/// CopyAndMoveSuccessors - Clone the specified node and move its scheduled -/// successors to the newly created node. -SUnit *ScheduleDAGRRList::CopyAndMoveSuccessors(SUnit *SU) { - DOUT << "Duplicating SU # " << SU->NodeNum << "\n"; + // FIXME: Calculate height / depth and propagate the changes? + LoadSU->Depth = NewSU->Depth = SU->Depth; + LoadSU->Height = NewSU->Height = SU->Height; + ComputeLatency(LoadSU); + ComputeLatency(NewSU); + + SUnit *ChainPred = NULL; + SmallVector ChainSuccs; + SmallVector LoadPreds; + SmallVector NodePreds; + SmallVector NodeSuccs; + for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end(); + I != E; ++I) { + if (I->isCtrl) + ChainPred = I->Dep; + else if (I->Dep->Node && I->Dep->Node->isOperand(LoadNode)) + LoadPreds.push_back(SDep(I->Dep, I->Reg, I->Cost, false, false)); + else + NodePreds.push_back(SDep(I->Dep, I->Reg, I->Cost, false, false)); + } + for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end(); + I != E; ++I) { + if (I->isCtrl) + ChainSuccs.push_back(SDep(I->Dep, I->Reg, I->Cost, + I->isCtrl, I->isSpecial)); + else + NodeSuccs.push_back(SDep(I->Dep, I->Reg, I->Cost, + I->isCtrl, I->isSpecial)); + } - SUnit *NewSU = Clone(SU); + SU->removePred(ChainPred, true, false); + LoadSU->addPred(ChainPred, true, false); + for (unsigned i = 0, e = LoadPreds.size(); i != e; ++i) { + SDep *Pred = &LoadPreds[i]; + SU->removePred(Pred->Dep, Pred->isCtrl, Pred->isSpecial); + LoadSU->addPred(Pred->Dep, Pred->isCtrl, Pred->isSpecial, + Pred->Reg, Pred->Cost); + } + for (unsigned i = 0, e = NodePreds.size(); i != e; ++i) { + SDep *Pred = &NodePreds[i]; + SU->removePred(Pred->Dep, Pred->isCtrl, Pred->isSpecial); + NewSU->addPred(Pred->Dep, Pred->isCtrl, Pred->isSpecial, + Pred->Reg, Pred->Cost); + } + for (unsigned i = 0, e = NodeSuccs.size(); i != e; ++i) { + SDep *Succ = &NodeSuccs[i]; + Succ->Dep->removePred(SU, Succ->isCtrl, Succ->isSpecial); + Succ->Dep->addPred(NewSU, Succ->isCtrl, Succ->isSpecial, + Succ->Reg, Succ->Cost); + } + for (unsigned i = 0, e = ChainSuccs.size(); i != e; ++i) { + SDep *Succ = &ChainSuccs[i]; + Succ->Dep->removePred(SU, Succ->isCtrl, Succ->isSpecial); + Succ->Dep->addPred(LoadSU, Succ->isCtrl, Succ->isSpecial, + Succ->Reg, Succ->Cost); + } + NewSU->addPred(LoadSU, false, false); + + AvailableQueue->addNode(LoadSU); + AvailableQueue->addNode(NewSU); + + ++NumUnfolds; + + if (NewSU->NumSuccsLeft == 0) { + NewSU->isAvailable = true; + return NewSU; + } else + SU = NewSU; + } + + DOUT << "Duplicating SU # " << SU->NodeNum << "\n"; + NewSU = Clone(SU); // New SUnit has the exact same predecessors. for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end(); @@ -452,6 +566,7 @@ const TargetRegisterClass *DestRC, const TargetRegisterClass *SrcRC, SmallVector &Copies) { + abort(); SUnit *CopyFromSU = NewSUnit(NULL); CopyFromSU->CopySrcRC = SrcRC; CopyFromSU->CopyDstRC = DestRC; @@ -640,10 +755,8 @@ assert(LRegs.size() == 1 && "Can't handle this yet!"); unsigned Reg = LRegs[0]; SUnit *LRDef = LiveRegDefs[Reg]; - SUnit *NewDef; - if (isSafeToCopy(LRDef->Node)) - NewDef = CopyAndMoveSuccessors(LRDef); - else { + SUnit *NewDef = CopyAndMoveSuccessors(LRDef); + if (!NewDef) { // Issue expensive cross register class copies. MVT::ValueType VT = getPhysicalRegisterVT(LRDef->Node, Reg, TII); const TargetRegisterClass *RC = From evan.cheng at apple.com Thu Oct 4 20:39:40 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:39:40 -0000 Subject: [llvm-commits] [llvm] r42626 - /llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Message-ID: <200710050139.l951deU9021571@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:39:40 2007 New Revision: 42626 URL: http://llvm.org/viewvc/llvm-project?rev=42626&view=rev Log: Oops. Didn't mean to leave this in. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp?rev=42626&r1=42625&r2=42626&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Thu Oct 4 20:39:40 2007 @@ -566,7 +566,6 @@ const TargetRegisterClass *DestRC, const TargetRegisterClass *SrcRC, SmallVector &Copies) { - abort(); SUnit *CopyFromSU = NewSUnit(NULL); CopyFromSU->CopySrcRC = SrcRC; CopyFromSU->CopyDstRC = DestRC; From evan.cheng at apple.com Thu Oct 4 20:42:35 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:42:35 -0000 Subject: [llvm-commits] [llvm] r42627 - /llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Message-ID: <200710050142.l951gZaM021672@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:42:35 2007 New Revision: 42627 URL: http://llvm.org/viewvc/llvm-project?rev=42627&view=rev Log: Chain producing nodes cannot be moved, not chain reading nodes. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp?rev=42627&r1=42626&r2=42627&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Thu Oct 4 20:42:35 2007 @@ -397,17 +397,19 @@ return NULL; SUnit *NewSU; - for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) - if (N->getValueType(i) == MVT::Flag) - return NULL; bool TryUnfold = false; + for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) { + MVT::ValueType VT = N->getValueType(i); + if (VT == MVT::Flag) + return NULL; + else if (VT == MVT::Other) + TryUnfold = true; + } for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) { const SDOperand &Op = N->getOperand(i); MVT::ValueType VT = Op.Val->getValueType(Op.ResNo); if (VT == MVT::Flag) return NULL; - else if (VT == MVT::Other) - TryUnfold = true; } if (TryUnfold) { From evan.cheng at apple.com Thu Oct 4 20:44:22 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 01:44:22 -0000 Subject: [llvm-commits] [llvm] r42628 - /llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll Message-ID: <200710050144.l951iMhF021771@zion.cs.uiuc.edu> Author: evancheng Date: Thu Oct 4 20:44:22 2007 New Revision: 42628 URL: http://llvm.org/viewvc/llvm-project?rev=42628&view=rev Log: New test case. Added: llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll Added: llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll?rev=42628&view=auto ============================================================================== --- llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll (added) +++ llvm/trunk/test/CodeGen/X86/2007-10-04-AvoidEFLAGSCopy.ll Thu Oct 4 20:44:22 2007 @@ -0,0 +1,20 @@ +; RUN: llvm-as < %s | llc | not grep pushf + + %struct.gl_texture_image = type { i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i8*, i8* } + %struct.gl_texture_object = type { i32, i32, i32, float, [4 x i32], i32, i32, i32, i32, i32, float, [11 x %struct.gl_texture_image*], [1024 x i8], i32, i32, i32, i8, i8*, i8, void (%struct.gl_texture_object*, i32, float*, float*, float*, float*, i8*, i8*, i8*, i8*)*, %struct.gl_texture_object* } + +define fastcc void @sample_3d_linear(%struct.gl_texture_object* %tObj, %struct.gl_texture_image* %img, float %s, float %t, float %r, i8* %red, i8* %green, i8* %blue, i8* %alpha) { +entry: + %tmp15 = load i32* null, align 4 ; [#uses=1] + %tmp16 = icmp eq i32 %tmp15, 10497 ; [#uses=1] + %tmp2152 = call float @floorf( float 0.000000e+00 ) ; [#uses=0] + br i1 %tmp16, label %cond_true, label %cond_false + +cond_true: ; preds = %entry + ret void + +cond_false: ; preds = %entry + ret void +} + +declare float @floorf(float) From dberlin at dberlin.org Fri Oct 5 00:10:54 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Fri, 05 Oct 2007 05:10:54 -0000 Subject: [llvm-commits] [llvm] r42629 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710050510.l955As2N028253@zion.cs.uiuc.edu> Author: dannyb Date: Fri Oct 5 00:10:53 2007 New Revision: 42629 URL: http://llvm.org/viewvc/llvm-project?rev=42629&view=rev Log: Fix intersectWithComplement bug noticed by Curtis Dunham Optimize |= case for empty RHS Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42629&r1=42628&r2=42629&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Fri Oct 5 00:10:53 2007 @@ -580,8 +580,8 @@ ElementListIter Iter1 = Elements.begin(); ElementListConstIter Iter2 = RHS.Elements.begin(); - // Check if both bitmaps are empty - if (Elements.empty() && RHS.Elements.empty()) + // If RHS is empty, we are done + if (RHS.Elements.empty()) return false; while (Iter2 != RHS.Elements.end()) { @@ -671,9 +671,7 @@ } ++Iter2; } else { - ElementListIter IterTmp = Iter1; ++Iter1; - Elements.erase(IterTmp); } } CurrElementIter = Elements.begin(); From evan.cheng at apple.com Fri Oct 5 02:06:19 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 5 Oct 2007 00:06:19 -0700 Subject: [llvm-commits] [llvm] r40628 - in /llvm/trunk: lib/Target/X86/X86InstrFPStack.td lib/Target/X86/X86InstrFormats.td lib/Target/X86/X86InstrInfo.td lib/Target/X86/X86InstrMMX.td lib/Target/X86/X86InstrSSE.td lib/Target/X86/X86InstrX86-64.td test/CodeGen/X86/2007-07-31-VInsertBug.ll In-Reply-To: References: <200707310804.l6V844EB007549@zion.cs.uiuc.edu> <31A3D1FE-643E-4ABA-896D-DDC1B7404B60@apple.com> Message-ID: <3AD4F30D-122A-4B9C-9707-C9E60ACDC761@apple.com> I'll look at this. Evan On Sep 28, 2007, at 12:17 PM, Chris Lattner wrote: > Evan, ping? > > -Chris > > On Aug 1, 2007, at 5:47 PM, Chris Lattner wrote: > >> On Jul 31, 2007, at 1:04 AM, Evan Cheng wrote: >>> Author: evancheng >>> Date: Tue Jul 31 03:04:03 2007 >>> New Revision: 40628 >>> >>> URL: http://llvm.org/viewvc/llvm-project?rev=40628&view=rev >>> Log: >>> Redo and generalize previously removed opt for pinsrw: (vextract >>> (v4i32 bc (v4f32 s2v (f32 load ))), 0) -> (i32 load ) >> >> >>> + >>> +// (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 >>> load $addr) >>> +def : Pat<(vector_extract >>> + (bc_v4i32 (v4f32 (scalar_to_vector (loadf32 addr: >>> $src)))), (iPTR 0)), >>> + (MOV32rm addr:$src)>; >>> +def : Pat<(vector_extract >>> + (bc_v2i64 (v2f64 (scalar_to_vector (loadf64 addr: >>> $src)))), (iPTR 0)), >>> + (MOV64rm addr:$src)>; >> >> Would it be possible (and would it make sense) to do this in the dag >> combiner rather than as a target-specific pattern? >> >> -Chris >> _______________________________________________ >> llvm-commits mailing list >> llvm-commits at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From evan.cheng at apple.com Fri Oct 5 03:04:02 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 08:04:02 -0000 Subject: [llvm-commits] [test-suite] r42630 - /test-suite/trunk/Makefile.programs Message-ID: <200710050804.l958420V008347@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 03:04:01 2007 New Revision: 42630 URL: http://llvm.org/viewvc/llvm-project?rev=42630&view=rev Log: Testing convertToThreeeAddress as X86 llcbeta. Modified: test-suite/trunk/Makefile.programs Modified: test-suite/trunk/Makefile.programs URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/Makefile.programs?rev=42630&r1=42629&r2=42630&view=diff ============================================================================== --- test-suite/trunk/Makefile.programs (original) +++ test-suite/trunk/Makefile.programs Fri Oct 5 03:04:01 2007 @@ -221,7 +221,8 @@ LLCBETAOPTION := -sched=simple endif ifeq ($(ARCH),x86) -LLCBETAOPTION := -regalloc=local -fast +LLCBETAOPTION := -enable-x86-conv-3-addr +#-regalloc=local -fast #-disable-rematerialization #-disable-fp-elim #-regalloc=bigblock -fast From evan.cheng at apple.com Fri Oct 5 03:04:02 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 08:04:02 -0000 Subject: [llvm-commits] [llvm] r42630 - /llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Message-ID: <200710050804.l95842x8008353@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 03:04:01 2007 New Revision: 42630 URL: http://llvm.org/viewvc/llvm-project?rev=42630&view=rev Log: Testing convertToThreeeAddress as X86 llcbeta. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42630&r1=42629&r2=42630&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 03:04:01 2007 @@ -22,8 +22,15 @@ #include "llvm/CodeGen/LiveVariables.h" #include "llvm/CodeGen/SSARegMap.h" #include "llvm/Target/TargetOptions.h" +#include "llvm/Support/CommandLine.h" using namespace llvm; +namespace { + cl::opt + EnableConvert3Addr("enable-x86-conv-3-addr", + cl::desc("Enable convertToThreeAddress for X86")); +} + X86InstrInfo::X86InstrInfo(X86TargetMachine &tm) : TargetInstrInfo(X86Insts, array_lengthof(X86Insts)), TM(tm), RI(tm, *this) { @@ -144,6 +151,21 @@ return true; } +/// hasLiveCondCodeDef - True if MI has a condition code def, e.g. EFLAGS, that +/// is not marked dead. +static bool hasLiveCondCodeDef(MachineInstr *MI) { + if (!EnableConvert3Addr) + return true; + for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { + MachineOperand &MO = MI->getOperand(i); + if (MO.isRegister() && MO.isDef() && + MO.getReg() == X86::EFLAGS && !MO.isDead()) { + return true; + } + } + return false; +} + /// convertToThreeAddress - This method must be implemented by targets that /// set the M_CONVERTIBLE_TO_3_ADDR flag. When this flag is set, the target /// may be able to convert a two-address instruction into a true @@ -169,7 +191,7 @@ bool DisableLEA16 = true; switch (MI->getOpcode()) { - default: return 0; + default: break; // All others need to check for live condition code defs. case X86::SHUFPSrri: { assert(MI->getNumOperands() == 4 && "Unknown shufps instruction!"); if (!TM.getSubtarget().hasSSE2()) return 0; @@ -253,10 +275,7 @@ } } - // FIXME: None of these instructions are promotable to LEAs without - // additional information. In particular, LEA doesn't set the flags that - // add and inc do. :( - if (0) + if (!hasLiveCondCodeDef(MI)) switch (MI->getOpcode()) { case X86::INC32r: case X86::INC64_32r: From baldrick at free.fr Fri Oct 5 05:37:01 2007 From: baldrick at free.fr (Duncan Sands) Date: Fri, 05 Oct 2007 10:37:01 -0000 Subject: [llvm-commits] [llvm] r42631 - /llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Message-ID: <200710051037.l95Ab2uC013450@zion.cs.uiuc.edu> Author: baldrick Date: Fri Oct 5 05:37:00 2007 New Revision: 42631 URL: http://llvm.org/viewvc/llvm-project?rev=42631&view=rev Log: Fix typo in comment. Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h?rev=42631&r1=42630&r2=42631&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h (original) +++ llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Fri Oct 5 05:37:00 2007 @@ -1512,7 +1512,7 @@ // AddrMode - unindexed, pre-indexed, post-indexed. ISD::MemIndexedMode AddrMode; - // IsTruncStore - True is the op does a truncation before store. + // IsTruncStore - True if the op does a truncation before store. bool IsTruncStore; // StoredVT - VT of the value after truncation. From djg at cray.com Fri Oct 5 09:07:57 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 14:07:57 -0000 Subject: [llvm-commits] [llvm] r42632 - /llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Message-ID: <200710051407.l95E7vHn020119@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 09:07:56 2007 New Revision: 42632 URL: http://llvm.org/viewvc/llvm-project?rev=42632&view=rev Log: Define target-indepenent SDNode types for multiply and divide that produce two results. Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h?rev=42632&r1=42631&r2=42632&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h (original) +++ llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Fri Oct 5 09:07:56 2007 @@ -233,6 +233,15 @@ // Simple integer binary arithmetic operators. ADD, SUB, MUL, SDIV, UDIV, SREM, UREM, + + // SMUL_LOHI/UMUL_LOHI - Multiply two integers of type iN, producing + // a signed/unsigned value of type i[2*n], and return the full value as + // two results, each of type iN. + SMUL_LOHI, UMUL_LOHI, + + // SDIVREM/UDIVREM - Divide two integers and produce both a quotient and + // remainder result. + SDIVREM, UDIVREM, // CARRY_FALSE - This node is used when folding other nodes, // like ADDC/SUBC, which indicate the carry result is always false. From djg at cray.com Fri Oct 5 09:09:33 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 14:09:33 -0000 Subject: [llvm-commits] [llvm] r42633 - /llvm/trunk/include/llvm/CodeGen/SelectionDAG.h Message-ID: <200710051409.l95E9XfZ020154@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 09:09:33 2007 New Revision: 42633 URL: http://llvm.org/viewvc/llvm-project?rev=42633&view=rev Log: SMUL_LOHI and UMUL_LOHI are commutative. Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAG.h Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAG.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SelectionDAG.h?rev=42633&r1=42632&r2=42633&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/SelectionDAG.h (original) +++ llvm/trunk/include/llvm/CodeGen/SelectionDAG.h Fri Oct 5 09:09:33 2007 @@ -457,6 +457,8 @@ case ISD::MUL: case ISD::MULHU: case ISD::MULHS: + case ISD::SMUL_LOHI: + case ISD::UMUL_LOHI: case ISD::FADD: case ISD::FMUL: case ISD::AND: From djg at cray.com Fri Oct 5 09:11:04 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 14:11:04 -0000 Subject: [llvm-commits] [llvm] r42634 - /llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Message-ID: <200710051411.l95EB4p8020196@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 09:11:04 2007 New Revision: 42634 URL: http://llvm.org/viewvc/llvm-project?rev=42634&view=rev Log: Provide names for MUL_LOHI and DIVREM operators. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=42634&r1=42633&r2=42634&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Fri Oct 5 09:11:04 2007 @@ -3571,6 +3571,10 @@ case ISD::UDIV: return "udiv"; case ISD::SREM: return "srem"; case ISD::UREM: return "urem"; + case ISD::SMUL_LOHI: return "smul_lohi"; + case ISD::UMUL_LOHI: return "umul_lohi"; + case ISD::SDIVREM: return "sdivrem"; + case ISD::UDIVREM: return "divrem"; case ISD::AND: return "and"; case ISD::OR: return "or"; case ISD::XOR: return "xor"; From djg at cray.com Fri Oct 5 09:11:58 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 14:11:58 -0000 Subject: [llvm-commits] [llvm] r42635 - /llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Message-ID: <200710051411.l95EBw5O020233@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 09:11:58 2007 New Revision: 42635 URL: http://llvm.org/viewvc/llvm-project?rev=42635&view=rev Log: Fix a typo in a comment. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp?rev=42635&r1=42634&r2=42635&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp Fri Oct 5 09:11:58 2007 @@ -1301,7 +1301,7 @@ I != E; ++I) { if (I->isCtrl) continue; SUnit *SuccSU = I->Dep; - // Don't constraint nodes with implicit defs. It can create cycles + // Don't constrain nodes with implicit defs. It can create cycles // plus it may increase register pressures. if (SuccSU == SU || SuccSU->hasPhysRegDefs) continue; From djg at cray.com Fri Oct 5 09:17:22 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 14:17:22 -0000 Subject: [llvm-commits] [llvm] r42636 - /llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Message-ID: <200710051417.l95EHMkQ020380@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 09:17:22 2007 New Revision: 42636 URL: http://llvm.org/viewvc/llvm-project?rev=42636&view=rev Log: Legalize support for MUL_LOHI and DIVREM. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=42636&r1=42635&r2=42636&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Fri Oct 5 09:17:22 2007 @@ -2630,6 +2630,16 @@ } break; + case ISD::SMUL_LOHI: + case ISD::UMUL_LOHI: + case ISD::SDIVREM: + case ISD::UDIVREM: + // These nodes will only be produced by target-specific lowering, so + // they shouldn't be here if they aren't legal. + assert(TLI.isOperationLegal(Node->getValueType(0), Node->getValueType(0)) && + "This must be legal!"); + break; + case ISD::FCOPYSIGN: // FCOPYSIGN does not require LHS/RHS to match type! Tmp1 = LegalizeOp(Node->getOperand(0)); // LHS switch (getTypeAction(Node->getOperand(1).getValueType())) { From djg at cray.com Fri Oct 5 10:54:58 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 15:54:58 -0000 Subject: [llvm-commits] [llvm] r42637 - /llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp Message-ID: <200710051554.l95Fswwq023862@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 10:54:58 2007 New Revision: 42637 URL: http://llvm.org/viewvc/llvm-project?rev=42637&view=rev Log: Change a space to a tab in the assembly output of a .globl directive for consistency. Modified: llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp Modified: llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp?rev=42637&r1=42636&r2=42637&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/X86/X86ATTAsmPrinter.cpp Fri Oct 5 10:54:58 2007 @@ -126,7 +126,7 @@ O << "\t.weak_definition\t" << CurrentFnName << "\n"; } else if (Subtarget->isTargetCygMing()) { EmitAlignment(4, F); // FIXME: This should be parameterized somewhere. - O << "\t.globl " << CurrentFnName << "\n"; + O << "\t.globl\t" << CurrentFnName << "\n"; O << "\t.linkonce discard\n"; } else { EmitAlignment(4, F); // FIXME: This should be parameterized somewhere. From djg at cray.com Fri Oct 5 10:58:41 2007 From: djg at cray.com (Dan Gohman) Date: Fri, 05 Oct 2007 15:58:41 -0000 Subject: [llvm-commits] [llvm] r42638 - /llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp Message-ID: <200710051558.l95FwfH9024048@zion.cs.uiuc.edu> Author: djg Date: Fri Oct 5 10:58:41 2007 New Revision: 42638 URL: http://llvm.org/viewvc/llvm-project?rev=42638&view=rev Log: Change a few more spaces to tabs in assembly output. Modified: llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp Modified: llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp?rev=42638&r1=42637&r2=42638&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp (original) +++ llvm/trunk/lib/Target/X86/X86AsmPrinter.cpp Fri Oct 5 10:58:41 2007 @@ -213,7 +213,7 @@ case GlobalValue::LinkOnceLinkage: case GlobalValue::WeakLinkage: if (Subtarget->isTargetDarwin()) { - O << "\t.globl " << name << "\n" + O << "\t.globl\t" << name << "\n" << "\t.weak_definition " << name << "\n"; SwitchToDataSection(".section __DATA,__const_coal,coalesced", I); } else if (Subtarget->isTargetCygMing()) { @@ -221,7 +221,7 @@ name + ",\"aw\""); SwitchToDataSection(SectionName.c_str(), I); - O << "\t.globl " << name << "\n" + O << "\t.globl\t" << name << "\n" << "\t.linkonce same_size\n"; } else { std::string SectionName("\t.section\t.llvm.linkonce.d." + @@ -239,7 +239,7 @@ // their name or something. For now, just emit them as external. case GlobalValue::ExternalLinkage: // If external or appending, declare as a global symbol - O << "\t.globl " << name << "\n"; + O << "\t.globl\t" << name << "\n"; // FALL THROUGH case GlobalValue::InternalLinkage: { if (I->isConstant()) { From sabre at nondot.org Fri Oct 5 12:40:39 2007 From: sabre at nondot.org (Chris Lattner) Date: Fri, 05 Oct 2007 17:40:39 -0000 Subject: [llvm-commits] [llvm] r42639 - /llvm/trunk/include/llvm/ADT/DenseSet.h Message-ID: <200710051740.l95Heddo029366@zion.cs.uiuc.edu> Author: lattner Date: Fri Oct 5 12:40:38 2007 New Revision: 42639 URL: http://llvm.org/viewvc/llvm-project?rev=42639&view=rev Log: Mark count method const. Modified: llvm/trunk/include/llvm/ADT/DenseSet.h Modified: llvm/trunk/include/llvm/ADT/DenseSet.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/DenseSet.h?rev=42639&r1=42638&r2=42639&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/DenseSet.h (original) +++ llvm/trunk/include/llvm/ADT/DenseSet.h Fri Oct 5 12:40:38 2007 @@ -38,7 +38,7 @@ TheMap.clear(); } - bool count(const ValueT &V) { + bool count(const ValueT &V) const { return TheMap.count(V); } From evan.cheng at apple.com Fri Oct 5 12:59:57 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 17:59:57 -0000 Subject: [llvm-commits] [llvm] r42640 - in /llvm/trunk/lib/Target/X86: X86InstrInfo.td X86InstrX86-64.td Message-ID: <200710051759.l95Hxv6X030523@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 12:59:57 2007 New Revision: 42640 URL: http://llvm.org/viewvc/llvm-project?rev=42640&view=rev Log: ADC and SBB uses EFLAGS. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.td llvm/trunk/lib/Target/X86/X86InstrX86-64.td Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.td?rev=42640&r1=42639&r2=42640&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.td Fri Oct 5 12:59:57 2007 @@ -1853,6 +1853,7 @@ [(store (add (load addr:$dst), i32immSExt8:$src2), addr:$dst)]>; } +let Uses = [EFLAGS] in { let isCommutable = 1 in { // X = ADC Y, Z --> X = ADC Z, Y def ADC32rr : I<0x11, MRMDestReg, (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "adc{l}\t{$src2, $dst|$dst, $src2}", @@ -1879,6 +1880,7 @@ "adc{l}\t{$src2, $dst|$dst, $src2}", [(store (adde (load addr:$dst), i32immSExt8:$src2), addr:$dst)]>; } +} // Uses = [EFLAGS] def SUB8rr : I<0x28, MRMDestReg, (outs GR8 :$dst), (ins GR8 :$src1, GR8 :$src2), "sub{b}\t{$src2, $dst|$dst, $src2}", @@ -1945,6 +1947,7 @@ [(store (sub (load addr:$dst), i32immSExt8:$src2), addr:$dst)]>; } +let Uses = [EFLAGS] in { def SBB32rr : I<0x19, MRMDestReg, (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "sbb{l}\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (sube GR32:$src1, GR32:$src2))]>; @@ -1972,6 +1975,7 @@ def SBB32ri8 : Ii8<0x83, MRM3r, (outs GR32:$dst), (ins GR32:$src1, i32i8imm:$src2), "sbb{l}\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (sube GR32:$src1, i32immSExt8:$src2))]>; +} // Uses = [EFLAGS] } // Defs = [EFLAGS] let Defs = [EFLAGS] in { Modified: llvm/trunk/lib/Target/X86/X86InstrX86-64.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrX86-64.td?rev=42640&r1=42639&r2=42640&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrX86-64.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrX86-64.td Fri Oct 5 12:59:57 2007 @@ -262,6 +262,7 @@ "add{q}\t{$src2, $dst|$dst, $src2}", [(store (add (load addr:$dst), i64immSExt8:$src2), addr:$dst)]>; +let Uses = [EFLAGS] in { let isTwoAddress = 1 in { let isCommutable = 1 in def ADC64rr : RI<0x11, MRMDestReg, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), @@ -289,6 +290,7 @@ def ADC64mi8 : RIi8<0x83, MRM2m, (outs), (ins i64mem:$dst, i64i8imm :$src2), "adc{q}\t{$src2, $dst|$dst, $src2}", [(store (adde (load addr:$dst), i64immSExt8:$src2), addr:$dst)]>; +} // Uses = [EFLAGS] let isTwoAddress = 1 in { def SUB64rr : RI<0x29, MRMDestReg, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), @@ -317,6 +319,7 @@ "sub{q}\t{$src2, $dst|$dst, $src2}", [(store (sub (load addr:$dst), i64immSExt8:$src2), addr:$dst)]>; +let Uses = [EFLAGS] in { let isTwoAddress = 1 in { def SBB64rr : RI<0x19, MRMDestReg, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "sbb{q}\t{$src2, $dst|$dst, $src2}", @@ -343,6 +346,7 @@ def SBB64mi8 : RIi8<0x83, MRM3m, (outs), (ins i64mem:$dst, i64i8imm :$src2), "sbb{q}\t{$src2, $dst|$dst, $src2}", [(store (sube (load addr:$dst), i64immSExt8:$src2), addr:$dst)]>; +} // Uses = [EFLAGS] } // Defs = [EFLAGS] // Unsigned multiplication From evan.cheng at apple.com Fri Oct 5 13:20:36 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 18:20:36 -0000 Subject: [llvm-commits] [llvm] r42642 - in /llvm/trunk/lib/Target/X86: X86InstrInfo.cpp X86InstrX86-64.td Message-ID: <200710051820.l95IKaaV031541@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 13:20:36 2007 New Revision: 42642 URL: http://llvm.org/viewvc/llvm-project?rev=42642&view=rev Log: Add support to convert more 64-bit instructions to 3-address instructions. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp llvm/trunk/lib/Target/X86/X86InstrX86-64.td Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42642&r1=42641&r2=42642&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 13:20:36 2007 @@ -277,39 +277,54 @@ if (!hasLiveCondCodeDef(MI)) switch (MI->getOpcode()) { + case X86::INC64r: case X86::INC32r: - case X86::INC64_32r: + case X86::INC64_32r: { assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); - NewMI = addRegOffset(BuildMI(get(X86::LEA32r), Dest), Src, 1); + unsigned Opc = MI->getOpcode() == X86::INC64r ? X86::LEA64r : X86::LEA32r; + NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, 1); break; + } case X86::INC16r: case X86::INC64_16r: if (DisableLEA16) return 0; assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, 1); break; + case X86::DEC64r: case X86::DEC32r: - case X86::DEC64_32r: + case X86::DEC64_32r: { assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); - NewMI = addRegOffset(BuildMI(get(X86::LEA32r), Dest), Src, -1); + unsigned Opc = MI->getOpcode() == X86::DEC64r ? X86::LEA64r : X86::LEA32r; + NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, -1); break; + } case X86::DEC16r: case X86::DEC64_16r: if (DisableLEA16) return 0; assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, -1); break; - case X86::ADD32rr: + case X86::ADD64rr: + case X86::ADD32rr: { assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - NewMI = addRegReg(BuildMI(get(X86::LEA32r), Dest), Src, - MI->getOperand(2).getReg()); + unsigned Opc = MI->getOpcode() == X86::ADD64rr ? X86::LEA64r : X86::LEA32r; + NewMI = addRegReg(BuildMI(get(Opc), Dest), Src, MI->getOperand(2).getReg()); break; + } case X86::ADD16rr: if (DisableLEA16) return 0; assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); NewMI = addRegReg(BuildMI(get(X86::LEA16r), Dest), Src, MI->getOperand(2).getReg()); break; + case X86::ADD64ri32: + case X86::ADD64ri8: + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + if (MI->getOperand(2).isImmediate()) + NewMI = addRegOffset(BuildMI(get(X86::LEA64r), Dest), Src, + MI->getOperand(2).getImmedValue()); + break; case X86::ADD32ri: case X86::ADD32ri8: assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); @@ -328,6 +343,7 @@ case X86::SHL16ri: if (DisableLEA16) return 0; case X86::SHL32ri: + case X86::SHL64ri: assert(MI->getNumOperands() >= 3 && MI->getOperand(2).isImmediate() && "Unknown shl instruction!"); unsigned ShAmt = MI->getOperand(2).getImmedValue(); @@ -335,7 +351,8 @@ X86AddressMode AM; AM.Scale = 1 << ShAmt; AM.IndexReg = Src; - unsigned Opc = MI->getOpcode() == X86::SHL32ri ? X86::LEA32r :X86::LEA16r; + unsigned Opc = MI->getOpcode() == X86::SHL64ri ? X86::LEA64r + : (MI->getOpcode() == X86::SHL32ri ? X86::LEA32r : X86::LEA16r); NewMI = addFullAddress(BuildMI(get(Opc), Dest), AM); } break; Modified: llvm/trunk/lib/Target/X86/X86InstrX86-64.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrX86-64.td?rev=42642&r1=42641&r2=42642&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrX86-64.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrX86-64.td Fri Oct 5 13:20:36 2007 @@ -454,6 +454,7 @@ def SHL64rCL : RI<0xD3, MRM4r, (outs GR64:$dst), (ins GR64:$src), "shl{q}\t{%cl, $dst|$dst, %CL}", [(set GR64:$dst, (shl GR64:$src, CL))]>; +let isConvertibleToThreeAddress = 1 in // Can transform into LEA. def SHL64ri : RIi8<0xC1, MRM4r, (outs GR64:$dst), (ins GR64:$src1, i8imm:$src2), "shl{q}\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (shl GR64:$src1, (i8 imm:$src2)))]>; From dalej at apple.com Fri Oct 5 15:04:43 2007 From: dalej at apple.com (Dale Johannesen) Date: Fri, 05 Oct 2007 20:04:43 -0000 Subject: [llvm-commits] [llvm] r42644 - in /llvm/trunk: include/llvm/CodeGen/RuntimeLibcalls.h lib/CodeGen/SelectionDAG/LegalizeDAG.cpp lib/CodeGen/SelectionDAG/TargetLowering.cpp lib/Target/PowerPC/PPCCallingConv.td lib/Target/PowerPC/PPCISelLowering.cpp test/CodeGen/PowerPC/ppcf128-1-opt.ll test/CodeGen/PowerPC/ppcf128-1.ll Message-ID: <200710052004.l95K4h1K004890@zion.cs.uiuc.edu> Author: johannes Date: Fri Oct 5 15:04:43 2007 New Revision: 42644 URL: http://llvm.org/viewvc/llvm-project?rev=42644&view=rev Log: First round of ppc long double. call/return and basic arithmetic works. Rename RTLIB long double functions to distinguish different flavors of long double; the lib functions have different names, alas. Added: llvm/trunk/test/CodeGen/PowerPC/ppcf128-1-opt.ll llvm/trunk/test/CodeGen/PowerPC/ppcf128-1.ll Modified: llvm/trunk/include/llvm/CodeGen/RuntimeLibcalls.h llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp llvm/trunk/lib/CodeGen/SelectionDAG/TargetLowering.cpp llvm/trunk/lib/Target/PowerPC/PPCCallingConv.td llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp Modified: llvm/trunk/include/llvm/CodeGen/RuntimeLibcalls.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/RuntimeLibcalls.h?rev=42644&r1=42643&r2=42644&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/RuntimeLibcalls.h (original) +++ llvm/trunk/include/llvm/CodeGen/RuntimeLibcalls.h Fri Oct 5 15:04:43 2007 @@ -18,9 +18,8 @@ namespace llvm { namespace RTLIB { /// RTLIB::Libcall enum - This enum defines all of the runtime library calls - /// the backend can emit. "LD" is used for all long double types, since - /// these functions will have the same interface on different targets even - /// though the data is not in the same format. + /// the backend can emit. The various long double types cannot be merged, + /// because 80-bit library functions use "xf" and 128-bit use "tf". /// enum Libcall { // Integer @@ -46,22 +45,29 @@ // FLOATING POINT ADD_F32, ADD_F64, + ADD_PPCF128, SUB_F32, SUB_F64, + SUB_PPCF128, MUL_F32, MUL_F64, + MUL_PPCF128, DIV_F32, DIV_F64, + DIV_PPCF128, REM_F32, REM_F64, + REM_PPCF128, NEG_F32, NEG_F64, POWI_F32, POWI_F64, - POWI_LD, + POWI_F80, + POWI_PPCF128, SQRT_F32, SQRT_F64, - SQRT_LD, + SQRT_F80, + SQRT_PPCF128, SIN_F32, SIN_F64, COS_F32, @@ -74,18 +80,21 @@ FPTOSINT_F32_I64, FPTOSINT_F64_I32, FPTOSINT_F64_I64, - FPTOSINT_LD_I64, + FPTOSINT_F80_I64, + FPTOSINT_PPCF128_I64, FPTOUINT_F32_I32, FPTOUINT_F32_I64, FPTOUINT_F64_I32, FPTOUINT_F64_I64, - FPTOUINT_LD_I32, - FPTOUINT_LD_I64, + FPTOUINT_F80_I32, + FPTOUINT_F80_I64, + FPTOUINT_PPCF128_I64, SINTTOFP_I32_F32, SINTTOFP_I32_F64, SINTTOFP_I64_F32, SINTTOFP_I64_F64, - SINTTOFP_I64_LD, + SINTTOFP_I64_F80, + SINTTOFP_I64_PPCF128, UINTTOFP_I32_F32, UINTTOFP_I32_F64, UINTTOFP_I64_F32, Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=42644&r1=42643&r2=42644&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Fri Oct 5 15:04:43 2007 @@ -3055,7 +3055,10 @@ switch(Node->getOpcode()) { case ISD::FSQRT: LC = VT == MVT::f32 ? RTLIB::SQRT_F32 : - VT == MVT::f64 ? RTLIB::SQRT_F64 : RTLIB::SQRT_LD; + VT == MVT::f64 ? RTLIB::SQRT_F64 : + VT == MVT::f80 ? RTLIB::SQRT_F80 : + VT == MVT::ppcf128 ? RTLIB::SQRT_PPCF128 : + RTLIB::UNKNOWN_LIBCALL; break; case ISD::FSIN: LC = VT == MVT::f32 ? RTLIB::SIN_F32 : RTLIB::SIN_F64; @@ -3079,7 +3082,9 @@ RTLIB::Libcall LC = Node->getValueType(0) == MVT::f32 ? RTLIB::POWI_F32 : Node->getValueType(0) == MVT::f64 ? RTLIB::POWI_F64 : - RTLIB::POWI_LD; + Node->getValueType(0) == MVT::f80 ? RTLIB::POWI_F80 : + Node->getValueType(0) == MVT::ppcf128 ? RTLIB::POWI_PPCF128 : + RTLIB::UNKNOWN_LIBCALL; SDOperand Dummy; Result = ExpandLibCall(TLI.getLibcallName(LC), Node, false/*sign irrelevant*/, Dummy); @@ -3261,9 +3266,13 @@ else if (OVT == MVT::f64) LC = (VT == MVT::i32) ? RTLIB::FPTOSINT_F64_I32 : RTLIB::FPTOSINT_F64_I64; - else if (OVT == MVT::f80 || OVT == MVT::f128 || OVT == MVT::ppcf128) { + else if (OVT == MVT::f80) { assert(VT == MVT::i64); - LC = RTLIB::FPTOSINT_LD_I64; + LC = RTLIB::FPTOSINT_F80_I64; + } + else if (OVT == MVT::ppcf128) { + assert(VT == MVT::i64); + LC = RTLIB::FPTOSINT_PPCF128_I64; } break; } @@ -3275,9 +3284,13 @@ else if (OVT == MVT::f64) LC = (VT == MVT::i32) ? RTLIB::FPTOUINT_F64_I32 : RTLIB::FPTOSINT_F64_I64; - else if (OVT == MVT::f80 || OVT == MVT::f128 || OVT == MVT::ppcf128) { + else if (OVT == MVT::f80) { LC = (VT == MVT::i32) - ? RTLIB::FPTOUINT_LD_I32 : RTLIB::FPTOUINT_LD_I64; + ? RTLIB::FPTOUINT_F80_I32 : RTLIB::FPTOUINT_F80_I64; + } + else if (OVT == MVT::ppcf128) { + assert(VT == MVT::i64); + LC = RTLIB::FPTOUINT_PPCF128_I64; } break; } @@ -5375,13 +5388,15 @@ } } - RTLIB::Libcall LC; + RTLIB::Libcall LC = RTLIB::UNKNOWN_LIBCALL; if (Node->getOperand(0).getValueType() == MVT::f32) LC = RTLIB::FPTOSINT_F32_I64; else if (Node->getOperand(0).getValueType() == MVT::f64) LC = RTLIB::FPTOSINT_F64_I64; - else - LC = RTLIB::FPTOSINT_LD_I64; + else if (Node->getOperand(0).getValueType() == MVT::f80) + LC = RTLIB::FPTOSINT_F80_I64; + else if (Node->getOperand(0).getValueType() == MVT::ppcf128) + LC = RTLIB::FPTOSINT_PPCF128_I64; Lo = ExpandLibCall(TLI.getLibcallName(LC), Node, false/*sign irrelevant*/, Hi); break; @@ -5410,10 +5425,10 @@ LC = RTLIB::FPTOUINT_F32_I64; else if (Node->getOperand(0).getValueType() == MVT::f64) LC = RTLIB::FPTOUINT_F64_I64; - else if (Node->getOperand(0).getValueType() == MVT::f80 || - Node->getOperand(0).getValueType() == MVT::f128 || - Node->getOperand(0).getValueType() == MVT::ppcf128) - LC = RTLIB::FPTOUINT_LD_I64; + else if (Node->getOperand(0).getValueType() == MVT::f80) + LC = RTLIB::FPTOUINT_F80_I64; + else if (Node->getOperand(0).getValueType() == MVT::ppcf128) + LC = RTLIB::FPTOUINT_PPCF128_I64; Lo = ExpandLibCall(TLI.getLibcallName(LC), Node, false/*sign irrelevant*/, Hi); break; @@ -5679,23 +5694,35 @@ break; case ISD::FADD: - Lo = ExpandLibCall(TLI.getLibcallName((VT == MVT::f32) - ? RTLIB::ADD_F32 : RTLIB::ADD_F64), + Lo = ExpandLibCall(TLI.getLibcallName(VT == MVT::f32 ? RTLIB::ADD_F32 : + VT == MVT::f64 ? RTLIB::ADD_F64 : + VT == MVT::ppcf128 ? + RTLIB::ADD_PPCF128 : + RTLIB::UNKNOWN_LIBCALL), Node, false, Hi); break; case ISD::FSUB: - Lo = ExpandLibCall(TLI.getLibcallName((VT == MVT::f32) - ? RTLIB::SUB_F32 : RTLIB::SUB_F64), + Lo = ExpandLibCall(TLI.getLibcallName(VT == MVT::f32 ? RTLIB::SUB_F32 : + VT == MVT::f64 ? RTLIB::SUB_F64 : + VT == MVT::ppcf128 ? + RTLIB::SUB_PPCF128 : + RTLIB::UNKNOWN_LIBCALL), Node, false, Hi); break; case ISD::FMUL: - Lo = ExpandLibCall(TLI.getLibcallName((VT == MVT::f32) - ? RTLIB::MUL_F32 : RTLIB::MUL_F64), + Lo = ExpandLibCall(TLI.getLibcallName(VT == MVT::f32 ? RTLIB::MUL_F32 : + VT == MVT::f64 ? RTLIB::MUL_F64 : + VT == MVT::ppcf128 ? + RTLIB::MUL_PPCF128 : + RTLIB::UNKNOWN_LIBCALL), Node, false, Hi); break; case ISD::FDIV: - Lo = ExpandLibCall(TLI.getLibcallName((VT == MVT::f32) - ? RTLIB::DIV_F32 : RTLIB::DIV_F64), + Lo = ExpandLibCall(TLI.getLibcallName(VT == MVT::f32 ? RTLIB::DIV_F32 : + VT == MVT::f64 ? RTLIB::DIV_F64 : + VT == MVT::ppcf128 ? + RTLIB::DIV_PPCF128 : + RTLIB::UNKNOWN_LIBCALL), Node, false, Hi); break; case ISD::FP_EXTEND: @@ -5707,7 +5734,10 @@ case ISD::FPOWI: Lo = ExpandLibCall(TLI.getLibcallName((VT == MVT::f32) ? RTLIB::POWI_F32 : (VT == MVT::f64) ? RTLIB::POWI_F64 : - RTLIB::POWI_LD), + (VT == MVT::f80) ? RTLIB::POWI_F80 : + (VT == MVT::ppcf128) ? + RTLIB::POWI_PPCF128 : + RTLIB::UNKNOWN_LIBCALL), Node, false, Hi); break; case ISD::FSQRT: @@ -5717,7 +5747,10 @@ switch(Node->getOpcode()) { case ISD::FSQRT: LC = (VT == MVT::f32) ? RTLIB::SQRT_F32 : - (VT == MVT::f64) ? RTLIB::SQRT_F64 : RTLIB::SQRT_LD; + (VT == MVT::f64) ? RTLIB::SQRT_F64 : + (VT == MVT::f80) ? RTLIB::SQRT_F80 : + (VT == MVT::ppcf128) ? RTLIB::SQRT_PPCF128 : + RTLIB::UNKNOWN_LIBCALL; break; case ISD::FSIN: LC = (VT == MVT::f32) ? RTLIB::SIN_F32 : RTLIB::SIN_F64; @@ -5768,9 +5801,13 @@ LC = isSigned ? RTLIB::SINTTOFP_I64_F32 : RTLIB::UINTTOFP_I64_F32; else if (VT == MVT::f64) LC = isSigned ? RTLIB::SINTTOFP_I64_F64 : RTLIB::UINTTOFP_I64_F64; - else if (VT == MVT::f80 || VT == MVT::f128 || VT == MVT::ppcf128) { + else if (VT == MVT::f80) { + assert(isSigned); + LC = RTLIB::SINTTOFP_I64_F80; + } + else if (VT == MVT::ppcf128) { assert(isSigned); - LC = RTLIB::SINTTOFP_I64_LD; + LC = RTLIB::SINTTOFP_I64_PPCF128; } } else { if (VT == MVT::f32) Modified: llvm/trunk/lib/CodeGen/SelectionDAG/TargetLowering.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/TargetLowering.cpp?rev=42644&r1=42643&r2=42644&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/TargetLowering.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/TargetLowering.cpp Fri Oct 5 15:04:43 2007 @@ -46,22 +46,29 @@ Names[RTLIB::NEG_I64] = "__negdi2"; Names[RTLIB::ADD_F32] = "__addsf3"; Names[RTLIB::ADD_F64] = "__adddf3"; + Names[RTLIB::ADD_PPCF128] = "__gcc_qadd"; Names[RTLIB::SUB_F32] = "__subsf3"; Names[RTLIB::SUB_F64] = "__subdf3"; + Names[RTLIB::SUB_PPCF128] = "__gcc_qsub"; Names[RTLIB::MUL_F32] = "__mulsf3"; Names[RTLIB::MUL_F64] = "__muldf3"; + Names[RTLIB::MUL_PPCF128] = "__gcc_qmul"; Names[RTLIB::DIV_F32] = "__divsf3"; Names[RTLIB::DIV_F64] = "__divdf3"; + Names[RTLIB::DIV_PPCF128] = "__gcc_qdiv"; Names[RTLIB::REM_F32] = "fmodf"; Names[RTLIB::REM_F64] = "fmod"; + Names[RTLIB::REM_PPCF128] = "fmodl"; Names[RTLIB::NEG_F32] = "__negsf2"; Names[RTLIB::NEG_F64] = "__negdf2"; Names[RTLIB::POWI_F32] = "__powisf2"; Names[RTLIB::POWI_F64] = "__powidf2"; - Names[RTLIB::POWI_LD] = "__powixf2"; + Names[RTLIB::POWI_F80] = "__powixf2"; + Names[RTLIB::POWI_PPCF128] = "__powitf2"; Names[RTLIB::SQRT_F32] = "sqrtf"; Names[RTLIB::SQRT_F64] = "sqrt"; - Names[RTLIB::SQRT_LD] = "sqrtl"; + Names[RTLIB::SQRT_F80] = "sqrtl"; + Names[RTLIB::SQRT_PPCF128] = "sqrtl"; Names[RTLIB::SIN_F32] = "sinf"; Names[RTLIB::SIN_F64] = "sin"; Names[RTLIB::COS_F32] = "cosf"; @@ -72,18 +79,21 @@ Names[RTLIB::FPTOSINT_F32_I64] = "__fixsfdi"; Names[RTLIB::FPTOSINT_F64_I32] = "__fixdfsi"; Names[RTLIB::FPTOSINT_F64_I64] = "__fixdfdi"; - Names[RTLIB::FPTOSINT_LD_I64] = "__fixxfdi"; + Names[RTLIB::FPTOSINT_F80_I64] = "__fixxfdi"; + Names[RTLIB::FPTOSINT_PPCF128_I64] = "__fixtfdi"; Names[RTLIB::FPTOUINT_F32_I32] = "__fixunssfsi"; Names[RTLIB::FPTOUINT_F32_I64] = "__fixunssfdi"; Names[RTLIB::FPTOUINT_F64_I32] = "__fixunsdfsi"; Names[RTLIB::FPTOUINT_F64_I64] = "__fixunsdfdi"; - Names[RTLIB::FPTOUINT_LD_I32] = "__fixunsxfsi"; - Names[RTLIB::FPTOUINT_LD_I64] = "__fixunsxfdi"; + Names[RTLIB::FPTOUINT_F80_I32] = "__fixunsxfsi"; + Names[RTLIB::FPTOUINT_F80_I64] = "__fixunsxfdi"; + Names[RTLIB::FPTOUINT_PPCF128_I64] = "__fixunstfdi"; Names[RTLIB::SINTTOFP_I32_F32] = "__floatsisf"; Names[RTLIB::SINTTOFP_I32_F64] = "__floatsidf"; Names[RTLIB::SINTTOFP_I64_F32] = "__floatdisf"; Names[RTLIB::SINTTOFP_I64_F64] = "__floatdidf"; - Names[RTLIB::SINTTOFP_I64_LD] = "__floatdixf"; + Names[RTLIB::SINTTOFP_I64_F80] = "__floatdixf"; + Names[RTLIB::SINTTOFP_I64_PPCF128] = "__floatditf"; Names[RTLIB::UINTTOFP_I32_F32] = "__floatunsisf"; Names[RTLIB::UINTTOFP_I32_F64] = "__floatunsidf"; Names[RTLIB::UINTTOFP_I64_F32] = "__floatundisf"; @@ -222,6 +232,14 @@ } } + // ppcf128 type is really two f64's. + if (!isTypeLegal(MVT::ppcf128)) { + NumRegistersForVT[MVT::ppcf128] = 2*NumRegistersForVT[MVT::f64]; + RegisterTypeForVT[MVT::ppcf128] = MVT::f64; + TransformToType[MVT::ppcf128] = MVT::f64; + ValueTypeActions.setTypeAction(MVT::ppcf128, Expand); + } + // Decide how to handle f64. If the target does not have native f64 support, // expand it to i64 and we will be generating soft float library calls. if (!isTypeLegal(MVT::f64)) { Modified: llvm/trunk/lib/Target/PowerPC/PPCCallingConv.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCCallingConv.td?rev=42644&r1=42643&r2=42644&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCCallingConv.td (original) +++ llvm/trunk/lib/Target/PowerPC/PPCCallingConv.td Fri Oct 5 15:04:43 2007 @@ -25,7 +25,8 @@ CCIfType<[i32], CCAssignToReg<[R3, R4]>>, CCIfType<[i64], CCAssignToReg<[X3, X4]>>, - CCIfType<[f32, f64], CCAssignToReg<[F1]>>, + CCIfType<[f32], CCAssignToReg<[F1]>>, + CCIfType<[f64], CCAssignToReg<[F1, F2]>>, // Vector types are always returned in V2. CCIfType<[v16i8, v8i16, v4i32, v4f32], CCAssignToReg<[V2]>> Modified: llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp?rev=42644&r1=42643&r2=42644&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp (original) +++ llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp Fri Oct 5 15:04:43 2007 @@ -1812,8 +1812,20 @@ NumResults = 1; NodeTys.push_back(MVT::i64); break; - case MVT::f32: case MVT::f64: + if (Op.Val->getValueType(1) == MVT::f64) { + Chain = DAG.getCopyFromReg(Chain, PPC::F1, MVT::f64, InFlag).getValue(1); + ResultVals[0] = Chain.getValue(0); + Chain = DAG.getCopyFromReg(Chain, PPC::F2, MVT::f64, + Chain.getValue(2)).getValue(1); + ResultVals[1] = Chain.getValue(0); + NumResults = 2; + NodeTys.push_back(MVT::f64); + NodeTys.push_back(MVT::f64); + break; + } + // else fall through + case MVT::f32: Chain = DAG.getCopyFromReg(Chain, PPC::F1, Op.Val->getValueType(0), InFlag).getValue(1); ResultVals[0] = Chain.getValue(0); Added: llvm/trunk/test/CodeGen/PowerPC/ppcf128-1-opt.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ppcf128-1-opt.ll?rev=42644&view=auto ============================================================================== --- llvm/trunk/test/CodeGen/PowerPC/ppcf128-1-opt.ll (added) +++ llvm/trunk/test/CodeGen/PowerPC/ppcf128-1-opt.ll Fri Oct 5 15:04:43 2007 @@ -0,0 +1,29 @@ +; RUN: llvm-as < %s | llc > %t +; ModuleID = '' +target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128" +target triple = "powerpc-apple-darwin8" + +define ppc_fp128 @plus(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %tmp3 = add ppc_fp128 %x, %y ; [#uses=1] + ret ppc_fp128 %tmp3 +} + +define ppc_fp128 @minus(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %tmp3 = sub ppc_fp128 %x, %y ; [#uses=1] + ret ppc_fp128 %tmp3 +} + +define ppc_fp128 @times(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %tmp3 = mul ppc_fp128 %x, %y ; [#uses=1] + ret ppc_fp128 %tmp3 +} + +define ppc_fp128 @divide(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %tmp3 = fdiv ppc_fp128 %x, %y ; [#uses=1] + ret ppc_fp128 %tmp3 +} + Added: llvm/trunk/test/CodeGen/PowerPC/ppcf128-1.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ppcf128-1.ll?rev=42644&view=auto ============================================================================== --- llvm/trunk/test/CodeGen/PowerPC/ppcf128-1.ll (added) +++ llvm/trunk/test/CodeGen/PowerPC/ppcf128-1.ll Fri Oct 5 15:04:43 2007 @@ -0,0 +1,92 @@ +; RUN: llvm-as < %s | opt -std-compile-opts | llc > %t +; ModuleID = 'ld3.c' +target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128" +target triple = "powerpc-apple-darwin8" + +define ppc_fp128 @plus(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %x_addr = alloca ppc_fp128 ; [#uses=2] + %y_addr = alloca ppc_fp128 ; [#uses=2] + %retval = alloca ppc_fp128, align 16 ; [#uses=2] + %tmp = alloca ppc_fp128, align 16 ; [#uses=2] + %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] + store ppc_fp128 %x, ppc_fp128* %x_addr + store ppc_fp128 %y, ppc_fp128* %y_addr + %tmp1 = load ppc_fp128* %x_addr, align 16 ; [#uses=1] + %tmp2 = load ppc_fp128* %y_addr, align 16 ; [#uses=1] + %tmp3 = add ppc_fp128 %tmp1, %tmp2 ; [#uses=1] + store ppc_fp128 %tmp3, ppc_fp128* %tmp, align 16 + %tmp4 = load ppc_fp128* %tmp, align 16 ; [#uses=1] + store ppc_fp128 %tmp4, ppc_fp128* %retval, align 16 + br label %return + +return: ; preds = %entry + %retval5 = load ppc_fp128* %retval ; [#uses=1] + ret ppc_fp128 %retval5 +} + +define ppc_fp128 @minus(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %x_addr = alloca ppc_fp128 ; [#uses=2] + %y_addr = alloca ppc_fp128 ; [#uses=2] + %retval = alloca ppc_fp128, align 16 ; [#uses=2] + %tmp = alloca ppc_fp128, align 16 ; [#uses=2] + %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] + store ppc_fp128 %x, ppc_fp128* %x_addr + store ppc_fp128 %y, ppc_fp128* %y_addr + %tmp1 = load ppc_fp128* %x_addr, align 16 ; [#uses=1] + %tmp2 = load ppc_fp128* %y_addr, align 16 ; [#uses=1] + %tmp3 = sub ppc_fp128 %tmp1, %tmp2 ; [#uses=1] + store ppc_fp128 %tmp3, ppc_fp128* %tmp, align 16 + %tmp4 = load ppc_fp128* %tmp, align 16 ; [#uses=1] + store ppc_fp128 %tmp4, ppc_fp128* %retval, align 16 + br label %return + +return: ; preds = %entry + %retval5 = load ppc_fp128* %retval ; [#uses=1] + ret ppc_fp128 %retval5 +} + +define ppc_fp128 @times(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %x_addr = alloca ppc_fp128 ; [#uses=2] + %y_addr = alloca ppc_fp128 ; [#uses=2] + %retval = alloca ppc_fp128, align 16 ; [#uses=2] + %tmp = alloca ppc_fp128, align 16 ; [#uses=2] + %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] + store ppc_fp128 %x, ppc_fp128* %x_addr + store ppc_fp128 %y, ppc_fp128* %y_addr + %tmp1 = load ppc_fp128* %x_addr, align 16 ; [#uses=1] + %tmp2 = load ppc_fp128* %y_addr, align 16 ; [#uses=1] + %tmp3 = mul ppc_fp128 %tmp1, %tmp2 ; [#uses=1] + store ppc_fp128 %tmp3, ppc_fp128* %tmp, align 16 + %tmp4 = load ppc_fp128* %tmp, align 16 ; [#uses=1] + store ppc_fp128 %tmp4, ppc_fp128* %retval, align 16 + br label %return + +return: ; preds = %entry + %retval5 = load ppc_fp128* %retval ; [#uses=1] + ret ppc_fp128 %retval5 +} + +define ppc_fp128 @divide(ppc_fp128 %x, ppc_fp128 %y) { +entry: + %x_addr = alloca ppc_fp128 ; [#uses=2] + %y_addr = alloca ppc_fp128 ; [#uses=2] + %retval = alloca ppc_fp128, align 16 ; [#uses=2] + %tmp = alloca ppc_fp128, align 16 ; [#uses=2] + %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] + store ppc_fp128 %x, ppc_fp128* %x_addr + store ppc_fp128 %y, ppc_fp128* %y_addr + %tmp1 = load ppc_fp128* %x_addr, align 16 ; [#uses=1] + %tmp2 = load ppc_fp128* %y_addr, align 16 ; [#uses=1] + %tmp3 = fdiv ppc_fp128 %tmp1, %tmp2 ; [#uses=1] + store ppc_fp128 %tmp3, ppc_fp128* %tmp, align 16 + %tmp4 = load ppc_fp128* %tmp, align 16 ; [#uses=1] + store ppc_fp128 %tmp4, ppc_fp128* %retval, align 16 + br label %return + +return: ; preds = %entry + %retval5 = load ppc_fp128* %retval ; [#uses=1] + ret ppc_fp128 %retval5 +} From evan.cheng at apple.com Fri Oct 5 15:34:26 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 20:34:26 -0000 Subject: [llvm-commits] [llvm] r42647 - /llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Message-ID: <200710052034.l95KYQja006086@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 15:34:26 2007 New Revision: 42647 URL: http://llvm.org/viewvc/llvm-project?rev=42647&view=rev Log: In 64-bit mode, avoid using leal with 32-bit 32-bit address size, e.g. leal 1(%ecx), %edi, which requires 67H prefix. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42647&r1=42646&r2=42647&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 15:34:26 2007 @@ -190,8 +190,8 @@ // we have better subtarget support, enable the 16-bit LEA generation here. bool DisableLEA16 = true; - switch (MI->getOpcode()) { - default: break; // All others need to check for live condition code defs. + unsigned MIOpc = MI->getOpcode(); + switch (MIOpc) { case X86::SHUFPSrri: { assert(MI->getNumOperands() == 4 && "Unknown shufps instruction!"); if (!TM.getSubtarget().hasSSE2()) return 0; @@ -273,96 +273,105 @@ } break; } - } - - if (!hasLiveCondCodeDef(MI)) - switch (MI->getOpcode()) { - case X86::INC64r: - case X86::INC32r: - case X86::INC64_32r: { - assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); - unsigned Opc = MI->getOpcode() == X86::INC64r ? X86::LEA64r : X86::LEA32r; - NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, 1); - break; - } - case X86::INC16r: - case X86::INC64_16r: - if (DisableLEA16) return 0; - assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); - NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, 1); - break; - case X86::DEC64r: - case X86::DEC32r: - case X86::DEC64_32r: { - assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); - unsigned Opc = MI->getOpcode() == X86::DEC64r ? X86::LEA64r : X86::LEA32r; - NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, -1); - break; - } - case X86::DEC16r: - case X86::DEC64_16r: - if (DisableLEA16) return 0; - assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); - NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, -1); - break; - case X86::ADD64rr: - case X86::ADD32rr: { - assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - unsigned Opc = MI->getOpcode() == X86::ADD64rr ? X86::LEA64r : X86::LEA32r; - NewMI = addRegReg(BuildMI(get(Opc), Dest), Src, MI->getOperand(2).getReg()); - break; - } - case X86::ADD16rr: - if (DisableLEA16) return 0; - assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - NewMI = addRegReg(BuildMI(get(X86::LEA16r), Dest), Src, - MI->getOperand(2).getReg()); - break; - case X86::ADD64ri32: - case X86::ADD64ri8: - assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - if (MI->getOperand(2).isImmediate()) - NewMI = addRegOffset(BuildMI(get(X86::LEA64r), Dest), Src, - MI->getOperand(2).getImmedValue()); - break; - case X86::ADD32ri: - case X86::ADD32ri8: - assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - if (MI->getOperand(2).isImmediate()) - NewMI = addRegOffset(BuildMI(get(X86::LEA32r), Dest), Src, - MI->getOperand(2).getImmedValue()); - break; - case X86::ADD16ri: - case X86::ADD16ri8: - if (DisableLEA16) return 0; - assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); - if (MI->getOperand(2).isImmediate()) - NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, - MI->getOperand(2).getImmedValue()); - break; - case X86::SHL16ri: - if (DisableLEA16) return 0; - case X86::SHL32ri: - case X86::SHL64ri: - assert(MI->getNumOperands() >= 3 && MI->getOperand(2).isImmediate() && - "Unknown shl instruction!"); - unsigned ShAmt = MI->getOperand(2).getImmedValue(); - if (ShAmt == 1 || ShAmt == 2 || ShAmt == 3) { - X86AddressMode AM; - AM.Scale = 1 << ShAmt; - AM.IndexReg = Src; - unsigned Opc = MI->getOpcode() == X86::SHL64ri ? X86::LEA64r - : (MI->getOpcode() == X86::SHL32ri ? X86::LEA32r : X86::LEA16r); - NewMI = addFullAddress(BuildMI(get(Opc), Dest), AM); + default: { + // The following opcodes also sets the condition code register(s). Only + // convert them to equivalent lea if the condition code register def's + // are dead! + if (hasLiveCondCodeDef(MI)) + return 0; + + switch (MIOpc) { + default: return 0; + case X86::INC64r: + case X86::INC32r: + case X86::INC64_32r: { + assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); + unsigned Opc = MIOpc == X86::INC64r ? X86::LEA64r + : (MIOpc == X86::INC64_32r ? X86::LEA64_32r : X86::LEA32r); + NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, 1); + break; + } + case X86::INC16r: + case X86::INC64_16r: + if (DisableLEA16) return 0; + assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); + NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, 1); + break; + case X86::DEC64r: + case X86::DEC32r: + case X86::DEC64_32r: { + assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); + unsigned Opc = MIOpc == X86::DEC64r ? X86::LEA64r + : (MIOpc == X86::DEC64_32r ? X86::LEA64_32r : X86::LEA32r); + NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, -1); + break; + } + case X86::DEC16r: + case X86::DEC64_16r: + if (DisableLEA16) return 0; + assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); + NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, -1); + break; + case X86::ADD64rr: + case X86::ADD32rr: { + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + unsigned Opc = MIOpc == X86::ADD64rr ? X86::LEA64r : X86::LEA32r; + NewMI = addRegReg(BuildMI(get(Opc), Dest), Src, + MI->getOperand(2).getReg()); + break; + } + case X86::ADD16rr: + if (DisableLEA16) return 0; + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + NewMI = addRegReg(BuildMI(get(X86::LEA16r), Dest), Src, + MI->getOperand(2).getReg()); + break; + case X86::ADD64ri32: + case X86::ADD64ri8: + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + if (MI->getOperand(2).isImmediate()) + NewMI = addRegOffset(BuildMI(get(X86::LEA64r), Dest), Src, + MI->getOperand(2).getImmedValue()); + break; + case X86::ADD32ri: + case X86::ADD32ri8: + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + if (MI->getOperand(2).isImmediate()) + NewMI = addRegOffset(BuildMI(get(X86::LEA32r), Dest), Src, + MI->getOperand(2).getImmedValue()); + break; + case X86::ADD16ri: + case X86::ADD16ri8: + if (DisableLEA16) return 0; + assert(MI->getNumOperands() >= 3 && "Unknown add instruction!"); + if (MI->getOperand(2).isImmediate()) + NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, + MI->getOperand(2).getImmedValue()); + break; + case X86::SHL16ri: + if (DisableLEA16) return 0; + case X86::SHL32ri: + case X86::SHL64ri: { + assert(MI->getNumOperands() >= 3 && MI->getOperand(2).isImmediate() && + "Unknown shl instruction!"); + unsigned ShAmt = MI->getOperand(2).getImmedValue(); + if (ShAmt == 1 || ShAmt == 2 || ShAmt == 3) { + X86AddressMode AM; + AM.Scale = 1 << ShAmt; + AM.IndexReg = Src; + unsigned Opc = MIOpc == X86::SHL64ri ? X86::LEA64r + : (MIOpc == X86::SHL32ri ? X86::LEA32r : X86::LEA16r); + NewMI = addFullAddress(BuildMI(get(Opc), Dest), AM); + } + break; + } } - break; } - - if (NewMI) { - NewMI->copyKillDeadInfo(MI); - LV.instructionChanged(MI, NewMI); // Update live variables - MFI->insert(MBBI, NewMI); // Insert the new inst } + + NewMI->copyKillDeadInfo(MI); + LV.instructionChanged(MI, NewMI); // Update live variables + MFI->insert(MBBI, NewMI); // Insert the new inst return NewMI; } From evan.cheng at apple.com Fri Oct 5 16:55:32 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 21:55:32 -0000 Subject: [llvm-commits] [llvm] r42653 - /llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Message-ID: <200710052155.l95LtWgD009023@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 16:55:32 2007 New Revision: 42653 URL: http://llvm.org/viewvc/llvm-project?rev=42653&view=rev Log: INC64_32r -> LEA64_32r is better than INC64_32r -> LEA32r, but it still can cause performance degradation. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42653&r1=42652&r2=42653&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 16:55:32 2007 @@ -283,11 +283,9 @@ switch (MIOpc) { default: return 0; case X86::INC64r: - case X86::INC32r: - case X86::INC64_32r: { + case X86::INC32r: { assert(MI->getNumOperands() >= 2 && "Unknown inc instruction!"); - unsigned Opc = MIOpc == X86::INC64r ? X86::LEA64r - : (MIOpc == X86::INC64_32r ? X86::LEA64_32r : X86::LEA32r); + unsigned Opc = MIOpc == X86::INC64r ? X86::LEA64r : X86::LEA32r; NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, 1); break; } @@ -298,11 +296,9 @@ NewMI = addRegOffset(BuildMI(get(X86::LEA16r), Dest), Src, 1); break; case X86::DEC64r: - case X86::DEC32r: - case X86::DEC64_32r: { + case X86::DEC32r: { assert(MI->getNumOperands() >= 2 && "Unknown dec instruction!"); - unsigned Opc = MIOpc == X86::DEC64r ? X86::LEA64r - : (MIOpc == X86::DEC64_32r ? X86::LEA64_32r : X86::LEA32r); + unsigned Opc = MIOpc == X86::DEC64r ? X86::LEA64r : X86::LEA32r; NewMI = addRegOffset(BuildMI(get(Opc), Dest), Src, -1); break; } From dpatel at apple.com Fri Oct 5 17:29:34 2007 From: dpatel at apple.com (Devang Patel) Date: Fri, 05 Oct 2007 22:29:34 -0000 Subject: [llvm-commits] [llvm] r42654 - in /llvm/trunk: lib/Transforms/Scalar/LoopUnswitch.cpp lib/VMCore/Dominators.cpp test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll Message-ID: <200710052229.l95MTYp2010695@zion.cs.uiuc.edu> Author: dpatel Date: Fri Oct 5 17:29:34 2007 New Revision: 42654 URL: http://llvm.org/viewvc/llvm-project?rev=42654&view=rev Log: Fix bug in updating dominance frontier after loop unswitch when frontier includes basic blocks that are not inside loop. Added: llvm/trunk/test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp llvm/trunk/lib/VMCore/Dominators.cpp Modified: llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp?rev=42654&r1=42653&r2=42654&view=diff ============================================================================== --- llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp (original) +++ llvm/trunk/lib/Transforms/Scalar/LoopUnswitch.cpp Fri Oct 5 17:29:34 2007 @@ -41,7 +41,6 @@ #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/ADT/Statistic.h" #include "llvm/ADT/SmallPtrSet.h" -#include "llvm/ADT/PostOrderIterator.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Compiler.h" #include "llvm/Support/Debug.h" @@ -71,6 +70,18 @@ bool OptimizeForSize; bool redoLoop; + + DominanceFrontier *DF; + DominatorTree *DT; + + /// LoopDF - Loop's dominance frontier. This set is a collection of + /// loop exiting blocks' DF member blocks. However this does set does not + /// includes basic blocks that are inside loop. + SmallPtrSet LoopDF; + + /// OrigLoopExitMap - This is used to map loop exiting block with + /// corresponding loop exit block, before updating CFG. + DenseMap OrigLoopExitMap; public: static char ID; // Pass ID, replacement for typeid explicit LoopUnswitch(bool Os = false) : @@ -106,8 +117,13 @@ /// Split all of the edges from inside the loop to their exit blocks. Update /// the appropriate Phi nodes as we do so. - void SplitExitEdges(const SmallVector &ExitBlocks, + void SplitExitEdges(Loop *L, const SmallVector &ExitBlocks, SmallVector &MiddleBlocks); + + /// If BB's dominance frontier has a member that is not part of loop L then + /// remove it. Add NewDFMember in BB's dominance frontier. + void ReplaceLoopExternalDFMember(Loop *L, BasicBlock *BB, + BasicBlock *NewDFMember); bool UnswitchIfProfitable(Value *LoopCond, Constant *Val,Loop *L); unsigned getLoopUnswitchCost(Loop *L, Value *LIC); @@ -158,13 +174,16 @@ if (Value *RHS = FindLIVLoopCondition(BO->getOperand(1), L, Changed)) return RHS; } - - return 0; + + return 0; } bool LoopUnswitch::runOnLoop(Loop *L, LPPassManager &LPM_Ref) { LI = &getAnalysis(); LPM = &LPM_Ref; + DF = getAnalysisToUpdate(); + DT = getAnalysisToUpdate(); + bool Changed = false; do { @@ -601,11 +620,36 @@ ++NumTrivial; } +/// ReplaceLoopExternalDFMember - +/// If BB's dominance frontier has a member that is not part of loop L then +/// remove it. Add NewDFMember in BB's dominance frontier. +void LoopUnswitch::ReplaceLoopExternalDFMember(Loop *L, BasicBlock *BB, + BasicBlock *NewDFMember) { + + DominanceFrontier::iterator DFI = DF->find(BB); + if (DFI == DF->end()) + return; + + DominanceFrontier::DomSetType &DFSet = DFI->second; + for (DominanceFrontier::DomSetType::iterator DI = DFSet.begin(), + DE = DFSet.end(); DI != DE; ++DI) { + BasicBlock *B = *DI; + if (L->contains(B)) + continue; + + DF->removeFromFrontier(DFI, B); + LoopDF.insert(B); + } + + DF->addToFrontier(DFI, NewDFMember); +} + /// SplitExitEdges - /// Split all of the edges from inside the loop to their exit blocks. Update /// the appropriate Phi nodes as we do so. -void LoopUnswitch::SplitExitEdges(const SmallVector &ExitBlocks, +void LoopUnswitch::SplitExitEdges(Loop *L, const SmallVector &ExitBlocks, SmallVector &MiddleBlocks) { + for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i) { BasicBlock *ExitBlock = ExitBlocks[i]; std::vector Preds(pred_begin(ExitBlock), pred_end(ExitBlock)); @@ -622,6 +666,8 @@ EndBlock = ExitBlock; } + OrigLoopExitMap[StartBlock] = EndBlock; + std::set InsertedPHIs; PHINode* OldLCSSA = 0; for (BasicBlock::iterator I = EndBlock->begin(); @@ -647,8 +693,24 @@ OldLCSSA->replaceAllUsesWith(NewLCSSA); NewLCSSA->addIncoming(OldLCSSA, MiddleBlock); } + + if (DF && DT) { + // StartBlock -- > MiddleBlock -- > EndBlock + // StartBlock is loop exiting block. EndBlock will become merge point + // of two loop exits after loop unswitch. + + // If StartBlock's DF member includes a block that is not loop member + // then replace that DF member with EndBlock. + + // If MiddleBlock's DF member includes a block that is not loop member + // tnen replace that DF member with EndBlock. + + ReplaceLoopExternalDFMember(L, StartBlock, EndBlock); + ReplaceLoopExternalDFMember(L, MiddleBlock, EndBlock); + } } } + } /// UnswitchNontrivialCondition - We determined that the loop is profitable @@ -683,7 +745,7 @@ // Split all of the edges from inside the loop to their exit blocks. Update // the appropriate Phi nodes as we do so. SmallVector MiddleBlocks; - SplitExitEdges(ExitBlocks, MiddleBlocks); + SplitExitEdges(L, ExitBlocks, MiddleBlocks); // The exit blocks may have been changed due to edge splitting, recompute. ExitBlocks.clear(); @@ -692,9 +754,6 @@ // Add exit blocks to the loop blocks. LoopBlocks.insert(LoopBlocks.end(), ExitBlocks.begin(), ExitBlocks.end()); - DominanceFrontier *DF = getAnalysisToUpdate(); - DominatorTree *DT = getAnalysisToUpdate(); - // Next step, clone all of the basic blocks that make up the loop (including // the loop preheader and exit blocks), keeping track of the mapping between // the instructions and blocks. @@ -778,6 +837,9 @@ // Update dominator info if (DF && DT) { + SmallVector ExitingBlocks; + L->getExitingBlocks(ExitingBlocks); + // Clone dominator info for all cloned basic block. for (unsigned i = 0, e = LoopBlocks.size(); i != e; ++i) { BasicBlock *LBB = LoopBlocks[i]; @@ -785,29 +847,57 @@ CloneDomInfo(NBB, LBB, NewPreheader, OrigPreheader, OrigHeader, DT, DF, ValueMap); - // Remove any OutSiders from LBB and NBB's dominance frontier. - DominanceFrontier::iterator LBBI = DF->find(LBB); - if (LBBI != DF->end()) { - DominanceFrontier::DomSetType &LBSet = LBBI->second; - for (DominanceFrontier::DomSetType::iterator LI = LBSet.begin(), - LE = LBSet.end(); LI != LE; /* NULL */) { - BasicBlock *B = *LI++; - if (OutSiders.count(B)) - DF->removeFromFrontier(LBBI, B); - } - } + // If LBB's dominance frontier includes DFMember + // such that DFMember is also a member of LoopDF then + // - Remove DFMember from LBB's dominance frontier + // - Copy loop exiting blocks', that are dominated by BB, dominance frontier + // member in BB's dominance frontier - // Remove any OutSiders from LBB and NBB's dominance frontier. + DominanceFrontier::iterator LBBI = DF->find(LBB); DominanceFrontier::iterator NBBI = DF->find(NBB); - if (NBBI != DF->end()) { - DominanceFrontier::DomSetType NBSet = NBBI->second; - for (DominanceFrontier::DomSetType::iterator NI = NBSet.begin(), - NE = NBSet.end(); NI != NE; /* NULL */) { - BasicBlock *B = *NI++; - if (OutSiders.count(B)) + if (LBBI == DF->end()) + continue; + + DominanceFrontier::DomSetType &LBSet = LBBI->second; + for (DominanceFrontier::DomSetType::iterator LI = LBSet.begin(), + LE = LBSet.end(); LI != LE; /* NULL */) { + BasicBlock *B = *LI++; + if (B == LBB && B == L->getHeader()) + continue; + bool removeB = false; + if (!LoopDF.count(B)) + continue; + + // If LBB dominates loop exits then insert loop exit block's DF + // into B's DF. + for(SmallVector::iterator LExitI = ExitingBlocks.begin(), + LExitE = ExitingBlocks.end(); LExitI != LExitE; ++LExitI) { + BasicBlock *E = *LExitI; + + if (!DT->dominates(LBB,E)) + continue; + + DenseMap::iterator DFBI = + OrigLoopExitMap.find(E); + if (DFBI == OrigLoopExitMap.end()) + continue; + + BasicBlock *DFB = DFBI->second; + DF->addToFrontier(LBBI, DFB); + DF->addToFrontier(NBBI, DFB); + removeB = true; + } + + // If B's replacement is inserted in DF then now is the time to remove B. + if (removeB) { + DF->removeFromFrontier(LBBI, B); + if (L->contains(B)) + DF->removeFromFrontier(NBBI, cast(ValueMap[B])); + else DF->removeFromFrontier(NBBI, B); } } + } // MiddleBlocks are dominated by original pre header. SplitEdge updated Modified: llvm/trunk/lib/VMCore/Dominators.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Dominators.cpp?rev=42654&r1=42653&r2=42654&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Dominators.cpp (original) +++ llvm/trunk/lib/VMCore/Dominators.cpp Fri Oct 5 17:29:34 2007 @@ -422,8 +422,11 @@ } if (NewBBI != end()) { - DominanceFrontier::DomSetType NewBBSet = NewBBI->second; - NewBBSet.insert(Set.begin(), Set.end()); + for (DominanceFrontier::DomSetType::iterator SetI = Set.begin(), + E = Set.end(); SetI != E; ++SetI) { + BasicBlock *SB = *SetI; + addToFrontier(NewBBI, SB); + } } else addBasicBlock(NewBB, Set); } Added: llvm/trunk/test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll?rev=42654&view=auto ============================================================================== --- llvm/trunk/test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll (added) +++ llvm/trunk/test/Transforms/LoopUnswitch/2007-10-04-DomFrontier.ll Fri Oct 5 17:29:34 2007 @@ -0,0 +1,29 @@ +; RUN: llvm-as < %s | opt -licm -loop-unroll -disable-output + + at resonant = external global i32 ; [#uses=2] + +define void @weightadj() { +entry: + br label %bb + +bb: ; preds = %bb158, %entry + store i32 0, i32* @resonant, align 4 + br i1 false, label %g.exit, label %bb158 + +g.exit: ; preds = %bb68, %bb + br i1 false, label %bb68, label %cond_true + +cond_true: ; preds = %g.exit + store i32 1, i32* @resonant, align 4 + br label %bb68 + +bb68: ; preds = %cond_true, %g.exit + %tmp71 = icmp slt i32 0, 0 ; [#uses=1] + br i1 %tmp71, label %g.exit, label %bb158 + +bb158: ; preds = %bb68, %bb + br i1 false, label %bb, label %return + +return: ; preds = %bb158 + ret void +} From evan.cheng at apple.com Fri Oct 5 17:31:10 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 22:31:10 -0000 Subject: [llvm-commits] [test-suite] r42655 - /test-suite/trunk/Makefile.programs Message-ID: <200710052231.l95MVAVT010742@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 17:31:10 2007 New Revision: 42655 URL: http://llvm.org/viewvc/llvm-project?rev=42655&view=rev Log: Enable convertToThreeAddress for X86 by default. Modified: test-suite/trunk/Makefile.programs Modified: test-suite/trunk/Makefile.programs URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/Makefile.programs?rev=42655&r1=42654&r2=42655&view=diff ============================================================================== --- test-suite/trunk/Makefile.programs (original) +++ test-suite/trunk/Makefile.programs Fri Oct 5 17:31:10 2007 @@ -221,8 +221,7 @@ LLCBETAOPTION := -sched=simple endif ifeq ($(ARCH),x86) -LLCBETAOPTION := -enable-x86-conv-3-addr -#-regalloc=local -fast +LLCBETAOPTION := -regalloc=local -fast #-disable-rematerialization #-disable-fp-elim #-regalloc=bigblock -fast From evan.cheng at apple.com Fri Oct 5 17:31:10 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 22:31:10 -0000 Subject: [llvm-commits] [llvm] r42655 - in /llvm/trunk: lib/Target/X86/X86InstrInfo.cpp test/CodeGen/X86/2006-05-02-InstrSched2.ll test/CodeGen/X86/2006-05-11-InstrSched.ll test/CodeGen/X86/loop-strength-reduce2.ll Message-ID: <200710052231.l95MVAD8010750@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 17:31:10 2007 New Revision: 42655 URL: http://llvm.org/viewvc/llvm-project?rev=42655&view=rev Log: Enable convertToThreeAddress for X86 by default. Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp llvm/trunk/test/CodeGen/X86/2006-05-02-InstrSched2.ll llvm/trunk/test/CodeGen/X86/2006-05-11-InstrSched.ll llvm/trunk/test/CodeGen/X86/loop-strength-reduce2.ll Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42655&r1=42654&r2=42655&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 17:31:10 2007 @@ -22,15 +22,8 @@ #include "llvm/CodeGen/LiveVariables.h" #include "llvm/CodeGen/SSARegMap.h" #include "llvm/Target/TargetOptions.h" -#include "llvm/Support/CommandLine.h" using namespace llvm; -namespace { - cl::opt - EnableConvert3Addr("enable-x86-conv-3-addr", - cl::desc("Enable convertToThreeAddress for X86")); -} - X86InstrInfo::X86InstrInfo(X86TargetMachine &tm) : TargetInstrInfo(X86Insts, array_lengthof(X86Insts)), TM(tm), RI(tm, *this) { @@ -154,8 +147,6 @@ /// hasLiveCondCodeDef - True if MI has a condition code def, e.g. EFLAGS, that /// is not marked dead. static bool hasLiveCondCodeDef(MachineInstr *MI) { - if (!EnableConvert3Addr) - return true; for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { MachineOperand &MO = MI->getOperand(i); if (MO.isRegister() && MO.isDef() && Modified: llvm/trunk/test/CodeGen/X86/2006-05-02-InstrSched2.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2006-05-02-InstrSched2.ll?rev=42655&r1=42654&r2=42655&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/X86/2006-05-02-InstrSched2.ll (original) +++ llvm/trunk/test/CodeGen/X86/2006-05-02-InstrSched2.ll Fri Oct 5 17:31:10 2007 @@ -1,5 +1,5 @@ ; RUN: llvm-upgrade < %s | llvm-as | llc -march=x86 -stats |& \ -; RUN: grep asm-printer | grep 16 +; RUN: grep asm-printer | grep 15 void %_ZN9__gnu_cxx9hashtableISt4pairIKPKciES3_NS_4hashIS3_EESt10_Select1stIS5_E5eqstrSaIiEE14find_or_insertERKS5__cond_true456.i(sbyte* %tmp435.i, uint* %tmp449.i.out) { newFuncRoot: Modified: llvm/trunk/test/CodeGen/X86/2006-05-11-InstrSched.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2006-05-11-InstrSched.ll?rev=42655&r1=42654&r2=42655&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/X86/2006-05-11-InstrSched.ll (original) +++ llvm/trunk/test/CodeGen/X86/2006-05-11-InstrSched.ll Fri Oct 5 17:31:10 2007 @@ -1,5 +1,5 @@ ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -stats |&\ -; RUN: grep {asm-printer} | grep 33 +; RUN: grep {asm-printer} | grep 32 target datalayout = "e-p:32:32" define void @foo(i32* %mc, i32* %bp, i32* %ms, i32* %xmb, i32* %mpp, i32* %tpmm, i32* %ip, i32* %tpim, i32* %dpp, i32* %tpdm, i32* %bpi, i32 %M) { Modified: llvm/trunk/test/CodeGen/X86/loop-strength-reduce2.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/loop-strength-reduce2.ll?rev=42655&r1=42654&r2=42655&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/X86/loop-strength-reduce2.ll (original) +++ llvm/trunk/test/CodeGen/X86/loop-strength-reduce2.ll Fri Oct 5 17:31:10 2007 @@ -1,4 +1,4 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -mtriple=i686-apple-darwin -relocation-model=pic | not grep lea +; RUN: llvm-upgrade < %s | llvm-as | llc -mtriple=i686-apple-darwin -relocation-model=pic | grep '\$pb' | grep mov ; ; Make sure the PIC label flags2-"L1$pb" is not moved up to the preheader. From evan.cheng at apple.com Fri Oct 5 17:34:59 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 22:34:59 -0000 Subject: [llvm-commits] [llvm] r42656 - /llvm/trunk/lib/Target/X86/README.txt Message-ID: <200710052234.l95MYxCM010919@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 17:34:59 2007 New Revision: 42656 URL: http://llvm.org/viewvc/llvm-project?rev=42656&view=rev Log: This is done. Modified: llvm/trunk/lib/Target/X86/README.txt Modified: llvm/trunk/lib/Target/X86/README.txt URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42656&r1=42655&r2=42656&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/README.txt (original) +++ llvm/trunk/lib/Target/X86/README.txt Fri Oct 5 17:34:59 2007 @@ -290,10 +290,6 @@ //===---------------------------------------------------------------------===// -Enable X86InstrInfo::convertToThreeAddress(). - -//===---------------------------------------------------------------------===// - We are currently lowering large (1MB+) memmove/memcpy to rep/stosl and rep/movsl We should leave these as libcalls for everything over a much lower threshold, since libc is hand tuned for medium and large mem ops (avoiding RFO for large From gordonhenriksen at mac.com Fri Oct 5 17:36:48 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Fri, 5 Oct 2007 18:36:48 -0400 Subject: [llvm-commits] [cfe-commits] r42652 - in /cfe/trunk/www: get_involved.html index.html In-Reply-To: <200710052144.l95Li4W0008520@zion.cs.uiuc.edu> References: <200710052144.l95Li4W0008520@zion.cs.uiuc.edu> Message-ID: Hi Kevin, Since you're using non-ASCII characters, could you please declare the character set in the section? I believe this is correct for this file: On Oct 5, 2007, at 17:44, Kevin wrote: > Author: kevin > Date: Fri Oct 5 16:44:03 2007 > New Revision: 42652 > > URL: http://llvm.org/viewvc/llvm-project?rev=42652&view=rev > Log: > A message on the main page about bulding clang. > > Modified: > cfe/trunk/www/get_involved.html > cfe/trunk/www/index.html > > Modified: cfe/trunk/www/get_involved.html > URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/www/ > get_involved.html?rev=42652&r1=42651&r2=42652&view=diff > > ====================================================================== > ======== > --- cfe/trunk/www/get_involved.html (original) > +++ cfe/trunk/www/get_involved.html Fri Oct 5 16:44:03 2007 > @@ -32,7 +32,7 @@ >
  • cfe-commits - This list is for patch submission/ > discussion. >
  • cfe-dev - This list is for everything else clang related. > > -

    Building clang / working with the code

    > +

    Building clang / working with the code  a>

    > If you would like to check out and build the project, the current > scheme is:
    >
      >
    1. Check out llvm > > Modified: cfe/trunk/www/index.html > URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/www/index.html? > rev=42652&r1=42651&r2=42652&view=diff > > ====================================================================== > ======== > --- cfe/trunk/www/index.html (original) > +++ cfe/trunk/www/index.html Fri Oct 5 16:44:03 2007 > @@ -41,6 +41,8 @@ >
    2. BSD License > > Of course this is only a rough outline of the goals and features > of Clang. To get a true sense of what the new LLVM front-end is > all about, as well as why you might want to considering using it, > see the Features section. > +

      Try Clang

      > + If you are a compiler developer and want to try out Clang, then > build instructions are available on the href="get_involved.html#build">Get Involved page. >

      Get Involved

      > If you are interested in working on Clang, then check out the href="get_involved.html">Get Involved section. > > > > _______________________________________________ > cfe-commits mailing list > cfe-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits ? Gordon -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20071005/2bbb2aad/attachment.html From asl at math.spbu.ru Fri Oct 5 17:39:24 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Fri, 05 Oct 2007 22:39:24 -0000 Subject: [llvm-commits] [test-suite] r42657 - /test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c Message-ID: <200710052239.l95MdOt0011080@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 17:39:24 2007 New Revision: 42657 URL: http://llvm.org/viewvc/llvm-project?rev=42657&view=rev Log: Fix misr on mingw32 Modified: test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c Modified: test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c?rev=42657&r1=42656&r2=42657&view=diff ============================================================================== --- test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c (original) +++ test-suite/trunk/SingleSource/Benchmarks/McGill/misr.c Fri Oct 5 17:39:24 2007 @@ -40,6 +40,11 @@ #define TRUE 1 #define FALSE 0 +#ifdef __MINGW32__ +# define seed48(x) srand(x) +# define lrand48() ((rand() << 16) ^ (rand())) +#endif + int reg_len; typedef struct cells { @@ -217,7 +222,7 @@ misr_type *temp; int different, savef_free, savefaulty; int rem, quot, h, i, j; - long rand; + long randv; double randprob; different = FALSE; @@ -231,7 +236,7 @@ savefaulty = 0; for (i=0; if_free; savefaulty += temp->faulty; } - temp->f_free = ((temp->next->f_free + rand) & BIN_MASK); + temp->f_free = ((temp->next->f_free + randv) & BIN_MASK); randprob = ((double)(lrand48() % 1000) / 1000); - if (prob > randprob) rand ^= BIN_MASK; - temp->faulty = ((temp->next->faulty + rand) & BIN_MASK); + if (prob > randprob) randv ^= BIN_MASK; + temp->faulty = ((temp->next->faulty + randv) & BIN_MASK); temp = temp->next; - rand >>= 1; + randv >>= 1; } } - rand = lrand48(); + randv = lrand48(); for (j=0; jf_free; savefaulty += temp->faulty; } - temp->f_free = ((temp->next->f_free + rand) & BIN_MASK); - randprob = ((double)(lrand48() % 1000) / 1000); if (prob > randprob) rand ^= BIN_MASK; - temp->faulty = ((temp->next->faulty + rand) & BIN_MASK); + temp->f_free = ((temp->next->f_free + randv) & BIN_MASK); + randprob = ((double)(lrand48() % 1000) / 1000); + if (prob > randprob) randv ^= BIN_MASK; + temp->faulty = ((temp->next->faulty + randv) & BIN_MASK); temp = temp->next; - rand >>= 1; + randv >>= 1; } - rand = lrand48(); + randv = lrand48(); if (structure[reg_len - 1] == '1') { savef_free += temp->f_free; savefaulty += temp->faulty; } - temp->f_free = ((savef_free + rand) & BIN_MASK); + temp->f_free = ((savef_free + randv) & BIN_MASK); randprob = ((double)(lrand48() % 10000) / 10000); - if (prob > randprob) rand ^= BIN_MASK; - temp->faulty = ((savefaulty + rand) & BIN_MASK); + if (prob > randprob) randv ^= BIN_MASK; + temp->faulty = ((savefaulty + randv) & BIN_MASK); temp = present; } From asl at math.spbu.ru Fri Oct 5 17:50:25 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Fri, 05 Oct 2007 22:50:25 -0000 Subject: [llvm-commits] [test-suite] r42658 - in /test-suite/trunk/SingleSource/Benchmarks/Misc: ReedSolomon.c flops.c mandel.c Message-ID: <200710052250.l95MoPHH011738@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 17:50:24 2007 New Revision: 42658 URL: http://llvm.org/viewvc/llvm-project?rev=42658&view=rev Log: Fix 'Misc' on mingw32 Modified: test-suite/trunk/SingleSource/Benchmarks/Misc/ReedSolomon.c test-suite/trunk/SingleSource/Benchmarks/Misc/flops.c test-suite/trunk/SingleSource/Benchmarks/Misc/mandel.c Modified: test-suite/trunk/SingleSource/Benchmarks/Misc/ReedSolomon.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/Benchmarks/Misc/ReedSolomon.c?rev=42658&r1=42657&r2=42658&view=diff ============================================================================== --- test-suite/trunk/SingleSource/Benchmarks/Misc/ReedSolomon.c (original) +++ test-suite/trunk/SingleSource/Benchmarks/Misc/ReedSolomon.c Fri Oct 5 17:50:24 2007 @@ -423,6 +423,10 @@ #define LENGTH 150000 #endif +#ifdef __MINGW32__ +#define random() rand() +#endif + for (i=0; i struct timeval tnow; Modified: test-suite/trunk/SingleSource/Benchmarks/Misc/mandel.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/Benchmarks/Misc/mandel.c?rev=42658&r1=42657&r2=42658&view=diff ============================================================================== --- test-suite/trunk/SingleSource/Benchmarks/Misc/mandel.c (original) +++ test-suite/trunk/SingleSource/Benchmarks/Misc/mandel.c Fri Oct 5 17:50:24 2007 @@ -18,7 +18,7 @@ #define I 1.0iF -#if defined(__FreeBSD__) || defined(__OpenBSD__) +#if defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__MINGW32__) #include #elif defined(__APPLE__) #include From asl at math.spbu.ru Fri Oct 5 18:00:17 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Fri, 05 Oct 2007 23:00:17 -0000 Subject: [llvm-commits] [test-suite] r42659 - /test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c Message-ID: <200710052300.l95N0HaS012294@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 18:00:17 2007 New Revision: 42659 URL: http://llvm.org/viewvc/llvm-project?rev=42659&view=rev Log: Fix smg2000 on mingw32 Modified: test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c Modified: test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c?rev=42659&r1=42658&r2=42659&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/ASCI_Purple/SMG2000/timer.c Fri Oct 5 18:00:17 2007 @@ -19,7 +19,9 @@ */ #include +#ifndef __MINGW32__ #include +#endif #ifdef TIMER_USE_MPI #include "mpi.h" #endif From sabre at nondot.org Fri Oct 5 18:01:44 2007 From: sabre at nondot.org (Chris Lattner) Date: Fri, 5 Oct 2007 18:01:44 -0500 Subject: [llvm-commits] CVS: llvm-www/img/PhotoDale.png Message-ID: <200710052301.l95N1itE012385@zion.cs.uiuc.edu> Changes in directory llvm-www/img: PhotoDale.png added (r1.1) --- Log message: add dale --- Diffs of the changes: (+0 -0) PhotoDale.png | 0 1 files changed Index: llvm-www/img/PhotoDale.png From sabre at nondot.org Fri Oct 5 18:01:44 2007 From: sabre at nondot.org (Chris Lattner) Date: Fri, 5 Oct 2007 18:01:44 -0500 Subject: [llvm-commits] CVS: llvm-www/developers.txt Message-ID: <200710052301.l95N1iMO012386@zion.cs.uiuc.edu> Changes in directory llvm-www: developers.txt updated: 1.14 -> 1.15 --- Log message: add dale --- Diffs of the changes: (+1 -0) developers.txt | 1 + 1 files changed, 1 insertion(+) Index: llvm-www/developers.txt diff -u llvm-www/developers.txt:1.14 llvm-www/developers.txt:1.15 --- llvm-www/developers.txt:1.14 Tue May 15 17:52:47 2007 +++ llvm-www/developers.txt Fri Oct 5 18:01:07 2007 @@ -11,6 +11,7 @@ Alkis Evlogimenos href=http://alkis.evlogimenos.com img=PhotoAlkis.jpg width=200 height=170 alt=alkis Brian Gaeke href=http://netfiles.uiuc.edu/gaeke/www/ img=PhotoBrian.png width=155 height=163 alt=brg Patrick Jenkins href=http://www.patjenk.com img=PhotoPatJenk.jpg width=200 width=173 height=130 alt=PatJenk +Dale Johannesen href=http://llvm.org/ img=img/PhotoDale.png width=132 height=107 alt=Dale Brad Jones href=http://www.nondot.org/~kungfoomaster/ img=PhotoBrad.jpg width=200 height=171 alt=KungFooMaster Anton Korobeynikov href=mailto:asl_at_math_dot_spbu_dot_ru img=PhotoAnton.jpg width=239 height=207 alt=Anton Jim Laskey href=mailto:jlaskey at mac.com img=PhotoJim.jpg width=128 height=128 alt=Wickund From sabre at nondot.org Fri Oct 5 18:02:36 2007 From: sabre at nondot.org (Chris Lattner) Date: Fri, 5 Oct 2007 18:02:36 -0500 Subject: [llvm-commits] CVS: llvm-www/developers.txt Message-ID: <200710052302.l95N2agT012434@zion.cs.uiuc.edu> Changes in directory llvm-www: developers.txt updated: 1.15 -> 1.16 --- Log message: fix path --- Diffs of the changes: (+1 -1) developers.txt | 2 +- 1 files changed, 1 insertion(+), 1 deletion(-) Index: llvm-www/developers.txt diff -u llvm-www/developers.txt:1.15 llvm-www/developers.txt:1.16 --- llvm-www/developers.txt:1.15 Fri Oct 5 18:01:07 2007 +++ llvm-www/developers.txt Fri Oct 5 18:02:18 2007 @@ -11,7 +11,7 @@ Alkis Evlogimenos href=http://alkis.evlogimenos.com img=PhotoAlkis.jpg width=200 height=170 alt=alkis Brian Gaeke href=http://netfiles.uiuc.edu/gaeke/www/ img=PhotoBrian.png width=155 height=163 alt=brg Patrick Jenkins href=http://www.patjenk.com img=PhotoPatJenk.jpg width=200 width=173 height=130 alt=PatJenk -Dale Johannesen href=http://llvm.org/ img=img/PhotoDale.png width=132 height=107 alt=Dale +Dale Johannesen href=http://llvm.org/ img=PhotoDale.png width=132 height=107 alt=Dale Brad Jones href=http://www.nondot.org/~kungfoomaster/ img=PhotoBrad.jpg width=200 height=171 alt=KungFooMaster Anton Korobeynikov href=mailto:asl_at_math_dot_spbu_dot_ru img=PhotoAnton.jpg width=239 height=207 alt=Anton Jim Laskey href=mailto:jlaskey at mac.com img=PhotoJim.jpg width=128 height=128 alt=Wickund From asl at math.spbu.ru Fri Oct 5 18:07:16 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Fri, 05 Oct 2007 23:07:16 -0000 Subject: [llvm-commits] [test-suite] r42660 - in /test-suite/trunk/MultiSource/Benchmarks: Fhourstones-3.1/Game.h Fhourstones-3.1/SearchGame.c Fhourstones/time.c Message-ID: <200710052307.l95N7GtH012741@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 18:07:16 2007 New Revision: 42660 URL: http://llvm.org/viewvc/llvm-project?rev=42660&view=rev Log: Clean Fhoursstones: remove time-related routines, unbreak mingw32 Modified: test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/Game.h test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/SearchGame.c test-suite/trunk/MultiSource/Benchmarks/Fhourstones/time.c Modified: test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/Game.h URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/Game.h?rev=42660&r1=42659&r2=42660&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/Game.h (original) +++ test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/Game.h Fri Oct 5 18:07:16 2007 @@ -42,7 +42,8 @@ #define TOP (BOTTOM << HEIGHT) #include -typedef u_int64_t uint64; +#include +typedef uint64_t uint64; typedef int64_t int64; uint64 color[2]; // black and white bitboard Modified: test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/SearchGame.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/SearchGame.c?rev=42660&r1=42659&r2=42660&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/SearchGame.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Fhourstones-3.1/SearchGame.c Fri Oct 5 18:07:16 2007 @@ -12,16 +12,14 @@ #include "TransGame.h" #include -#include #define BOOKPLY 0 // full-width search up to this depth #define REPORTPLY -1 uint64 millisecs() { - struct rusage rusage; - getrusage(RUSAGE_SELF,&rusage); - return rusage.ru_utime.tv_sec * 1000 + rusage.ru_utime.tv_usec / 1000; + static int64 Time = 0; + return Time++; // No time count for LLVM } int history[2][SIZE1]; Modified: test-suite/trunk/MultiSource/Benchmarks/Fhourstones/time.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Fhourstones/time.c?rev=42660&r1=42659&r2=42660&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Fhourstones/time.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Fhourstones/time.c Fri Oct 5 18:07:16 2007 @@ -3,17 +3,11 @@ /* add more platforms if necessary */ #ifdef UNIX -#include -#include int64 millisecs() { - struct rusage rusage; static int64 Time = 0; return ++Time; /* DO NOT TIME FOR LLVM */ - - getrusage(RUSAGE_SELF,&rusage); - return rusage.ru_utime.tv_sec * 1000 + rusage.ru_utime.tv_usec / 1000; } #endif From evan.cheng at apple.com Fri Oct 5 18:13:21 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 23:13:21 -0000 Subject: [llvm-commits] [test-suite] r42661 - /test-suite/trunk/Makefile.programs Message-ID: <200710052313.l95NDLn9012928@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 18:13:21 2007 New Revision: 42661 URL: http://llvm.org/viewvc/llvm-project?rev=42661&view=rev Log: Commute x86 cmove instructions by swapping the operands and change the condition to its inverse. Testing this as llcbeta Modified: test-suite/trunk/Makefile.programs Modified: test-suite/trunk/Makefile.programs URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/Makefile.programs?rev=42661&r1=42660&r2=42661&view=diff ============================================================================== --- test-suite/trunk/Makefile.programs (original) +++ test-suite/trunk/Makefile.programs Fri Oct 5 18:13:21 2007 @@ -221,7 +221,8 @@ LLCBETAOPTION := -sched=simple endif ifeq ($(ARCH),x86) -LLCBETAOPTION := -regalloc=local -fast +LLCBETAOPTION := -enable-x86-commute-cmove +#-regalloc=local -fast #-disable-rematerialization #-disable-fp-elim #-regalloc=bigblock -fast From evan.cheng at apple.com Fri Oct 5 18:13:21 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 23:13:21 -0000 Subject: [llvm-commits] [llvm] r42661 - in /llvm/trunk/lib/Target/X86: X86InstrInfo.cpp X86InstrInfo.td X86InstrX86-64.td Message-ID: <200710052313.l95NDLI5012935@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 18:13:21 2007 New Revision: 42661 URL: http://llvm.org/viewvc/llvm-project?rev=42661&view=rev Log: Commute x86 cmove instructions by swapping the operands and change the condition to its inverse. Testing this as llcbeta Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp llvm/trunk/lib/Target/X86/X86InstrInfo.td llvm/trunk/lib/Target/X86/X86InstrX86-64.td Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.cpp?rev=42661&r1=42660&r2=42661&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.cpp Fri Oct 5 18:13:21 2007 @@ -22,8 +22,15 @@ #include "llvm/CodeGen/LiveVariables.h" #include "llvm/CodeGen/SSARegMap.h" #include "llvm/Target/TargetOptions.h" +#include "llvm/Support/CommandLine.h" using namespace llvm; +namespace { + cl::opt + EnableCommuteCMove("enable-x86-commute-cmove", + cl::desc("Commute conditional moves by inverting conditions")); +} + X86InstrInfo::X86InstrInfo(X86TargetMachine &tm) : TargetInstrInfo(X86Insts, array_lengthof(X86Insts)), TM(tm), RI(tm, *this) { @@ -366,7 +373,6 @@ /// commute them. /// MachineInstr *X86InstrInfo::commuteInstruction(MachineInstr *MI) const { - // FIXME: Can commute cmoves by changing the condition! switch (MI->getOpcode()) { case X86::SHRD16rri8: // A = SHRD16rri8 B, C, I -> A = SHLD16rri8 C, B, (16-I) case X86::SHLD16rri8: // A = SHLD16rri8 B, C, I -> A = SHRD16rri8 C, B, (16-I) @@ -394,6 +400,100 @@ return BuildMI(get(Opc), A).addReg(C, false, false, CisKill) .addReg(B, false, false, BisKill).addImm(Size-Amt); } + case X86::CMOVB16rr: + case X86::CMOVB32rr: + case X86::CMOVB64rr: + case X86::CMOVAE16rr: + case X86::CMOVAE32rr: + case X86::CMOVAE64rr: + case X86::CMOVE16rr: + case X86::CMOVE32rr: + case X86::CMOVE64rr: + case X86::CMOVNE16rr: + case X86::CMOVNE32rr: + case X86::CMOVNE64rr: + case X86::CMOVBE16rr: + case X86::CMOVBE32rr: + case X86::CMOVBE64rr: + case X86::CMOVA16rr: + case X86::CMOVA32rr: + case X86::CMOVA64rr: + case X86::CMOVL16rr: + case X86::CMOVL32rr: + case X86::CMOVL64rr: + case X86::CMOVGE16rr: + case X86::CMOVGE32rr: + case X86::CMOVGE64rr: + case X86::CMOVLE16rr: + case X86::CMOVLE32rr: + case X86::CMOVLE64rr: + case X86::CMOVG16rr: + case X86::CMOVG32rr: + case X86::CMOVG64rr: + case X86::CMOVS16rr: + case X86::CMOVS32rr: + case X86::CMOVS64rr: + case X86::CMOVNS16rr: + case X86::CMOVNS32rr: + case X86::CMOVNS64rr: + case X86::CMOVP16rr: + case X86::CMOVP32rr: + case X86::CMOVP64rr: + case X86::CMOVNP16rr: + case X86::CMOVNP32rr: + case X86::CMOVNP64rr: { + if (!EnableCommuteCMove) + return 0; + unsigned Opc = 0; + switch (MI->getOpcode()) { + default: break; + case X86::CMOVB16rr: Opc = X86::CMOVAE16rr; break; + case X86::CMOVB32rr: Opc = X86::CMOVAE32rr; break; + case X86::CMOVB64rr: Opc = X86::CMOVAE64rr; break; + case X86::CMOVAE16rr: Opc = X86::CMOVB16rr; break; + case X86::CMOVAE32rr: Opc = X86::CMOVB32rr; break; + case X86::CMOVAE64rr: Opc = X86::CMOVB64rr; break; + case X86::CMOVE16rr: Opc = X86::CMOVNE16rr; break; + case X86::CMOVE32rr: Opc = X86::CMOVNE32rr; break; + case X86::CMOVE64rr: Opc = X86::CMOVNE64rr; break; + case X86::CMOVNE16rr: Opc = X86::CMOVE16rr; break; + case X86::CMOVNE32rr: Opc = X86::CMOVE32rr; break; + case X86::CMOVNE64rr: Opc = X86::CMOVE64rr; break; + case X86::CMOVBE16rr: Opc = X86::CMOVA16rr; break; + case X86::CMOVBE32rr: Opc = X86::CMOVA32rr; break; + case X86::CMOVBE64rr: Opc = X86::CMOVA64rr; break; + case X86::CMOVA16rr: Opc = X86::CMOVBE16rr; break; + case X86::CMOVA32rr: Opc = X86::CMOVBE32rr; break; + case X86::CMOVA64rr: Opc = X86::CMOVBE64rr; break; + case X86::CMOVL16rr: Opc = X86::CMOVGE16rr; break; + case X86::CMOVL32rr: Opc = X86::CMOVGE32rr; break; + case X86::CMOVL64rr: Opc = X86::CMOVGE64rr; break; + case X86::CMOVGE16rr: Opc = X86::CMOVL16rr; break; + case X86::CMOVGE32rr: Opc = X86::CMOVL32rr; break; + case X86::CMOVGE64rr: Opc = X86::CMOVL64rr; break; + case X86::CMOVLE16rr: Opc = X86::CMOVG16rr; break; + case X86::CMOVLE32rr: Opc = X86::CMOVG32rr; break; + case X86::CMOVLE64rr: Opc = X86::CMOVG64rr; break; + case X86::CMOVG16rr: Opc = X86::CMOVLE16rr; break; + case X86::CMOVG32rr: Opc = X86::CMOVLE32rr; break; + case X86::CMOVG64rr: Opc = X86::CMOVLE64rr; break; + case X86::CMOVS16rr: Opc = X86::CMOVNS16rr; break; + case X86::CMOVS32rr: Opc = X86::CMOVNS32rr; break; + case X86::CMOVS64rr: Opc = X86::CMOVNS32rr; break; + case X86::CMOVNS16rr: Opc = X86::CMOVS16rr; break; + case X86::CMOVNS32rr: Opc = X86::CMOVS32rr; break; + case X86::CMOVNS64rr: Opc = X86::CMOVS64rr; break; + case X86::CMOVP16rr: Opc = X86::CMOVNP16rr; break; + case X86::CMOVP32rr: Opc = X86::CMOVNP32rr; break; + case X86::CMOVP64rr: Opc = X86::CMOVNP32rr; break; + case X86::CMOVNP16rr: Opc = X86::CMOVP16rr; break; + case X86::CMOVNP32rr: Opc = X86::CMOVP32rr; break; + case X86::CMOVNP64rr: Opc = X86::CMOVP64rr; break; + } + + MI->setInstrDescriptor(get(Opc)); + // Fallthrough intended. + } default: return TargetInstrInfo::commuteInstruction(MI); } Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.td?rev=42661&r1=42660&r2=42661&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrInfo.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrInfo.td Fri Oct 5 18:13:21 2007 @@ -647,30 +647,19 @@ // Conditional moves let Uses = [EFLAGS] in { +let isCommutable = 1 in { def CMOVB16rr : I<0x42, MRMSrcReg, // if , TB, OpSize; -def CMOVB16rm : I<0x42, MRMSrcMem, // if , - TB, OpSize; def CMOVB32rr : I<0x42, MRMSrcReg, // if , TB; -def CMOVB32rm : I<0x42, MRMSrcMem, // if , - TB; def CMOVAE16rr: I<0x43, MRMSrcReg, // if >=u, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), @@ -678,324 +667,327 @@ [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_AE, EFLAGS))]>, TB, OpSize; -def CMOVAE16rm: I<0x43, MRMSrcMem, // if >=u, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovae\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_AE, EFLAGS))]>, - TB, OpSize; def CMOVAE32rr: I<0x43, MRMSrcReg, // if >=u, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovae\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_AE, EFLAGS))]>, TB; -def CMOVAE32rm: I<0x43, MRMSrcMem, // if >=u, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovae\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_AE, EFLAGS))]>, - TB; - def CMOVE16rr : I<0x44, MRMSrcReg, // if ==, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmove\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_E, EFLAGS))]>, TB, OpSize; -def CMOVE16rm : I<0x44, MRMSrcMem, // if ==, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmove\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_E, EFLAGS))]>, - TB, OpSize; def CMOVE32rr : I<0x44, MRMSrcReg, // if ==, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmove\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_E, EFLAGS))]>, TB; -def CMOVE32rm : I<0x44, MRMSrcMem, // if ==, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmove\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_E, EFLAGS))]>, - TB; - def CMOVNE16rr: I<0x45, MRMSrcReg, // if !=, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovne\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_NE, EFLAGS))]>, TB, OpSize; -def CMOVNE16rm: I<0x45, MRMSrcMem, // if !=, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovne\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_NE, EFLAGS))]>, - TB, OpSize; def CMOVNE32rr: I<0x45, MRMSrcReg, // if !=, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovne\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_NE, EFLAGS))]>, TB; -def CMOVNE32rm: I<0x45, MRMSrcMem, // if !=, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovne\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_NE, EFLAGS))]>, - TB; - def CMOVBE16rr: I<0x46, MRMSrcReg, // if <=u, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovbe\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_BE, EFLAGS))]>, TB, OpSize; -def CMOVBE16rm: I<0x46, MRMSrcMem, // if <=u, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovbe\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_BE, EFLAGS))]>, - TB, OpSize; def CMOVBE32rr: I<0x46, MRMSrcReg, // if <=u, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovbe\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_BE, EFLAGS))]>, TB; -def CMOVBE32rm: I<0x46, MRMSrcMem, // if <=u, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovbe\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_BE, EFLAGS))]>, - TB; - def CMOVA16rr : I<0x47, MRMSrcReg, // if >u, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmova\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_A, EFLAGS))]>, TB, OpSize; -def CMOVA16rm : I<0x47, MRMSrcMem, // if >u, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmova\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_A, EFLAGS))]>, - TB, OpSize; def CMOVA32rr : I<0x47, MRMSrcReg, // if >u, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmova\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_A, EFLAGS))]>, TB; -def CMOVA32rm : I<0x47, MRMSrcMem, // if >u, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmova\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_A, EFLAGS))]>, - TB; - def CMOVL16rr : I<0x4C, MRMSrcReg, // if , TB, OpSize; -def CMOVL16rm : I<0x4C, MRMSrcMem, // if , - TB, OpSize; def CMOVL32rr : I<0x4C, MRMSrcReg, // if , TB; -def CMOVL32rm : I<0x4C, MRMSrcMem, // if , - TB; - def CMOVGE16rr: I<0x4D, MRMSrcReg, // if >=s, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovge\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_GE, EFLAGS))]>, TB, OpSize; -def CMOVGE16rm: I<0x4D, MRMSrcMem, // if >=s, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovge\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_GE, EFLAGS))]>, - TB, OpSize; def CMOVGE32rr: I<0x4D, MRMSrcReg, // if >=s, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovge\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_GE, EFLAGS))]>, TB; -def CMOVGE32rm: I<0x4D, MRMSrcMem, // if >=s, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovge\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_GE, EFLAGS))]>, - TB; - def CMOVLE16rr: I<0x4E, MRMSrcReg, // if <=s, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovle\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_LE, EFLAGS))]>, TB, OpSize; -def CMOVLE16rm: I<0x4E, MRMSrcMem, // if <=s, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovle\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_LE, EFLAGS))]>, - TB, OpSize; def CMOVLE32rr: I<0x4E, MRMSrcReg, // if <=s, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovle\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_LE, EFLAGS))]>, TB; -def CMOVLE32rm: I<0x4E, MRMSrcMem, // if <=s, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovle\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_LE, EFLAGS))]>, - TB; - def CMOVG16rr : I<0x4F, MRMSrcReg, // if >s, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovg\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_G, EFLAGS))]>, TB, OpSize; -def CMOVG16rm : I<0x4F, MRMSrcMem, // if >s, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovg\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_G, EFLAGS))]>, - TB, OpSize; def CMOVG32rr : I<0x4F, MRMSrcReg, // if >s, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovg\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_G, EFLAGS))]>, TB; -def CMOVG32rm : I<0x4F, MRMSrcMem, // if >s, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovg\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_G, EFLAGS))]>, - TB; - def CMOVS16rr : I<0x48, MRMSrcReg, // if signed, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovs\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_S, EFLAGS))]>, TB, OpSize; -def CMOVS16rm : I<0x48, MRMSrcMem, // if signed, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovs\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_S, EFLAGS))]>, - TB, OpSize; def CMOVS32rr : I<0x48, MRMSrcReg, // if signed, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovs\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_S, EFLAGS))]>, TB; -def CMOVS32rm : I<0x48, MRMSrcMem, // if signed, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovs\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_S, EFLAGS))]>, - TB; - def CMOVNS16rr: I<0x49, MRMSrcReg, // if !signed, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovns\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_NS, EFLAGS))]>, TB, OpSize; -def CMOVNS16rm: I<0x49, MRMSrcMem, // if !signed, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovns\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_NS, EFLAGS))]>, - TB, OpSize; def CMOVNS32rr: I<0x49, MRMSrcReg, // if !signed, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovns\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_NS, EFLAGS))]>, TB; -def CMOVNS32rm: I<0x49, MRMSrcMem, // if !signed, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovns\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_NS, EFLAGS))]>, - TB; - def CMOVP16rr : I<0x4A, MRMSrcReg, // if parity, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovp\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_P, EFLAGS))]>, TB, OpSize; -def CMOVP16rm : I<0x4A, MRMSrcMem, // if parity, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovp\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_P, EFLAGS))]>, - TB, OpSize; def CMOVP32rr : I<0x4A, MRMSrcReg, // if parity, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovp\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_P, EFLAGS))]>, TB; -def CMOVP32rm : I<0x4A, MRMSrcMem, // if parity, GR32 = [mem32] - (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), - "cmovp\t{$src2, $dst|$dst, $src2}", - [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), - X86_COND_P, EFLAGS))]>, - TB; - def CMOVNP16rr : I<0x4B, MRMSrcReg, // if !parity, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), "cmovnp\t{$src2, $dst|$dst, $src2}", [(set GR16:$dst, (X86cmov GR16:$src1, GR16:$src2, X86_COND_NP, EFLAGS))]>, TB, OpSize; -def CMOVNP16rm : I<0x4B, MRMSrcMem, // if !parity, GR16 = [mem16] - (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), - "cmovnp\t{$src2, $dst|$dst, $src2}", - [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), - X86_COND_NP, EFLAGS))]>, - TB, OpSize; def CMOVNP32rr : I<0x4B, MRMSrcReg, // if !parity, GR32 = GR32 (outs GR32:$dst), (ins GR32:$src1, GR32:$src2), "cmovnp\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, GR32:$src2, X86_COND_NP, EFLAGS))]>, TB; +} // isCommutable = 1 + def CMOVNP32rm : I<0x4B, MRMSrcMem, // if !parity, GR32 = [mem32] (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), "cmovnp\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), X86_COND_NP, EFLAGS))]>, TB; + +def CMOVB16rm : I<0x42, MRMSrcMem, // if , + TB, OpSize; +def CMOVB32rm : I<0x42, MRMSrcMem, // if , + TB; +def CMOVAE16rm: I<0x43, MRMSrcMem, // if >=u, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovae\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_AE, EFLAGS))]>, + TB, OpSize; +def CMOVAE32rm: I<0x43, MRMSrcMem, // if >=u, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovae\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_AE, EFLAGS))]>, + TB; +def CMOVE16rm : I<0x44, MRMSrcMem, // if ==, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmove\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_E, EFLAGS))]>, + TB, OpSize; +def CMOVE32rm : I<0x44, MRMSrcMem, // if ==, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmove\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_E, EFLAGS))]>, + TB; +def CMOVNE16rm: I<0x45, MRMSrcMem, // if !=, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovne\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_NE, EFLAGS))]>, + TB, OpSize; +def CMOVNE32rm: I<0x45, MRMSrcMem, // if !=, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovne\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_NE, EFLAGS))]>, + TB; +def CMOVBE16rm: I<0x46, MRMSrcMem, // if <=u, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovbe\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_BE, EFLAGS))]>, + TB, OpSize; +def CMOVBE32rm: I<0x46, MRMSrcMem, // if <=u, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovbe\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_BE, EFLAGS))]>, + TB; +def CMOVA16rm : I<0x47, MRMSrcMem, // if >u, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmova\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_A, EFLAGS))]>, + TB, OpSize; +def CMOVA32rm : I<0x47, MRMSrcMem, // if >u, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmova\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_A, EFLAGS))]>, + TB; +def CMOVL16rm : I<0x4C, MRMSrcMem, // if , + TB, OpSize; +def CMOVL32rm : I<0x4C, MRMSrcMem, // if , + TB; +def CMOVGE16rm: I<0x4D, MRMSrcMem, // if >=s, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovge\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_GE, EFLAGS))]>, + TB, OpSize; +def CMOVGE32rm: I<0x4D, MRMSrcMem, // if >=s, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovge\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_GE, EFLAGS))]>, + TB; +def CMOVLE16rm: I<0x4E, MRMSrcMem, // if <=s, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovle\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_LE, EFLAGS))]>, + TB, OpSize; +def CMOVLE32rm: I<0x4E, MRMSrcMem, // if <=s, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovle\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_LE, EFLAGS))]>, + TB; +def CMOVG16rm : I<0x4F, MRMSrcMem, // if >s, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovg\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_G, EFLAGS))]>, + TB, OpSize; +def CMOVG32rm : I<0x4F, MRMSrcMem, // if >s, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovg\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_G, EFLAGS))]>, + TB; +def CMOVS16rm : I<0x48, MRMSrcMem, // if signed, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovs\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_S, EFLAGS))]>, + TB, OpSize; +def CMOVS32rm : I<0x48, MRMSrcMem, // if signed, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovs\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_S, EFLAGS))]>, + TB; +def CMOVNS16rm: I<0x49, MRMSrcMem, // if !signed, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovns\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_NS, EFLAGS))]>, + TB, OpSize; +def CMOVNS32rm: I<0x49, MRMSrcMem, // if !signed, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovns\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_NS, EFLAGS))]>, + TB; +def CMOVP16rm : I<0x4A, MRMSrcMem, // if parity, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovp\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_P, EFLAGS))]>, + TB, OpSize; +def CMOVP32rm : I<0x4A, MRMSrcMem, // if parity, GR32 = [mem32] + (outs GR32:$dst), (ins GR32:$src1, i32mem:$src2), + "cmovp\t{$src2, $dst|$dst, $src2}", + [(set GR32:$dst, (X86cmov GR32:$src1, (loadi32 addr:$src2), + X86_COND_P, EFLAGS))]>, + TB; +def CMOVNP16rm : I<0x4B, MRMSrcMem, // if !parity, GR16 = [mem16] + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), + "cmovnp\t{$src2, $dst|$dst, $src2}", + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 addr:$src2), + X86_COND_NP, EFLAGS))]>, + TB, OpSize; } // Uses = [EFLAGS] Modified: llvm/trunk/lib/Target/X86/X86InstrX86-64.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrX86-64.td?rev=42661&r1=42660&r2=42661&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrX86-64.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrX86-64.td Fri Oct 5 18:13:21 2007 @@ -778,141 +778,144 @@ // Conditional moves let Uses = [EFLAGS], isTwoAddress = 1 in { +let isCommutable = 1 in { def CMOVB64rr : RI<0x42, MRMSrcReg, // if , TB; -def CMOVB64rm : RI<0x42, MRMSrcMem, // if , TB; def CMOVAE64rr: RI<0x43, MRMSrcReg, // if >=u, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovae\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_AE, EFLAGS))]>, TB; -def CMOVAE64rm: RI<0x43, MRMSrcMem, // if >=u, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovae\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_AE, EFLAGS))]>, TB; def CMOVE64rr : RI<0x44, MRMSrcReg, // if ==, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmove\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_E, EFLAGS))]>, TB; -def CMOVE64rm : RI<0x44, MRMSrcMem, // if ==, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmove\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_E, EFLAGS))]>, TB; def CMOVNE64rr: RI<0x45, MRMSrcReg, // if !=, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovne\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_NE, EFLAGS))]>, TB; -def CMOVNE64rm: RI<0x45, MRMSrcMem, // if !=, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovne\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_NE, EFLAGS))]>, TB; def CMOVBE64rr: RI<0x46, MRMSrcReg, // if <=u, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovbe\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_BE, EFLAGS))]>, TB; -def CMOVBE64rm: RI<0x46, MRMSrcMem, // if <=u, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovbe\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_BE, EFLAGS))]>, TB; def CMOVA64rr : RI<0x47, MRMSrcReg, // if >u, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmova\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_A, EFLAGS))]>, TB; -def CMOVA64rm : RI<0x47, MRMSrcMem, // if >u, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmova\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_A, EFLAGS))]>, TB; def CMOVL64rr : RI<0x4C, MRMSrcReg, // if , TB; -def CMOVL64rm : RI<0x4C, MRMSrcMem, // if , TB; def CMOVGE64rr: RI<0x4D, MRMSrcReg, // if >=s, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovge\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_GE, EFLAGS))]>, TB; -def CMOVGE64rm: RI<0x4D, MRMSrcMem, // if >=s, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovge\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_GE, EFLAGS))]>, TB; def CMOVLE64rr: RI<0x4E, MRMSrcReg, // if <=s, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovle\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_LE, EFLAGS))]>, TB; -def CMOVLE64rm: RI<0x4E, MRMSrcMem, // if <=s, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovle\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_LE, EFLAGS))]>, TB; def CMOVG64rr : RI<0x4F, MRMSrcReg, // if >s, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovg\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_G, EFLAGS))]>, TB; -def CMOVG64rm : RI<0x4F, MRMSrcMem, // if >s, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovg\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_G, EFLAGS))]>, TB; def CMOVS64rr : RI<0x48, MRMSrcReg, // if signed, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovs\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_S, EFLAGS))]>, TB; -def CMOVS64rm : RI<0x48, MRMSrcMem, // if signed, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovs\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_S, EFLAGS))]>, TB; def CMOVNS64rr: RI<0x49, MRMSrcReg, // if !signed, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovns\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_NS, EFLAGS))]>, TB; -def CMOVNS64rm: RI<0x49, MRMSrcMem, // if !signed, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovns\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_NS, EFLAGS))]>, TB; def CMOVP64rr : RI<0x4A, MRMSrcReg, // if parity, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovp\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_P, EFLAGS))]>, TB; -def CMOVP64rm : RI<0x4A, MRMSrcMem, // if parity, GR64 = [mem64] - (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), - "cmovp\t{$src2, $dst|$dst, $src2}", - [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), - X86_COND_P, EFLAGS))]>, TB; def CMOVNP64rr : RI<0x4B, MRMSrcReg, // if !parity, GR64 = GR64 (outs GR64:$dst), (ins GR64:$src1, GR64:$src2), "cmovnp\t{$src2, $dst|$dst, $src2}", [(set GR64:$dst, (X86cmov GR64:$src1, GR64:$src2, X86_COND_NP, EFLAGS))]>, TB; +} // isCommutable = 1 + +def CMOVB64rm : RI<0x42, MRMSrcMem, // if , TB; +def CMOVAE64rm: RI<0x43, MRMSrcMem, // if >=u, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovae\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_AE, EFLAGS))]>, TB; +def CMOVE64rm : RI<0x44, MRMSrcMem, // if ==, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmove\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_E, EFLAGS))]>, TB; +def CMOVNE64rm: RI<0x45, MRMSrcMem, // if !=, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovne\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_NE, EFLAGS))]>, TB; +def CMOVBE64rm: RI<0x46, MRMSrcMem, // if <=u, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovbe\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_BE, EFLAGS))]>, TB; +def CMOVA64rm : RI<0x47, MRMSrcMem, // if >u, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmova\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_A, EFLAGS))]>, TB; +def CMOVL64rm : RI<0x4C, MRMSrcMem, // if , TB; +def CMOVGE64rm: RI<0x4D, MRMSrcMem, // if >=s, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovge\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_GE, EFLAGS))]>, TB; +def CMOVLE64rm: RI<0x4E, MRMSrcMem, // if <=s, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovle\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_LE, EFLAGS))]>, TB; +def CMOVG64rm : RI<0x4F, MRMSrcMem, // if >s, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovg\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_G, EFLAGS))]>, TB; +def CMOVS64rm : RI<0x48, MRMSrcMem, // if signed, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovs\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_S, EFLAGS))]>, TB; +def CMOVNS64rm: RI<0x49, MRMSrcMem, // if !signed, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovns\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_NS, EFLAGS))]>, TB; +def CMOVP64rm : RI<0x4A, MRMSrcMem, // if parity, GR64 = [mem64] + (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), + "cmovp\t{$src2, $dst|$dst, $src2}", + [(set GR64:$dst, (X86cmov GR64:$src1, (loadi64 addr:$src2), + X86_COND_P, EFLAGS))]>, TB; def CMOVNP64rm : RI<0x4B, MRMSrcMem, // if !parity, GR64 = [mem64] (outs GR64:$dst), (ins GR64:$src1, i64mem:$src2), "cmovnp\t{$src2, $dst|$dst, $src2}", From evan.cheng at apple.com Fri Oct 5 18:33:09 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Fri, 05 Oct 2007 23:33:09 -0000 Subject: [llvm-commits] [llvm] r42664 - /llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Message-ID: <200710052333.l95NXAb1013723@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 18:33:09 2007 New Revision: 42664 URL: http://llvm.org/viewvc/llvm-project?rev=42664&view=rev Log: Test case for 3-address conversion. Added: llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Added: llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll?rev=42664&view=auto ============================================================================== --- llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll (added) +++ llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Fri Oct 5 18:33:09 2007 @@ -0,0 +1,32 @@ +; RUN: llvm-as < %s | llc | grep lea + + %struct.anon = type { [3 x double], double, %struct.node*, [64 x %struct.bnode*], [64 x %struct.bnode*] } + %struct.bnode = type { i16, double, [3 x double], i32, i32, [3 x double], [3 x double], [3 x double], double, %struct.bnode*, %struct.bnode* } + %struct.node = type { i16, double, [3 x double], i32, i32 } + +define fastcc void @old_main() { +entry: + %tmp44 = malloc %struct.anon ; <%struct.anon*> [#uses=2] + store double 4.000000e+00, double* null, align 4 + br label %bb41 + +bb41: ; preds = %uniform_testdata.exit, %entry + %i.0110 = phi i32 [ 0, %entry ], [ %tmp48, %uniform_testdata.exit ] ; [#uses=2] + %tmp48 = add i32 %i.0110, 1 ; [#uses=1] + br i1 false, label %uniform_testdata.exit, label %bb33.preheader.i + +bb33.preheader.i: ; preds = %bb41 + ret void + +uniform_testdata.exit: ; preds = %bb41 + %tmp57 = getelementptr %struct.anon* %tmp44, i32 0, i32 3, i32 %i.0110 ; <%struct.bnode**> [#uses=1] + store %struct.bnode* null, %struct.bnode** %tmp57, align 4 + br i1 false, label %bb154, label %bb41 + +bb154: ; preds = %bb154, %uniform_testdata.exit + br i1 false, label %bb166, label %bb154 + +bb166: ; preds = %bb154 + %tmp169 = getelementptr %struct.anon* %tmp44, i32 0, i32 3, i32 0 ; <%struct.bnode**> [#uses=0] + ret void +} From asl at math.spbu.ru Fri Oct 5 18:37:11 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Fri, 05 Oct 2007 23:37:11 -0000 Subject: [llvm-commits] [test-suite] r42665 - in /test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft: ft.c graph.c Message-ID: <200710052337.l95NbB2r013860@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 18:37:11 2007 New Revision: 42665 URL: http://llvm.org/viewvc/llvm-project?rev=42665&view=rev Log: Time reporting is useless here, remove it. This unbreaks mingw32 also :) Modified: test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/ft.c test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/graph.c Modified: test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/ft.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/ft.c?rev=42665&r1=42664&r2=42665&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/ft.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/ft.c Fri Oct 5 18:37:11 2007 @@ -40,7 +40,6 @@ #include #include -#include #include "Fheap.h" #include "graph.h" @@ -51,12 +50,15 @@ #define DEFAULT_N_VERTEX 10 #define DEFAULT_N_EDGE 9 +#ifdef __MINGW32__ +#define srandom(x) srand(x) +#endif + /* * Local functions. */ void PrintMST(Vertices * graph); Vertices * MST(Vertices * graph); -void PrintRUsage(struct rusage * rUBuf1, struct rusage * rUBuf2); /* * Local variables. @@ -69,8 +71,6 @@ int nVertex; int nEdge; Vertices * graph; - struct rusage * rUBuf1; - struct rusage * rUBuf2; nVertex = DEFAULT_N_VERTEX; nEdge = DEFAULT_N_EDGE; @@ -88,11 +88,6 @@ } } - rUBuf1 = (struct rusage *)malloc(sizeof(struct rusage)); - assert(rUBuf1 != NULL); - rUBuf2 = (struct rusage *)malloc(sizeof(struct rusage)); - assert(rUBuf1 != NULL); - if(debug) { printf("Generating a connected graph ... "); @@ -105,12 +100,8 @@ printf("done\nFinding the mininmum spanning tree ... "); } - getrusage(RUSAGE_SELF, rUBuf1); - graph = MST(graph); - getrusage(RUSAGE_SELF, rUBuf2); - if(debug) { printf("done\nThe graph:\n"); @@ -123,7 +114,6 @@ { printf("Time spent in finding the mininum spanning tree:\n"); } - PrintRUsage(rUBuf1, rUBuf2); #ifdef PLUS_STATS PrintDerefStats(stderr); PrintHeapSize(stderr); @@ -200,34 +190,3 @@ } -void -PrintRUsage(struct rusage * rUBuf1, struct rusage * rUBuf2) -{ - long sec; - long usec; - - sec = (*rUBuf2).ru_utime.tv_sec - (*rUBuf1).ru_utime.tv_sec; - usec = (*rUBuf2).ru_utime.tv_usec - (*rUBuf1).ru_utime.tv_usec; - if(usec < 0) - { - sec--; - usec += 1000000; - } - assert(sec >= 0); - assert(usec >= 0); - - /*fprintf(stderr, "User time: %d%06d usec\n", sec, usec); */ - - sec = (*rUBuf2).ru_stime.tv_sec - (*rUBuf1).ru_stime.tv_sec; - usec = (*rUBuf2).ru_stime.tv_usec - (*rUBuf1).ru_stime.tv_usec; - - if(usec < 0) - { - sec--; - usec += 1000000; - } - assert(sec >= 0); - assert(usec >= 0); - - /*fprintf(stderr, "Sys time: %d%06d usec\n", sec, usec); */ -} Modified: test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/graph.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/graph.c?rev=42665&r1=42664&r2=42665&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/graph.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Ptrdist/ft/graph.c Fri Oct 5 18:37:11 2007 @@ -36,6 +36,10 @@ #define TRUE 1 #define FALSE 0 +#ifdef __MINGW32__ +#define random() rand() +#endif + #define GET_WEIGHT (random() + 1) % MAX_WEIGHT /* From clattner at apple.com Fri Oct 5 18:45:17 2007 From: clattner at apple.com (Chris Lattner) Date: Fri, 5 Oct 2007 16:45:17 -0700 Subject: [llvm-commits] [llvm] r42661 - in /llvm/trunk/lib/Target/X86: X86InstrInfo.cpp X86InstrInfo.td X86InstrX86-64.td In-Reply-To: <200710052313.l95NDLI5012935@zion.cs.uiuc.edu> References: <200710052313.l95NDLI5012935@zion.cs.uiuc.edu> Message-ID: <54364F06-F5DE-457D-BC58-C8BF09A30196@apple.com> > URL: http://llvm.org/viewvc/llvm-project?rev=42661&view=rev > Log: > Commute x86 cmove instructions by swapping the operands and change > the condition > to its inverse. > Testing this as llcbeta Random crazy idea: > +def CMOVB16rm : I<0x42, MRMSrcMem, // if + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), > + "cmovb\t{$src2, $dst|$dst, $src2}", > + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 > addr:$src2), > + X86_COND_B, EFLAGS))]>, > + TB, OpSize; Instead of having one instruction for each comparison code, would it make sense to have one "CMOV16rm" that took the condcode as an immediate argument? If you had that, you could change the comparison just by twiddling the immediate operand and wouldn't have a bazillion CMOV instructions. -Chris From gordonhenriksen at mac.com Fri Oct 5 18:59:36 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Fri, 05 Oct 2007 23:59:36 -0000 Subject: [llvm-commits] [llvm] r42666 - in /llvm/trunk: Xcode/LLVM.xcodeproj/project.pbxproj include/llvm-c/Core.h include/llvm/Support/CHelpers.h lib/Bitcode/Writer/BitWriter.cpp lib/VMCore/Core.cpp Message-ID: <200710052359.l95NxbdM014885@zion.cs.uiuc.edu> Author: gordon Date: Fri Oct 5 18:59:36 2007 New Revision: 42666 URL: http://llvm.org/viewvc/llvm-project?rev=42666&view=rev Log: Removing the silly CHelpers header by rolling wrap and unwrap into the C bindings headers themselves, hidden behind #ifdef __cplusplus. Removed: llvm/trunk/include/llvm/Support/CHelpers.h Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp llvm/trunk/lib/VMCore/Core.cpp Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj?rev=42666&r1=42665&r2=42666&view=diff ============================================================================== --- llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj (original) +++ llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj Fri Oct 5 18:59:36 2007 @@ -170,8 +170,6 @@ 9FE25D940CAB16FB005383FC /* RegisterCoalescer.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = RegisterCoalescer.cpp; sourceTree = ""; }; 9FE25D950CAB1724005383FC /* APFloat.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = APFloat.cpp; sourceTree = ""; }; 9FE25D960CAB1759005383FC /* TargetCallingConv.td */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = TargetCallingConv.td; sourceTree = ""; }; - 9FE25D970CAB17F9005383FC /* DominatorCalculation.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = DominatorCalculation.h; path = ../lib/VMCore/DominatorCalculation.h; sourceTree = SOURCE_ROOT; }; - 9FE25D980CAB17F9005383FC /* DominatorInternals.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = DominatorInternals.cpp; path = ../lib/VMCore/DominatorInternals.cpp; sourceTree = SOURCE_ROOT; }; 9FE4508B0C77A77000C4FEA4 /* ARMCodeEmitter.cpp */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.cpp.cpp; path = ARMCodeEmitter.cpp; sourceTree = ""; }; 9FE4508C0C77A77000C4FEA4 /* ARMGenAsmWriter.inc */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.pascal; path = ARMGenAsmWriter.inc; sourceTree = ""; }; 9FE4508D0C77A77000C4FEA4 /* ARMGenDAGISel.inc */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.pascal; path = ARMGenDAGISel.inc; sourceTree = ""; }; @@ -1345,8 +1343,6 @@ 9F77937D0C73C4F400551F9C /* ConstantFold.h */, DE66EC6008ABE86A00323D32 /* Constants.cpp */, 9FD3E5900CA0129D00E54D15 /* Core.cpp */, - 9FE25D970CAB17F9005383FC /* DominatorCalculation.h */, - 9FE25D980CAB17F9005383FC /* DominatorInternals.cpp */, DE66EC6108ABE86A00323D32 /* Dominators.cpp */, DE66EC6208ABE86A00323D32 /* Function.cpp */, DE66EC6308ABE86A00323D32 /* Globals.cpp */, Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42666&r1=42665&r2=42666&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Fri Oct 5 18:59:36 2007 @@ -23,12 +23,23 @@ |* with C++ due to name mangling. So in addition to C, this interface enables *| |* tools written in such languages. *| |* *| +|* When included into a C++ source file, also declares 'wrap' and 'unwrap' *| +|* helpers to perform opaque reference<-->pointer conversions. These helpers *| +|* are shorter and more tightly typed than writing the casts by hand when *| +|* authoring bindings. In assert builds, they will do runtime type checking. *| +|* *| \*===----------------------------------------------------------------------===*/ #ifndef LLVM_C_CORE_H #define LLVM_C_CORE_H #ifdef __cplusplus + +/* Need these includes to support the LLVM 'cast' template for the C++ 'wrap' + and 'unwrap' conversion functions. */ +#include "llvm/Module.h" +#include "llvm/Support/LLVMBuilder.h" + extern "C" { #endif @@ -412,6 +423,94 @@ #ifdef __cplusplus } -#endif -#endif +namespace llvm { + /* Opaque module conversions + */ + inline Module *unwrap(LLVMModuleRef M) { + return reinterpret_cast(M); + } + + inline LLVMModuleRef wrap(Module *M) { + return reinterpret_cast(M); + } + + /* Opaque type conversions + */ + inline Type *unwrap(LLVMTypeRef Ty) { + return reinterpret_cast(Ty); + } + + template + inline T *unwrap(LLVMTypeRef Ty) { + return cast(unwrap(Ty)); + } + + inline Type **unwrap(LLVMTypeRef* Tys) { + return reinterpret_cast(Tys); + } + + inline LLVMTypeRef wrap(const Type *Ty) { + return reinterpret_cast(const_cast(Ty)); + } + + inline LLVMTypeRef *wrap(const Type **Tys) { + return reinterpret_cast(const_cast(Tys)); + } + + /* Opaque value conversions + */ + inline Value *unwrap(LLVMValueRef Val) { + return reinterpret_cast(Val); + } + + template + inline T *unwrap(LLVMValueRef Val) { + return cast(unwrap(Val)); + } + + inline Value **unwrap(LLVMValueRef *Vals) { + return reinterpret_cast(Vals); + } + + template + inline T **unwrap(LLVMValueRef *Vals, unsigned Length) { + #if DEBUG + for (LLVMValueRef *I = Vals, E = Vals + Length; I != E; ++I) + cast(*I); + #endif + return reinterpret_cast(Vals); + } + + inline LLVMValueRef wrap(const Value *Val) { + return reinterpret_cast(const_cast(Val)); + } + + inline LLVMValueRef *wrap(const Value **Vals) { + return reinterpret_cast(const_cast(Vals)); + } + + /* Basic block conversions + */ + inline BasicBlock *unwrap(LLVMBasicBlockRef BBRef) { + return reinterpret_cast(BBRef); + } + + inline LLVMBasicBlockRef wrap(const BasicBlock *BB) { + return reinterpret_cast(const_cast(BB)); + } + + /* Opaque builder conversions. + */ + inline LLVMBuilder *unwrap(LLVMBuilderRef B) { + return reinterpret_cast(B); + } + + inline LLVMBuilderRef wrap(LLVMBuilder *B) { + return reinterpret_cast(B); + } +} + +#endif /* !defined(__cplusplus) */ + +#endif /* !defined(LLVM_C_CORE_H) */ Removed: llvm/trunk/include/llvm/Support/CHelpers.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/CHelpers.h?rev=42665&view=auto ============================================================================== --- llvm/trunk/include/llvm/Support/CHelpers.h (original) +++ llvm/trunk/include/llvm/Support/CHelpers.h (removed) @@ -1,104 +0,0 @@ -//===-- Support/CHelpers.h - Utilities for writing C bindings -------------===// -// -// The LLVM Compiler Infrastructure -// -// This file was developed by the LLVM research group and is distributed under -// the University of Illinois Open Source License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// These opaque reference<-->pointer conversions are shorter and more tightly -// typed than writing the casts by hand in C bindings. In assert builds, they -// will do type checking. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_SUPPORT_CHELPERS_H -#define LLVM_SUPPORT_CHELPERS_H - -#include "llvm/Module.h" -#include "llvm/Type.h" -#include "llvm/Value.h" - -typedef struct LLVMOpaqueModule *LLVMModuleRef; -typedef struct LLVMOpaqueType *LLVMTypeRef; -typedef struct LLVMOpaqueValue *LLVMValueRef; - -namespace llvm { - /// Opaque module conversions - /// - inline Module *unwrap(LLVMModuleRef M) { - return reinterpret_cast(M); - } - - inline LLVMModuleRef wrap(Module *M) { - return reinterpret_cast(M); - } - - /// Opaque type conversions - /// - inline Type *unwrap(LLVMTypeRef Ty) { - return reinterpret_cast(Ty); - } - - template - inline T *unwrap(LLVMTypeRef Ty) { - return cast(unwrap(Ty)); - } - - inline Type **unwrap(LLVMTypeRef* Tys) { - return reinterpret_cast(Tys); - } - - inline LLVMTypeRef wrap(const Type *Ty) { - return reinterpret_cast(const_cast(Ty)); - } - - inline LLVMTypeRef *wrap(const Type **Tys) { - return reinterpret_cast(const_cast(Tys)); - } - - /// Opaque value conversions - /// - inline Value *unwrap(LLVMValueRef Val) { - return reinterpret_cast(Val); - } - - template - inline T *unwrap(LLVMValueRef Val) { - return cast(unwrap(Val)); - } - - inline Value **unwrap(LLVMValueRef *Vals) { - return reinterpret_cast(Vals); - } - - template - inline T **unwrap(LLVMValueRef *Vals, unsigned Length) { - #if DEBUG - for (LLVMValueRef *I = Vals, E = Vals + Length; I != E; ++I) - cast(*I); - #endif - return reinterpret_cast(Vals); - } - - inline LLVMValueRef wrap(const Value *Val) { - return reinterpret_cast(const_cast(Val)); - } - - inline LLVMValueRef *wrap(const Value **Vals) { - return reinterpret_cast(const_cast(Vals)); - } - - /// Basic block conversions - /// - inline BasicBlock *unwrap(LLVMBasicBlockRef BBRef) { - return reinterpret_cast(BBRef); - } - - inline LLVMBasicBlockRef wrap(const BasicBlock *BB) { - return reinterpret_cast(const_cast(BB)); - } -} - -#endif Modified: llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp?rev=42666&r1=42665&r2=42666&view=diff ============================================================================== --- llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp (original) +++ llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp Fri Oct 5 18:59:36 2007 @@ -9,7 +9,6 @@ #include "llvm-c/BitWriter.h" #include "llvm/Bitcode/ReaderWriter.h" -#include "llvm/Support/CHelpers.h" #include using namespace llvm; Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42666&r1=42665&r2=42666&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Fri Oct 5 18:59:36 2007 @@ -17,25 +17,11 @@ #include "llvm/Constants.h" #include "llvm/DerivedTypes.h" #include "llvm/GlobalVariable.h" -#include "llvm/Support/CHelpers.h" -#include "llvm/Support/LLVMBuilder.h" #include "llvm/TypeSymbolTable.h" #include using namespace llvm; -namespace { - /// Opaque builder conversions. - /// - inline LLVMBuilder *unwrap(LLVMBuilderRef B) { - return reinterpret_cast(B); - } - - inline LLVMBuilderRef wrap(LLVMBuilder *B) { - return reinterpret_cast(B); - } -} - /*===-- Operations on modules ---------------------------------------------===*/ From asl at math.spbu.ru Fri Oct 5 19:04:11 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 00:04:11 -0000 Subject: [llvm-commits] [test-suite] r42667 - in /test-suite/trunk/MultiSource: Applications/SIBsim4/sim4.init.c Applications/lambda-0.1.3/lambda.cc Benchmarks/FreeBench/analyzer/analyzer.c Benchmarks/FreeBench/distray/distray.c Benchmarks/FreeBench/fourinarow/fourinarow.c Benchmarks/MallocBench/espresso/utility.c Benchmarks/Olden/bh/util.c Benchmarks/Olden/em3d/util.c Benchmarks/Olden/tsp/build.c Benchmarks/Olden/tsp/tsp.h Benchmarks/Olden/voronoi/newvor.c Benchmarks/sim/sim.c Message-ID: <200710060004.l9604CIu015082@zion.cs.uiuc.edu> Author: asl Date: Fri Oct 5 19:04:11 2007 New Revision: 42667 URL: http://llvm.org/viewvc/llvm-project?rev=42667&view=rev Log: Another bunch of mingw32 fixes here and there Modified: test-suite/trunk/MultiSource/Applications/SIBsim4/sim4.init.c test-suite/trunk/MultiSource/Applications/lambda-0.1.3/lambda.cc test-suite/trunk/MultiSource/Benchmarks/FreeBench/analyzer/analyzer.c test-suite/trunk/MultiSource/Benchmarks/FreeBench/distray/distray.c test-suite/trunk/MultiSource/Benchmarks/FreeBench/fourinarow/fourinarow.c test-suite/trunk/MultiSource/Benchmarks/MallocBench/espresso/utility.c test-suite/trunk/MultiSource/Benchmarks/Olden/bh/util.c test-suite/trunk/MultiSource/Benchmarks/Olden/em3d/util.c test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/build.c test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/tsp.h test-suite/trunk/MultiSource/Benchmarks/Olden/voronoi/newvor.c test-suite/trunk/MultiSource/Benchmarks/sim/sim.c Modified: test-suite/trunk/MultiSource/Applications/SIBsim4/sim4.init.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Applications/SIBsim4/sim4.init.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Applications/SIBsim4/sim4.init.c (original) +++ test-suite/trunk/MultiSource/Applications/SIBsim4/sim4.init.c Fri Oct 5 19:04:11 2007 @@ -119,7 +119,9 @@ if (setlocale(LC_ALL, "POSIX") == NULL) fprintf(stderr, "%s: Warning: could not set locale to POSIX\n", argv[0]); signal(SIGSEGV, bug_handler); +#ifndef __MINGW32__ signal(SIGBUS, bug_handler); +#endif /* Default options. */ options.C = DEFAULT_C; options.cutoff = DIST_CUTOFF; Modified: test-suite/trunk/MultiSource/Applications/lambda-0.1.3/lambda.cc URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Applications/lambda-0.1.3/lambda.cc?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Applications/lambda-0.1.3/lambda.cc (original) +++ test-suite/trunk/MultiSource/Applications/lambda-0.1.3/lambda.cc Fri Oct 5 19:04:11 2007 @@ -38,11 +38,7 @@ #include #endif -#ifdef WIN32 -void -#else int -#endif main(int argc, char** argv) { printf( @@ -233,8 +229,6 @@ } } tokstr.close(); -#ifndef WIN32 return 0; -#endif } Modified: test-suite/trunk/MultiSource/Benchmarks/FreeBench/analyzer/analyzer.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/FreeBench/analyzer/analyzer.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/FreeBench/analyzer/analyzer.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/FreeBench/analyzer/analyzer.c Fri Oct 5 19:04:11 2007 @@ -2,7 +2,6 @@ #include #include #include -#include #include #include "version.h" #include "types.h" Modified: test-suite/trunk/MultiSource/Benchmarks/FreeBench/distray/distray.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/FreeBench/distray/distray.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/FreeBench/distray/distray.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/FreeBench/distray/distray.c Fri Oct 5 19:04:11 2007 @@ -2,7 +2,6 @@ #include #include #include -#include #include /* Simple raytracer Modified: test-suite/trunk/MultiSource/Benchmarks/FreeBench/fourinarow/fourinarow.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/FreeBench/fourinarow/fourinarow.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/FreeBench/fourinarow/fourinarow.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/FreeBench/fourinarow/fourinarow.c Fri Oct 5 19:04:11 2007 @@ -31,7 +31,6 @@ #include #include #include -#include #include /* 64 bit type needed */ Modified: test-suite/trunk/MultiSource/Benchmarks/MallocBench/espresso/utility.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/MallocBench/espresso/utility.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/MallocBench/espresso/utility.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/MallocBench/espresso/utility.c Fri Oct 5 19:04:11 2007 @@ -36,8 +36,12 @@ #endif #endif +#ifdef __MINGW32__ +#define IBMPC +#endif + /* default */ -#if !defined(BSD) && !defined(UNIX10) && !defined(UNIX60) && !defined(UNIX100) && !defined(UNIX50) +#if !defined(BSD) && !defined(UNIX10) && !defined(UNIX60) && !defined(UNIX100) && !defined(UNIX50) && !defined(IBMPC) #define BSD #endif Modified: test-suite/trunk/MultiSource/Benchmarks/Olden/bh/util.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Olden/bh/util.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Olden/bh/util.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Olden/bh/util.c Fri Oct 5 19:04:11 2007 @@ -43,24 +43,6 @@ /* - * CPUTIME: compute CPU time in min. - */ - -#include -#include - -# define HZ 60.0 - -double cputime() -{ - struct tms buffer; - - if (times(&buffer) == -1) - error("times() call failed\n"); - return (buffer.tms_utime / (60.0 * HZ)); -} - -/* * ERROR: scream and die quickly. */ Modified: test-suite/trunk/MultiSource/Benchmarks/Olden/em3d/util.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Olden/em3d/util.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Olden/em3d/util.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Olden/em3d/util.c Fri Oct 5 19:04:11 2007 @@ -7,7 +7,13 @@ #define chatting printf #endif +#ifdef __MINGW32__ +#define srand48(x) srand(x) +#define lrand48() (rand() << 16 | rand()) +#define drand48() (1.0*rand() / RAND_MAX) +#else extern double drand48(); +#endif static int percentcheck=0,numlocal=0; Modified: test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/build.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/build.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/build.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/build.c Fri Oct 5 19:04:11 2007 @@ -17,6 +17,10 @@ #ifdef TORONTO #include +#ifdef __MINGW32__ +#define drand48() (1.0 * rand() / RAND_MAX) +#define srand48(x) srand(x) +#endif #else extern double drand48(); extern void srand48(long seedval); Modified: test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/tsp.h URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/tsp.h?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/tsp.h (original) +++ test-suite/trunk/MultiSource/Benchmarks/Olden/tsp/tsp.h Fri Oct 5 19:04:11 2007 @@ -7,7 +7,6 @@ extern int flag; -double drand48(void); int atoi(const char *); int dealwithargs(int argc, char *argv[]); Modified: test-suite/trunk/MultiSource/Benchmarks/Olden/voronoi/newvor.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/Olden/voronoi/newvor.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/Olden/voronoi/newvor.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/Olden/voronoi/newvor.c Fri Oct 5 19:04:11 2007 @@ -165,7 +165,7 @@ void delete_all_edges() { next_edge= 0; avail_edge = NYL;} -#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) +#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__MINGW32__) #define MEMALIGN_IS_NOT_AVAILABLE #endif Modified: test-suite/trunk/MultiSource/Benchmarks/sim/sim.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/sim/sim.c?rev=42667&r1=42666&r2=42667&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/sim/sim.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/sim/sim.c Fri Oct 5 19:04:11 2007 @@ -88,6 +88,10 @@ #include #define long int +#ifdef __MINGW32__ +#undef UNIX +#endif + /* ckalloc - allocate space; check for success */ #if 1 #define ckalloc(x) malloc(x) From gordonhenriksen at mac.com Fri Oct 5 19:08:50 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 00:08:50 -0000 Subject: [llvm-commits] [llvm] r42668 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp Message-ID: <200710060008.l9608o1H015293@zion.cs.uiuc.edu> Author: gordon Date: Fri Oct 5 19:08:49 2007 New Revision: 42668 URL: http://llvm.org/viewvc/llvm-project?rev=42668&view=rev Log: Wrapping Value::dump. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42668&r1=42667&r2=42668&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Fri Oct 5 19:08:49 2007 @@ -181,6 +181,7 @@ external type_of : llvalue -> lltype = "llvm_type_of" external value_name : llvalue -> string = "llvm_value_name" external set_value_name : string -> llvalue -> unit = "llvm_set_value_name" +external dump_value : llvalue -> unit = "llvm_dump_value" (*--... Operations on constants of (mostly) any type .......................--*) external is_constant : llvalue -> bool = "llvm_is_constant" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42668&r1=42667&r2=42668&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Fri Oct 5 19:08:49 2007 @@ -164,6 +164,7 @@ external type_of : llvalue -> lltype = "llvm_type_of" external value_name : llvalue -> string = "llvm_value_name" external set_value_name : string -> llvalue -> unit = "llvm_set_value_name" +external dump_value : llvalue -> unit = "llvm_dump_value" (*--... Operations on constants of (mostly) any type .......................--*) external is_constant : llvalue -> bool = "llvm_is_constant" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42668&r1=42667&r2=42668&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Fri Oct 5 19:08:49 2007 @@ -219,6 +219,12 @@ return Val_unit; } +/* llvalue -> unit */ +CAMLprim value llvm_dump_value(LLVMValueRef Val) { + LLVMDumpValue(Val); + return Val_unit; +} + /*--... Operations on constants of (mostly) any type .......................--*/ /* llvalue -> bool */ Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42668&r1=42667&r2=42668&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Fri Oct 5 19:08:49 2007 @@ -225,6 +225,7 @@ LLVMTypeRef LLVMTypeOf(LLVMValueRef Val); const char *LLVMGetValueName(LLVMValueRef Val); void LLVMSetValueName(LLVMValueRef Val, const char *Name); +void LLVMDumpValue(LLVMValueRef Val); /* Operations on constants of any type */ LLVMValueRef LLVMGetNull(LLVMTypeRef Ty); /* all zeroes */ Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42668&r1=42667&r2=42668&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Fri Oct 5 19:08:49 2007 @@ -194,6 +194,10 @@ unwrap(Val)->setName(Name); } +void LLVMDumpValue(LLVMValueRef Val) { + unwrap(Val)->dump(); +} + /*--.. Operations on constants of any type .................................--*/ LLVMValueRef LLVMGetNull(LLVMTypeRef Ty) { From neil at daikokuya.co.uk Fri Oct 5 19:24:48 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sat, 06 Oct 2007 00:24:48 -0000 Subject: [llvm-commits] [llvm] r42669 - in /llvm/trunk: include/llvm/ADT/APInt.h lib/Support/APFloat.cpp lib/Support/APInt.cpp Message-ID: <200710060024.l960OmPT015966@zion.cs.uiuc.edu> Author: neil Date: Fri Oct 5 19:24:48 2007 New Revision: 42669 URL: http://llvm.org/viewvc/llvm-project?rev=42669&view=rev Log: Generalize tcFullMultiply so that the operands can be of differing part widths. Also, return the number of parts actually required to hold the result's value. Remove an over-cautious condition from rounding of float->hex conversion. Modified: llvm/trunk/include/llvm/ADT/APInt.h llvm/trunk/lib/Support/APFloat.cpp llvm/trunk/lib/Support/APInt.cpp Modified: llvm/trunk/include/llvm/ADT/APInt.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APInt.h?rev=42669&r1=42668&r2=42669&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APInt.h (original) +++ llvm/trunk/include/llvm/ADT/APInt.h Fri Oct 5 19:24:48 2007 @@ -1114,10 +1114,12 @@ static int tcMultiply(integerPart *, const integerPart *, const integerPart *, unsigned); - /// DST = LHS * RHS, where DST has twice the width as the operands. - /// No overflow occurs. DST must be disjoint from both operands. - static void tcFullMultiply(integerPart *, const integerPart *, - const integerPart *, unsigned); + /// DST = LHS * RHS, where DST has width the sum of the widths of + /// the operands. No overflow occurs. DST must be disjoint from + /// both operands. Returns the number of parts required to hold the + /// result. + static unsigned int tcFullMultiply(integerPart *, const integerPart *, + const integerPart *, unsigned, unsigned); /// If RHS is zero LHS and REMAINDER are left unchanged, return one. /// Otherwise set LHS to LHS / RHS with the fractional part Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42669&r1=42668&r2=42669&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Fri Oct 5 19:24:48 2007 @@ -519,7 +519,7 @@ partsCount = partCount(); APInt::tcFullMultiply(fullSignificand, lhsSignificand, - rhs.significandParts(), partsCount); + rhs.significandParts(), partsCount, partsCount); lost_fraction = lfExactlyZero; omsb = APInt::tcMSB(fullSignificand, newPartsCount) + 1; @@ -1795,7 +1795,7 @@ /* hexDigits of zero means use the required number for the precision. Otherwise, see if we are truncating. If we are, - found out if we need to round away from zero. */ + find out if we need to round away from zero. */ if (hexDigits) { if (hexDigits < outputDigits) { /* We are dropping non-zero bits, so need to check how to round. @@ -1845,7 +1845,8 @@ do { q--; *q = hexDigitChars[hexDigitValue (*q) + 1]; - } while (*q == '0' && q > p); + } while (*q == '0'); + assert (q >= p); } else { /* Add trailing zeroes. */ memset (dst, '0', outputDigits); Modified: llvm/trunk/lib/Support/APInt.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APInt.cpp?rev=42669&r1=42668&r2=42669&view=diff ============================================================================== --- llvm/trunk/lib/Support/APInt.cpp (original) +++ llvm/trunk/lib/Support/APInt.cpp Fri Oct 5 19:24:48 2007 @@ -2363,25 +2363,32 @@ return overflow; } -/* DST = LHS * RHS, where DST has twice the width as the operands. No - overflow occurs. DST must be disjoint from both operands. */ -void +/* DST = LHS * RHS, where DST has width the sum of the widths of the + operands. No overflow occurs. DST must be disjoint from both + operands. Returns the number of parts required to hold the + result. */ +unsigned int APInt::tcFullMultiply(integerPart *dst, const integerPart *lhs, - const integerPart *rhs, unsigned int parts) + const integerPart *rhs, unsigned int lhsParts, + unsigned int rhsParts) { - unsigned int i; - int overflow; + /* Put the narrower number on the LHS for less loops below. */ + if (lhsParts > rhsParts) { + return tcFullMultiply (dst, rhs, lhs, rhsParts, lhsParts); + } else { + unsigned int n; - assert(dst != lhs && dst != rhs); + assert(dst != lhs && dst != rhs); - overflow = 0; - tcSet(dst, 0, parts); + tcSet(dst, 0, rhsParts); - for(i = 0; i < parts; i++) - overflow |= tcMultiplyPart(&dst[i], lhs, rhs[i], 0, parts, - parts + 1, true); + for(n = 0; n < lhsParts; n++) + tcMultiplyPart(&dst[n], rhs, lhs[n], 0, rhsParts, rhsParts + 1, true); - assert(!overflow); + n = lhsParts + rhsParts; + + return n - (dst[n - 1] == 0); + } } /* If RHS is zero LHS and REMAINDER are left unchanged, return one. From neil at daikokuya.co.uk Fri Oct 5 19:43:46 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sat, 06 Oct 2007 00:43:46 -0000 Subject: [llvm-commits] [llvm] r42670 - /llvm/trunk/lib/Support/APInt.cpp Message-ID: <200710060043.l960hk8V016710@zion.cs.uiuc.edu> Author: neil Date: Fri Oct 5 19:43:45 2007 New Revision: 42670 URL: http://llvm.org/viewvc/llvm-project?rev=42670&view=rev Log: Fix and clarify some comments. Modified: llvm/trunk/lib/Support/APInt.cpp Modified: llvm/trunk/lib/Support/APInt.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APInt.cpp?rev=42670&r1=42669&r2=42670&view=diff ============================================================================== --- llvm/trunk/lib/Support/APInt.cpp (original) +++ llvm/trunk/lib/Support/APInt.cpp Fri Oct 5 19:43:45 2007 @@ -2029,22 +2029,22 @@ return ~(integerPart) 0 >> (integerPartWidth - bits); } - /* Returns the value of the lower nibble of PART. */ + /* Returns the value of the lower half of PART. */ inline integerPart lowHalf(integerPart part) { return part & lowBitMask(integerPartWidth / 2); } - /* Returns the value of the upper nibble of PART. */ + /* Returns the value of the upper half of PART. */ inline integerPart highHalf(integerPart part) { return part >> (integerPartWidth / 2); } - /* Returns the bit number of the most significant bit of a part. If - the input number has no bits set -1U is returned. */ + /* Returns the bit number of the most significant set bit of a part. + If the input number has no bits set -1U is returned. */ unsigned int partMSB(integerPart value) { @@ -2068,8 +2068,8 @@ return msb; } - /* Returns the bit number of the least significant bit of a part. - If the input number has no bits set -1U is returned. */ + /* Returns the bit number of the least significant set bit of a + part. If the input number has no bits set -1U is returned. */ unsigned int partLSB(integerPart value) { @@ -2144,8 +2144,8 @@ parts[bit / integerPartWidth] |= (integerPart) 1 << (bit % integerPartWidth); } -/* Returns the bit number of the least significant bit of a number. - If the input number has no bits set -1U is returned. */ +/* Returns the bit number of the least significant set bit of a + number. If the input number has no bits set -1U is returned. */ unsigned int APInt::tcLSB(const integerPart *parts, unsigned int n) { @@ -2162,8 +2162,8 @@ return -1U; } -/* Returns the bit number of the most significant bit of a number. If - the input number has no bits set -1U is returned. */ +/* Returns the bit number of the most significant set bit of a number. + If the input number has no bits set -1U is returned. */ unsigned int APInt::tcMSB(const integerPart *parts, unsigned int n) { @@ -2240,8 +2240,8 @@ tcIncrement(dst, parts); } -/* DST += SRC * MULTIPLIER + PART if add is true - DST = SRC * MULTIPLIER + PART if add is false +/* DST += SRC * MULTIPLIER + CARRY if add is true + DST = SRC * MULTIPLIER + CARRY if add is false Requires 0 <= DSTPARTS <= SRCPARTS + 1. If DST overlaps SRC they must start at the same point, i.e. DST == SRC. From dalej at apple.com Fri Oct 5 20:22:39 2007 From: dalej at apple.com (Dale Johannesen) Date: Sat, 06 Oct 2007 01:22:39 -0000 Subject: [llvm-commits] [llvm] r42671 - /llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Message-ID: <200710060122.l961Md37018011@zion.cs.uiuc.edu> Author: johannes Date: Fri Oct 5 20:22:39 2007 New Revision: 42671 URL: http://llvm.org/viewvc/llvm-project?rev=42671&view=rev Log: Make test work on non-x86 hosts. Modified: llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Modified: llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll?rev=42671&r1=42670&r2=42671&view=diff ============================================================================== --- llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll (original) +++ llvm/trunk/test/CodeGen/X86/2007-10-05-3AddrConvert.ll Fri Oct 5 20:22:39 2007 @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc | grep lea +; RUN: llvm-as < %s | llc -march=x86 | grep lea %struct.anon = type { [3 x double], double, %struct.node*, [64 x %struct.bnode*], [64 x %struct.bnode*] } %struct.bnode = type { i16, double, [3 x double], i32, i32, [3 x double], [3 x double], [3 x double], double, %struct.bnode*, %struct.bnode* } From dalej at apple.com Fri Oct 5 20:24:12 2007 From: dalej at apple.com (Dale Johannesen) Date: Sat, 06 Oct 2007 01:24:12 -0000 Subject: [llvm-commits] [llvm] r42672 - in /llvm/trunk/lib: CodeGen/SelectionDAG/LegalizeDAG.cpp Target/PowerPC/PPCISelLowering.cpp Message-ID: <200710060124.l961OCUw018066@zion.cs.uiuc.edu> Author: johannes Date: Fri Oct 5 20:24:11 2007 New Revision: 42672 URL: http://llvm.org/viewvc/llvm-project?rev=42672&view=rev Log: Next powerpc long double bits. Comparisons work, although not well, and shortening FP converts. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=42672&r1=42671&r2=42672&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Fri Oct 5 20:24:11 2007 @@ -3314,32 +3314,42 @@ MVT::ValueType newVT = Op.getValueType(); MVT::ValueType oldVT = Op.getOperand(0).getValueType(); if (TLI.getConvertAction(oldVT, newVT) == TargetLowering::Expand) { - // The only way we can lower this is to turn it into a STORE, - // LOAD pair, targetting a temporary location (a stack slot). - - // NOTE: there is a choice here between constantly creating new stack - // slots and always reusing the same one. We currently always create - // new ones, as reuse may inhibit scheduling. - MVT::ValueType slotVT = - (Node->getOpcode() == ISD::FP_EXTEND) ? oldVT : newVT; - const Type *Ty = MVT::getTypeForValueType(slotVT); - uint64_t TySize = TLI.getTargetData()->getTypeSize(Ty); - unsigned Align = TLI.getTargetData()->getPrefTypeAlignment(Ty); - MachineFunction &MF = DAG.getMachineFunction(); - int SSFI = - MF.getFrameInfo()->CreateStackObject(TySize, Align); - SDOperand StackSlot = DAG.getFrameIndex(SSFI, TLI.getPointerTy()); - if (Node->getOpcode() == ISD::FP_EXTEND) { - Result = DAG.getStore(DAG.getEntryNode(), Node->getOperand(0), - StackSlot, NULL, 0); - Result = DAG.getExtLoad(ISD::EXTLOAD, newVT, - Result, StackSlot, NULL, 0, oldVT); + if (Node->getOpcode() == ISD::FP_ROUND && oldVT == MVT::ppcf128) { + SDOperand Lo, Hi; + ExpandOp(Node->getOperand(0), Lo, Hi); + if (newVT == MVT::f64) + Result = Hi; + else + Result = DAG.getNode(ISD::FP_ROUND, newVT, Hi); + break; } else { - Result = DAG.getTruncStore(DAG.getEntryNode(), Node->getOperand(0), - StackSlot, NULL, 0, newVT); - Result = DAG.getLoad(newVT, Result, StackSlot, NULL, 0, newVT); + // The only other way we can lower this is to turn it into a STORE, + // LOAD pair, targetting a temporary location (a stack slot). + + // NOTE: there is a choice here between constantly creating new stack + // slots and always reusing the same one. We currently always create + // new ones, as reuse may inhibit scheduling. + MVT::ValueType slotVT = + (Node->getOpcode() == ISD::FP_EXTEND) ? oldVT : newVT; + const Type *Ty = MVT::getTypeForValueType(slotVT); + uint64_t TySize = TLI.getTargetData()->getTypeSize(Ty); + unsigned Align = TLI.getTargetData()->getPrefTypeAlignment(Ty); + MachineFunction &MF = DAG.getMachineFunction(); + int SSFI = + MF.getFrameInfo()->CreateStackObject(TySize, Align); + SDOperand StackSlot = DAG.getFrameIndex(SSFI, TLI.getPointerTy()); + if (Node->getOpcode() == ISD::FP_EXTEND) { + Result = DAG.getStore(DAG.getEntryNode(), Node->getOperand(0), + StackSlot, NULL, 0); + Result = DAG.getExtLoad(ISD::EXTLOAD, newVT, + Result, StackSlot, NULL, 0, oldVT); + } else { + Result = DAG.getTruncStore(DAG.getEntryNode(), Node->getOperand(0), + StackSlot, NULL, 0, newVT); + Result = DAG.getLoad(newVT, Result, StackSlot, NULL, 0, newVT); + } + break; } - break; } } // FALL THROUGH @@ -3995,7 +4005,7 @@ void SelectionDAGLegalize::LegalizeSetCCOperands(SDOperand &LHS, SDOperand &RHS, SDOperand &CC) { - SDOperand Tmp1, Tmp2, Result; + SDOperand Tmp1, Tmp2, Tmp3, Result; switch (getTypeAction(LHS.getValueType())) { case Legal: @@ -4126,8 +4136,27 @@ SDOperand LHSLo, LHSHi, RHSLo, RHSHi; ExpandOp(LHS, LHSLo, LHSHi); - ExpandOp(RHS, RHSLo, RHSHi); - switch (cast(CC)->get()) { + ExpandOp(RHS, RHSLo, RHSHi); + ISD::CondCode CCCode = cast(CC)->get(); + + if (VT==MVT::ppcf128) { + // FIXME: This generated code sucks. We want to generate + // FCMP crN, hi1, hi2 + // BNE crN, L: + // FCMP crN, lo1, lo2 + // The following can be improved, but not that much. + Tmp1 = DAG.getSetCC(TLI.getSetCCResultTy(), LHSHi, RHSHi, ISD::SETEQ); + Tmp2 = DAG.getSetCC(TLI.getSetCCResultTy(), LHSLo, RHSLo, CCCode); + Tmp3 = DAG.getNode(ISD::AND, Tmp1.getValueType(), Tmp1, Tmp2); + Tmp1 = DAG.getSetCC(TLI.getSetCCResultTy(), LHSHi, RHSHi, ISD::SETNE); + Tmp2 = DAG.getSetCC(TLI.getSetCCResultTy(), LHSHi, RHSHi, CCCode); + Tmp1 = DAG.getNode(ISD::AND, Tmp1.getValueType(), Tmp1, Tmp2); + Tmp1 = DAG.getNode(ISD::OR, Tmp1.getValueType(), Tmp1, Tmp3); + Tmp2 = SDOperand(); + break; + } + + switch (CCCode) { case ISD::SETEQ: case ISD::SETNE: if (RHSLo == RHSHi) @@ -4159,7 +4188,6 @@ // FIXME: This generated code sucks. ISD::CondCode LowCC; - ISD::CondCode CCCode = cast(CC)->get(); switch (CCCode) { default: assert(0 && "Unknown integer setcc!"); case ISD::SETLT: Modified: llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp?rev=42672&r1=42671&r2=42672&view=diff ============================================================================== --- llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp (original) +++ llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp Fri Oct 5 20:24:11 2007 @@ -73,6 +73,10 @@ setOperationAction(ISD::ConstantFP, MVT::f64, Expand); setOperationAction(ISD::ConstantFP, MVT::f32, Expand); + // Shortening conversions involving ppcf128 get expanded (2 regs -> 1 reg) + setConvertAction(MVT::ppcf128, MVT::f64, Expand); + setConvertAction(MVT::ppcf128, MVT::f32, Expand); + // PowerPC has no intrinsics for these particular operations setOperationAction(ISD::MEMMOVE, MVT::Other, Expand); setOperationAction(ISD::MEMSET, MVT::Other, Expand); From isanbard at gmail.com Fri Oct 5 20:45:22 2007 From: isanbard at gmail.com (Bill Wendling) Date: Fri, 5 Oct 2007 18:45:22 -0700 Subject: [llvm-commits] [cfe-commits] r42652 - in /cfe/trunk/www: get_involved.html index.html In-Reply-To: References: <200710052144.l95Li4W0008520@zion.cs.uiuc.edu> Message-ID: I would prefer that they not be there at all. :-) -bw On Oct 5, 2007, at 3:36 PM, Gordon Henriksen wrote: > Hi Kevin, > > Since you're using non-ASCII characters, could you please declare > the character set in the section? I believe this is correct > for this file: > > > > On Oct 5, 2007, at 17:44, Kevin wrote: > >> Author: kevin >> Date: Fri Oct 5 16:44:03 2007 >> New Revision: 42652 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=42652&view=rev >> Log: >> A message on the main page about bulding clang. >> >> Modified: >> cfe/trunk/www/get_involved.html >> cfe/trunk/www/index.html >> >> Modified: cfe/trunk/www/get_involved.html >> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/www/ >> get_involved.html?rev=42652&r1=42651&r2=42652&view=diff >> >> ===================================================================== >> ========= >> --- cfe/trunk/www/get_involved.html (original) >> +++ cfe/trunk/www/get_involved.html Fri Oct 5 16:44:03 2007 >> @@ -32,7 +32,7 @@ >>
    3. cfe-commits - This list is for patch submission/ >> discussion. >>
    4. cfe-dev - This list is for everything else clang related. >> >> -

      Building clang / working with the code

      >> +

      Building clang / working with the code > a>

      >> If you would like to check out and build the project, the current >> scheme is:
      >>
        >>
      1. Check out llvm >> >> Modified: cfe/trunk/www/index.html >> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/www/index.html? >> rev=42652&r1=42651&r2=42652&view=diff >> >> ===================================================================== >> ========= >> --- cfe/trunk/www/index.html (original) >> +++ cfe/trunk/www/index.html Fri Oct 5 16:44:03 2007 >> @@ -41,6 +41,8 @@ >>
      2. BSD License >> >> Of course this is only a rough outline of the goals and >> features of Clang. To get a true sense of what the new LLVM front- >> end is all about, as well as why you might want to considering >> using it, see the Features section. >> +

        Try Clang

        >> + If you are a compiler developer and want to try out Clang, then >> build instructions are available on the > href="get_involved.html#build">Get Involved page. >>

        Get Involved

        >> If you are interested in working on Clang, then check out the >> Get Involved section. >> >> >> >> _______________________________________________ >> cfe-commits mailing list >> cfe-commits at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits > > > > ? Gordon > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20071005/136ef69f/attachment.html From evan.cheng at apple.com Fri Oct 5 21:46:29 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Sat, 06 Oct 2007 02:46:29 -0000 Subject: [llvm-commits] [llvm] r42677 - in /llvm/trunk/lib: CodeGen/SelectionDAG/DAGCombiner.cpp Target/X86/X86InstrSSE.td Message-ID: <200710060246.l962kT9R020741@zion.cs.uiuc.edu> Author: evancheng Date: Fri Oct 5 21:46:29 2007 New Revision: 42677 URL: http://llvm.org/viewvc/llvm-project?rev=42677&view=rev Log: Added DAG xforms. e.g. (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr) Remove x86 specific patterns. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp llvm/trunk/lib/Target/X86/X86InstrSSE.td Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=42677&r1=42676&r2=42677&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Fri Oct 5 21:46:29 2007 @@ -263,6 +263,7 @@ SDOperand visitLOAD(SDNode *N); SDOperand visitSTORE(SDNode *N); SDOperand visitINSERT_VECTOR_ELT(SDNode *N); + SDOperand visitEXTRACT_VECTOR_ELT(SDNode *N); SDOperand visitBUILD_VECTOR(SDNode *N); SDOperand visitCONCAT_VECTORS(SDNode *N); SDOperand visitVECTOR_SHUFFLE(SDNode *N); @@ -682,6 +683,7 @@ case ISD::LOAD: return visitLOAD(N); case ISD::STORE: return visitSTORE(N); case ISD::INSERT_VECTOR_ELT: return visitINSERT_VECTOR_ELT(N); + case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N); case ISD::BUILD_VECTOR: return visitBUILD_VECTOR(N); case ISD::CONCAT_VECTORS: return visitCONCAT_VECTORS(N); case ISD::VECTOR_SHUFFLE: return visitVECTOR_SHUFFLE(N); @@ -2907,9 +2909,8 @@ return DAG.getNode(ISD::BIT_CONVERT, VT, N0.getOperand(0)); // fold (conv (load x)) -> (load (conv*)x) - // If the resultant load doesn't need a higher alignment than the original! - if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() && - ISD::isUNINDEXEDLoad(N0.Val) && + // If the resultant load doesn't need a higher alignment than the original! + if (ISD::isNormalLoad(N0.Val) && N0.hasOneUse() && TLI.isOperationLegal(ISD::LOAD, VT)) { LoadSDNode *LN0 = cast(N0); unsigned Align = TLI.getTargetMachine().getTargetData()-> @@ -3901,6 +3902,54 @@ return SDOperand(); } +SDOperand DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) { + SDOperand InVec = N->getOperand(0); + SDOperand EltNo = N->getOperand(1); + + // (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) + // (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr) + if (isa(EltNo)) { + unsigned Elt = cast(EltNo)->getValue(); + bool NewLoad = false; + if (Elt == 0) { + MVT::ValueType VT = InVec.getValueType(); + MVT::ValueType EVT = MVT::getVectorElementType(VT); + MVT::ValueType LVT = EVT; + unsigned NumElts = MVT::getVectorNumElements(VT); + if (InVec.getOpcode() == ISD::BIT_CONVERT) { + MVT::ValueType BCVT = InVec.getOperand(0).getValueType(); + if (NumElts != MVT::getVectorNumElements(BCVT)) + return SDOperand(); + InVec = InVec.getOperand(0); + EVT = MVT::getVectorElementType(BCVT); + NewLoad = true; + } + if (InVec.getOpcode() == ISD::SCALAR_TO_VECTOR && + InVec.getOperand(0).getValueType() == EVT && + ISD::isNormalLoad(InVec.getOperand(0).Val) && + InVec.getOperand(0).hasOneUse()) { + LoadSDNode *LN0 = cast(InVec.getOperand(0)); + unsigned Align = LN0->getAlignment(); + if (NewLoad) { + // Check the resultant load doesn't need a higher alignment than the + // original load. + unsigned NewAlign = TLI.getTargetMachine().getTargetData()-> + getABITypeAlignment(MVT::getTypeForValueType(LVT)); + if (!TLI.isOperationLegal(ISD::LOAD, LVT) || NewAlign > Align) + return SDOperand(); + Align = NewAlign; + } + + return DAG.getLoad(LVT, LN0->getChain(), LN0->getBasePtr(), + LN0->getSrcValue(), LN0->getSrcValueOffset(), + LN0->isVolatile(), Align); + } + } + } + return SDOperand(); +} + + SDOperand DAGCombiner::visitBUILD_VECTOR(SDNode *N) { unsigned NumInScalars = N->getNumOperands(); MVT::ValueType VT = N->getValueType(0); Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=42677&r1=42676&r2=42677&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original) +++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Fri Oct 5 21:46:29 2007 @@ -2942,11 +2942,3 @@ (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; def : Pat<(store (v16i8 VR128:$src), addr:$dst), (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; - -// (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr) -def : Pat<(vector_extract - (bc_v4i32 (v4f32 (scalar_to_vector (loadf32 addr:$src)))), (iPTR 0)), - (MOV32rm addr:$src)>, Requires<[HasSSE2]>; -def : Pat<(vector_extract - (bc_v2i64 (v2f64 (scalar_to_vector (loadf64 addr:$src)))), (iPTR 0)), - (MOV64rm addr:$src)>, Requires<[HasSSE2, In64BitMode]>; From clattner at apple.com Sat Oct 6 00:11:52 2007 From: clattner at apple.com (Chris Lattner) Date: Fri, 5 Oct 2007 22:11:52 -0700 Subject: [llvm-commits] [llvm] r42677 - in /llvm/trunk/lib: CodeGen/SelectionDAG/DAGCombiner.cpp Target/X86/X86InstrSSE.td In-Reply-To: <200710060246.l962kT9R020741@zion.cs.uiuc.edu> References: <200710060246.l962kT9R020741@zion.cs.uiuc.edu> Message-ID: <44ADE7F2-D285-425A-B78A-41A874D3F698@apple.com> Evan, it looks like you missed a commit. Would a similar approach be suitable to implement this X86/README-SSE.txt entry?: + +These functions should produce the same code: + +#include + +typedef long long __m128i __attribute__ ((__vector_size__ (16))); + +int foo(__m128i* val) { + return __builtin_ia32_vec_ext_v4si(*val, 1); +} +int bar(__m128i* val) { + union vs { + __m128i *_v; + int* _s; + } v = {val}; + return v._s[1]; +} + +We currently produce (with -m64): + +_foo: + pshufd $1, (%rdi), %xmm0 + movd %xmm0, %eax + ret +_bar: + movl 4(%rdi), %eax + ret + -Chris On Oct 5, 2007, at 7:46 PM, Evan Cheng wrote: > Author: evancheng > Date: Fri Oct 5 21:46:29 2007 > New Revision: 42677 > > URL: http://llvm.org/viewvc/llvm-project?rev=42677&view=rev > Log: > Added DAG xforms. e.g. > (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) > (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load > $addr) > Remove x86 specific patterns. > > Modified: > llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp > llvm/trunk/lib/Target/X86/X86InstrSSE.td > > Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/ > SelectionDAG/DAGCombiner.cpp?rev=42677&r1=42676&r2=42677&view=diff > > ====================================================================== > ======== > --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original) > +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Fri Oct 5 > 21:46:29 2007 > @@ -263,6 +263,7 @@ > SDOperand visitLOAD(SDNode *N); > SDOperand visitSTORE(SDNode *N); > SDOperand visitINSERT_VECTOR_ELT(SDNode *N); > + SDOperand visitEXTRACT_VECTOR_ELT(SDNode *N); > SDOperand visitBUILD_VECTOR(SDNode *N); > SDOperand visitCONCAT_VECTORS(SDNode *N); > SDOperand visitVECTOR_SHUFFLE(SDNode *N); > @@ -682,6 +683,7 @@ > case ISD::LOAD: return visitLOAD(N); > case ISD::STORE: return visitSTORE(N); > case ISD::INSERT_VECTOR_ELT: return visitINSERT_VECTOR_ELT(N); > + case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N); > case ISD::BUILD_VECTOR: return visitBUILD_VECTOR(N); > case ISD::CONCAT_VECTORS: return visitCONCAT_VECTORS(N); > case ISD::VECTOR_SHUFFLE: return visitVECTOR_SHUFFLE(N); > @@ -2907,9 +2909,8 @@ > return DAG.getNode(ISD::BIT_CONVERT, VT, N0.getOperand(0)); > > // fold (conv (load x)) -> (load (conv*)x) > - // If the resultant load doesn't need a higher alignment than > the original! > - if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() && > - ISD::isUNINDEXEDLoad(N0.Val) && > + // If the resultant load doesn't need a higher alignment than > the original! > + if (ISD::isNormalLoad(N0.Val) && N0.hasOneUse() && > TLI.isOperationLegal(ISD::LOAD, VT)) { > LoadSDNode *LN0 = cast(N0); > unsigned Align = TLI.getTargetMachine().getTargetData()-> > @@ -3901,6 +3902,54 @@ > return SDOperand(); > } > > +SDOperand DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) { > + SDOperand InVec = N->getOperand(0); > + SDOperand EltNo = N->getOperand(1); > + > + // (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) > + // (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 > load $addr) > + if (isa(EltNo)) { > + unsigned Elt = cast(EltNo)->getValue(); > + bool NewLoad = false; > + if (Elt == 0) { > + MVT::ValueType VT = InVec.getValueType(); > + MVT::ValueType EVT = MVT::getVectorElementType(VT); > + MVT::ValueType LVT = EVT; > + unsigned NumElts = MVT::getVectorNumElements(VT); > + if (InVec.getOpcode() == ISD::BIT_CONVERT) { > + MVT::ValueType BCVT = InVec.getOperand(0).getValueType(); > + if (NumElts != MVT::getVectorNumElements(BCVT)) > + return SDOperand(); > + InVec = InVec.getOperand(0); > + EVT = MVT::getVectorElementType(BCVT); > + NewLoad = true; > + } > + if (InVec.getOpcode() == ISD::SCALAR_TO_VECTOR && > + InVec.getOperand(0).getValueType() == EVT && > + ISD::isNormalLoad(InVec.getOperand(0).Val) && > + InVec.getOperand(0).hasOneUse()) { > + LoadSDNode *LN0 = cast(InVec.getOperand(0)); > + unsigned Align = LN0->getAlignment(); > + if (NewLoad) { > + // Check the resultant load doesn't need a higher > alignment than the > + // original load. > + unsigned NewAlign = TLI.getTargetMachine().getTargetData > ()-> > + getABITypeAlignment(MVT::getTypeForValueType(LVT)); > + if (!TLI.isOperationLegal(ISD::LOAD, LVT) || NewAlign > > Align) > + return SDOperand(); > + Align = NewAlign; > + } > + > + return DAG.getLoad(LVT, LN0->getChain(), LN0->getBasePtr(), > + LN0->getSrcValue(), LN0- > >getSrcValueOffset(), > + LN0->isVolatile(), Align); > + } > + } > + } > + return SDOperand(); > +} > + > + > SDOperand DAGCombiner::visitBUILD_VECTOR(SDNode *N) { > unsigned NumInScalars = N->getNumOperands(); > MVT::ValueType VT = N->getValueType(0); > > Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td > URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ > X86InstrSSE.td?rev=42677&r1=42676&r2=42677&view=diff > > ====================================================================== > ======== > --- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original) > +++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Fri Oct 5 21:46:29 2007 > @@ -2942,11 +2942,3 @@ > (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; > def : Pat<(store (v16i8 VR128:$src), addr:$dst), > (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; > - > -// (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 > load $addr) > -def : Pat<(vector_extract > - (bc_v4i32 (v4f32 (scalar_to_vector (loadf32 addr: > $src)))), (iPTR 0)), > - (MOV32rm addr:$src)>, Requires<[HasSSE2]>; > -def : Pat<(vector_extract > - (bc_v2i64 (v2f64 (scalar_to_vector (loadf64 addr: > $src)))), (iPTR 0)), > - (MOV64rm addr:$src)>, Requires<[HasSSE2, In64BitMode]>; > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From sabre at nondot.org Sat Oct 6 01:08:17 2007 From: sabre at nondot.org (Chris Lattner) Date: Sat, 06 Oct 2007 06:08:17 -0000 Subject: [llvm-commits] [llvm] r42686 - /llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Message-ID: <200710060608.l9668HKd026356@zion.cs.uiuc.edu> Author: lattner Date: Sat Oct 6 01:08:17 2007 New Revision: 42686 URL: http://llvm.org/viewvc/llvm-project?rev=42686&view=rev Log: revert evan's patch until the header is committed Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=42686&r1=42685&r2=42686&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Sat Oct 6 01:08:17 2007 @@ -263,7 +263,6 @@ SDOperand visitLOAD(SDNode *N); SDOperand visitSTORE(SDNode *N); SDOperand visitINSERT_VECTOR_ELT(SDNode *N); - SDOperand visitEXTRACT_VECTOR_ELT(SDNode *N); SDOperand visitBUILD_VECTOR(SDNode *N); SDOperand visitCONCAT_VECTORS(SDNode *N); SDOperand visitVECTOR_SHUFFLE(SDNode *N); @@ -683,7 +682,6 @@ case ISD::LOAD: return visitLOAD(N); case ISD::STORE: return visitSTORE(N); case ISD::INSERT_VECTOR_ELT: return visitINSERT_VECTOR_ELT(N); - case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N); case ISD::BUILD_VECTOR: return visitBUILD_VECTOR(N); case ISD::CONCAT_VECTORS: return visitCONCAT_VECTORS(N); case ISD::VECTOR_SHUFFLE: return visitVECTOR_SHUFFLE(N); @@ -2909,8 +2907,9 @@ return DAG.getNode(ISD::BIT_CONVERT, VT, N0.getOperand(0)); // fold (conv (load x)) -> (load (conv*)x) - // If the resultant load doesn't need a higher alignment than the original! - if (ISD::isNormalLoad(N0.Val) && N0.hasOneUse() && + // If the resultant load doesn't need a higher alignment than the original! + if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() && + ISD::isUNINDEXEDLoad(N0.Val) && TLI.isOperationLegal(ISD::LOAD, VT)) { LoadSDNode *LN0 = cast(N0); unsigned Align = TLI.getTargetMachine().getTargetData()-> @@ -3902,54 +3901,6 @@ return SDOperand(); } -SDOperand DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) { - SDOperand InVec = N->getOperand(0); - SDOperand EltNo = N->getOperand(1); - - // (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) - // (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr) - if (isa(EltNo)) { - unsigned Elt = cast(EltNo)->getValue(); - bool NewLoad = false; - if (Elt == 0) { - MVT::ValueType VT = InVec.getValueType(); - MVT::ValueType EVT = MVT::getVectorElementType(VT); - MVT::ValueType LVT = EVT; - unsigned NumElts = MVT::getVectorNumElements(VT); - if (InVec.getOpcode() == ISD::BIT_CONVERT) { - MVT::ValueType BCVT = InVec.getOperand(0).getValueType(); - if (NumElts != MVT::getVectorNumElements(BCVT)) - return SDOperand(); - InVec = InVec.getOperand(0); - EVT = MVT::getVectorElementType(BCVT); - NewLoad = true; - } - if (InVec.getOpcode() == ISD::SCALAR_TO_VECTOR && - InVec.getOperand(0).getValueType() == EVT && - ISD::isNormalLoad(InVec.getOperand(0).Val) && - InVec.getOperand(0).hasOneUse()) { - LoadSDNode *LN0 = cast(InVec.getOperand(0)); - unsigned Align = LN0->getAlignment(); - if (NewLoad) { - // Check the resultant load doesn't need a higher alignment than the - // original load. - unsigned NewAlign = TLI.getTargetMachine().getTargetData()-> - getABITypeAlignment(MVT::getTypeForValueType(LVT)); - if (!TLI.isOperationLegal(ISD::LOAD, LVT) || NewAlign > Align) - return SDOperand(); - Align = NewAlign; - } - - return DAG.getLoad(LVT, LN0->getChain(), LN0->getBasePtr(), - LN0->getSrcValue(), LN0->getSrcValueOffset(), - LN0->isVolatile(), Align); - } - } - } - return SDOperand(); -} - - SDOperand DAGCombiner::visitBUILD_VECTOR(SDNode *N) { unsigned NumInScalars = N->getNumOperands(); MVT::ValueType VT = N->getValueType(0); From sabre at nondot.org Sat Oct 6 01:13:42 2007 From: sabre at nondot.org (Chris Lattner) Date: Sat, 06 Oct 2007 06:13:42 -0000 Subject: [llvm-commits] [llvm] r42687 - /llvm/trunk/lib/Support/APFloat.cpp Message-ID: <200710060613.l966Dgxc026507@zion.cs.uiuc.edu> Author: lattner Date: Sat Oct 6 01:13:42 2007 New Revision: 42687 URL: http://llvm.org/viewvc/llvm-project?rev=42687&view=rev Log: silence warnings in no-assert build. Modified: llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42687&r1=42686&r2=42687&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sat Oct 6 01:13:42 2007 @@ -1913,19 +1913,18 @@ } else if (category==fcInfinity) { myexponent = 0x7fff; mysignificand = 0x8000000000000000ULL; - } else if (category==fcNaN) { + } else { + assert(category == fcNaN && "Unknown category"); myexponent = 0x7fff; mysignificand = significandParts()[0]; - } else - assert(0); + } uint64_t words[2]; words[0] = (((uint64_t)sign & 1) << 63) | ((myexponent & 0x7fff) << 48) | ((mysignificand >>16) & 0xffffffffffffLL); words[1] = mysignificand & 0xffff; - APInt api(80, 2, words); - return api; + return APInt(80, 2, words); } APInt @@ -1947,16 +1946,15 @@ } else if (category==fcInfinity) { myexponent = 0x7ff; mysignificand = 0; - } else if (category==fcNaN) { + } else { + assert(category == fcNaN && "Unknown category!"); myexponent = 0x7ff; mysignificand = *significandParts(); - } else - assert(0); + } - APInt api(64, (((((uint64_t)sign & 1) << 63) | - ((myexponent & 0x7ff) << 52) | - (mysignificand & 0xfffffffffffffLL)))); - return api; + return APInt(64, (((((uint64_t)sign & 1) << 63) | + ((myexponent & 0x7ff) << 52) | + (mysignificand & 0xfffffffffffffLL)))); } APInt @@ -1978,15 +1976,14 @@ } else if (category==fcInfinity) { myexponent = 0xff; mysignificand = 0; - } else if (category==fcNaN) { + } else { + assert(category == fcNaN && "Unknown category!"); myexponent = 0xff; mysignificand = *significandParts(); - } else - assert(0); + } - APInt api(32, (((sign&1) << 31) | ((myexponent&0xff) << 23) | - (mysignificand & 0x7fffff))); - return api; + return APInt(32, (((sign&1) << 31) | ((myexponent&0xff) << 23) | + (mysignificand & 0x7fffff))); } APInt @@ -1994,13 +1991,13 @@ { if (semantics == (const llvm::fltSemantics* const)&IEEEsingle) return convertFloatAPFloatToAPInt(); - else if (semantics == (const llvm::fltSemantics* const)&IEEEdouble) + + if (semantics == (const llvm::fltSemantics* const)&IEEEdouble) return convertDoubleAPFloatToAPInt(); - else if (semantics == (const llvm::fltSemantics* const)&x87DoubleExtended) - return convertF80LongDoubleAPFloatToAPInt(); - assert(0); - abort(); + assert(semantics == (const llvm::fltSemantics* const)&x87DoubleExtended && + "unknown format!"); + return convertF80LongDoubleAPFloatToAPInt(); } float From neil at daikokuya.co.uk Sat Oct 6 02:29:25 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sat, 06 Oct 2007 07:29:25 -0000 Subject: [llvm-commits] [llvm] r42690 - /llvm/trunk/lib/Support/APFloat.cpp Message-ID: <200710060729.l967TPPs028237@zion.cs.uiuc.edu> Author: neil Date: Sat Oct 6 02:29:25 2007 New Revision: 42690 URL: http://llvm.org/viewvc/llvm-project?rev=42690&view=rev Log: Cleaner, more general exponent output. Modified: llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42690&r1=42689&r2=42690&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sat Oct 6 02:29:25 2007 @@ -252,32 +252,33 @@ return result; } - /* Write out a decimal exponent. */ + /* Write out an unsigned decimal integer. */ static char * - writeDecimalExponent (char *dst, int exponent) + writeUnsignedDecimal (char *dst, unsigned int n) { - assert (exponent >= -65536 && exponent <= 65535); + char buff[40], *p; - if (exponent < 0) { - *dst++ = '-'; - exponent = -exponent; - } - - if (exponent == 0) { - *dst++ = '0'; - } else { - char buff[12], *p; + p = buff; + do + *p++ = '0' + n % 10; + while (n /= 10); + + do + *dst++ = *--p; + while (p != buff); - p = buff; - while (exponent) { - *p++ = '0' + exponent % 10; - exponent /= 10; - } + return dst; + } - do - *dst++ = *--p; - while (p != buff); - } + /* Write out a signed decimal integer. */ + static char * + writeSignedDecimal (char *dst, int value) + { + if (value < 0) { + *dst++ = '-'; + dst = writeUnsignedDecimal(dst, -(unsigned) value); + } else + dst = writeUnsignedDecimal(dst, value); return dst; } @@ -1865,7 +1866,7 @@ /* Finally output the exponent. */ *dst++ = upperCase ? 'P': 'p'; - return writeDecimalExponent (dst, exponent); + return writeSignedDecimal (dst, exponent); } // For good performance it is desirable for different APFloats From evan.cheng at apple.com Sat Oct 6 03:09:29 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Sat, 06 Oct 2007 08:09:29 -0000 Subject: [llvm-commits] [llvm] r42691 - /llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Message-ID: <200710060809.l9689TQs003966@zion.cs.uiuc.edu> Author: evancheng Date: Sat Oct 6 03:09:29 2007 New Revision: 42691 URL: http://llvm.org/viewvc/llvm-project?rev=42691&view=rev Log: Oops. Forgot this. Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Modified: llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h?rev=42691&r1=42690&r2=42691&view=diff ============================================================================== --- llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h (original) +++ llvm/trunk/include/llvm/CodeGen/SelectionDAGNodes.h Sat Oct 6 03:09:29 2007 @@ -1651,6 +1651,16 @@ }; namespace ISD { + /// isNormalLoad - Returns true if the specified node is a non-extending + /// and unindexed load. + inline bool isNormalLoad(const SDNode *N) { + if (N->getOpcode() != ISD::LOAD) + return false; + const LoadSDNode *Ld = cast(N); + return Ld->getExtensionType() == ISD::NON_EXTLOAD && + Ld->getAddressingMode() == ISD::UNINDEXED; + } + /// isNON_EXTLoad - Returns true if the specified node is a non-extending /// load. inline bool isNON_EXTLoad(const SDNode *N) { From evan.cheng at apple.com Sat Oct 6 03:19:55 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Sat, 06 Oct 2007 08:19:55 -0000 Subject: [llvm-commits] [llvm] r42692 - /llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Message-ID: <200710060819.l968Jtmd004240@zion.cs.uiuc.edu> Author: evancheng Date: Sat Oct 6 03:19:55 2007 New Revision: 42692 URL: http://llvm.org/viewvc/llvm-project?rev=42692&view=rev Log: Reapply 42677. Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=42692&r1=42691&r2=42692&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Sat Oct 6 03:19:55 2007 @@ -263,6 +263,7 @@ SDOperand visitLOAD(SDNode *N); SDOperand visitSTORE(SDNode *N); SDOperand visitINSERT_VECTOR_ELT(SDNode *N); + SDOperand visitEXTRACT_VECTOR_ELT(SDNode *N); SDOperand visitBUILD_VECTOR(SDNode *N); SDOperand visitCONCAT_VECTORS(SDNode *N); SDOperand visitVECTOR_SHUFFLE(SDNode *N); @@ -682,6 +683,7 @@ case ISD::LOAD: return visitLOAD(N); case ISD::STORE: return visitSTORE(N); case ISD::INSERT_VECTOR_ELT: return visitINSERT_VECTOR_ELT(N); + case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N); case ISD::BUILD_VECTOR: return visitBUILD_VECTOR(N); case ISD::CONCAT_VECTORS: return visitCONCAT_VECTORS(N); case ISD::VECTOR_SHUFFLE: return visitVECTOR_SHUFFLE(N); @@ -2907,9 +2909,8 @@ return DAG.getNode(ISD::BIT_CONVERT, VT, N0.getOperand(0)); // fold (conv (load x)) -> (load (conv*)x) - // If the resultant load doesn't need a higher alignment than the original! - if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() && - ISD::isUNINDEXEDLoad(N0.Val) && + // If the resultant load doesn't need a higher alignment than the original! + if (ISD::isNormalLoad(N0.Val) && N0.hasOneUse() && TLI.isOperationLegal(ISD::LOAD, VT)) { LoadSDNode *LN0 = cast(N0); unsigned Align = TLI.getTargetMachine().getTargetData()-> @@ -3901,6 +3902,54 @@ return SDOperand(); } +SDOperand DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) { + SDOperand InVec = N->getOperand(0); + SDOperand EltNo = N->getOperand(1); + + // (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) + // (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr) + if (isa(EltNo)) { + unsigned Elt = cast(EltNo)->getValue(); + bool NewLoad = false; + if (Elt == 0) { + MVT::ValueType VT = InVec.getValueType(); + MVT::ValueType EVT = MVT::getVectorElementType(VT); + MVT::ValueType LVT = EVT; + unsigned NumElts = MVT::getVectorNumElements(VT); + if (InVec.getOpcode() == ISD::BIT_CONVERT) { + MVT::ValueType BCVT = InVec.getOperand(0).getValueType(); + if (NumElts != MVT::getVectorNumElements(BCVT)) + return SDOperand(); + InVec = InVec.getOperand(0); + EVT = MVT::getVectorElementType(BCVT); + NewLoad = true; + } + if (InVec.getOpcode() == ISD::SCALAR_TO_VECTOR && + InVec.getOperand(0).getValueType() == EVT && + ISD::isNormalLoad(InVec.getOperand(0).Val) && + InVec.getOperand(0).hasOneUse()) { + LoadSDNode *LN0 = cast(InVec.getOperand(0)); + unsigned Align = LN0->getAlignment(); + if (NewLoad) { + // Check the resultant load doesn't need a higher alignment than the + // original load. + unsigned NewAlign = TLI.getTargetMachine().getTargetData()-> + getABITypeAlignment(MVT::getTypeForValueType(LVT)); + if (!TLI.isOperationLegal(ISD::LOAD, LVT) || NewAlign > Align) + return SDOperand(); + Align = NewAlign; + } + + return DAG.getLoad(LVT, LN0->getChain(), LN0->getBasePtr(), + LN0->getSrcValue(), LN0->getSrcValueOffset(), + LN0->isVolatile(), Align); + } + } + } + return SDOperand(); +} + + SDOperand DAGCombiner::visitBUILD_VECTOR(SDNode *N) { unsigned NumInScalars = N->getNumOperands(); MVT::ValueType VT = N->getValueType(0); From asl at math.spbu.ru Sat Oct 6 05:37:40 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 10:37:40 -0000 Subject: [llvm-commits] [test-suite] r42693 - in /test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast: config.h toast.c Message-ID: <200710061037.l96AbeXV008896@zion.cs.uiuc.edu> Author: asl Date: Sat Oct 6 05:37:39 2007 New Revision: 42693 URL: http://llvm.org/viewvc/llvm-project?rev=42693&view=rev Log: Unbreak mediabench on mingw32 Modified: test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/config.h test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/toast.c Modified: test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/config.h URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/config.h?rev=42693&r1=42692&r2=42693&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/config.h (original) +++ test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/config.h Sat Oct 6 05:37:39 2007 @@ -18,8 +18,13 @@ /*efine HAS_ERRNO_DECL 1 /* errno.h declares errno */ #define HAS_FSTAT 1 /* fstat syscall */ +#ifndef __MINGW32__ #define HAS_FCHMOD 1 /* fchmod syscall */ #define HAS_FCHOWN 1 /* fchown syscall */ +#else +#undef HAS_FCHMOD +#undef HAS_FCHOWN +#endif #define HAS_STRING_H 1 /* /usr/include/string.h */ /*efine HAS_STRINGS_H 1 /* /usr/include/strings.h */ Modified: test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/toast.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/toast.c?rev=42693&r1=42692&r2=42693&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/toast.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/mediabench/gsm/toast/toast.c Sat Oct 6 05:37:39 2007 @@ -338,8 +338,10 @@ if (!instat.st_nlink) return; /* couldn't stat in */ #ifdef HAS_FCHOWN (void)fchown(fileno(out), instat.st_uid, instat.st_gid); -#else +#else +#ifndef __MINGW32__ (void)chown(outname, instat.st_uid, instat.st_gid); +#endif #endif /* HAS_FCHOWN */ } From asl at math.spbu.ru Sat Oct 6 05:55:21 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 10:55:21 -0000 Subject: [llvm-commits] [test-suite] r42694 - in /test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm: config.h toast.c Message-ID: <200710061055.l96AtMYO009463@zion.cs.uiuc.edu> Author: asl Date: Sat Oct 6 05:55:20 2007 New Revision: 42694 URL: http://llvm.org/viewvc/llvm-project?rev=42694&view=rev Log: Unbreak toast again for mingw32. Funny, but we have two toast's in in llvm-test. Maybe we should drop one? Modified: test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/config.h test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/toast.c Modified: test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/config.h URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/config.h?rev=42694&r1=42693&r2=42694&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/config.h (original) +++ test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/config.h Sat Oct 6 05:55:20 2007 @@ -18,8 +18,13 @@ /*efine HAS_ERRNO_DECL 1 /* errno.h declares errno */ #define HAS_FSTAT 1 /* fstat syscall */ +#ifndef __MINGW32__ #define HAS_FCHMOD 1 /* fchmod syscall */ #define HAS_FCHOWN 1 /* fchown syscall */ +#else +#undef HAS_FCHMOD +#undef HAS_FCHOWN +#endif #define HAS_STRING_H 1 /* /usr/include/string.h */ /*efine HAS_STRINGS_H 1 /* /usr/include/strings.h */ Modified: test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/toast.c URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/toast.c?rev=42694&r1=42693&r2=42694&view=diff ============================================================================== --- test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/toast.c (original) +++ test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm/toast.c Sat Oct 6 05:55:20 2007 @@ -339,7 +339,9 @@ #ifdef HAS_FCHOWN (void)fchown(fileno(out), instat.st_uid, instat.st_gid); #else +#ifndef __MINGW32__ (void)chown(outname, instat.st_uid, instat.st_gid); +#endif #endif /* HAS_FCHOWN */ } From asl at math.spbu.ru Sat Oct 6 08:07:11 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 13:07:11 -0000 Subject: [llvm-commits] [llvm] r42695 - /llvm/trunk/include/llvm/Pass.h Message-ID: <200710061307.l96D7BrQ014630@zion.cs.uiuc.edu> Author: asl Date: Sat Oct 6 08:07:11 2007 New Revision: 42695 URL: http://llvm.org/viewvc/llvm-project?rev=42695&view=rev Log: Include DataTypes.h to provide intptr_t type. This fixes OpenBSD compilation of this file. Inspired by Laurence Tratt's patch Modified: llvm/trunk/include/llvm/Pass.h Modified: llvm/trunk/include/llvm/Pass.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Pass.h?rev=42695&r1=42694&r2=42695&view=diff ============================================================================== --- llvm/trunk/include/llvm/Pass.h (original) +++ llvm/trunk/include/llvm/Pass.h Sat Oct 6 08:07:11 2007 @@ -29,6 +29,7 @@ #ifndef LLVM_PASS_H #define LLVM_PASS_H +#include "llvm/Support/DataTypes.h" #include "llvm/Support/Streams.h" #include #include From gordonhenriksen at mac.com Sat Oct 6 09:29:37 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 14:29:37 -0000 Subject: [llvm-commits] [llvm] r42696 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h include/llvm/Constants.h lib/VMCore/Constants.cpp lib/VMCore/Core.cpp test/Bindings/Ocaml/vmcore.ml Message-ID: <200710061429.l96ETbt7018104@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 09:29:36 2007 New Revision: 42696 URL: http://llvm.org/viewvc/llvm-project?rev=42696&view=rev Log: Adding C and Ocaml bindings for ConstantExpr. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/include/llvm/Constants.h llvm/trunk/lib/VMCore/Constants.cpp llvm/trunk/lib/VMCore/Core.cpp llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sat Oct 6 09:29:36 2007 @@ -209,6 +209,51 @@ external make_vector_constant : llvalue array -> llvalue = "llvm_make_vector_constant" +(*--... Constant expressions ...............................................--*) +external sizeof : lltype -> llvalue = "LLVMSizeOf" +external const_neg : llvalue -> llvalue = "LLVMConstNeg" +external const_not : llvalue -> llvalue = "LLVMConstNot" +external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" +external const_sub : llvalue -> llvalue -> llvalue = "LLVMConstSub" +external const_mul : llvalue -> llvalue -> llvalue = "LLVMConstMul" +external const_udiv : llvalue -> llvalue -> llvalue = "LLVMConstUDiv" +external const_sdiv : llvalue -> llvalue -> llvalue = "LLVMConstSDiv" +external const_fdiv : llvalue -> llvalue -> llvalue = "LLVMConstFDiv" +external const_urem : llvalue -> llvalue -> llvalue = "LLVMConstURem" +external const_srem : llvalue -> llvalue -> llvalue = "LLVMConstSRem" +external const_frem : llvalue -> llvalue -> llvalue = "LLVMConstFRem" +external const_and : llvalue -> llvalue -> llvalue = "LLVMConstAnd" +external const_or : llvalue -> llvalue -> llvalue = "LLVMConstOr" +external const_xor : llvalue -> llvalue -> llvalue = "LLVMConstXor" +external const_icmp : int_predicate -> llvalue -> llvalue -> llvalue + = "llvm_const_icmp" +external const_fcmp : real_predicate -> llvalue -> llvalue -> llvalue + = "llvm_const_fcmp" +external const_shl : llvalue -> llvalue -> llvalue = "LLVMConstShl" +external const_lshr : llvalue -> llvalue -> llvalue = "LLVMConstLShr" +external const_ashr : llvalue -> llvalue -> llvalue = "LLVMConstAShr" +external const_gep : llvalue -> llvalue array -> llvalue = "llvm_const_gep" +external const_trunc : llvalue -> lltype -> llvalue = "LLVMConstTrunc" +external const_sext : llvalue -> lltype -> llvalue = "LLVMConstSExt" +external const_zext : llvalue -> lltype -> llvalue = "LLVMConstZExt" +external const_fptrunc : llvalue -> lltype -> llvalue = "LLVMConstFPTrunc" +external const_fpext : llvalue -> lltype -> llvalue = "LLVMConstFPExt" +external const_uitofp : llvalue -> lltype -> llvalue = "LLVMConstUIToFP" +external const_sitofp : llvalue -> lltype -> llvalue = "LLVMConstSIToFP" +external const_fptoui : llvalue -> lltype -> llvalue = "LLVMConstFPToUI" +external const_fptosi : llvalue -> lltype -> llvalue = "LLVMConstFPToSI" +external const_ptrtoint : llvalue -> lltype -> llvalue = "LLVMConstPtrToInt" +external const_inttoptr : llvalue -> lltype -> llvalue = "LLVMConstIntToPtr" +external const_bitcast : llvalue -> lltype -> llvalue = "LLVMConstBitCast" +external const_select : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstSelect" +external const_extractelement : llvalue -> llvalue -> llvalue + = "LLVMConstExtractElement" +external const_insertelement : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstInsertElement" +external const_shufflevector : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstShuffleVector" + (*--... Operations on global variables, functions, and aliases (globals) ...--*) external is_declaration : llvalue -> bool = "llvm_is_declaration" external linkage : llvalue -> linkage = "llvm_linkage" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sat Oct 6 09:29:36 2007 @@ -192,6 +192,51 @@ external make_vector_constant : llvalue array -> llvalue = "llvm_make_vector_constant" +(*--... Constant expressions ...............................................--*) +external sizeof : lltype -> llvalue = "LLVMSizeOf" +external const_neg : llvalue -> llvalue = "LLVMConstNeg" +external const_not : llvalue -> llvalue = "LLVMConstNot" +external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" +external const_sub : llvalue -> llvalue -> llvalue = "LLVMConstSub" +external const_mul : llvalue -> llvalue -> llvalue = "LLVMConstMul" +external const_udiv : llvalue -> llvalue -> llvalue = "LLVMConstUDiv" +external const_sdiv : llvalue -> llvalue -> llvalue = "LLVMConstSDiv" +external const_fdiv : llvalue -> llvalue -> llvalue = "LLVMConstFDiv" +external const_urem : llvalue -> llvalue -> llvalue = "LLVMConstURem" +external const_srem : llvalue -> llvalue -> llvalue = "LLVMConstSRem" +external const_frem : llvalue -> llvalue -> llvalue = "LLVMConstFRem" +external const_and : llvalue -> llvalue -> llvalue = "LLVMConstAnd" +external const_or : llvalue -> llvalue -> llvalue = "LLVMConstOr" +external const_xor : llvalue -> llvalue -> llvalue = "LLVMConstXor" +external const_icmp : int_predicate -> llvalue -> llvalue -> llvalue + = "llvm_const_icmp" +external const_fcmp : real_predicate -> llvalue -> llvalue -> llvalue + = "llvm_const_fcmp" +external const_shl : llvalue -> llvalue -> llvalue = "LLVMConstShl" +external const_lshr : llvalue -> llvalue -> llvalue = "LLVMConstLShr" +external const_ashr : llvalue -> llvalue -> llvalue = "LLVMConstAShr" +external const_gep : llvalue -> llvalue array -> llvalue = "llvm_const_gep" +external const_trunc : llvalue -> lltype -> llvalue = "LLVMConstTrunc" +external const_sext : llvalue -> lltype -> llvalue = "LLVMConstSExt" +external const_zext : llvalue -> lltype -> llvalue = "LLVMConstZExt" +external const_fptrunc : llvalue -> lltype -> llvalue = "LLVMConstFPTrunc" +external const_fpext : llvalue -> lltype -> llvalue = "LLVMConstFPExt" +external const_uitofp : llvalue -> lltype -> llvalue = "LLVMConstUIToFP" +external const_sitofp : llvalue -> lltype -> llvalue = "LLVMConstSIToFP" +external const_fptoui : llvalue -> lltype -> llvalue = "LLVMConstFPToUI" +external const_fptosi : llvalue -> lltype -> llvalue = "LLVMConstFPToSI" +external const_ptrtoint : llvalue -> lltype -> llvalue = "LLVMConstPtrToInt" +external const_inttoptr : llvalue -> lltype -> llvalue = "LLVMConstIntToPtr" +external const_bitcast : llvalue -> lltype -> llvalue = "LLVMConstBitCast" +external const_select : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstSelect" +external const_extractelement : llvalue -> llvalue -> llvalue + = "LLVMConstExtractElement" +external const_insertelement : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstInsertElement" +external const_shufflevector : llvalue -> llvalue -> llvalue -> llvalue + = "LLVMConstShuffleVector" + (*--... Operations on global variables, functions, and aliases (globals) ...--*) external is_declaration : llvalue -> bool = "llvm_is_declaration" external linkage : llvalue -> linkage = "llvm_linkage" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 09:29:36 2007 @@ -139,10 +139,10 @@ /*--... Operations on struct types .........................................--*/ /* lltype array -> bool -> lltype */ -CAMLprim value llvm_make_struct_type(value ElementTypes, value Packed) { - return (value) LLVMCreateStructType((LLVMTypeRef *) ElementTypes, - Wosize_val(ElementTypes), - Bool_val(Packed)); +CAMLprim LLVMTypeRef llvm_make_struct_type(value ElementTypes, value Packed) { + return LLVMCreateStructType((LLVMTypeRef *) ElementTypes, + Wosize_val(ElementTypes), + Bool_val(Packed)); } /* lltype -> lltype array */ @@ -160,8 +160,8 @@ /*--... Operations on array, pointer, and vector types .....................--*/ /* lltype -> int -> lltype */ -CAMLprim value llvm_make_array_type(LLVMTypeRef ElementTy, value Count) { - return (value) LLVMCreateArrayType(ElementTy, Int_val(Count)); +CAMLprim LLVMTypeRef llvm_make_array_type(LLVMTypeRef ElementTy, value Count) { + return LLVMCreateArrayType(ElementTy, Int_val(Count)); } /* lltype -> lltype */ @@ -291,9 +291,31 @@ } /* llvalue array -> llvalue */ -CAMLprim value llvm_make_vector_constant(value ElementVals) { - return (value) LLVMGetVectorConstant((LLVMValueRef*) Op_val(ElementVals), - Wosize_val(ElementVals)); +CAMLprim LLVMValueRef llvm_make_vector_constant(value ElementVals) { + return LLVMGetVectorConstant((LLVMValueRef*) Op_val(ElementVals), + Wosize_val(ElementVals)); +} + +/*--... Constant expressions ...............................................--*/ + +/* int_predicate -> llvalue -> llvalue -> llvalue */ +CAMLprim LLVMValueRef llvm_const_icmp(value Pred, + LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return LLVMConstICmp(Int_val(Pred) + LLVMIntEQ, LHSConstant, RHSConstant); +} + +/* real_predicate -> llvalue -> llvalue -> llvalue */ +CAMLprim LLVMValueRef llvm_const_fcmp(value Pred, + LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return LLVMConstFCmp(Int_val(Pred), LHSConstant, RHSConstant); +} + +/* llvalue -> llvalue array -> llvalue */ +CAMLprim LLVMValueRef llvm_const_gep(LLVMValueRef ConstantVal, value Indices) { + return LLVMConstGEP(ConstantVal, (LLVMValueRef*) Op_val(Indices), + Wosize_val(Indices)); } /*--... Operations on global variables, functions, and aliases (globals) ...--*/ Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sat Oct 6 09:29:36 2007 @@ -250,6 +250,55 @@ LLVMValueRef LLVMGetVectorConstant(LLVMValueRef *ScalarConstantVals, unsigned Size); +/* Constant expressions */ +LLVMValueRef LLVMSizeOf(LLVMTypeRef Ty); +LLVMValueRef LLVMConstNeg(LLVMValueRef ConstantVal); +LLVMValueRef LLVMConstNot(LLVMValueRef ConstantVal); +LLVMValueRef LLVMConstAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstUDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstSDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstFDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstURem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstSRem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstFRem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstAnd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstOr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstXor(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstICmp(LLVMIntPredicate Predicate, + LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstFCmp(LLVMRealPredicate Predicate, + LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstShl(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstLShr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstAShr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstGEP(LLVMValueRef ConstantVal, + LLVMValueRef *ConstantIndices, unsigned NumIndices); +LLVMValueRef LLVMConstTrunc(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstSExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstZExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstFPTrunc(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstFPExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstUIToFP(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstSIToFP(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstFPToUI(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstFPToSI(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstPtrToInt(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstIntToPtr(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstBitCast(LLVMValueRef ConstantVal, LLVMTypeRef ToType); +LLVMValueRef LLVMConstSelect(LLVMValueRef ConstantCondition, + LLVMValueRef ConstantIfTrue, + LLVMValueRef ConstantIfFalse); +LLVMValueRef LLVMConstExtractElement(LLVMValueRef VectorConstant, + LLVMValueRef IndexConstant); +LLVMValueRef LLVMConstInsertElement(LLVMValueRef VectorConstant, + LLVMValueRef ElementValueConstant, + LLVMValueRef IndexConstant); +LLVMValueRef LLVMConstShuffleVector(LLVMValueRef VectorAConstant, + LLVMValueRef VectorBConstant, + LLVMValueRef MaskConstant); + /* Operations on global variables, functions, and aliases (globals) */ int LLVMIsDeclaration(LLVMValueRef Global); LLVMLinkage LLVMGetLinkage(LLVMValueRef Global); Modified: llvm/trunk/include/llvm/Constants.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Constants.h?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/include/llvm/Constants.h (original) +++ llvm/trunk/include/llvm/Constants.h Sat Oct 6 09:29:36 2007 @@ -599,7 +599,7 @@ } /// getSizeOf constant expr - computes the size of a type in a target - /// independent way (Note: the return type is a ULong). + /// independent way (Note: the return type is an i64). /// static Constant *getSizeOf(const Type *Ty); @@ -628,8 +628,8 @@ static Constant *getAnd(Constant *C1, Constant *C2); static Constant *getOr(Constant *C1, Constant *C2); static Constant *getXor(Constant *C1, Constant *C2); - static Constant* getICmp(unsigned short pred, Constant* LHS, Constant* RHS); - static Constant* getFCmp(unsigned short pred, Constant* LHS, Constant* RHS); + static Constant *getICmp(unsigned short pred, Constant *LHS, Constant *RHS); + static Constant *getFCmp(unsigned short pred, Constant *LHS, Constant *RHS); static Constant *getShl(Constant *C1, Constant *C2); static Constant *getLShr(Constant *C1, Constant *C2); static Constant *getAShr(Constant *C1, Constant *C2); Modified: llvm/trunk/lib/VMCore/Constants.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Constants.cpp?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Constants.cpp (original) +++ llvm/trunk/lib/VMCore/Constants.cpp Sat Oct 6 09:29:36 2007 @@ -533,7 +533,7 @@ C); } Constant *ConstantExpr::getNot(Constant *C) { - assert(isa(C) && "Cannot NOT a nonintegral type!"); + assert(isa(C->getType()) && "Cannot NOT a nonintegral value!"); return get(Instruction::Xor, C, ConstantInt::getAllOnesValue(C->getType())); } @@ -1675,7 +1675,7 @@ } Constant *ConstantExpr::getSizeOf(const Type *Ty) { - // sizeof is implemented as: (ulong) gep (Ty*)null, 1 + // sizeof is implemented as: (i64) gep (Ty*)null, 1 Constant *GEPIdx = ConstantInt::get(Type::Int32Ty, 1); Constant *GEP = getGetElementPtr(getNullValue(PointerType::get(Ty)), &GEPIdx, 1); Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sat Oct 6 09:29:36 2007 @@ -266,6 +266,207 @@ Size)); } +/*--.. Constant expressions ................................................--*/ + +LLVMValueRef LLVMSizeOf(LLVMTypeRef Ty) { + return wrap(ConstantExpr::getSizeOf(unwrap(Ty))); +} + +LLVMValueRef LLVMConstNeg(LLVMValueRef ConstantVal) { + return wrap(ConstantExpr::getNeg(unwrap(ConstantVal))); +} + +LLVMValueRef LLVMConstNot(LLVMValueRef ConstantVal) { + return wrap(ConstantExpr::getNot(unwrap(ConstantVal))); +} + +LLVMValueRef LLVMConstAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getAdd(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getSub(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getMul(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstUDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getUDiv(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstSDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getSDiv(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstFDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getFDiv(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstURem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getURem(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstSRem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getSRem(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstFRem(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getFRem(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstAnd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getAnd(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstOr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getOr(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstXor(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getXor(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstICmp(LLVMIntPredicate Predicate, + LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getICmp(Predicate, + unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstFCmp(LLVMRealPredicate Predicate, + LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getFCmp(Predicate, + unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstShl(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getShl(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstLShr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getLShr(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstAShr(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getAShr(unwrap(LHSConstant), + unwrap(RHSConstant))); +} + +LLVMValueRef LLVMConstGEP(LLVMValueRef ConstantVal, + LLVMValueRef *ConstantIndices, unsigned NumIndices) { + return wrap(ConstantExpr::getGetElementPtr(unwrap(ConstantVal), + unwrap(ConstantIndices, + NumIndices), + NumIndices)); +} + +LLVMValueRef LLVMConstTrunc(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getTrunc(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstSExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getSExt(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstZExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getZExt(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstFPTrunc(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getFPTrunc(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstFPExt(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getFPExtend(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstUIToFP(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getUIToFP(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstSIToFP(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getSIToFP(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstFPToUI(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getFPToUI(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstFPToSI(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getFPToSI(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstPtrToInt(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getPtrToInt(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstIntToPtr(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getIntToPtr(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstBitCast(LLVMValueRef ConstantVal, LLVMTypeRef ToType) { + return wrap(ConstantExpr::getBitCast(unwrap(ConstantVal), + unwrap(ToType))); +} + +LLVMValueRef LLVMConstSelect(LLVMValueRef ConstantCondition, + LLVMValueRef ConstantIfTrue, + LLVMValueRef ConstantIfFalse) { + return wrap(ConstantExpr::getSelect(unwrap(ConstantCondition), + unwrap(ConstantIfTrue), + unwrap(ConstantIfFalse))); +} + +LLVMValueRef LLVMConstExtractElement(LLVMValueRef VectorConstant, + LLVMValueRef IndexConstant) { + return wrap(ConstantExpr::getExtractElement(unwrap(VectorConstant), + unwrap(IndexConstant))); +} + +LLVMValueRef LLVMConstInsertElement(LLVMValueRef VectorConstant, + LLVMValueRef ElementValueConstant, + LLVMValueRef IndexConstant) { + return wrap(ConstantExpr::getInsertElement(unwrap(VectorConstant), + unwrap(ElementValueConstant), + unwrap(IndexConstant))); +} + +LLVMValueRef LLVMConstShuffleVector(LLVMValueRef VectorAConstant, + LLVMValueRef VectorBConstant, + LLVMValueRef MaskConstant) { + return wrap(ConstantExpr::getShuffleVector(unwrap(VectorAConstant), + unwrap(VectorBConstant), + unwrap(MaskConstant))); +} + /*--.. Operations on global variables, functions, and aliases (globals) ....--*/ int LLVMIsDeclaration(LLVMValueRef Global) { Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42696&r1=42695&r2=42696&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sat Oct 6 09:29:36 2007 @@ -227,7 +227,108 @@ let c = make_undef i1_type in ignore (define_global "Const12" c m); insist (i1_type = type_of c); - insist (is_undef c) + insist (is_undef c); + + group "constant arithmetic"; + (* RUN: grep {ConstNeg.*sub} < %t.ll + * RUN: grep {ConstNot.*xor} < %t.ll + * RUN: grep {ConstAdd.*add} < %t.ll + * RUN: grep {ConstSub.*sub} < %t.ll + * RUN: grep {ConstMul.*mul} < %t.ll + * RUN: grep {ConstUDiv.*udiv} < %t.ll + * RUN: grep {ConstSDiv.*sdiv} < %t.ll + * RUN: grep {ConstFDiv.*fdiv} < %t.ll + * RUN: grep {ConstURem.*urem} < %t.ll + * RUN: grep {ConstSRem.*srem} < %t.ll + * RUN: grep {ConstFRem.*frem} < %t.ll + * RUN: grep {ConstAnd.*and} < %t.ll + * RUN: grep {ConstOr.*or} < %t.ll + * RUN: grep {ConstXor.*xor} < %t.ll + * RUN: grep {ConstICmp.*icmp} < %t.ll + * RUN: grep {ConstFCmp.*fcmp} < %t.ll + *) + let void_ptr = make_pointer_type i8_type in + let five = make_int_constant i64_type 5 false in + let ffive = const_uitofp five double_type in + let foldbomb_gv = define_global "FoldBomb" (make_null i8_type) m in + let foldbomb = const_ptrtoint foldbomb_gv i64_type in + let ffoldbomb = const_uitofp foldbomb double_type in + ignore (define_global "ConstNeg" (const_neg foldbomb) m); + ignore (define_global "ConstNot" (const_not foldbomb) m); + ignore (define_global "ConstAdd" (const_add foldbomb five) m); + ignore (define_global "ConstSub" (const_sub foldbomb five) m); + ignore (define_global "ConstMul" (const_mul foldbomb five) m); + ignore (define_global "ConstUDiv" (const_udiv foldbomb five) m); + ignore (define_global "ConstSDiv" (const_sdiv foldbomb five) m); + ignore (define_global "ConstFDiv" (const_fdiv ffoldbomb ffive) m); + ignore (define_global "ConstURem" (const_urem foldbomb five) m); + ignore (define_global "ConstSRem" (const_srem foldbomb five) m); + ignore (define_global "ConstFRem" (const_frem ffoldbomb ffive) m); + ignore (define_global "ConstAnd" (const_and foldbomb five) m); + ignore (define_global "ConstOr" (const_or foldbomb five) m); + ignore (define_global "ConstXor" (const_xor foldbomb five) m); + ignore (define_global "ConstICmp" (const_icmp Icmp_sle foldbomb five) m); + ignore (define_global "ConstFCmp" (const_fcmp Fcmp_ole ffoldbomb ffive) m); + + group "constant casts"; + (* RUN: grep {ConstTrunc.*trunc} < %t.ll + * RUN: grep {ConstSExt.*sext} < %t.ll + * RUN: grep {ConstZExt.*zext} < %t.ll + * RUN: grep {ConstFPTrunc.*fptrunc} < %t.ll + * RUN: grep {ConstFPExt.*fpext} < %t.ll + * RUN: grep {ConstUIToFP.*uitofp} < %t.ll + * RUN: grep {ConstSIToFP.*sitofp} < %t.ll + * RUN: grep {ConstFPToUI.*fptoui} < %t.ll + * RUN: grep {ConstFPToSI.*fptosi} < %t.ll + * RUN: grep {ConstPtrToInt.*ptrtoint} < %t.ll + * RUN: grep {ConstIntToPtr.*inttoptr} < %t.ll + * RUN: grep {ConstBitCast.*bitcast} < %t.ll + *) + let i128_type = make_integer_type 128 in + ignore (define_global "ConstTrunc" (const_trunc (const_add foldbomb five) + i8_type) m); + ignore (define_global "ConstSExt" (const_sext foldbomb i128_type) m); + ignore (define_global "ConstZExt" (const_zext foldbomb i128_type) m); + ignore (define_global "ConstFPTrunc" (const_fptrunc ffoldbomb float_type) m); + ignore (define_global "ConstFPExt" (const_fpext ffoldbomb fp128_type) m); + ignore (define_global "ConstUIToFP" (const_uitofp foldbomb double_type) m); + ignore (define_global "ConstSIToFP" (const_sitofp foldbomb double_type) m); + ignore (define_global "ConstFPToUI" (const_fptoui ffoldbomb i32_type) m); + ignore (define_global "ConstFPToSI" (const_fptosi ffoldbomb i32_type) m); + ignore (define_global "ConstPtrToInt" (const_ptrtoint + (const_gep (make_null (make_pointer_type i8_type)) + [| make_int_constant i32_type 1 false |]) + i32_type) m); + ignore (define_global "ConstIntToPtr" (const_inttoptr (const_add foldbomb five) + void_ptr) m); + ignore (define_global "ConstBitCast" (const_bitcast ffoldbomb i64_type) m); + + group "misc constants"; + (* RUN: grep {ConstSizeOf.*getelementptr.*null} < %t.ll + * RUN: grep {ConstGEP.*getelementptr} < %t.ll + * RUN: grep {ConstSelect.*select} < %t.ll + * RUN: grep {ConstExtractElement.*extractelement} < %t.ll + * RUN: grep {ConstInsertElement.*insertelement} < %t.ll + * RUN: grep {ConstShuffleVector.*shufflevector} < %t.ll + *) + ignore (define_global "ConstSizeOf" (sizeof (make_pointer_type i8_type)) m); + ignore (define_global "ConstGEP" (const_gep foldbomb_gv [| five |]) m); + ignore (define_global "ConstSelect" (const_select + (const_icmp Icmp_sle foldbomb five) + (make_int_constant i8_type (-1) true) + (make_int_constant i8_type 0 true)) m); + let zero = make_int_constant i32_type 0 false in + let one = make_int_constant i32_type 1 false in + ignore (define_global "ConstExtractElement" (const_extractelement + (make_vector_constant [| zero; one; zero; one |]) + (const_trunc foldbomb i32_type)) m); + ignore (define_global "ConstInsertElement" (const_insertelement + (make_vector_constant [| zero; one; zero; one |]) + zero (const_trunc foldbomb i32_type)) m); + ignore (define_global "ConstShuffleVector" (const_shufflevector + (make_vector_constant [| zero; one |]) + (make_vector_constant [| one; zero |]) + (const_bitcast foldbomb (make_vector_type i32_type 2))) m) (*===-- Global Values -----------------------------------------------------===*) From gordonhenriksen at mac.com Sat Oct 6 10:11:06 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 15:11:06 -0000 Subject: [llvm-commits] [llvm] r42697 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp test/Bindings/Ocaml/vmcore.ml Message-ID: <200710061511.l96FB6hr019845@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 10:11:06 2007 New Revision: 42697 URL: http://llvm.org/viewvc/llvm-project?rev=42697&view=rev Log: Adopting a uniform naming convention for constant constructors in bindings. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sat Oct 6 10:11:06 2007 @@ -185,32 +185,29 @@ (*--... Operations on constants of (mostly) any type .......................--*) external is_constant : llvalue -> bool = "llvm_is_constant" -external make_null : lltype -> llvalue = "LLVMGetNull" -external make_all_ones : (*int|vec*)lltype -> llvalue = "LLVMGetAllOnes" -external make_undef : lltype -> llvalue = "LLVMGetUndef" +external const_null : lltype -> llvalue = "LLVMConstNull" +external const_all_ones : (*int|vec*)lltype -> llvalue = "LLVMConstAllOnes" +external undef : lltype -> llvalue = "LLVMGetUndef" external is_null : llvalue -> bool = "llvm_is_null" external is_undef : llvalue -> bool = "llvm_is_undef" (*--... Operations on scalar constants .....................................--*) -external make_int_constant : lltype -> int -> bool -> llvalue - = "llvm_make_int_constant" -external make_int64_constant : lltype -> Int64.t -> bool -> llvalue - = "llvm_make_int64_constant" -external make_real_constant : lltype -> float -> llvalue - = "llvm_make_real_constant" +external const_int : lltype -> int -> llvalue = "llvm_const_int" +external const_of_int64 : lltype -> Int64.t -> bool -> llvalue + = "llvm_const_of_int64" +external const_float : lltype -> float -> llvalue = "llvm_const_float" (*--... Operations on composite constants ..................................--*) -external make_string_constant : string -> bool -> llvalue - = "llvm_make_string_constant" -external make_array_constant : lltype -> llvalue array -> llvalue - = "llvm_make_array_constant" -external make_struct_constant : llvalue array -> bool -> llvalue - = "llvm_make_struct_constant" -external make_vector_constant : llvalue array -> llvalue - = "llvm_make_vector_constant" +external const_string : string -> llvalue = "llvm_const_string" +external const_stringz : string -> llvalue = "llvm_const_stringz" +external const_array : lltype -> llvalue array -> llvalue = "llvm_const_array" +external const_struct : llvalue array -> llvalue = "llvm_const_struct" +external const_packed_struct : llvalue array -> llvalue + = "llvm_const_packed_struct" +external const_vector : llvalue array -> llvalue = "llvm_const_vector" (*--... Constant expressions ...............................................--*) -external sizeof : lltype -> llvalue = "LLVMSizeOf" +external size_of : lltype -> llvalue = "LLVMSizeOf" external const_neg : llvalue -> llvalue = "LLVMConstNeg" external const_not : llvalue -> llvalue = "LLVMConstNot" external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sat Oct 6 10:11:06 2007 @@ -168,32 +168,29 @@ (*--... Operations on constants of (mostly) any type .......................--*) external is_constant : llvalue -> bool = "llvm_is_constant" -external make_null : lltype -> llvalue = "LLVMGetNull" -external make_all_ones : (*int|vec*)lltype -> llvalue = "LLVMGetAllOnes" -external make_undef : lltype -> llvalue = "LLVMGetUndef" +external const_null : lltype -> llvalue = "LLVMConstNull" +external const_all_ones : (*int|vec*)lltype -> llvalue = "LLVMConstAllOnes" +external undef : lltype -> llvalue = "LLVMGetUndef" external is_null : llvalue -> bool = "llvm_is_null" external is_undef : llvalue -> bool = "llvm_is_undef" (*--... Operations on scalar constants .....................................--*) -external make_int_constant : lltype -> int -> bool -> llvalue - = "llvm_make_int_constant" -external make_int64_constant : lltype -> Int64.t -> bool -> llvalue - = "llvm_make_int64_constant" -external make_real_constant : lltype -> float -> llvalue - = "llvm_make_real_constant" +external const_int : lltype -> int -> llvalue = "llvm_const_int" +external const_of_int64 : lltype -> Int64.t -> bool -> llvalue + = "llvm_const_of_int64" +external const_float : lltype -> float -> llvalue = "llvm_const_float" (*--... Operations on composite constants ..................................--*) -external make_string_constant : string -> bool -> llvalue - = "llvm_make_string_constant" -external make_array_constant : lltype -> llvalue array -> llvalue - = "llvm_make_array_constant" -external make_struct_constant : llvalue array -> bool -> llvalue - = "llvm_make_struct_constant" -external make_vector_constant : llvalue array -> llvalue - = "llvm_make_vector_constant" +external const_string : string -> llvalue = "llvm_const_string" +external const_stringz : string -> llvalue = "llvm_const_stringz" +external const_array : lltype -> llvalue array -> llvalue = "llvm_const_array" +external const_struct : llvalue array -> llvalue = "llvm_const_struct" +external const_packed_struct : llvalue array -> llvalue + = "llvm_const_packed_struct" +external const_vector : llvalue array -> llvalue = "llvm_const_vector" (*--... Constant expressions ...............................................--*) -external sizeof : lltype -> llvalue = "LLVMSizeOf" +external size_of : lltype -> llvalue = "LLVMSizeOf" external const_neg : llvalue -> llvalue = "LLVMConstNeg" external const_not : llvalue -> llvalue = "LLVMConstNot" external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 10:11:06 2007 @@ -244,56 +244,57 @@ /*--... Operations on scalar constants .....................................--*/ -/* lltype -> int -> bool -> llvalue */ -CAMLprim LLVMValueRef llvm_make_int_constant(LLVMTypeRef IntTy, value N, - value SExt) { - /* GCC warns if we use the ternary operator. */ - unsigned long long N2; - if (Bool_val(SExt)) - N2 = (value) Int_val(N); - else - N2 = (mlsize_t) Int_val(N); - - return LLVMGetIntConstant(IntTy, N2, Bool_val(SExt)); +/* lltype -> int -> llvalue */ +CAMLprim LLVMValueRef llvm_const_int(LLVMTypeRef IntTy, value N) { + return LLVMConstInt(IntTy, (long long) Int_val(N), 1); } /* lltype -> Int64.t -> bool -> llvalue */ -CAMLprim LLVMValueRef llvm_make_int64_constant(LLVMTypeRef IntTy, value N, - value SExt) { - return LLVMGetIntConstant(IntTy, Int64_val(N), Bool_val(SExt)); +CAMLprim LLVMValueRef llvm_const_of_int64(LLVMTypeRef IntTy, value N, + value SExt) { + return LLVMConstInt(IntTy, Int64_val(N), Bool_val(SExt)); } /* lltype -> float -> llvalue */ -CAMLprim LLVMValueRef llvm_make_real_constant(LLVMTypeRef RealTy, value N) { - return LLVMGetRealConstant(RealTy, Double_val(N)); +CAMLprim LLVMValueRef llvm_const_float(LLVMTypeRef RealTy, value N) { + return LLVMConstReal(RealTy, Double_val(N)); } /*--... Operations on composite constants ..................................--*/ -/* string -> bool -> llvalue */ -CAMLprim LLVMValueRef llvm_make_string_constant(value Str, value NullTerminate) { - return LLVMGetStringConstant(String_val(Str), string_length(Str), - Bool_val(NullTerminate) == 0); +/* string -> llvalue */ +CAMLprim LLVMValueRef llvm_const_string(value Str, value NullTerminate) { + return LLVMConstString(String_val(Str), string_length(Str), 1); +} + +/* string -> llvalue */ +CAMLprim LLVMValueRef llvm_const_stringz(value Str, value NullTerminate) { + return LLVMConstString(String_val(Str), string_length(Str), 0); } /* lltype -> llvalue array -> llvalue */ -CAMLprim LLVMValueRef llvm_make_array_constant(LLVMTypeRef ElementTy, +CAMLprim LLVMValueRef llvm_const_array(LLVMTypeRef ElementTy, value ElementVals) { - return LLVMGetArrayConstant(ElementTy, (LLVMValueRef*) Op_val(ElementVals), - Wosize_val(ElementVals)); + return LLVMConstArray(ElementTy, (LLVMValueRef*) Op_val(ElementVals), + Wosize_val(ElementVals)); +} + +/* llvalue array -> llvalue */ +CAMLprim LLVMValueRef llvm_const_struct(value ElementVals) { + return LLVMConstStruct((LLVMValueRef *) Op_val(ElementVals), + Wosize_val(ElementVals), 0); } -/* llvalue array -> bool -> llvalue */ -CAMLprim LLVMValueRef llvm_make_struct_constant(value ElementVals, - value Packed) { - return LLVMGetStructConstant((LLVMValueRef *) Op_val(ElementVals), - Wosize_val(ElementVals), Bool_val(Packed)); +/* llvalue array -> llvalue */ +CAMLprim LLVMValueRef llvm_const_packed_struct(value ElementVals) { + return LLVMConstStruct((LLVMValueRef *) Op_val(ElementVals), + Wosize_val(ElementVals), 1); } /* llvalue array -> llvalue */ -CAMLprim LLVMValueRef llvm_make_vector_constant(value ElementVals) { - return LLVMGetVectorConstant((LLVMValueRef*) Op_val(ElementVals), - Wosize_val(ElementVals)); +CAMLprim LLVMValueRef llvm_const_vector(value ElementVals) { + return LLVMConstVector((LLVMValueRef*) Op_val(ElementVals), + Wosize_val(ElementVals)); } /*--... Constant expressions ...............................................--*/ Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sat Oct 6 10:11:06 2007 @@ -228,27 +228,26 @@ void LLVMDumpValue(LLVMValueRef Val); /* Operations on constants of any type */ -LLVMValueRef LLVMGetNull(LLVMTypeRef Ty); /* all zeroes */ -LLVMValueRef LLVMGetAllOnes(LLVMTypeRef Ty); /* only for int/vector */ +LLVMValueRef LLVMConstNull(LLVMTypeRef Ty); /* all zeroes */ +LLVMValueRef LLVMConstAllOnes(LLVMTypeRef Ty); /* only for int/vector */ LLVMValueRef LLVMGetUndef(LLVMTypeRef Ty); int LLVMIsConstant(LLVMValueRef Val); int LLVMIsNull(LLVMValueRef Val); int LLVMIsUndef(LLVMValueRef Val); /* Operations on scalar constants */ -LLVMValueRef LLVMGetIntConstant(LLVMTypeRef IntTy, unsigned long long N, - int SignExtend); -LLVMValueRef LLVMGetRealConstant(LLVMTypeRef RealTy, double N); +LLVMValueRef LLVMConstInt(LLVMTypeRef IntTy, unsigned long long N, + int SignExtend); +LLVMValueRef LLVMConstReal(LLVMTypeRef RealTy, double N); /* Operations on composite constants */ -LLVMValueRef LLVMGetStringConstant(const char *Str, unsigned Length, - int DontNullTerminate); -LLVMValueRef LLVMGetArrayConstant(LLVMTypeRef ArrayTy, - LLVMValueRef *ConstantVals, unsigned Length); -LLVMValueRef LLVMGetStructConstant(LLVMValueRef *ConstantVals, unsigned Count, - int packed); -LLVMValueRef LLVMGetVectorConstant(LLVMValueRef *ScalarConstantVals, - unsigned Size); +LLVMValueRef LLVMConstString(const char *Str, unsigned Length, + int DontNullTerminate); +LLVMValueRef LLVMConstArray(LLVMTypeRef ArrayTy, + LLVMValueRef *ConstantVals, unsigned Length); +LLVMValueRef LLVMConstStruct(LLVMValueRef *ConstantVals, unsigned Count, + int packed); +LLVMValueRef LLVMConstVector(LLVMValueRef *ScalarConstantVals, unsigned Size); /* Constant expressions */ LLVMValueRef LLVMSizeOf(LLVMTypeRef Ty); Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sat Oct 6 10:11:06 2007 @@ -200,11 +200,11 @@ /*--.. Operations on constants of any type .................................--*/ -LLVMValueRef LLVMGetNull(LLVMTypeRef Ty) { +LLVMValueRef LLVMConstNull(LLVMTypeRef Ty) { return wrap(Constant::getNullValue(unwrap(Ty))); } -LLVMValueRef LLVMGetAllOnes(LLVMTypeRef Ty) { +LLVMValueRef LLVMConstAllOnes(LLVMTypeRef Ty) { return wrap(Constant::getAllOnesValue(unwrap(Ty))); } @@ -228,40 +228,39 @@ /*--.. Operations on scalar constants ......................................--*/ -LLVMValueRef LLVMGetIntConstant(LLVMTypeRef IntTy, unsigned long long N, - int SignExtend) { +LLVMValueRef LLVMConstInt(LLVMTypeRef IntTy, unsigned long long N, + int SignExtend) { return wrap(ConstantInt::get(unwrap(IntTy), N, SignExtend != 0)); } -LLVMValueRef LLVMGetRealConstant(LLVMTypeRef RealTy, double N) { +LLVMValueRef LLVMConstReal(LLVMTypeRef RealTy, double N) { return wrap(ConstantFP::get(unwrap(RealTy), APFloat(N))); } /*--.. Operations on composite constants ...................................--*/ -LLVMValueRef LLVMGetStringConstant(const char *Str, unsigned Length, - int DontNullTerminate) { +LLVMValueRef LLVMConstString(const char *Str, unsigned Length, + int DontNullTerminate) { /* Inverted the sense of AddNull because ', 0)' is a better mnemonic for null termination than ', 1)'. */ return wrap(ConstantArray::get(std::string(Str, Length), DontNullTerminate == 0)); } -LLVMValueRef LLVMGetArrayConstant(LLVMTypeRef ElementTy, - LLVMValueRef *ConstantVals, unsigned Length) { +LLVMValueRef LLVMConstArray(LLVMTypeRef ElementTy, + LLVMValueRef *ConstantVals, unsigned Length) { return wrap(ConstantArray::get(ArrayType::get(unwrap(ElementTy), Length), unwrap(ConstantVals, Length), Length)); } -LLVMValueRef LLVMGetStructConstant(LLVMValueRef *ConstantVals, unsigned Count, - int Packed) { +LLVMValueRef LLVMConstStruct(LLVMValueRef *ConstantVals, unsigned Count, + int Packed) { return wrap(ConstantStruct::get(unwrap(ConstantVals, Count), Count, Packed != 0)); } -LLVMValueRef LLVMGetVectorConstant(LLVMValueRef *ScalarConstantVals, - unsigned Size) { +LLVMValueRef LLVMConstVector(LLVMValueRef *ScalarConstantVals, unsigned Size) { return wrap(ConstantVector::get(unwrap(ScalarConstantVals, Size), Size)); } Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42697&r1=42696&r2=42697&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sat Oct 6 10:11:06 2007 @@ -140,7 +140,7 @@ (* RUN: grep {Const01.*i32.*-1} < %t.ll *) group "int"; - let c = make_int_constant i32_type (-1) true in + let c = const_int i32_type (-1) in ignore (define_global "Const01" c m); insist (i32_type = type_of c); insist (is_constant c); @@ -148,62 +148,62 @@ (* RUN: grep {Const02.*i64.*-1} < %t.ll *) group "sext int"; - let c = make_int_constant i64_type (-1) true in + let c = const_int i64_type (-1) in ignore (define_global "Const02" c m); insist (i64_type = type_of c); (* RUN: grep {Const03.*i64.*4294967295} < %t.ll *) group "zext int64"; - let c = make_int64_constant i64_type (Int64.of_string "4294967295") false in + let c = const_of_int64 i64_type (Int64.of_string "4294967295") false in ignore (define_global "Const03" c m); insist (i64_type = type_of c); (* RUN: grep {Const04.*"cruel\\\\00world"} < %t.ll *) group "string"; - let c = make_string_constant "cruel\000world" false in + let c = const_string "cruel\000world" in ignore (define_global "Const04" c m); insist ((make_array_type i8_type 11) = type_of c); (* RUN: grep {Const05.*"hi\\\\00again\\\\00"} < %t.ll *) - group "string w/ null"; - let c = make_string_constant "hi\000again" true in + group "stringz"; + let c = const_stringz "hi\000again" in ignore (define_global "Const05" c m); insist ((make_array_type i8_type 9) = type_of c); (* RUN: grep {Const06.*3.1459} < %t.ll *) group "real"; - let c = make_real_constant double_type 3.1459 in + let c = const_float double_type 3.1459 in ignore (define_global "Const06" c m); insist (double_type = type_of c); - let one = make_int_constant i16_type 1 true in - let two = make_int_constant i16_type 2 true in - let three = make_int_constant i32_type 3 true in - let four = make_int_constant i32_type 4 true in + let one = const_int i16_type 1 in + let two = const_int i16_type 2 in + let three = const_int i32_type 3 in + let four = const_int i32_type 4 in (* RUN: grep {Const07.*\\\[ i32 3, i32 4 \\\]} < %t.ll *) group "array"; - let c = make_array_constant i32_type [| three; four |] in + let c = const_array i32_type [| three; four |] in ignore (define_global "Const07" c m); insist ((make_array_type i32_type 2) = (type_of c)); (* RUN: grep {Const08.*< i16 1, i16 2.* >} < %t.ll *) group "vector"; - let c = make_vector_constant [| one; two; one; two; - one; two; one; two |] in + let c = const_vector [| one; two; one; two; + one; two; one; two |] in ignore (define_global "Const08" c m); insist ((make_vector_type i16_type 8) = (type_of c)); (* RUN: grep {Const09.*\{ i16, i16, i32, i32 \} \{} < %t.ll *) group "structure"; - let c = make_struct_constant [| one; two; three; four |] false in + let c = const_struct [| one; two; three; four |] in ignore (define_global "Const09" c m); insist ((make_struct_type [| i16_type; i16_type; i32_type; i32_type |] false) = (type_of c)); @@ -211,20 +211,20 @@ (* RUN: grep {Const10.*zeroinit} < %t.ll *) group "null"; - let c = make_null (make_struct_type [| i1_type; i8_type; - i64_type; double_type |] true) in + let c = const_null (make_struct_type [| i1_type; i8_type; + i64_type; double_type |] true) in ignore (define_global "Const10" c m); (* RUN: grep {Const11.*-1} < %t.ll *) group "all ones"; - let c = make_all_ones i64_type in + let c = const_all_ones i64_type in ignore (define_global "Const11" c m); (* RUN: grep {Const12.*undef} < %t.ll *) group "undef"; - let c = make_undef i1_type in + let c = undef i1_type in ignore (define_global "Const12" c m); insist (i1_type = type_of c); insist (is_undef c); @@ -248,9 +248,9 @@ * RUN: grep {ConstFCmp.*fcmp} < %t.ll *) let void_ptr = make_pointer_type i8_type in - let five = make_int_constant i64_type 5 false in + let five = const_int i64_type 5 in let ffive = const_uitofp five double_type in - let foldbomb_gv = define_global "FoldBomb" (make_null i8_type) m in + let foldbomb_gv = define_global "FoldBomb" (const_null i8_type) m in let foldbomb = const_ptrtoint foldbomb_gv i64_type in let ffoldbomb = const_uitofp foldbomb double_type in ignore (define_global "ConstNeg" (const_neg foldbomb) m); @@ -296,8 +296,8 @@ ignore (define_global "ConstFPToUI" (const_fptoui ffoldbomb i32_type) m); ignore (define_global "ConstFPToSI" (const_fptosi ffoldbomb i32_type) m); ignore (define_global "ConstPtrToInt" (const_ptrtoint - (const_gep (make_null (make_pointer_type i8_type)) - [| make_int_constant i32_type 1 false |]) + (const_gep (const_null (make_pointer_type i8_type)) + [| const_int i32_type 1 |]) i32_type) m); ignore (define_global "ConstIntToPtr" (const_inttoptr (const_add foldbomb five) void_ptr) m); @@ -311,23 +311,23 @@ * RUN: grep {ConstInsertElement.*insertelement} < %t.ll * RUN: grep {ConstShuffleVector.*shufflevector} < %t.ll *) - ignore (define_global "ConstSizeOf" (sizeof (make_pointer_type i8_type)) m); + ignore (define_global "ConstSizeOf" (size_of (make_pointer_type i8_type)) m); ignore (define_global "ConstGEP" (const_gep foldbomb_gv [| five |]) m); ignore (define_global "ConstSelect" (const_select (const_icmp Icmp_sle foldbomb five) - (make_int_constant i8_type (-1) true) - (make_int_constant i8_type 0 true)) m); - let zero = make_int_constant i32_type 0 false in - let one = make_int_constant i32_type 1 false in + (const_int i8_type (-1)) + (const_int i8_type 0)) m); + let zero = const_int i32_type 0 in + let one = const_int i32_type 1 in ignore (define_global "ConstExtractElement" (const_extractelement - (make_vector_constant [| zero; one; zero; one |]) + (const_vector [| zero; one; zero; one |]) (const_trunc foldbomb i32_type)) m); ignore (define_global "ConstInsertElement" (const_insertelement - (make_vector_constant [| zero; one; zero; one |]) + (const_vector [| zero; one; zero; one |]) zero (const_trunc foldbomb i32_type)) m); ignore (define_global "ConstShuffleVector" (const_shufflevector - (make_vector_constant [| zero; one |]) - (make_vector_constant [| one; zero |]) + (const_vector [| zero; one |]) + (const_vector [| one; zero |]) (const_bitcast foldbomb (make_vector_type i32_type 2))) m) @@ -335,7 +335,7 @@ let test_global_values () = let (++) x f = f x; x in - let zero32 = make_null i32_type in + let zero32 = const_null i32_type in (* RUN: grep {GVal01} < %t.ll *) @@ -378,7 +378,7 @@ let test_global_variables () = let (++) x f = f x; x in - let fourty_two32 = make_int_constant i32_type 42 false in + let fourty_two32 = const_int i32_type 42 in (* RUN: grep {GVar01.*external} < %t.ll *) @@ -501,7 +501,7 @@ let bb = entry_block fn in insist (bb = block_of_value (value_of_block bb)); insist (value_is_block (value_of_block bb)); - insist (not (value_is_block (make_null i32_type))) + insist (not (value_is_block (const_null i32_type))) (*===-- Builder -----------------------------------------------------------===*) @@ -703,15 +703,15 @@ let inst46 = build_icmp Icmp_eq p1 p2 "Inst46" atentry in ignore (build_select inst46 p1 p2 "Inst47" atentry); ignore (build_va_arg - (make_null (make_pointer_type (make_pointer_type i8_type))) + (const_null (make_pointer_type (make_pointer_type i8_type))) i32_type "Inst48" atentry); (* Set up some vector vregs. *) - let one = make_int_constant i32_type (-1) true in - let zero = make_int_constant i32_type 1 true in - let t1 = make_vector_constant [| one; zero; one; zero |] in - let t2 = make_vector_constant [| zero; one; zero; one |] in - let t3 = make_vector_constant [| one; one; zero; zero |] in + let one = const_int i32_type (-1) in + let zero = const_int i32_type 1 in + let t1 = const_vector [| one; zero; one; zero |] in + let t2 = const_vector [| zero; one; zero; one |] in + let t3 = const_vector [| one; one; zero; zero |] in let vec1 = build_insertelement t1 p1 p2 "Vec1" atentry in let vec2 = build_insertelement t2 p1 p2 "Vec2" atentry in let vec3 = build_insertelement t3 p1 p2 "Vec3" atentry in From gordonhenriksen at mac.com Sat Oct 6 11:05:21 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 16:05:21 -0000 Subject: [llvm-commits] [llvm] r42698 - in /llvm/trunk: bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp Message-ID: <200710061605.l96G5LEs021853@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 11:05:20 2007 New Revision: 42698 URL: http://llvm.org/viewvc/llvm-project?rev=42698&view=rev Log: Adopting a uniform naming convention for type constructors in bindings. Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42698&r1=42697&r2=42698&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 11:05:20 2007 @@ -74,7 +74,7 @@ /* int -> lltype */ CAMLprim LLVMTypeRef llvm_make_integer_type(value Width) { - return LLVMCreateIntType(Int_val(Width)); + return LLVMIntType(Int_val(Width)); } /* lltype -> int */ @@ -114,9 +114,9 @@ /* lltype -> lltype array -> bool -> lltype */ CAMLprim LLVMTypeRef llvm_make_function_type(LLVMTypeRef RetTy, value ParamTys, value IsVarArg) { - return LLVMCreateFunctionType(RetTy, (LLVMTypeRef *) ParamTys, - Wosize_val(ParamTys), - Bool_val(IsVarArg)); + return LLVMFunctionType(RetTy, (LLVMTypeRef *) ParamTys, + Wosize_val(ParamTys), + Bool_val(IsVarArg)); } /* lltype -> bool */ @@ -140,9 +140,9 @@ /* lltype array -> bool -> lltype */ CAMLprim LLVMTypeRef llvm_make_struct_type(value ElementTypes, value Packed) { - return LLVMCreateStructType((LLVMTypeRef *) ElementTypes, - Wosize_val(ElementTypes), - Bool_val(Packed)); + return LLVMStructType((LLVMTypeRef *) ElementTypes, + Wosize_val(ElementTypes), + Bool_val(Packed)); } /* lltype -> lltype array */ @@ -161,17 +161,17 @@ /* lltype -> int -> lltype */ CAMLprim LLVMTypeRef llvm_make_array_type(LLVMTypeRef ElementTy, value Count) { - return LLVMCreateArrayType(ElementTy, Int_val(Count)); + return LLVMArrayType(ElementTy, Int_val(Count)); } /* lltype -> lltype */ CAMLprim LLVMTypeRef llvm_make_pointer_type(LLVMTypeRef ElementTy) { - return LLVMCreatePointerType(ElementTy); + return LLVMPointerType(ElementTy); } /* lltype -> int -> lltype */ CAMLprim LLVMTypeRef llvm_make_vector_type(LLVMTypeRef ElementTy, value Count) { - return LLVMCreateVectorType(ElementTy, Int_val(Count)); + return LLVMVectorType(ElementTy, Int_val(Count)); } /* lltype -> lltype */ @@ -197,7 +197,7 @@ /* unit -> lltype */ CAMLprim LLVMTypeRef llvm_make_opaque_type(value Unit) { - return LLVMCreateOpaqueType(); + return LLVMOpaqueType(); } Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42698&r1=42697&r2=42698&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sat Oct 6 11:05:20 2007 @@ -164,7 +164,7 @@ LLVMTypeRef LLVMInt16Type(); LLVMTypeRef LLVMInt32Type(); LLVMTypeRef LLVMInt64Type(); -LLVMTypeRef LLVMCreateIntType(unsigned NumBits); +LLVMTypeRef LLVMIntType(unsigned NumBits); unsigned LLVMGetIntTypeWidth(LLVMTypeRef IntegerTy); /* Operations on real types */ @@ -175,25 +175,25 @@ LLVMTypeRef LLVMPPCFP128Type(); /* Operations on function types */ -LLVMTypeRef LLVMCreateFunctionType(LLVMTypeRef ReturnType, - LLVMTypeRef *ParamTypes, unsigned ParamCount, - int IsVarArg); +LLVMTypeRef LLVMFunctionType(LLVMTypeRef ReturnType, + LLVMTypeRef *ParamTypes, unsigned ParamCount, + int IsVarArg); int LLVMIsFunctionVarArg(LLVMTypeRef FunctionTy); LLVMTypeRef LLVMGetReturnType(LLVMTypeRef FunctionTy); unsigned LLVMCountParamTypes(LLVMTypeRef FunctionTy); void LLVMGetParamTypes(LLVMTypeRef FunctionTy, LLVMTypeRef *Dest); /* Operations on struct types */ -LLVMTypeRef LLVMCreateStructType(LLVMTypeRef *ElementTypes, - unsigned ElementCount, int Packed); +LLVMTypeRef LLVMStructType(LLVMTypeRef *ElementTypes, unsigned ElementCount, + int Packed); unsigned LLVMCountStructElementTypes(LLVMTypeRef StructTy); void LLVMGetStructElementTypes(LLVMTypeRef StructTy, LLVMTypeRef *Dest); int LLVMIsPackedStruct(LLVMTypeRef StructTy); /* Operations on array, pointer, and vector types (sequence types) */ -LLVMTypeRef LLVMCreateArrayType(LLVMTypeRef ElementType, unsigned ElementCount); -LLVMTypeRef LLVMCreatePointerType(LLVMTypeRef ElementType); -LLVMTypeRef LLVMCreateVectorType(LLVMTypeRef ElementType,unsigned ElementCount); +LLVMTypeRef LLVMArrayType(LLVMTypeRef ElementType, unsigned ElementCount); +LLVMTypeRef LLVMPointerType(LLVMTypeRef ElementType); +LLVMTypeRef LLVMVectorType(LLVMTypeRef ElementType, unsigned ElementCount); LLVMTypeRef LLVMGetElementType(LLVMTypeRef Ty); unsigned LLVMGetArrayLength(LLVMTypeRef ArrayTy); @@ -202,7 +202,7 @@ /* Operations on other types */ LLVMTypeRef LLVMVoidType(); LLVMTypeRef LLVMLabelType(); -LLVMTypeRef LLVMCreateOpaqueType(); +LLVMTypeRef LLVMOpaqueType(); /*===-- Values ------------------------------------------------------------===*/ Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42698&r1=42697&r2=42698&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sat Oct 6 11:05:20 2007 @@ -68,7 +68,7 @@ LLVMTypeRef LLVMInt32Type() { return (LLVMTypeRef) Type::Int32Ty; } LLVMTypeRef LLVMInt64Type() { return (LLVMTypeRef) Type::Int64Ty; } -LLVMTypeRef LLVMCreateIntType(unsigned NumBits) { +LLVMTypeRef LLVMIntType(unsigned NumBits) { return wrap(IntegerType::get(NumBits)); } @@ -86,9 +86,9 @@ /*--.. Operations on function types ........................................--*/ -LLVMTypeRef LLVMCreateFunctionType(LLVMTypeRef ReturnType, - LLVMTypeRef *ParamTypes, unsigned ParamCount, - int IsVarArg) { +LLVMTypeRef LLVMFunctionType(LLVMTypeRef ReturnType, + LLVMTypeRef *ParamTypes, unsigned ParamCount, + int IsVarArg) { std::vector Tys; for (LLVMTypeRef *I = ParamTypes, *E = ParamTypes + ParamCount; I != E; ++I) Tys.push_back(unwrap(*I)); @@ -117,8 +117,8 @@ /*--.. Operations on struct types ..........................................--*/ -LLVMTypeRef LLVMCreateStructType(LLVMTypeRef *ElementTypes, - unsigned ElementCount, int Packed) { +LLVMTypeRef LLVMStructType(LLVMTypeRef *ElementTypes, + unsigned ElementCount, int Packed) { std::vector Tys; for (LLVMTypeRef *I = ElementTypes, *E = ElementTypes + ElementCount; I != E; ++I) @@ -144,15 +144,15 @@ /*--.. Operations on array, pointer, and vector types (sequence types) .....--*/ -LLVMTypeRef LLVMCreateArrayType(LLVMTypeRef ElementType, unsigned ElementCount){ +LLVMTypeRef LLVMArrayType(LLVMTypeRef ElementType, unsigned ElementCount){ return wrap(ArrayType::get(unwrap(ElementType), ElementCount)); } -LLVMTypeRef LLVMCreatePointerType(LLVMTypeRef ElementType) { +LLVMTypeRef LLVMPointerType(LLVMTypeRef ElementType) { return wrap(PointerType::get(unwrap(ElementType))); } -LLVMTypeRef LLVMCreateVectorType(LLVMTypeRef ElementType,unsigned ElementCount){ +LLVMTypeRef LLVMVectorType(LLVMTypeRef ElementType,unsigned ElementCount){ return wrap(VectorType::get(unwrap(ElementType), ElementCount)); } @@ -173,7 +173,7 @@ LLVMTypeRef LLVMVoidType() { return (LLVMTypeRef) Type::VoidTy; } LLVMTypeRef LLVMLabelType() { return (LLVMTypeRef) Type::LabelTy; } -LLVMTypeRef LLVMCreateOpaqueType() { +LLVMTypeRef LLVMOpaqueType() { return wrap(llvm::OpaqueType::get()); } From asl at math.spbu.ru Sat Oct 6 11:17:49 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 16:17:49 -0000 Subject: [llvm-commits] [llvm] r42699 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Message-ID: <200710061617.l96GHnve022368@zion.cs.uiuc.edu> Author: asl Date: Sat Oct 6 11:17:49 2007 New Revision: 42699 URL: http://llvm.org/viewvc/llvm-project?rev=42699&view=rev Log: Move merge code into new helper function. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42699&r1=42698&r2=42699&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Sat Oct 6 11:17:49 2007 @@ -1477,6 +1477,29 @@ } } +// mergeSPUpdatesUp - Merge two stack-manipulating instructions upper iterator. +static +void mergeSPUpdatesUp(MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI, + unsigned StackPtr, uint64_t *NumBytes = NULL) { + if (MBBI != MBB.begin()) { + MachineBasicBlock::iterator PI = prior(MBBI); + unsigned Opc = PI->getOpcode(); + if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || + Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && + PI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes += PI->getOperand(2).getImm(); + MBB.erase(PI); + } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || + Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && + PI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes -= PI->getOperand(2).getImm(); + MBB.erase(PI); + } + } +} + void X86RegisterInfo::emitPrologue(MachineFunction &MF) const { MachineBasicBlock &MBB = MF.front(); // Prolog goes in entry BB MachineFrameInfo *MFI = MF.getFrameInfo(); @@ -1562,26 +1585,10 @@ MBB.insert(MBBI, MI); } } else { - // If there is an ADD32ri or SUB32ri of ESP immediately after this + // If there is an ADD32ri or SUB32ri of ESP immediately before this // instruction, merge the two instructions. - if (MBBI != MBB.end()) { - MachineBasicBlock::iterator NI = next(MBBI); - unsigned Opc = MBBI->getOpcode(); - if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || - Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && - MBBI->getOperand(0).getReg() == StackPtr) { - NumBytes -= MBBI->getOperand(2).getImm(); - MBB.erase(MBBI); - MBBI = NI; - } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || - Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && - MBBI->getOperand(0).getReg() == StackPtr) { - NumBytes += MBBI->getOperand(2).getImm(); - MBB.erase(MBBI); - MBBI = NI; - } - } - + mergeSPUpdatesUp(MBB, MBBI, StackPtr, &NumBytes); + if (NumBytes) emitSPUpdate(MBB, MBBI, StackPtr, -(int64_t)NumBytes, Is64Bit, TII); } @@ -1702,31 +1709,15 @@ --MBBI; } - if (NumBytes || MFI->hasVarSizedObjects()) { - // If there is an ADD32ri or SUB32ri of ESP immediately before this - // instruction, merge the two instructions. - if (MBBI != MBB.begin()) { - MachineBasicBlock::iterator PI = prior(MBBI); - unsigned Opc = PI->getOpcode(); - if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || - Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && - PI->getOperand(0).getReg() == StackPtr) { - NumBytes += PI->getOperand(2).getImm(); - MBB.erase(PI); - } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || - Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && - PI->getOperand(0).getReg() == StackPtr) { - NumBytes -= PI->getOperand(2).getImm(); - MBB.erase(PI); - } - } - } - - // If dynamic alloca is used, then reset esp to point to the last - // callee-saved slot before popping them off! - // Also, if it's main() on Cygwin/Mingw32 we aligned stack in the prologue, - revert - // stack changes back. Note: we're assuming, that frame pointer was forced - // for main() + // If there is an ADD32ri or SUB32ri of ESP immediately before this + // instruction, merge the two instructions. + if (NumBytes || MFI->hasVarSizedObjects()) + mergeSPUpdatesUp(MBB, MBBI, StackPtr, &NumBytes); + + // If dynamic alloca is used, then reset esp to point to the last callee-saved + // slot before popping them off! Also, if it's main() on Cygwin/Mingw32 we + // aligned stack in the prologue, - revert stack changes back. Note: we're + // assuming, that frame pointer was forced for main() if (MFI->hasVarSizedObjects() || (Fn->hasExternalLinkage() && Fn->getName() == "main" && Subtarget->isTargetCygMing())) { From asl at math.spbu.ru Sat Oct 6 11:39:43 2007 From: asl at math.spbu.ru (Anton Korobeynikov) Date: Sat, 06 Oct 2007 16:39:43 -0000 Subject: [llvm-commits] [llvm] r42700 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Message-ID: <200710061639.l96GdhRD023221@zion.cs.uiuc.edu> Author: asl Date: Sat Oct 6 11:39:43 2007 New Revision: 42700 URL: http://llvm.org/viewvc/llvm-project?rev=42700&view=rev Log: Oops, I really wanted to commit this part also :) Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42700&r1=42699&r2=42700&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Sat Oct 6 11:39:43 2007 @@ -1500,6 +1500,31 @@ } } +// mergeSPUpdatesUp - Merge two stack-manipulating instructions lower iterator. +static +void mergeSPUpdatesDown(MachineBasicBlock &MBB,MachineBasicBlock::iterator &MBBI, + unsigned StackPtr, uint64_t *NumBytes = NULL) { + if (MBBI != MBB.end()) { + MachineBasicBlock::iterator NI = next(MBBI); + unsigned Opc = NI->getOpcode(); + if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || + Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && + NI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes -= NI->getOperand(2).getImm(); + MBB.erase(NI); + MBBI = NI; + } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || + Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && + NI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes += NI->getOperand(2).getImm(); + MBB.erase(NI); + MBBI = NI; + } + } +} + void X86RegisterInfo::emitPrologue(MachineFunction &MF) const { MachineBasicBlock &MBB = MF.front(); // Prolog goes in entry BB MachineFrameInfo *MFI = MF.getFrameInfo(); @@ -1585,9 +1610,9 @@ MBB.insert(MBBI, MI); } } else { - // If there is an ADD32ri or SUB32ri of ESP immediately before this + // If there is an ADD32ri or SUB32ri of ESP immediately after this // instruction, merge the two instructions. - mergeSPUpdatesUp(MBB, MBBI, StackPtr, &NumBytes); + mergeSPUpdatesDown(MBB, MBBI, StackPtr, &NumBytes); if (NumBytes) emitSPUpdate(MBB, MBBI, StackPtr, -(int64_t)NumBytes, Is64Bit, TII); From gordonhenriksen at mac.com Sat Oct 6 11:56:09 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 16:56:09 -0000 Subject: [llvm-commits] [llvm] r42701 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c test/Bindings/Ocaml/vmcore.ml Message-ID: <200710061656.l96GuAdF024037@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 11:56:09 2007 New Revision: 42701 URL: http://llvm.org/viewvc/llvm-project?rev=42701&view=rev Log: Adopting a uniform naming convention for type constructors in bindings (part le deux). Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42701&r1=42700&r2=42701&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sat Oct 6 11:56:09 2007 @@ -128,7 +128,7 @@ let i32_type = _i32_type () let i64_type = _i64_type () -external make_integer_type : int -> lltype = "llvm_make_integer_type" +external integer_type : int -> lltype = "llvm_integer_type" external integer_bitwidth : lltype -> int = "llvm_integer_bitwidth" (*--... Operations on real types ...........................................--*) @@ -146,29 +146,30 @@ (*--... Operations on function types .......................................--*) (* FIXME: handle parameter attributes *) -external make_function_type : lltype -> lltype array -> bool -> lltype - = "llvm_make_function_type" +external function_type : lltype -> lltype array -> lltype = "llvm_function_type" +external var_arg_function_type : lltype -> lltype array -> lltype + = "llvm_var_arg_function_type" external is_var_arg : lltype -> bool = "llvm_is_var_arg" external return_type : lltype -> lltype = "llvm_return_type" external param_types : lltype -> lltype array = "llvm_param_types" (*--... Operations on struct types .........................................--*) -external make_struct_type : lltype array -> bool -> lltype - = "llvm_make_struct_type" +external struct_type : lltype array -> lltype = "llvm_struct_type" +external packed_struct_type : lltype array -> lltype = "llvm_packed_struct_type" external element_types : lltype -> lltype array = "llvm_element_types" external is_packed : lltype -> bool = "llvm_is_packed" (*--... Operations on pointer, vector, and array types .....................--*) -external make_array_type : lltype -> int -> lltype = "llvm_make_array_type" -external make_pointer_type : lltype -> lltype = "llvm_make_pointer_type" -external make_vector_type : lltype -> int -> lltype = "llvm_make_vector_type" +external array_type : lltype -> int -> lltype = "llvm_array_type" +external pointer_type : lltype -> lltype = "llvm_pointer_type" +external vector_type : lltype -> int -> lltype = "llvm_vector_type" external element_type : lltype -> lltype = "llvm_element_type" external array_length : lltype -> int = "llvm_array_length" external vector_size : lltype -> int = "llvm_vector_size" (*--... Operations on other types ..........................................--*) -external make_opaque_type : unit -> lltype = "llvm_make_opaque_type" +external opaque_type : unit -> lltype = "llvm_opaque_type" external _void_type : unit -> lltype = "llvm_void_type" external _label_type : unit -> lltype = "llvm_label_type" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42701&r1=42700&r2=42701&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sat Oct 6 11:56:09 2007 @@ -121,7 +121,7 @@ val i16_type : lltype val i32_type : lltype val i64_type : lltype -external make_integer_type : int -> lltype = "llvm_make_integer_type" +external integer_type : int -> lltype = "llvm_integer_type" external integer_bitwidth : lltype -> int = "llvm_integer_bitwidth" (*--... Operations on real types ...........................................--*) @@ -133,29 +133,30 @@ (*--... Operations on function types .......................................--*) (* FIXME: handle parameter attributes *) -external make_function_type : lltype -> lltype array -> bool -> lltype - = "llvm_make_function_type" +external function_type : lltype -> lltype array -> lltype = "llvm_function_type" +external var_arg_function_type : lltype -> lltype array -> lltype + = "llvm_var_arg_function_type" external is_var_arg : lltype -> bool = "llvm_is_var_arg" external return_type : lltype -> lltype = "llvm_return_type" external param_types : lltype -> lltype array = "llvm_param_types" (*--... Operations on struct types .........................................--*) -external make_struct_type : lltype array -> bool -> lltype - = "llvm_make_struct_type" +external struct_type : lltype array -> lltype = "llvm_struct_type" +external packed_struct_type : lltype array -> lltype = "llvm_packed_struct_type" external element_types : lltype -> lltype array = "llvm_element_types" external is_packed : lltype -> bool = "llvm_is_packed" (*--... Operations on pointer, vector, and array types .....................--*) -external make_array_type : lltype -> int -> lltype = "llvm_make_array_type" -external make_pointer_type : lltype -> lltype = "llvm_make_pointer_type" -external make_vector_type : lltype -> int -> lltype = "llvm_make_vector_type" +external array_type : lltype -> int -> lltype = "llvm_array_type" +external pointer_type : lltype -> lltype = "llvm_pointer_type" +external vector_type : lltype -> int -> lltype = "llvm_vector_type" external element_type : lltype -> lltype = "llvm_element_type" external array_length : lltype -> int = "llvm_array_length" external vector_size : lltype -> int = "llvm_vector_size" (*--... Operations on other types ..........................................--*) -external make_opaque_type : unit -> lltype = "llvm_make_opaque_type" +external opaque_type : unit -> lltype = "llvm_opaque_type" val void_type : lltype val label_type : lltype Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42701&r1=42700&r2=42701&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 11:56:09 2007 @@ -73,7 +73,7 @@ CAMLprim LLVMTypeRef llvm_i64_type(value Unit) { return LLVMInt64Type(); } /* int -> lltype */ -CAMLprim LLVMTypeRef llvm_make_integer_type(value Width) { +CAMLprim LLVMTypeRef llvm_integer_type(value Width) { return LLVMIntType(Int_val(Width)); } @@ -111,12 +111,17 @@ /*--... Operations on function types .......................................--*/ -/* lltype -> lltype array -> bool -> lltype */ -CAMLprim LLVMTypeRef llvm_make_function_type(LLVMTypeRef RetTy, value ParamTys, - value IsVarArg) { +/* lltype -> lltype array -> lltype */ +CAMLprim LLVMTypeRef llvm_function_type(LLVMTypeRef RetTy, value ParamTys) { return LLVMFunctionType(RetTy, (LLVMTypeRef *) ParamTys, - Wosize_val(ParamTys), - Bool_val(IsVarArg)); + Wosize_val(ParamTys), 0); +} + +/* lltype -> lltype array -> lltype */ +CAMLprim LLVMTypeRef llvm_var_arg_function_type(LLVMTypeRef RetTy, + value ParamTys) { + return LLVMFunctionType(RetTy, (LLVMTypeRef *) ParamTys, + Wosize_val(ParamTys), 1); } /* lltype -> bool */ @@ -138,11 +143,16 @@ /*--... Operations on struct types .........................................--*/ -/* lltype array -> bool -> lltype */ -CAMLprim LLVMTypeRef llvm_make_struct_type(value ElementTypes, value Packed) { +/* lltype array -> lltype */ +CAMLprim LLVMTypeRef llvm_struct_type(value ElementTypes) { + return LLVMStructType((LLVMTypeRef *) ElementTypes, + Wosize_val(ElementTypes), 0); +} + +/* lltype array -> lltype */ +CAMLprim LLVMTypeRef llvm_packed_struct_type(value ElementTypes) { return LLVMStructType((LLVMTypeRef *) ElementTypes, - Wosize_val(ElementTypes), - Bool_val(Packed)); + Wosize_val(ElementTypes), 1); } /* lltype -> lltype array */ @@ -160,17 +170,17 @@ /*--... Operations on array, pointer, and vector types .....................--*/ /* lltype -> int -> lltype */ -CAMLprim LLVMTypeRef llvm_make_array_type(LLVMTypeRef ElementTy, value Count) { +CAMLprim LLVMTypeRef llvm_array_type(LLVMTypeRef ElementTy, value Count) { return LLVMArrayType(ElementTy, Int_val(Count)); } /* lltype -> lltype */ -CAMLprim LLVMTypeRef llvm_make_pointer_type(LLVMTypeRef ElementTy) { +CAMLprim LLVMTypeRef llvm_pointer_type(LLVMTypeRef ElementTy) { return LLVMPointerType(ElementTy); } /* lltype -> int -> lltype */ -CAMLprim LLVMTypeRef llvm_make_vector_type(LLVMTypeRef ElementTy, value Count) { +CAMLprim LLVMTypeRef llvm_vector_type(LLVMTypeRef ElementTy, value Count) { return LLVMVectorType(ElementTy, Int_val(Count)); } @@ -196,7 +206,7 @@ CAMLprim LLVMTypeRef llvm_label_type(value Unit) { return LLVMLabelType(); } /* unit -> lltype */ -CAMLprim LLVMTypeRef llvm_make_opaque_type(value Unit) { +CAMLprim LLVMTypeRef llvm_opaque_type(value Unit) { return LLVMOpaqueType(); } Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42701&r1=42700&r2=42701&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sat Oct 6 11:56:09 2007 @@ -61,7 +61,7 @@ (* RUN: grep {Ty04.*i42} < %t.ll *) group "i42"; - let ty = make_integer_type 42 in + let ty = integer_type 42 in insist (define_type_name "Ty04" ty m); (* RUN: grep {Ty05.*float} < %t.ll @@ -79,7 +79,7 @@ (* RUN: grep {Ty07.*i32.*i1, double} < %t.ll *) group "function"; - let ty = make_function_type i32_type [| i1_type; double_type |] false in + let ty = function_type i32_type [| i1_type; double_type |] in insist (define_type_name "Ty07" ty m); insist (Function_type = classify_type ty); insist (not (is_var_arg ty)); @@ -88,15 +88,15 @@ (* RUN: grep {Ty08.*\.\.\.} < %t.ll *) - group "vararg"; - let ty = make_function_type void_type [| i32_type |] true in + group "var arg function"; + let ty = var_arg_function_type void_type [| i32_type |] in insist (define_type_name "Ty08" ty m); insist (is_var_arg ty); (* RUN: grep {Ty09.*\\\[7 x i8\\\]} < %t.ll *) group "array"; - let ty = make_array_type i8_type 7 in + let ty = array_type i8_type 7 in insist (define_type_name "Ty09" ty m); insist (7 = array_length ty); insist (i8_type == element_type ty); @@ -105,7 +105,7 @@ (* RUN: grep {Ty10.*float\*} < %t.ll *) group "pointer"; - let ty = make_pointer_type float_type in + let ty = pointer_type float_type in insist (define_type_name "Ty10" ty m); insist (float_type == element_type ty); insist (Pointer_type == classify_type ty); @@ -113,7 +113,7 @@ (* RUN: grep {Ty11.*\<4 x i16\>} < %t.ll *) group "vector"; - let ty = make_vector_type i16_type 4 in + let ty = vector_type i16_type 4 in insist (define_type_name "Ty11" ty m); insist (i16_type == element_type ty); insist (4 = vector_size ty); @@ -121,15 +121,15 @@ (* RUN: grep {Ty12.*opaque} < %t.ll *) group "opaque"; - let ty = make_opaque_type () in + let ty = opaque_type () in insist (define_type_name "Ty12" ty m); insist (ty == ty); - insist (ty <> make_opaque_type ()); + insist (ty <> opaque_type ()); (* RUN: grep -v {Ty13} < %t.ll *) group "delete"; - let ty = make_opaque_type () in + let ty = opaque_type () in insist (define_type_name "Ty13" ty m); delete_type_name "Ty13" m @@ -164,14 +164,14 @@ group "string"; let c = const_string "cruel\000world" in ignore (define_global "Const04" c m); - insist ((make_array_type i8_type 11) = type_of c); + insist ((array_type i8_type 11) = type_of c); (* RUN: grep {Const05.*"hi\\\\00again\\\\00"} < %t.ll *) group "stringz"; let c = const_stringz "hi\000again" in ignore (define_global "Const05" c m); - insist ((make_array_type i8_type 9) = type_of c); + insist ((array_type i8_type 9) = type_of c); (* RUN: grep {Const06.*3.1459} < %t.ll *) @@ -190,7 +190,7 @@ group "array"; let c = const_array i32_type [| three; four |] in ignore (define_global "Const07" c m); - insist ((make_array_type i32_type 2) = (type_of c)); + insist ((array_type i32_type 2) = (type_of c)); (* RUN: grep {Const08.*< i16 1, i16 2.* >} < %t.ll *) @@ -198,21 +198,21 @@ let c = const_vector [| one; two; one; two; one; two; one; two |] in ignore (define_global "Const08" c m); - insist ((make_vector_type i16_type 8) = (type_of c)); + insist ((vector_type i16_type 8) = (type_of c)); (* RUN: grep {Const09.*\{ i16, i16, i32, i32 \} \{} < %t.ll *) group "structure"; let c = const_struct [| one; two; three; four |] in ignore (define_global "Const09" c m); - insist ((make_struct_type [| i16_type; i16_type; i32_type; i32_type |] false) + insist ((struct_type [| i16_type; i16_type; i32_type; i32_type |]) = (type_of c)); (* RUN: grep {Const10.*zeroinit} < %t.ll *) group "null"; - let c = const_null (make_struct_type [| i1_type; i8_type; - i64_type; double_type |] true) in + let c = const_null (packed_struct_type [| i1_type; i8_type; + i64_type; double_type |]) in ignore (define_global "Const10" c m); (* RUN: grep {Const11.*-1} < %t.ll @@ -247,7 +247,7 @@ * RUN: grep {ConstICmp.*icmp} < %t.ll * RUN: grep {ConstFCmp.*fcmp} < %t.ll *) - let void_ptr = make_pointer_type i8_type in + let void_ptr = pointer_type i8_type in let five = const_int i64_type 5 in let ffive = const_uitofp five double_type in let foldbomb_gv = define_global "FoldBomb" (const_null i8_type) m in @@ -284,7 +284,7 @@ * RUN: grep {ConstIntToPtr.*inttoptr} < %t.ll * RUN: grep {ConstBitCast.*bitcast} < %t.ll *) - let i128_type = make_integer_type 128 in + let i128_type = integer_type 128 in ignore (define_global "ConstTrunc" (const_trunc (const_add foldbomb five) i8_type) m); ignore (define_global "ConstSExt" (const_sext foldbomb i128_type) m); @@ -296,7 +296,7 @@ ignore (define_global "ConstFPToUI" (const_fptoui ffoldbomb i32_type) m); ignore (define_global "ConstFPToSI" (const_fptosi ffoldbomb i32_type) m); ignore (define_global "ConstPtrToInt" (const_ptrtoint - (const_gep (const_null (make_pointer_type i8_type)) + (const_gep (const_null (pointer_type i8_type)) [| const_int i32_type 1 |]) i32_type) m); ignore (define_global "ConstIntToPtr" (const_inttoptr (const_add foldbomb five) @@ -311,7 +311,7 @@ * RUN: grep {ConstInsertElement.*insertelement} < %t.ll * RUN: grep {ConstShuffleVector.*shufflevector} < %t.ll *) - ignore (define_global "ConstSizeOf" (size_of (make_pointer_type i8_type)) m); + ignore (define_global "ConstSizeOf" (size_of (pointer_type i8_type)) m); ignore (define_global "ConstGEP" (const_gep foldbomb_gv [| five |]) m); ignore (define_global "ConstSelect" (const_select (const_icmp Icmp_sle foldbomb five) @@ -328,7 +328,7 @@ ignore (define_global "ConstShuffleVector" (const_shufflevector (const_vector [| zero; one |]) (const_vector [| one; zero |]) - (const_bitcast foldbomb (make_vector_type i32_type 2))) m) + (const_bitcast foldbomb (vector_type i32_type 2))) m) (*===-- Global Values -----------------------------------------------------===*) @@ -414,8 +414,8 @@ (*===-- Functions ---------------------------------------------------------===*) let test_functions () = - let ty = make_function_type i32_type [| i32_type; i64_type |] false in - let pty = make_pointer_type ty in + let ty = function_type i32_type [| i32_type; i64_type |] in + let pty = pointer_type ty in (* RUN: grep {declare i32 @Fn1\(i32, i64\)} < %t.ll *) @@ -465,7 +465,7 @@ (*===-- Basic Blocks ------------------------------------------------------===*) let test_basic_blocks () = - let ty = make_function_type void_type [| |] false in + let ty = function_type void_type [| |] in (* RUN: grep {Bb1} < %t.ll *) @@ -513,14 +513,14 @@ begin (* RUN: grep {ret void} < %t.ll *) - let fty = make_function_type void_type [| |] false in + let fty = function_type void_type [| |] in let fn = declare_function "X6" fty m in let b = builder_at_end (append_block "Bb01" fn) in ignore (build_ret_void b) end; (* The rest of the tests will use one big function. *) - let fty = make_function_type i32_type [| i32_type; i32_type |] false in + let fty = function_type i32_type [| i32_type; i32_type |] in let fn = define_function "X7" fty m in let atentry = builder_at_end (entry_block fn) in let p1 = param fn 0 ++ set_value_name "P1" in @@ -647,7 +647,7 @@ end; group "casts"; begin - let void_ptr = make_pointer_type i8_type in + let void_ptr = pointer_type i8_type in (* RUN: grep {Inst28.*trunc.*P1.*i8} < %t.ll * RUN: grep {Inst29.*zext.*Inst28.*i32} < %t.ll @@ -703,7 +703,7 @@ let inst46 = build_icmp Icmp_eq p1 p2 "Inst46" atentry in ignore (build_select inst46 p1 p2 "Inst47" atentry); ignore (build_va_arg - (const_null (make_pointer_type (make_pointer_type i8_type))) + (const_null (pointer_type (pointer_type i8_type))) i32_type "Inst48" atentry); (* Set up some vector vregs. *) From gordonhenriksen at mac.com Sat Oct 6 12:10:44 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 17:10:44 -0000 Subject: [llvm-commits] [llvm] r42702 - in /llvm/trunk/bindings/ocaml/llvm: llvm.ml llvm.mli llvm_ocaml.c Message-ID: <200710061710.l96HAiek024632@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 12:10:44 2007 New Revision: 42702 URL: http://llvm.org/viewvc/llvm-project?rev=42702&view=rev Log: Deleting some unnecessary glue. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42702&r1=42701&r2=42702&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sat Oct 6 12:10:44 2007 @@ -150,7 +150,7 @@ external var_arg_function_type : lltype -> lltype array -> lltype = "llvm_var_arg_function_type" external is_var_arg : lltype -> bool = "llvm_is_var_arg" -external return_type : lltype -> lltype = "llvm_return_type" +external return_type : lltype -> lltype = "LLVMGetReturnType" external param_types : lltype -> lltype array = "llvm_param_types" (*--... Operations on struct types .........................................--*) @@ -161,10 +161,10 @@ (*--... Operations on pointer, vector, and array types .....................--*) external array_type : lltype -> int -> lltype = "llvm_array_type" -external pointer_type : lltype -> lltype = "llvm_pointer_type" +external pointer_type : lltype -> lltype = "LLVMPointerType" external vector_type : lltype -> int -> lltype = "llvm_vector_type" -external element_type : lltype -> lltype = "llvm_element_type" +external element_type : lltype -> lltype = "LLVMGetElementType" external array_length : lltype -> int = "llvm_array_length" external vector_size : lltype -> int = "llvm_vector_size" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42702&r1=42701&r2=42702&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sat Oct 6 12:10:44 2007 @@ -137,7 +137,7 @@ external var_arg_function_type : lltype -> lltype array -> lltype = "llvm_var_arg_function_type" external is_var_arg : lltype -> bool = "llvm_is_var_arg" -external return_type : lltype -> lltype = "llvm_return_type" +external return_type : lltype -> lltype = "LLVMGetReturnType" external param_types : lltype -> lltype array = "llvm_param_types" (*--... Operations on struct types .........................................--*) @@ -148,10 +148,10 @@ (*--... Operations on pointer, vector, and array types .....................--*) external array_type : lltype -> int -> lltype = "llvm_array_type" -external pointer_type : lltype -> lltype = "llvm_pointer_type" +external pointer_type : lltype -> lltype = "LLVMPointerType" external vector_type : lltype -> int -> lltype = "llvm_vector_type" -external element_type : lltype -> lltype = "llvm_element_type" +external element_type : lltype -> lltype = "LLVMGetElementType" external array_length : lltype -> int = "llvm_array_length" external vector_size : lltype -> int = "llvm_vector_size" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42702&r1=42701&r2=42702&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 12:10:44 2007 @@ -129,11 +129,6 @@ return Val_bool(LLVMIsFunctionVarArg(FunTy)); } -/* lltype -> lltype */ -CAMLprim LLVMTypeRef llvm_return_type(LLVMTypeRef FunTy) { - return LLVMGetReturnType(FunTy); -} - /* lltype -> lltype array */ CAMLprim value llvm_param_types(LLVMTypeRef FunTy) { value Tys = alloc(LLVMCountParamTypes(FunTy), 0); @@ -174,21 +169,11 @@ return LLVMArrayType(ElementTy, Int_val(Count)); } -/* lltype -> lltype */ -CAMLprim LLVMTypeRef llvm_pointer_type(LLVMTypeRef ElementTy) { - return LLVMPointerType(ElementTy); -} - /* lltype -> int -> lltype */ CAMLprim LLVMTypeRef llvm_vector_type(LLVMTypeRef ElementTy, value Count) { return LLVMVectorType(ElementTy, Int_val(Count)); } -/* lltype -> lltype */ -CAMLprim LLVMTypeRef llvm_element_type(LLVMTypeRef Ty) { - return LLVMGetElementType(Ty); -} - /* lltype -> int */ CAMLprim value llvm_array_length(LLVMTypeRef ArrayTy) { return Val_int(LLVMGetArrayLength(ArrayTy)); From gordonhenriksen at mac.com Sat Oct 6 16:00:37 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sat, 06 Oct 2007 21:00:37 -0000 Subject: [llvm-commits] [llvm] r42707 - in /llvm/trunk: Xcode/LLVM.xcodeproj/project.pbxproj bindings/ocaml/Makefile bindings/ocaml/Makefile.ocaml bindings/ocaml/analysis/ bindings/ocaml/analysis/Makefile bindings/ocaml/analysis/analysis_ocaml.c bindings/ocaml/analysis/llvm_analysis.ml bindings/ocaml/analysis/llvm_analysis.mli bindings/ocaml/bitwriter/Makefile bindings/ocaml/llvm/Makefile include/llvm-c/Analysis.h lib/Analysis/Analysis.cpp lib/Bitcode/Writer/BitWriter.cpp test/Bindings/Ocaml/analysis.ml Message-ID: <200710062100.l96L0bPa000996@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 16:00:36 2007 New Revision: 42707 URL: http://llvm.org/viewvc/llvm-project?rev=42707&view=rev Log: Bindings for the verifier. Added: llvm/trunk/bindings/ocaml/analysis/ (with props) llvm/trunk/bindings/ocaml/analysis/Makefile llvm/trunk/bindings/ocaml/analysis/analysis_ocaml.c llvm/trunk/bindings/ocaml/analysis/llvm_analysis.ml llvm/trunk/bindings/ocaml/analysis/llvm_analysis.mli llvm/trunk/include/llvm-c/Analysis.h llvm/trunk/lib/Analysis/Analysis.cpp llvm/trunk/test/Bindings/Ocaml/analysis.ml Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj llvm/trunk/bindings/ocaml/Makefile llvm/trunk/bindings/ocaml/Makefile.ocaml llvm/trunk/bindings/ocaml/bitwriter/Makefile llvm/trunk/bindings/ocaml/llvm/Makefile llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj (original) +++ llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj Sat Oct 6 16:00:36 2007 @@ -137,6 +137,11 @@ 9F7794870C73D51000551F9C /* LLVMBuilder.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = LLVMBuilder.h; sourceTree = ""; }; 9F7794880C73D51000551F9C /* MemoryBuffer.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = MemoryBuffer.h; sourceTree = ""; }; 9F7794890C73D51000551F9C /* Streams.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = Streams.h; sourceTree = ""; }; + 9F7C23E50CB81C2100498408 /* Analysis.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = Analysis.h; sourceTree = ""; }; + 9F7C23E60CB81C2B00498408 /* Analysis.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = Analysis.cpp; sourceTree = ""; }; + 9F7C240C0CB81ECD00498408 /* analysis_ocaml.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; path = analysis_ocaml.c; sourceTree = ""; }; + 9F7C240D0CB81ECD00498408 /* llvm_analysis.ml */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = llvm_analysis.ml; sourceTree = ""; }; + 9F7C240E0CB81ECD00498408 /* llvm_analysis.mli */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = llvm_analysis.mli; sourceTree = ""; }; 9FA638D90C77B184007F12AE /* AutoUpgrade.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = AutoUpgrade.h; sourceTree = ""; }; 9FA638DA0C77B184007F12AE /* GlobalAlias.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = GlobalAlias.h; sourceTree = ""; }; 9FA638DB0C77B1AB007F12AE /* APInt.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = APInt.h; sourceTree = ""; }; @@ -166,7 +171,6 @@ 9FE25D900CAB166D005383FC /* APFloat.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = APFloat.h; sourceTree = ""; }; 9FE25D910CAB166D005383FC /* SparseBitVector.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = SparseBitVector.h; sourceTree = ""; }; 9FE25D920CAB169F005383FC /* RegisterCoalescer.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = RegisterCoalescer.h; sourceTree = ""; }; - 9FE25D930CAB16D8005383FC /* PostDominatorCalculation.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = PostDominatorCalculation.h; sourceTree = ""; }; 9FE25D940CAB16FB005383FC /* RegisterCoalescer.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = RegisterCoalescer.cpp; sourceTree = ""; }; 9FE25D950CAB1724005383FC /* APFloat.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = APFloat.cpp; sourceTree = ""; }; 9FE25D960CAB1759005383FC /* TargetCallingConv.td */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = TargetCallingConv.td; sourceTree = ""; }; @@ -1156,6 +1160,16 @@ path = MSIL; sourceTree = ""; }; + 9F7C240B0CB81ECD00498408 /* analysis */ = { + isa = PBXGroup; + children = ( + 9F7C240C0CB81ECD00498408 /* analysis_ocaml.c */, + 9F7C240D0CB81ECD00498408 /* llvm_analysis.ml */, + 9F7C240E0CB81ECD00498408 /* llvm_analysis.mli */, + ); + path = analysis; + sourceTree = ""; + }; 9FA638E90C77B252007F12AE /* IPO */ = { isa = PBXGroup; children = ( @@ -1176,6 +1190,7 @@ 9FD3E56F0CA0116100E54D15 /* ocaml */ = { isa = PBXGroup; children = ( + 9F7C240B0CB81ECD00498408 /* analysis */, 9FD3E5700CA0116100E54D15 /* bitwriter */, 9FD3E57A0CA0116100E54D15 /* llvm */, ); @@ -1217,6 +1232,7 @@ CF8F1B480B64F7AB00BB4199 /* include/llvm-c */ = { isa = PBXGroup; children = ( + 9F7C23E50CB81C2100498408 /* Analysis.h */, 9FD3E58D0CA0125F00E54D15 /* BitWriter.h */, 9FD3E58E0CA0125F00E54D15 /* Core.h */, CF8F1B490B64F7AB00BB4199 /* LinkTimeOptimizer.h */, @@ -1381,15 +1397,13 @@ DE66ECBD08ABEC0700323D32 /* lib/Analysis */ = { isa = PBXGroup; children = ( - 9FE25D930CAB16D8005383FC /* PostDominatorCalculation.h */, - 9F68EB010C77AD02004AA152 /* LoopPass.cpp */, - 9F68EB020C77AD02004AA152 /* MemoryDependenceAnalysis.cpp */, DE66ED1A08ABEC0800323D32 /* IPA */, DE66ECBE08ABEC0700323D32 /* AliasAnalysis.cpp */, DE66ECBF08ABEC0700323D32 /* AliasAnalysisCounter.cpp */, DE66ECC008ABEC0700323D32 /* AliasAnalysisEvaluator.cpp */, CF8F1B4D0B64F80700BB4199 /* AliasDebugger.cpp */, DE66ECC108ABEC0700323D32 /* AliasSetTracker.cpp */, + 9F7C23E60CB81C2B00498408 /* Analysis.cpp */, DE66ECC208ABEC0700323D32 /* BasicAliasAnalysis.cpp */, DE66ECC308ABEC0700323D32 /* CFGPrinter.cpp */, CF73C0B0098A523C00627152 /* ConstantFolding.cpp */, @@ -1398,6 +1412,8 @@ DE66ED1908ABEC0800323D32 /* IntervalPartition.cpp */, DE66ED3308ABEC0800323D32 /* LoadValueNumbering.cpp */, DE66ED3408ABEC0800323D32 /* LoopInfo.cpp */, + 9F68EB010C77AD02004AA152 /* LoopPass.cpp */, + 9F68EB020C77AD02004AA152 /* MemoryDependenceAnalysis.cpp */, DE66ED3608ABEC0800323D32 /* PostDominators.cpp */, DE66ED3708ABEC0800323D32 /* ProfileInfo.cpp */, DE66ED3808ABEC0800323D32 /* ProfileInfoLoader.cpp */, Modified: llvm/trunk/bindings/ocaml/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/Makefile?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/Makefile (original) +++ llvm/trunk/bindings/ocaml/Makefile Sat Oct 6 16:00:36 2007 @@ -8,6 +8,6 @@ ##===----------------------------------------------------------------------===## LEVEL := ../.. -DIRS = llvm bitwriter +DIRS = llvm bitwriter analysis include $(LEVEL)/Makefile.common Modified: llvm/trunk/bindings/ocaml/Makefile.ocaml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/Makefile.ocaml?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/Makefile.ocaml (original) +++ llvm/trunk/bindings/ocaml/Makefile.ocaml Sat Oct 6 16:00:36 2007 @@ -10,6 +10,11 @@ # An ocaml library is a unique project type in the context of LLVM, so rules are # here rather than in Makefile.rules. # +# Reference materials on installing ocaml libraries: +# +# https://fedoraproject.org/wiki/Packaging/OCaml +# http://pkg-ocaml-maint.alioth.debian.org/ocaml_packaging_policy.txt +# ##===----------------------------------------------------------------------===## include $(LEVEL)/Makefile.config @@ -188,7 +193,6 @@ $(Verb) $(MKDIR) $(PROJ_libocamldir) $(Verb) $(DataInstall) $(OutputCMA) "$(DestCMA)" $(Verb) for i in $(UsedLibNames); do \ - $(EchoCmd) "Installing $(BuildMode) $(PROJ_libocamldir)/$$i"; \ ln -sf "$(PROJ_libdir)/$$i" "$(PROJ_libocamldir)/$$i"; \ done @@ -196,7 +200,6 @@ $(Echo) "Uninstalling $(DestCMA)" -$(Verb) $(RM) -f $(DestCMA) $(Verb) for i in $(UsedLibNames); do \ - $(EchoCmd) "Uninstalling $(PROJ_libocamldir)/$$i"; \ $(RM) -f "$(PROJ_libocamldir)/$$i"; \ done Propchange: llvm/trunk/bindings/ocaml/analysis/ ------------------------------------------------------------------------------ --- svn:ignore (added) +++ svn:ignore Sat Oct 6 16:00:36 2007 @@ -0,0 +1,2 @@ +Debug +Release Added: llvm/trunk/bindings/ocaml/analysis/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/analysis/Makefile?rev=42707&view=auto ============================================================================== --- llvm/trunk/bindings/ocaml/analysis/Makefile (added) +++ llvm/trunk/bindings/ocaml/analysis/Makefile Sat Oct 6 16:00:36 2007 @@ -0,0 +1,20 @@ +##===- bindings/ocaml/analysis/Makefile --------------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file was developed by Gordon Henriksen and is distributed under the +# University of Illinois Open Source License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## +# +# This is the makefile for the Objective Caml Llvm_analysis interface. +# +##===----------------------------------------------------------------------===## + +LEVEL := ../../.. +LIBRARYNAME := llvm_analysis +DONT_BUILD_RELINKED := 1 +UsedComponents := analysis +UsedOcamlInterfaces := llvm + +include ../Makefile.ocaml Added: llvm/trunk/bindings/ocaml/analysis/analysis_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/analysis/analysis_ocaml.c?rev=42707&view=auto ============================================================================== --- llvm/trunk/bindings/ocaml/analysis/analysis_ocaml.c (added) +++ llvm/trunk/bindings/ocaml/analysis/analysis_ocaml.c Sat Oct 6 16:00:36 2007 @@ -0,0 +1,60 @@ +/*===-- analysis_ocaml.c - LLVM Ocaml Glue ----------------------*- C++ -*-===*\ +|* *| +|* The LLVM Compiler Infrastructure *| +|* *| +|* This file was developed by Gordon Henriksen and is distributed under the *| +|* University of Illinois Open Source License. See LICENSE.TXT for details. *| +|* *| +|*===----------------------------------------------------------------------===*| +|* *| +|* This file glues LLVM's ocaml interface to its C interface. These functions *| +|* are by and large transparent wrappers to the corresponding C functions. *| +|* *| +|* Note that these functions intentionally take liberties with the CAMLparamX *| +|* macros, since most of the parameters are not GC heap objects. *| +|* *| +\*===----------------------------------------------------------------------===*/ + +#include "llvm-c/Analysis.h" +#include "caml/alloc.h" +#include "caml/mlvalues.h" +#include "caml/memory.h" + + +/* Llvm.llmodule -> string option */ +CAMLprim value llvm_verify_module(LLVMModuleRef M) { + CAMLparam0(); + CAMLlocal2(String, Option); + + char *Message; + int Result = LLVMVerifyModule(M, LLVMReturnStatusAction, &Message); + + if (0 == Result) { + Option = Val_int(0); + } else { + Option = alloc(1, 1); + String = copy_string(Message); + Store_field(Option, 0, String); + } + + LLVMDisposeVerifierMessage(Message); + + CAMLreturn(Option); +} + +/* Llvm.llvalue -> bool */ +CAMLprim value llvm_verify_function(LLVMValueRef Fn) { + return Val_bool(LLVMVerifyFunction(Fn, LLVMReturnStatusAction) == 0); +} + +/* Llvm.llmodule -> unit */ +CAMLprim value llvm_assert_valid_module(LLVMModuleRef M) { + LLVMVerifyModule(M, LLVMAbortProcessAction, 0); + return Val_unit; +} + +/* Llvm.llvalue -> unit */ +CAMLprim value llvm_assert_valid_function(LLVMValueRef Fn) { + LLVMVerifyFunction(Fn, LLVMAbortProcessAction); + return Val_unit; +} Added: llvm/trunk/bindings/ocaml/analysis/llvm_analysis.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/analysis/llvm_analysis.ml?rev=42707&view=auto ============================================================================== --- llvm/trunk/bindings/ocaml/analysis/llvm_analysis.ml (added) +++ llvm/trunk/bindings/ocaml/analysis/llvm_analysis.ml Sat Oct 6 16:00:36 2007 @@ -0,0 +1,24 @@ +(*===-- llvm_analysis.ml - LLVM Ocaml Interface -----------------*- C++ -*-===* + * + * The LLVM Compiler Infrastructure + * + * This file was developed by Gordon Henriksen and is distributed under the + * University of Illinois Open Source License. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------=== + * + * This interface provides an ocaml API for LLVM IR analyses, the classes in + * the Analysis library. + * + *===----------------------------------------------------------------------===*) + + +external verify_module : Llvm.llmodule -> string option = "llvm_verify_module" + +external verify_function : Llvm.llvalue -> bool = "llvm_verify_function" + +external assert_valid_module : Llvm.llmodule -> unit + = "llvm_assert_valid_module" + +external assert_valid_function : Llvm.llvalue -> unit + = "llvm_assert_valid_function" Added: llvm/trunk/bindings/ocaml/analysis/llvm_analysis.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/analysis/llvm_analysis.mli?rev=42707&view=auto ============================================================================== --- llvm/trunk/bindings/ocaml/analysis/llvm_analysis.mli (added) +++ llvm/trunk/bindings/ocaml/analysis/llvm_analysis.mli Sat Oct 6 16:00:36 2007 @@ -0,0 +1,24 @@ +(*===-- llvm_analysis.mli - LLVM Ocaml Interface ----------------*- C++ -*-===* + * + * The LLVM Compiler Infrastructure + * + * This file was developed by Gordon Henriksen and is distributed under the + * University of Illinois Open Source License. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------=== + * + * This interface provides an ocaml API for LLVM IR analyses, the classes in + * the Analysis library. + * + *===----------------------------------------------------------------------===*) + + +external verify_module : Llvm.llmodule -> string option = "llvm_verify_module" + +external verify_function : Llvm.llvalue -> bool = "llvm_verify_function" + +external assert_valid_module : Llvm.llmodule -> unit + = "llvm_assert_valid_module" + +external assert_valid_function : Llvm.llvalue -> unit + = "llvm_assert_valid_function" Modified: llvm/trunk/bindings/ocaml/bitwriter/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/bitwriter/Makefile?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/bitwriter/Makefile (original) +++ llvm/trunk/bindings/ocaml/bitwriter/Makefile Sat Oct 6 16:00:36 2007 @@ -1,17 +1,13 @@ -##===- bindings/ocaml/llvm/Makefile ------------------------*- Makefile -*-===## +##===- bindings/ocaml/bitwriter/Makefile -------------------*- Makefile -*-===## # # The LLVM Compiler Infrastructure # -# This file was developed by the LLVM research group and is distributed under -# the University of Illinois Open Source License. See LICENSE.TXT for details. +# This file was developed by Gordon Henriksen and is distributed under the +# University of Illinois Open Source License. See LICENSE.TXT for details. # ##===----------------------------------------------------------------------===## # -# This is the makefile for the llvm-ml interface. Reference materials on -# installing ocaml libraries: -# -# https://fedoraproject.org/wiki/Packaging/OCaml -# http://pkg-ocaml-maint.alioth.debian.org/ocaml_packaging_policy.txt +# This is the makefile for the Objective Caml Llvm_bitwriter interface. # ##===----------------------------------------------------------------------===## Modified: llvm/trunk/bindings/ocaml/llvm/Makefile URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/Makefile?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/Makefile (original) +++ llvm/trunk/bindings/ocaml/llvm/Makefile Sat Oct 6 16:00:36 2007 @@ -2,16 +2,12 @@ # # The LLVM Compiler Infrastructure # -# This file was developed by the LLVM research group and is distributed under -# the University of Illinois Open Source License. See LICENSE.TXT for details. +# This file was developed by Gordon Henriksen and is distributed under the +# University of Illinois Open Source License. See LICENSE.TXT for details. # ##===----------------------------------------------------------------------===## # -# This is the makefile for the llvm-ml interface. Reference materials on -# installing ocaml libraries: -# -# https://fedoraproject.org/wiki/Packaging/OCaml -# http://pkg-ocaml-maint.alioth.debian.org/ocaml_packaging_policy.txt +# This is the makefile for the Objective Caml Llvm interface. # ##===----------------------------------------------------------------------===## Added: llvm/trunk/include/llvm-c/Analysis.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Analysis.h?rev=42707&view=auto ============================================================================== --- llvm/trunk/include/llvm-c/Analysis.h (added) +++ llvm/trunk/include/llvm-c/Analysis.h Sat Oct 6 16:00:36 2007 @@ -0,0 +1,53 @@ +/*===-- llvm-c/Analysis.h - Analysis Library C Interface --------*- C++ -*-===*\ +|* *| +|* The LLVM Compiler Infrastructure *| +|* *| +|* This file was developed by Gordon Henriksen and is distributed under the *| +|* University of Illinois Open Source License. See LICENSE.TXT for details. *| +|* *| +|*===----------------------------------------------------------------------===*| +|* *| +|* This header declares the C interface to libLLVMAnalysis.a, which *| +|* implements various analyses of the LLVM IR. *| +|* *| +|* Many exotic languages can interoperate with C code but have a harder time *| +|* with C++ due to name mangling. So in addition to C, this interface enables *| +|* tools written in such languages. *| +|* *| +\*===----------------------------------------------------------------------===*/ + +#ifndef LLVM_C_ANALYSIS_H +#define LLVM_C_ANALYSIS_H + +#include "llvm-c/Core.h" + +#ifdef __cplusplus +extern "C" { +#endif + + +typedef enum { + LLVMAbortProcessAction, /* verifier will print to stderr and abort() */ + LLVMPrintMessageAction, /* verifier will print to stderr and return 1 */ + LLVMReturnStatusAction /* verifier will just return 1 */ +} LLVMVerifierFailureAction; + + +/* Verifies that a module is valid, taking the specified action if not. + Optionally returns a human-readable description of any invalid constructs. */ +int LLVMVerifyModule(LLVMModuleRef M, LLVMVerifierFailureAction Action, + char **OutMessage); + +/* Disposes of the message allocated by the verifier, if any. */ +void LLVMDisposeVerifierMessage(char *Message); + +/* Verifies that a single function is valid, taking the specified action. Useful + for debugging. */ +int LLVMVerifyFunction(LLVMValueRef Fn, LLVMVerifierFailureAction Action); + + +#ifdef __cplusplus +} +#endif + +#endif Added: llvm/trunk/lib/Analysis/Analysis.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/Analysis.cpp?rev=42707&view=auto ============================================================================== --- llvm/trunk/lib/Analysis/Analysis.cpp (added) +++ llvm/trunk/lib/Analysis/Analysis.cpp Sat Oct 6 16:00:36 2007 @@ -0,0 +1,38 @@ +//===-- Analysis.cpp ------------------------------------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file was developed by Gordon Henriksen and is distributed under the +// University of Illinois Open Source License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "llvm-c/Analysis.h" +#include "llvm/Analysis/Verifier.h" +#include + +using namespace llvm; + +int LLVMVerifyModule(LLVMModuleRef M, LLVMVerifierFailureAction Action, + char **OutMessages) { + std::string Messages; + + int Result = verifyModule(*unwrap(M), + static_cast(Action), + OutMessages? &Messages : 0); + + if (OutMessages) + *OutMessages = strdup(Messages.c_str()); + + return Result; +} + +void LLVMDisposeVerifierMessage(char *Message) { + free(Message); +} + +int LLVMVerifyFunction(LLVMValueRef Fn, LLVMVerifierFailureAction Action) { + return verifyFunction(*unwrap(Fn), + static_cast(Action)); +} + Modified: llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp?rev=42707&r1=42706&r2=42707&view=diff ============================================================================== --- llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp (original) +++ llvm/trunk/lib/Bitcode/Writer/BitWriter.cpp Sat Oct 6 16:00:36 2007 @@ -2,8 +2,8 @@ // // The LLVM Compiler Infrastructure // -// This file was developed by the LLVM research group and is distributed under -// the University of Illinois Open Source License. See LICENSE.TXT for details. +// This file was developed by Gordon Henriksen and is distributed under the +// University of Illinois Open Source License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// Added: llvm/trunk/test/Bindings/Ocaml/analysis.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/analysis.ml?rev=42707&view=auto ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/analysis.ml (added) +++ llvm/trunk/test/Bindings/Ocaml/analysis.ml Sat Oct 6 16:00:36 2007 @@ -0,0 +1,46 @@ +(* RUN: %ocamlc llvm.cma llvm_analysis.cma %s -o %t + * RUN: ./%t %t.bc + *) + +open Llvm +open Llvm_analysis + +(* Note that this takes a moment to link, so it's best to keep the number of + individual tests low. *) + +let test x = if not x then exit 1 else () + +let bomb msg = + prerr_endline msg; + exit 2 + +let _ = + let fty = function_type void_type [| |] in + let m = create_module "valid_m" in + let fn = define_function "valid_fn" fty m in + let at_entry = builder_at_end (entry_block fn) in + ignore (build_ret_void at_entry); + + + (* Test that valid constructs verify. *) + match verify_module m with + Some msg -> bomb "valid module failed verification!" + | None -> (); + + if not (verify_function fn) then bomb "valid function failed verification!"; + + + (* Test that invalid constructs do not verify. + A basic block can contain only one terminator instruction. *) + ignore (build_ret_void at_entry); + + match verify_module m with + Some msg -> () + | None -> bomb "invalid module passed verification!"; + + if verify_function fn then bomb "invalid function passed verification!"; + + + dispose_module m + + (* Don't bother to test assert_valid_{module,function}. *) From evan.cheng at apple.com Sat Oct 6 17:07:19 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Sat, 06 Oct 2007 15:07:19 -0700 Subject: [llvm-commits] [llvm] r42661 - in /llvm/trunk/lib/Target/X86: X86InstrInfo.cpp X86InstrInfo.td X86InstrX86-64.td In-Reply-To: <54364F06-F5DE-457D-BC58-C8BF09A30196@apple.com> References: <200710052313.l95NDLI5012935@zion.cs.uiuc.edu> <54364F06-F5DE-457D-BC58-C8BF09A30196@apple.com> Message-ID: <9456FE96-BACF-4DBC-A645-6047DEA8F855@apple.com> Not crazy at all. If I had any sense at all I would have done that back then. :) I'll make that clean up one of these days. Evan On Oct 5, 2007, at 4:45 PM, Chris Lattner wrote: >> URL: http://llvm.org/viewvc/llvm-project?rev=42661&view=rev >> Log: >> Commute x86 cmove instructions by swapping the operands and change >> the condition >> to its inverse. >> Testing this as llcbeta > > Random crazy idea: > >> +def CMOVB16rm : I<0x42, MRMSrcMem, // if > + (outs GR16:$dst), (ins GR16:$src1, i16mem:$src2), >> + "cmovb\t{$src2, $dst|$dst, $src2}", >> + [(set GR16:$dst, (X86cmov GR16:$src1, (loadi16 >> addr:$src2), >> + X86_COND_B, EFLAGS))]>, >> + TB, OpSize; > > Instead of having one instruction for each comparison code, would it > make sense to have one "CMOV16rm" that took the condcode as an > immediate argument? If you had that, you could change the comparison > just by twiddling the immediate operand and wouldn't have a bazillion > CMOV instructions. > > -Chris > > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From evan.cheng at apple.com Sat Oct 6 17:13:50 2007 From: evan.cheng at apple.com (Evan Cheng) Date: Sat, 06 Oct 2007 15:13:50 -0700 Subject: [llvm-commits] [llvm] r42677 - in /llvm/trunk/lib: CodeGen/SelectionDAG/DAGCombiner.cpp Target/X86/X86InstrSSE.td In-Reply-To: <44ADE7F2-D285-425A-B78A-41A874D3F698@apple.com> References: <200710060246.l962kT9R020741@zion.cs.uiuc.edu> <44ADE7F2-D285-425A-B78A-41A874D3F698@apple.com> Message-ID: <77D3FD56-30C6-4CC6-B9F2-CDF4B1EF6426@apple.com> On Oct 5, 2007, at 10:11 PM, Chris Lattner wrote: > Evan, it looks like you missed a commit. Would a similar approach be > suitable to implement this X86/README-SSE.txt entry?: Looks that way. Some of the x86 specific shuffle optmizations (and missing ones like this) probably should move to the combiner. Evan > > > + > +These functions should produce the same code: > + > +#include > + > +typedef long long __m128i __attribute__ ((__vector_size__ (16))); > + > +int foo(__m128i* val) { > + return __builtin_ia32_vec_ext_v4si(*val, 1); > +} > +int bar(__m128i* val) { > + union vs { > + __m128i *_v; > + int* _s; > + } v = {val}; > + return v._s[1]; > +} > + > +We currently produce (with -m64): > + > +_foo: > + pshufd $1, (%rdi), %xmm0 > + movd %xmm0, %eax > + ret > +_bar: > + movl 4(%rdi), %eax > + ret > + > > -Chris > > > On Oct 5, 2007, at 7:46 PM, Evan Cheng wrote: > >> Author: evancheng >> Date: Fri Oct 5 21:46:29 2007 >> New Revision: 42677 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=42677&view=rev >> Log: >> Added DAG xforms. e.g. >> (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) >> (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load >> $addr) >> Remove x86 specific patterns. >> >> Modified: >> llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp >> llvm/trunk/lib/Target/X86/X86InstrSSE.td >> >> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp >> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/ >> SelectionDAG/DAGCombiner.cpp?rev=42677&r1=42676&r2=42677&view=diff >> >> === >> =================================================================== >> ======== >> --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original) >> +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Fri Oct 5 >> 21:46:29 2007 >> @@ -263,6 +263,7 @@ >> SDOperand visitLOAD(SDNode *N); >> SDOperand visitSTORE(SDNode *N); >> SDOperand visitINSERT_VECTOR_ELT(SDNode *N); >> + SDOperand visitEXTRACT_VECTOR_ELT(SDNode *N); >> SDOperand visitBUILD_VECTOR(SDNode *N); >> SDOperand visitCONCAT_VECTORS(SDNode *N); >> SDOperand visitVECTOR_SHUFFLE(SDNode *N); >> @@ -682,6 +683,7 @@ >> case ISD::LOAD: return visitLOAD(N); >> case ISD::STORE: return visitSTORE(N); >> case ISD::INSERT_VECTOR_ELT: return visitINSERT_VECTOR_ELT(N); >> + case ISD::EXTRACT_VECTOR_ELT: return visitEXTRACT_VECTOR_ELT(N); >> case ISD::BUILD_VECTOR: return visitBUILD_VECTOR(N); >> case ISD::CONCAT_VECTORS: return visitCONCAT_VECTORS(N); >> case ISD::VECTOR_SHUFFLE: return visitVECTOR_SHUFFLE(N); >> @@ -2907,9 +2909,8 @@ >> return DAG.getNode(ISD::BIT_CONVERT, VT, N0.getOperand(0)); >> >> // fold (conv (load x)) -> (load (conv*)x) >> - // If the resultant load doesn't need a higher alignment than >> the original! >> - if (ISD::isNON_EXTLoad(N0.Val) && N0.hasOneUse() && >> - ISD::isUNINDEXEDLoad(N0.Val) && >> + // If the resultant load doesn't need a higher alignment than >> the original! >> + if (ISD::isNormalLoad(N0.Val) && N0.hasOneUse() && >> TLI.isOperationLegal(ISD::LOAD, VT)) { >> LoadSDNode *LN0 = cast(N0); >> unsigned Align = TLI.getTargetMachine().getTargetData()-> >> @@ -3901,6 +3902,54 @@ >> return SDOperand(); >> } >> >> +SDOperand DAGCombiner::visitEXTRACT_VECTOR_ELT(SDNode *N) { >> + SDOperand InVec = N->getOperand(0); >> + SDOperand EltNo = N->getOperand(1); >> + >> + // (vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) >> + // (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 >> load $addr) >> + if (isa(EltNo)) { >> + unsigned Elt = cast(EltNo)->getValue(); >> + bool NewLoad = false; >> + if (Elt == 0) { >> + MVT::ValueType VT = InVec.getValueType(); >> + MVT::ValueType EVT = MVT::getVectorElementType(VT); >> + MVT::ValueType LVT = EVT; >> + unsigned NumElts = MVT::getVectorNumElements(VT); >> + if (InVec.getOpcode() == ISD::BIT_CONVERT) { >> + MVT::ValueType BCVT = InVec.getOperand(0).getValueType(); >> + if (NumElts != MVT::getVectorNumElements(BCVT)) >> + return SDOperand(); >> + InVec = InVec.getOperand(0); >> + EVT = MVT::getVectorElementType(BCVT); >> + NewLoad = true; >> + } >> + if (InVec.getOpcode() == ISD::SCALAR_TO_VECTOR && >> + InVec.getOperand(0).getValueType() == EVT && >> + ISD::isNormalLoad(InVec.getOperand(0).Val) && >> + InVec.getOperand(0).hasOneUse()) { >> + LoadSDNode *LN0 = cast(InVec.getOperand(0)); >> + unsigned Align = LN0->getAlignment(); >> + if (NewLoad) { >> + // Check the resultant load doesn't need a higher >> alignment than the >> + // original load. >> + unsigned NewAlign = TLI.getTargetMachine().getTargetData >> ()-> >> + getABITypeAlignment(MVT::getTypeForValueType(LVT)); >> + if (!TLI.isOperationLegal(ISD::LOAD, LVT) || NewAlign > >> Align) >> + return SDOperand(); >> + Align = NewAlign; >> + } >> + >> + return DAG.getLoad(LVT, LN0->getChain(), LN0->getBasePtr(), >> + LN0->getSrcValue(), LN0- >>> getSrcValueOffset(), >> + LN0->isVolatile(), Align); >> + } >> + } >> + } >> + return SDOperand(); >> +} >> + >> + >> SDOperand DAGCombiner::visitBUILD_VECTOR(SDNode *N) { >> unsigned NumInScalars = N->getNumOperands(); >> MVT::ValueType VT = N->getValueType(0); >> >> Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td >> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/ >> X86InstrSSE.td?rev=42677&r1=42676&r2=42677&view=diff >> >> === >> =================================================================== >> ======== >> --- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original) >> +++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Fri Oct 5 21:46:29 2007 >> @@ -2942,11 +2942,3 @@ >> (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; >> def : Pat<(store (v16i8 VR128:$src), addr:$dst), >> (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; >> - >> -// (vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 >> load $addr) >> -def : Pat<(vector_extract >> - (bc_v4i32 (v4f32 (scalar_to_vector (loadf32 addr: >> $src)))), (iPTR 0)), >> - (MOV32rm addr:$src)>, Requires<[HasSSE2]>; >> -def : Pat<(vector_extract >> - (bc_v2i64 (v2f64 (scalar_to_vector (loadf64 addr: >> $src)))), (iPTR 0)), >> - (MOV64rm addr:$src)>, Requires<[HasSSE2, In64BitMode]>; >> >> >> _______________________________________________ >> llvm-commits mailing list >> llvm-commits at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits > > _______________________________________________ > llvm-commits mailing list > llvm-commits at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits From gordonhenriksen at mac.com Sat Oct 6 19:13:35 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sun, 07 Oct 2007 00:13:35 -0000 Subject: [llvm-commits] [llvm] r42713 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp test/Bindings/Ocaml/vmcore.ml Message-ID: <200710070013.l970Da4c007671@zion.cs.uiuc.edu> Author: gordon Date: Sat Oct 6 19:13:35 2007 New Revision: 42713 URL: http://llvm.org/viewvc/llvm-project?rev=42713&view=rev Log: C and Objective Caml bindings for PATypeHolder. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sat Oct 6 19:13:35 2007 @@ -16,6 +16,7 @@ (* These abstract types correlate directly to the LLVM VMCore classes. *) type llmodule type lltype +type lltypehandle type llvalue type llbasicblock (* These are actually values, but benefit from type checking. *) @@ -176,6 +177,11 @@ let void_type = _void_type () let label_type = _label_type () +(*--... Operations on type handles .........................................--*) +external handle_to_type : lltype -> lltypehandle = "llvm_handle_to_type" +external type_of_handle : lltypehandle -> lltype = "llvm_type_of_handle" +external refine_type : lltype -> lltype -> unit = "llvm_refine_type" + (*===-- Values ------------------------------------------------------------===*) Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sat Oct 6 19:13:35 2007 @@ -16,6 +16,7 @@ (* These abstract types correlate directly to the LLVM VMCore classes. *) type llmodule type lltype +type lltypehandle type llvalue type llbasicblock (* These are actually values, but benefit from type checking. *) @@ -160,6 +161,11 @@ val void_type : lltype val label_type : lltype +(*--... Operations on type handles .........................................--*) +external handle_to_type : lltype -> lltypehandle = "llvm_handle_to_type" +external type_of_handle : lltypehandle -> lltype = "llvm_type_of_handle" +external refine_type : lltype -> lltype -> unit = "llvm_refine_type" + (*===-- Values ------------------------------------------------------------===*) external type_of : llvalue -> lltype = "llvm_type_of" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sat Oct 6 19:13:35 2007 @@ -195,6 +195,38 @@ return LLVMOpaqueType(); } +/*--... Operations on type handles .........................................--*/ + +#define Typehandle_val(v) (*(LLVMTypeHandleRef *)(Data_custom_val(v))) + +void llvm_finalize_handle(value TH) { + LLVMDisposeTypeHandle(Typehandle_val(TH)); +} + +static struct custom_operations typehandle_ops = { + (char *) "LLVMTypeHandle", + llvm_finalize_handle, + custom_compare_default, + custom_hash_default, + custom_serialize_default, + custom_deserialize_default +}; + +CAMLprim value llvm_handle_to_type(LLVMTypeRef PATy) { + value TH = alloc_custom(&typehandle_ops, sizeof(LLVMBuilderRef), 0, 1); + Typehandle_val(TH) = LLVMCreateTypeHandle(PATy); + return TH; +} + +CAMLprim LLVMTypeRef llvm_type_of_handle(value TH) { + return LLVMResolveTypeHandle(Typehandle_val(TH)); +} + +CAMLprim value llvm_refine_type(LLVMTypeRef AbstractTy, LLVMTypeRef ConcreteTy){ + LLVMRefineAbstractType(AbstractTy, ConcreteTy); + return Val_unit; +} + /*===-- VALUES ------------------------------------------------------------===*/ Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sat Oct 6 19:13:35 2007 @@ -47,6 +47,7 @@ /* Opaque types. */ typedef struct LLVMOpaqueModule *LLVMModuleRef; typedef struct LLVMOpaqueType *LLVMTypeRef; +typedef struct LLVMOpaqueTypeHandle *LLVMTypeHandleRef; typedef struct LLVMOpaqueValue *LLVMValueRef; typedef struct LLVMOpaqueBasicBlock *LLVMBasicBlockRef; typedef struct LLVMOpaqueBuilder *LLVMBuilderRef; @@ -204,6 +205,12 @@ LLVMTypeRef LLVMLabelType(); LLVMTypeRef LLVMOpaqueType(); +/* Operations on type handles */ +LLVMTypeHandleRef LLVMCreateTypeHandle(LLVMTypeRef PotentiallyAbstractTy); +void LLVMRefineType(LLVMTypeRef AbstractTy, LLVMTypeRef ConcreteTy); +LLVMTypeRef LLVMResolveTypeHandle(LLVMTypeHandleRef TypeHandle); +void LLVMDisposeTypeHandle(LLVMTypeHandleRef TypeHandle); + /*===-- Values ------------------------------------------------------------===*/ @@ -558,6 +565,16 @@ inline LLVMBuilderRef wrap(LLVMBuilder *B) { return reinterpret_cast(B); } + + /* Opaque type handle conversions. + */ + inline PATypeHolder *unwrap(LLVMTypeHandleRef B) { + return reinterpret_cast(B); + } + + inline LLVMTypeHandleRef wrap(PATypeHolder *B) { + return reinterpret_cast(B); + } } #endif /* !defined(__cplusplus) */ Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sat Oct 6 19:13:35 2007 @@ -177,6 +177,24 @@ return wrap(llvm::OpaqueType::get()); } +/* Operations on type handles */ + +LLVMTypeHandleRef LLVMCreateTypeHandle(LLVMTypeRef PotentiallyAbstractTy) { + return wrap(new PATypeHolder(unwrap(PotentiallyAbstractTy))); +} + +void LLVMDisposeTypeHandle(LLVMTypeHandleRef TypeHandle) { + delete unwrap(TypeHandle); +} + +LLVMTypeRef LLVMResolveTypeHandle(LLVMTypeHandleRef TypeHandle) { + return wrap(unwrap(TypeHandle)->get()); +} + +void LLVMRefineType(LLVMTypeRef AbstractTy, LLVMTypeRef ConcreteTy) { + unwrap(AbstractTy)->refineAbstractTypeTo(unwrap(ConcreteTy)); +} + /*===-- Operations on values ----------------------------------------------===*/ Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42713&r1=42712&r2=42713&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sat Oct 6 19:13:35 2007 @@ -131,7 +131,17 @@ group "delete"; let ty = opaque_type () in insist (define_type_name "Ty13" ty m); - delete_type_name "Ty13" m + delete_type_name "Ty13" m; + + (* RUN: grep -v {RecursiveTy.*RecursiveTy} < %t.ll + *) + group "recursive"; + let ty = opaque_type () in + let th = handle_to_type ty in + refine_type ty (pointer_type ty); + let ty = type_of_handle th in + insist (define_type_name "RecursiveTy" ty m); + insist (ty == element_type ty) (*===-- Constants ---------------------------------------------------------===*) From clattner at apple.com Sat Oct 6 21:01:54 2007 From: clattner at apple.com (Chris Lattner) Date: Sat, 6 Oct 2007 19:01:54 -0700 Subject: [llvm-commits] [test-suite] r42694 - in /test-suite/trunk/MultiSource/Benchmarks/MiBench/telecomm-gsm: config.h toast.c In-Reply-To: <200710061055.l96AtMYO009463@zion.cs.uiuc.edu> References: <200710061055.l96AtMYO009463@zion.cs.uiuc.edu> Message-ID: On Oct 6, 2007, at 3:55 AM, Anton Korobeynikov wrote: > Author: asl > Date: Sat Oct 6 05:55:20 2007 > New Revision: 42694 > > URL: http://llvm.org/viewvc/llvm-project?rev=42694&view=rev > Log: > Unbreak toast again for mingw32. > Funny, but we have two toast's in in llvm-test. Maybe we should > drop one? Where is the other toast? -Chris From clattner at apple.com Sat Oct 6 21:03:09 2007 From: clattner at apple.com (Chris Lattner) Date: Sat, 6 Oct 2007 19:03:09 -0700 Subject: [llvm-commits] [llvm] r42699 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp In-Reply-To: <200710061617.l96GHnve022368@zion.cs.uiuc.edu> References: <200710061617.l96GHnve022368@zion.cs.uiuc.edu> Message-ID: <6979D9B6-71EA-4EF2-9E81-A6715A7B7978@apple.com> > URL: http://llvm.org/viewvc/llvm-project?rev=42699&view=rev > Log: > Move merge code into new helper function. Nice: > +// mergeSPUpdatesUp - Merge two stack-manipulating instructions > upper iterator. > +static > +void mergeSPUpdatesUp(MachineBasicBlock &MBB, > MachineBasicBlock::iterator &MBBI, > + unsigned StackPtr, uint64_t *NumBytes = NULL) { > + if (MBBI != MBB.begin()) { How about starting this out with: if (MBBI == MBB.begin()) return; to unnest the function body one level? -Chris From dberlin at dberlin.org Sat Oct 6 23:47:58 2007 From: dberlin at dberlin.org (Daniel Berlin) Date: Sun, 07 Oct 2007 04:47:58 -0000 Subject: [llvm-commits] [llvm] r42717 - /llvm/trunk/include/llvm/ADT/SparseBitVector.h Message-ID: <200710070447.l974lwEu015290@zion.cs.uiuc.edu> Author: dannyb Date: Sat Oct 6 23:47:57 2007 New Revision: 42717 URL: http://llvm.org/viewvc/llvm-project?rev=42717&view=rev Log: Some more problems noticed by Curtis Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h Modified: llvm/trunk/include/llvm/ADT/SparseBitVector.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/SparseBitVector.h?rev=42717&r1=42716&r2=42717&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/SparseBitVector.h (original) +++ llvm/trunk/include/llvm/ADT/SparseBitVector.h Sat Oct 6 23:47:57 2007 @@ -648,8 +648,8 @@ ElementListIter Iter1 = Elements.begin(); ElementListConstIter Iter2 = RHS.Elements.begin(); - // Check if they are both empty - if (Elements.empty() && RHS.Elements.empty()) + // If either our bitmap or RHS is empty, we are done + if (Elements.empty() || RHS.Elements.empty()) return false; // Loop through, intersecting as we go, erasing elements when necessary. @@ -692,8 +692,9 @@ ElementListConstIter Iter1 = RHS1.Elements.begin(); ElementListConstIter Iter2 = RHS2.Elements.begin(); - // Check if they are both empty. - if (RHS1.empty() && RHS2.empty()) + // If RHS1 is empty, we are done + // If RHS2 is empty, we still have to copy RHS1 + if (RHS1.Elements.empty()) return; // Loop through, intersecting as we go, erasing elements when necessary. @@ -716,6 +717,9 @@ ++Iter1; ++Iter2; } else { + SparseBitVectorElement *NewElement = + new SparseBitVectorElement(*Iter1); + Elements.push_back(NewElement); ++Iter1; } } From neil at daikokuya.co.uk Sun Oct 7 03:51:21 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sun, 07 Oct 2007 08:51:21 -0000 Subject: [llvm-commits] [llvm] r42729 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp Message-ID: <200710070851.l978pMMd030799@zion.cs.uiuc.edu> Author: neil Date: Sun Oct 7 03:51:21 2007 New Revision: 42729 URL: http://llvm.org/viewvc/llvm-project?rev=42729&view=rev Log: combineLostFractions does not need to be a member function Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42729&r1=42728&r2=42729&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Sun Oct 7 03:51:21 2007 @@ -268,7 +268,6 @@ bool roundAwayFromZero(roundingMode, lostFraction, unsigned int) const; opStatus convertFromUnsignedInteger(integerPart *, unsigned int, roundingMode); - lostFraction combineLostFractions(lostFraction, lostFraction); opStatus convertFromHexadecimalString(const char *, roundingMode); char *convertNormalToHexString(char *, unsigned int, bool, roundingMode) const; Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42729&r1=42728&r2=42729&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sun Oct 7 03:51:21 2007 @@ -222,6 +222,20 @@ return lost_fraction; } + /* Combine the effect of two lost fractions. */ + lostFraction + combineLostFractions(lostFraction moreSignificant, + lostFraction lessSignificant) + { + if(lessSignificant != lfExactlyZero) { + if(moreSignificant == lfExactlyZero) + moreSignificant = lfLessThanHalf; + else if(moreSignificant == lfExactlyHalf) + moreSignificant = lfMoreThanHalf; + } + + return moreSignificant; + } /* Zero at the end to avoid modular arithmetic when adding one; used when rounding up during hexadecimal output. */ @@ -429,21 +443,6 @@ return &significand.part; } -/* Combine the effect of two lost fractions. */ -lostFraction -APFloat::combineLostFractions(lostFraction moreSignificant, - lostFraction lessSignificant) -{ - if(lessSignificant != lfExactlyZero) { - if(moreSignificant == lfExactlyZero) - moreSignificant = lfLessThanHalf; - else if(moreSignificant == lfExactlyHalf) - moreSignificant = lfMoreThanHalf; - } - - return moreSignificant; -} - void APFloat::zeroSignificand() { @@ -1614,7 +1613,7 @@ partsCount = partCount(); bitPos = partsCount * integerPartWidth; - /* Skip leading zeroes and any(hexa)decimal point. */ + /* Skip leading zeroes and any (hexa)decimal point. */ p = skipLeadingZeroesAndAnyDot(p, &dot); firstSignificantDigit = p; From neil at daikokuya.co.uk Sun Oct 7 06:45:58 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sun, 07 Oct 2007 11:45:58 -0000 Subject: [llvm-commits] [llvm] r42732 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/CodeGen/SelectionDAG/LegalizeDAG.cpp lib/CodeGen/SelectionDAG/SelectionDAG.cpp lib/ExecutionEngine/ExecutionEngine.cpp lib/Support/APFloat.cpp lib/VMCore/ConstantFold.cpp Message-ID: <200710071146.l97Bk0bi020945@zion.cs.uiuc.edu> Author: neil Date: Sun Oct 7 06:45:55 2007 New Revision: 42732 URL: http://llvm.org/viewvc/llvm-project?rev=42732&view=rev Log: convertFromInteger, as originally written, expected sign-extended input. APInt unfortunately zero-extends signed integers, so Dale modified the function to expect zero-extended input. Make this assumption explicit in the function name. Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp llvm/trunk/lib/ExecutionEngine/ExecutionEngine.cpp llvm/trunk/lib/Support/APFloat.cpp llvm/trunk/lib/VMCore/ConstantFold.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Sun Oct 7 06:45:55 2007 @@ -194,8 +194,8 @@ opStatus convert(const fltSemantics &, roundingMode); opStatus convertToInteger(integerPart *, unsigned int, bool, roundingMode) const; - opStatus convertFromInteger(const integerPart *, unsigned int, bool, - roundingMode); + opStatus convertFromZeroExtendedInteger(const integerPart *, unsigned int, + bool, roundingMode); opStatus convertFromString(const char *, roundingMode); APInt convertToAPInt() const; double convertToDouble() const; Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Sun Oct 7 06:45:55 2007 @@ -3234,8 +3234,8 @@ const uint64_t zero[] = {0, 0}; APFloat apf = APFloat(APInt(MVT::getSizeInBits(VT), 2, zero)); uint64_t x = 1ULL << ShiftAmt; - (void)apf.convertFromInteger(&x, MVT::getSizeInBits(NVT), false, - APFloat::rmNearestTiesToEven); + (void)apf.convertFromZeroExtendedInteger + (&x, MVT::getSizeInBits(NVT), false, APFloat::rmNearestTiesToEven); Tmp2 = DAG.getConstantFP(apf, VT); Tmp3 = DAG.getSetCC(TLI.getSetCCResultTy(), Node->getOperand(0), Tmp2, ISD::SETLT); Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original) +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Sun Oct 7 06:45:55 2007 @@ -1595,7 +1595,7 @@ case ISD::SINT_TO_FP: { const uint64_t zero[] = {0, 0}; APFloat apf = APFloat(APInt(MVT::getSizeInBits(VT), 2, zero)); - (void)apf.convertFromInteger(&Val, + (void)apf.convertFromZeroExtendedInteger(&Val, MVT::getSizeInBits(Operand.getValueType()), Opcode==ISD::SINT_TO_FP, APFloat::rmNearestTiesToEven); Modified: llvm/trunk/lib/ExecutionEngine/ExecutionEngine.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/ExecutionEngine/ExecutionEngine.cpp?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/lib/ExecutionEngine/ExecutionEngine.cpp (original) +++ llvm/trunk/lib/ExecutionEngine/ExecutionEngine.cpp Sun Oct 7 06:45:55 2007 @@ -396,7 +396,7 @@ else if (CE->getType() == Type::X86_FP80Ty) { const uint64_t zero[] = {0, 0}; APFloat apf = APFloat(APInt(80, 2, zero)); - (void)apf.convertFromInteger(GV.IntVal.getRawData(), + (void)apf.convertFromZeroExtendedInteger(GV.IntVal.getRawData(), GV.IntVal.getBitWidth(), false, APFloat::rmNearestTiesToEven); GV.IntVal = apf.convertToAPInt(); @@ -412,7 +412,7 @@ else if (CE->getType() == Type::X86_FP80Ty) { const uint64_t zero[] = { 0, 0}; APFloat apf = APFloat(APInt(80, 2, zero)); - (void)apf.convertFromInteger(GV.IntVal.getRawData(), + (void)apf.convertFromZeroExtendedInteger(GV.IntVal.getRawData(), GV.IntVal.getBitWidth(), true, APFloat::rmNearestTiesToEven); GV.IntVal = apf.convertToAPInt(); Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sun Oct 7 06:45:55 2007 @@ -1247,8 +1247,8 @@ if (fs==opInvalidOp) return fs; - fs = V.convertFromInteger(x, parts * integerPartWidth, true, - rmNearestTiesToEven); + fs = V.convertFromZeroExtendedInteger(x, parts * integerPartWidth, true, + rmNearestTiesToEven); assert(fs==opOK); // should always work fs = V.multiply(rhs, rounding_mode); @@ -1576,9 +1576,11 @@ return normalize(rounding_mode, lost_fraction); } +/* FIXME: should this just take a const APInt reference? */ APFloat::opStatus -APFloat::convertFromInteger(const integerPart *parts, unsigned int width, - bool isSigned, roundingMode rounding_mode) +APFloat::convertFromZeroExtendedInteger(const integerPart *parts, + unsigned int width, bool isSigned, + roundingMode rounding_mode) { unsigned int partCount = partCountForBits(width); opStatus status; Modified: llvm/trunk/lib/VMCore/ConstantFold.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/ConstantFold.cpp?rev=42732&r1=42731&r2=42732&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/ConstantFold.cpp (original) +++ llvm/trunk/lib/VMCore/ConstantFold.cpp Sun Oct 7 06:45:55 2007 @@ -216,7 +216,7 @@ uint32_t BitWidth = cast(SrcTy)->getBitWidth(); APFloat apf = APFloat(APInt(DestTy->getPrimitiveSizeInBits(), 2, zero)); - (void)apf.convertFromInteger(api.getRawData(), BitWidth, + (void)apf.convertFromZeroExtendedInteger(api.getRawData(), BitWidth, opc==Instruction::SIToFP, APFloat::rmNearestTiesToEven); return ConstantFP::get(DestTy, apf); From neil at daikokuya.co.uk Sun Oct 7 07:07:54 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sun, 07 Oct 2007 12:07:54 -0000 Subject: [llvm-commits] [llvm] r42733 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp Message-ID: <200710071207.l97C7sxX021487@zion.cs.uiuc.edu> Author: neil Date: Sun Oct 7 07:07:53 2007 New Revision: 42733 URL: http://llvm.org/viewvc/llvm-project?rev=42733&view=rev Log: Reimplement convertFromUnsignedInteger so it is passed a const bignum. It used to modify its argument in-place. This interface is saner and the implementation more efficient. It will be needed for decimal->binary conversion. Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42733&r1=42732&r2=42733&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Sun Oct 7 07:07:53 2007 @@ -266,8 +266,8 @@ cmpResult compareAbsoluteValue(const APFloat &) const; opStatus handleOverflow(roundingMode); bool roundAwayFromZero(roundingMode, lostFraction, unsigned int) const; - opStatus convertFromUnsignedInteger(integerPart *, unsigned int, - roundingMode); + opStatus convertFromUnsignedParts(const integerPart *, unsigned int, + roundingMode); opStatus convertFromHexadecimalString(const char *, roundingMode); char *convertNormalToHexString(char *, unsigned int, bool, roundingMode) const; Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42733&r1=42732&r2=42733&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sun Oct 7 07:07:53 2007 @@ -1548,30 +1548,41 @@ return opInexact; } +/* Convert an unsigned integer SRC to a floating point number, + rounding according to ROUNDING_MODE. The sign of the floating + point number is not modified. */ APFloat::opStatus -APFloat::convertFromUnsignedInteger(integerPart *parts, - unsigned int partCount, - roundingMode rounding_mode) +APFloat::convertFromUnsignedParts(const integerPart *src, + unsigned int srcCount, + roundingMode rounding_mode) { - unsigned int msb, precision; + unsigned int dstCount; lostFraction lost_fraction; - - msb = APInt::tcMSB(parts, partCount) + 1; - precision = semantics->precision; + integerPart *dst; category = fcNormal; - exponent = precision - 1; + exponent = semantics->precision - 1; - if(msb > precision) { - exponent += (msb - precision); - lost_fraction = shiftRight(parts, partCount, msb - precision); - msb = precision; - } else - lost_fraction = lfExactlyZero; + dst = significandParts(); + dstCount = partCount(); - /* Copy the bit image. */ - zeroSignificand(); - APInt::tcAssign(significandParts(), parts, partCountForBits(msb)); + /* We need to capture the non-zero most significant parts. */ + while (srcCount > dstCount && src[srcCount - 1] == 0) + srcCount--; + + /* Copy the bit image of as many parts as we can. If we are wider, + zero-out remaining parts. */ + if (dstCount >= srcCount) { + APInt::tcAssign(dst, src, srcCount); + while (srcCount < dstCount) + dst[srcCount++] = 0; + lost_fraction = lfExactlyZero; + } else { + exponent += (srcCount - dstCount) * integerPartWidth; + APInt::tcAssign(dst, src + (srcCount - dstCount), dstCount); + lost_fraction = lostFractionThroughTruncation(src, srcCount, + dstCount * integerPartWidth); + } return normalize(rounding_mode, lost_fraction); } @@ -1594,7 +1605,7 @@ } APInt::tcAssign(copy, api.getRawData(), partCount); - status = convertFromUnsignedInteger(copy, partCount, rounding_mode); + status = convertFromUnsignedParts(copy, partCount, rounding_mode); return status; } From neil at daikokuya.co.uk Sun Oct 7 07:10:58 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sun, 07 Oct 2007 12:10:58 -0000 Subject: [llvm-commits] [llvm] r42734 - /llvm/trunk/lib/Support/APFloat.cpp Message-ID: <200710071210.l97CAw3K021561@zion.cs.uiuc.edu> Author: neil Date: Sun Oct 7 07:10:57 2007 New Revision: 42734 URL: http://llvm.org/viewvc/llvm-project?rev=42734&view=rev Log: Now that convertFromUnsignedParts has a sane, constant interface, convertFromZeroExtendedInteger can be simplified as it doesn't need to make a copy of the source bignum. Modified: llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42734&r1=42733&r2=42734&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sun Oct 7 07:10:57 2007 @@ -1594,9 +1594,7 @@ roundingMode rounding_mode) { unsigned int partCount = partCountForBits(width); - opStatus status; APInt api = APInt(width, partCount, parts); - integerPart *copy = new integerPart[partCount]; sign = false; if(isSigned && APInt::tcExtractBit(parts, width - 1)) { @@ -1604,9 +1602,7 @@ api = -api; } - APInt::tcAssign(copy, api.getRawData(), partCount); - status = convertFromUnsignedParts(copy, partCount, rounding_mode); - return status; + return convertFromUnsignedParts(api.getRawData(), partCount, rounding_mode); } APFloat::opStatus From neil at daikokuya.co.uk Sun Oct 7 07:15:41 2007 From: neil at daikokuya.co.uk (Neil Booth) Date: Sun, 07 Oct 2007 12:15:41 -0000 Subject: [llvm-commits] [llvm] r42735 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp Message-ID: <200710071215.l97CFfc1021662@zion.cs.uiuc.edu> Author: neil Date: Sun Oct 7 07:15:41 2007 New Revision: 42735 URL: http://llvm.org/viewvc/llvm-project?rev=42735&view=rev Log: Add back convertFromSignExtendedInteger. Modified: llvm/trunk/include/llvm/ADT/APFloat.h llvm/trunk/lib/Support/APFloat.cpp Modified: llvm/trunk/include/llvm/ADT/APFloat.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=42735&r1=42734&r2=42735&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/APFloat.h (original) +++ llvm/trunk/include/llvm/ADT/APFloat.h Sun Oct 7 07:15:41 2007 @@ -194,6 +194,8 @@ opStatus convert(const fltSemantics &, roundingMode); opStatus convertToInteger(integerPart *, unsigned int, bool, roundingMode) const; + opStatus convertFromSignExtendedInteger(const integerPart *, unsigned int, + bool, roundingMode); opStatus convertFromZeroExtendedInteger(const integerPart *, unsigned int, bool, roundingMode); opStatus convertFromString(const char *, roundingMode); Modified: llvm/trunk/lib/Support/APFloat.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=42735&r1=42734&r2=42735&view=diff ============================================================================== --- llvm/trunk/lib/Support/APFloat.cpp (original) +++ llvm/trunk/lib/Support/APFloat.cpp Sun Oct 7 07:15:41 2007 @@ -1587,6 +1587,36 @@ return normalize(rounding_mode, lost_fraction); } +/* Convert a two's complement integer SRC to a floating point number, + rounding according to ROUNDING_MODE. ISSIGNED is true if the + integer is signed, in which case it must be sign-extended. */ +APFloat::opStatus +APFloat::convertFromSignExtendedInteger(const integerPart *src, + unsigned int srcCount, + bool isSigned, + roundingMode rounding_mode) +{ + opStatus status; + + if (isSigned + && APInt::tcExtractBit(src, srcCount * integerPartWidth - 1)) { + integerPart *copy; + + /* If we're signed and negative negate a copy. */ + sign = true; + copy = new integerPart[srcCount]; + APInt::tcAssign(copy, src, srcCount); + APInt::tcNegate(copy, srcCount); + status = convertFromUnsignedParts(copy, srcCount, rounding_mode); + delete [] copy; + } else { + sign = false; + status = convertFromUnsignedParts(src, srcCount, rounding_mode); + } + + return status; +} + /* FIXME: should this just take a const APInt reference? */ APFloat::opStatus APFloat::convertFromZeroExtendedInteger(const integerPart *parts, From gordonhenriksen at mac.com Sun Oct 7 12:31:43 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Sun, 07 Oct 2007 17:31:43 -0000 Subject: [llvm-commits] [llvm] r42736 - in /llvm/trunk: Xcode/LLVM.xcodeproj/project.pbxproj bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp test/Bindings/Ocaml/vmcore.ml Message-ID: <200710071731.l97HVhjH001572@zion.cs.uiuc.edu> Author: gordon Date: Sun Oct 7 12:31:42 2007 New Revision: 42736 URL: http://llvm.org/viewvc/llvm-project?rev=42736&view=rev Log: C and Objective Caml bindings for GlobalVariable::isConstant. Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj (original) +++ llvm/trunk/Xcode/LLVM.xcodeproj/project.pbxproj Sun Oct 7 12:31:42 2007 @@ -142,6 +142,9 @@ 9F7C240C0CB81ECD00498408 /* analysis_ocaml.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; path = analysis_ocaml.c; sourceTree = ""; }; 9F7C240D0CB81ECD00498408 /* llvm_analysis.ml */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = llvm_analysis.ml; sourceTree = ""; }; 9F7C240E0CB81ECD00498408 /* llvm_analysis.mli */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = llvm_analysis.mli; sourceTree = ""; }; + 9F7C2C4F0CB9496C00498408 /* analysis.ml */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = analysis.ml; sourceTree = ""; }; + 9F7C2C520CB9496C00498408 /* bitwriter.ml */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = bitwriter.ml; sourceTree = ""; }; + 9F7C2C5D0CB9496C00498408 /* vmcore.ml */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = vmcore.ml; sourceTree = ""; }; 9FA638D90C77B184007F12AE /* AutoUpgrade.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = AutoUpgrade.h; sourceTree = ""; }; 9FA638DA0C77B184007F12AE /* GlobalAlias.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = GlobalAlias.h; sourceTree = ""; }; 9FA638DB0C77B1AB007F12AE /* APInt.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = APInt.h; sourceTree = ""; }; @@ -1071,6 +1074,7 @@ DE816FAC08CFB44C0093BDEF /* utils */, DE66F38D08ABF35C00323D32 /* docs */, 9FD3E56D0CA0116100E54D15 /* bindings */, + 9F7C2B690CB9496B00498408 /* test */, DE66F3FD08ABF37000323D32 /* examples */, DE66F38C08ABF35300323D32 /* CREDITS.TXT */, CFD99AA80AFE827B0068D19C /* LICENSE.TXT */, @@ -1170,6 +1174,33 @@ path = analysis; sourceTree = ""; }; + 9F7C2B690CB9496B00498408 /* test */ = { + isa = PBXGroup; + children = ( + 9F7C2C4B0CB9496C00498408 /* Bindings */, + ); + name = test; + path = ../test; + sourceTree = SOURCE_ROOT; + }; + 9F7C2C4B0CB9496C00498408 /* Bindings */ = { + isa = PBXGroup; + children = ( + 9F7C2C4C0CB9496C00498408 /* Ocaml */, + ); + path = Bindings; + sourceTree = ""; + }; + 9F7C2C4C0CB9496C00498408 /* Ocaml */ = { + isa = PBXGroup; + children = ( + 9F7C2C4F0CB9496C00498408 /* analysis.ml */, + 9F7C2C520CB9496C00498408 /* bitwriter.ml */, + 9F7C2C5D0CB9496C00498408 /* vmcore.ml */, + ); + path = Ocaml; + sourceTree = ""; + }; 9FA638E90C77B252007F12AE /* IPO */ = { isa = PBXGroup; children = ( Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sun Oct 7 12:31:42 2007 @@ -268,6 +268,9 @@ external set_visibility : visibility -> llvalue -> unit = "llvm_set_visibility" external alignment : llvalue -> int = "llvm_alignment" external set_alignment : int -> llvalue -> unit = "llvm_set_alignment" +external is_global_constant : llvalue -> bool = "llvm_is_global_constant" +external set_global_constant : bool -> llvalue -> unit + = "llvm_set_global_constant" (*--... Operations on global variables .....................................--*) external declare_global : lltype -> string -> llmodule -> llvalue Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sun Oct 7 12:31:42 2007 @@ -251,6 +251,9 @@ external set_visibility : visibility -> llvalue -> unit = "llvm_set_visibility" external alignment : llvalue -> int = "llvm_alignment" external set_alignment : int -> llvalue -> unit = "llvm_set_alignment" +external is_global_constant : llvalue -> bool = "llvm_is_global_constant" +external set_global_constant : bool -> llvalue -> unit + = "llvm_set_global_constant" (*--... Operations on global variables .....................................--*) external declare_global : lltype -> string -> llmodule -> llvalue Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sun Oct 7 12:31:42 2007 @@ -445,6 +445,17 @@ return Val_unit; } +/* llvalue -> bool */ +CAMLprim value llvm_is_global_constant(LLVMValueRef GlobalVar) { + return Val_bool(LLVMIsGlobalConstant(GlobalVar)); +} + +/* bool -> llvalue -> unit */ +CAMLprim value llvm_set_global_constant(value Flag, LLVMValueRef GlobalVar) { + LLVMSetGlobalConstant(GlobalVar, Bool_val(Flag)); + return Val_unit; +} + /*--... Operations on functions ............................................--*/ /* string -> lltype -> llmodule -> llvalue */ Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sun Oct 7 12:31:42 2007 @@ -324,6 +324,8 @@ void LLVMSetInitializer(LLVMValueRef GlobalVar, LLVMValueRef ConstantVal); int LLVMIsThreadLocal(LLVMValueRef GlobalVar); void LLVMSetThreadLocal(LLVMValueRef GlobalVar, int IsThreadLocal); +int LLVMIsGlobalConstant(LLVMValueRef GlobalVar); +void LLVMSetGlobalConstant(LLVMValueRef GlobalVar, int IsConstant); /* Operations on functions */ LLVMValueRef LLVMAddFunction(LLVMModuleRef M, const char *Name, Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sun Oct 7 12:31:42 2007 @@ -557,6 +557,17 @@ unwrap(GlobalVar)->setThreadLocal(IsThreadLocal != 0); } +int LLVMIsGlobalConstant(LLVMValueRef GlobalVar) { + bool res = unwrap(GlobalVar)->isConstant(); + fprintf(stderr, "LLVMIsConstantGlobal(0x%08x) = %s\n", + (int) GlobalVar, res? "YES" : "NO"); + return unwrap(GlobalVar)->isConstant(); +} + +void LLVMSetGlobalConstant(LLVMValueRef GlobalVar, int IsConstant) { + unwrap(GlobalVar)->setConstant(IsConstant != 0); +} + /*--.. Operations on functions .............................................--*/ LLVMValueRef LLVMAddFunction(LLVMModuleRef M, const char *Name, Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42736&r1=42735&r2=42736&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sun Oct 7 12:31:42 2007 @@ -418,7 +418,15 @@ *) group "delete"; let g = define_global "GVar05" fourty_two32 m in - delete_global g + delete_global g; + + (* RUN: grep -v {ConstGlobalVar.*constant} < %t.ll + *) + group "constant"; + let g = define_global "ConstGlobalVar" fourty_two32 m in + insist (not (is_global_constant g)); + set_global_constant true g; + insist (is_global_constant g) (*===-- Functions ---------------------------------------------------------===*) From sabre at nondot.org Sun Oct 7 12:57:11 2007 From: sabre at nondot.org (Chris Lattner) Date: Sun, 07 Oct 2007 17:57:11 -0000 Subject: [llvm-commits] [llvm] r42737 - /llvm/trunk/include/llvm/ADT/DenseMap.h Message-ID: <200710071757.l97HvBCZ002299@zion.cs.uiuc.edu> Author: lattner Date: Sun Oct 7 12:57:10 2007 New Revision: 42737 URL: http://llvm.org/viewvc/llvm-project?rev=42737&view=rev Log: Finish off PR1723, by working around some strange compiler bug. Modified: llvm/trunk/include/llvm/ADT/DenseMap.h Modified: llvm/trunk/include/llvm/ADT/DenseMap.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/DenseMap.h?rev=42737&r1=42736&r2=42737&view=diff ============================================================================== --- llvm/trunk/include/llvm/ADT/DenseMap.h (original) +++ llvm/trunk/include/llvm/ADT/DenseMap.h Sun Oct 7 12:57:10 2007 @@ -36,8 +36,8 @@ static inline T* getEmptyKey() { return reinterpret_cast(-1); } static inline T* getTombstoneKey() { return reinterpret_cast(-2); } static unsigned getHashValue(const T *PtrVal) { - return (unsigned(uintptr_t(PtrVal)) >> 4) ^ - (unsigned(uintptr_t(PtrVal)) >> 9); + return (unsigned((uintptr_t)PtrVal) >> 4) ^ + (unsigned((uintptr_t)PtrVal) >> 9); } static bool isEqual(const T *LHS, const T *RHS) { return LHS == RHS; } static bool isPod() { return true; } From sabre at nondot.org Sun Oct 7 16:53:12 2007 From: sabre at nondot.org (Chris Lattner) Date: Sun, 07 Oct 2007 21:53:12 -0000 Subject: [llvm-commits] [llvm] r42738 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Message-ID: <200710072153.l97LrCI5008602@zion.cs.uiuc.edu> Author: lattner Date: Sun Oct 7 16:53:12 2007 New Revision: 42738 URL: http://llvm.org/viewvc/llvm-project?rev=42738&view=rev Log: Fix many regressions on x86 by avoiding dereferencing the end iterator. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42738&r1=42737&r2=42738&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Sun Oct 7 16:53:12 2007 @@ -1481,47 +1481,50 @@ static void mergeSPUpdatesUp(MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI, unsigned StackPtr, uint64_t *NumBytes = NULL) { - if (MBBI != MBB.begin()) { - MachineBasicBlock::iterator PI = prior(MBBI); - unsigned Opc = PI->getOpcode(); - if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || - Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && - PI->getOperand(0).getReg() == StackPtr) { - if (NumBytes) - *NumBytes += PI->getOperand(2).getImm(); - MBB.erase(PI); - } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || - Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && - PI->getOperand(0).getReg() == StackPtr) { - if (NumBytes) - *NumBytes -= PI->getOperand(2).getImm(); - MBB.erase(PI); - } + if (MBBI == MBB.begin()) return; + + MachineBasicBlock::iterator PI = prior(MBBI); + unsigned Opc = PI->getOpcode(); + if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || + Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && + PI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes += PI->getOperand(2).getImm(); + MBB.erase(PI); + } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || + Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && + PI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes -= PI->getOperand(2).getImm(); + MBB.erase(PI); } } // mergeSPUpdatesUp - Merge two stack-manipulating instructions lower iterator. static -void mergeSPUpdatesDown(MachineBasicBlock &MBB,MachineBasicBlock::iterator &MBBI, +void mergeSPUpdatesDown(MachineBasicBlock &MBB, + MachineBasicBlock::iterator &MBBI, unsigned StackPtr, uint64_t *NumBytes = NULL) { - if (MBBI != MBB.end()) { - MachineBasicBlock::iterator NI = next(MBBI); - unsigned Opc = NI->getOpcode(); - if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || - Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && - NI->getOperand(0).getReg() == StackPtr) { - if (NumBytes) - *NumBytes -= NI->getOperand(2).getImm(); - MBB.erase(NI); - MBBI = NI; - } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || - Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && - NI->getOperand(0).getReg() == StackPtr) { - if (NumBytes) - *NumBytes += NI->getOperand(2).getImm(); - MBB.erase(NI); - MBBI = NI; - } + if (MBBI == MBB.end()) return; + + MachineBasicBlock::iterator NI = next(MBBI); + if (NI == MBB.end()) return; + + unsigned Opc = NI->getOpcode(); + if ((Opc == X86::ADD64ri32 || Opc == X86::ADD64ri8 || + Opc == X86::ADD32ri || Opc == X86::ADD32ri8) && + NI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes -= NI->getOperand(2).getImm(); + MBB.erase(NI); + MBBI = NI; + } else if ((Opc == X86::SUB64ri32 || Opc == X86::SUB64ri8 || + Opc == X86::SUB32ri || Opc == X86::SUB32ri8) && + NI->getOperand(0).getReg() == StackPtr) { + if (NumBytes) + *NumBytes += NI->getOperand(2).getImm(); + MBB.erase(NI); + MBBI = NI; } } From sabre at nondot.org Sun Oct 7 17:00:32 2007 From: sabre at nondot.org (Chris Lattner) Date: Sun, 07 Oct 2007 22:00:32 -0000 Subject: [llvm-commits] [llvm] r42739 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Message-ID: <200710072200.l97M0WoV008941@zion.cs.uiuc.edu> Author: lattner Date: Sun Oct 7 17:00:31 2007 New Revision: 42739 URL: http://llvm.org/viewvc/llvm-project?rev=42739&view=rev Log: disable this entirely: it is causing use of invalidated iterators and infinite looping. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=42739&r1=42738&r2=42739&view=diff ============================================================================== --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original) +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Sun Oct 7 17:00:31 2007 @@ -1505,6 +1505,8 @@ void mergeSPUpdatesDown(MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI, unsigned StackPtr, uint64_t *NumBytes = NULL) { + return; + if (MBBI == MBB.end()) return; MachineBasicBlock::iterator NI = next(MBBI); From gordonhenriksen at mac.com Sun Oct 7 22:45:09 2007 From: gordonhenriksen at mac.com (Gordon Henriksen) Date: Mon, 08 Oct 2007 03:45:09 -0000 Subject: [llvm-commits] [llvm] r42740 - in /llvm/trunk: bindings/ocaml/llvm/llvm.ml bindings/ocaml/llvm/llvm.mli bindings/ocaml/llvm/llvm_ocaml.c include/llvm-c/Core.h lib/VMCore/Core.cpp test/Bindings/Ocaml/vmcore.ml Message-ID: <200710080345.l983jASw026747@zion.cs.uiuc.edu> Author: gordon Date: Sun Oct 7 22:45:09 2007 New Revision: 42740 URL: http://llvm.org/viewvc/llvm-project?rev=42740&view=rev Log: C and Objective Caml bindings for getFunction and getNamedGlobal. Also enhanced the Objective Caml 'declare_*' functions to behave more or less like getOrInsertFunction. Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml llvm/trunk/bindings/ocaml/llvm/llvm.mli llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c llvm/trunk/include/llvm-c/Core.h llvm/trunk/lib/VMCore/Core.cpp llvm/trunk/test/Bindings/Ocaml/vmcore.ml Modified: llvm/trunk/bindings/ocaml/llvm/llvm.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.ml?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.ml (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.ml Sun Oct 7 22:45:09 2007 @@ -277,6 +277,8 @@ = "llvm_declare_global" external define_global : string -> llvalue -> llmodule -> llvalue = "llvm_define_global" +external lookup_global : string -> llmodule -> llvalue option + = "llvm_lookup_global" external delete_global : llvalue -> unit = "llvm_delete_global" external global_initializer : llvalue -> llvalue = "LLVMGetInitializer" external set_initializer : llvalue -> llvalue -> unit = "llvm_set_initializer" @@ -289,6 +291,8 @@ = "llvm_declare_function" external define_function : string -> lltype -> llmodule -> llvalue = "llvm_define_function" +external lookup_function : string -> llmodule -> llvalue option + = "llvm_lookup_function" external delete_function : llvalue -> unit = "llvm_delete_function" external params : llvalue -> llvalue array = "llvm_params" external param : llvalue -> int -> llvalue = "llvm_param" Modified: llvm/trunk/bindings/ocaml/llvm/llvm.mli URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm.mli?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm.mli (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm.mli Sun Oct 7 22:45:09 2007 @@ -260,6 +260,8 @@ = "llvm_declare_global" external define_global : string -> llvalue -> llmodule -> llvalue = "llvm_define_global" +external lookup_global : string -> llmodule -> llvalue option + = "llvm_lookup_global" external delete_global : llvalue -> unit = "llvm_delete_global" external global_initializer : llvalue -> llvalue = "LLVMGetInitializer" external set_initializer : llvalue -> llvalue -> unit = "llvm_set_initializer" @@ -272,6 +274,8 @@ = "llvm_declare_function" external define_function : string -> lltype -> llmodule -> llvalue = "llvm_define_function" +external lookup_function : string -> llmodule -> llvalue option + = "llvm_lookup_function" external delete_function : llvalue -> unit = "llvm_delete_function" external params : llvalue -> llvalue array = "llvm_params" external param : llvalue -> int -> llvalue = "llvm_param" Modified: llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c (original) +++ llvm/trunk/bindings/ocaml/llvm/llvm_ocaml.c Sun Oct 7 22:45:09 2007 @@ -20,7 +20,7 @@ #include "caml/custom.h" #include "caml/mlvalues.h" #include "caml/memory.h" -#include "llvm/Config/config.h" +#include "llvm/Config/config.h" /*===-- Modules -----------------------------------------------------------===*/ @@ -402,9 +402,27 @@ /* lltype -> string -> llmodule -> llvalue */ CAMLprim LLVMValueRef llvm_declare_global(LLVMTypeRef Ty, value Name, LLVMModuleRef M) { + LLVMValueRef GlobalVar; + if ((GlobalVar = LLVMGetNamedGlobal(M, String_val(Name)))) { + if (LLVMGetElementType(LLVMTypeOf(GlobalVar)) != Ty) + return LLVMConstBitCast(GlobalVar, LLVMPointerType(Ty)); + return GlobalVar; + } return LLVMAddGlobal(M, Ty, String_val(Name)); } +/* string -> llmodule -> llvalue option */ +CAMLprim value llvm_lookup_global(value Name, LLVMModuleRef M) { + CAMLparam1(Name); + LLVMValueRef GlobalVar; + if ((GlobalVar = LLVMGetNamedGlobal(M, String_val(Name)))) { + value Option = caml_alloc(1, 1); + Field(Option, 0) = (value) GlobalVar; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + /* string -> llvalue -> llmodule -> llvalue */ CAMLprim LLVMValueRef llvm_define_global(value Name, LLVMValueRef Initializer, LLVMModuleRef M) { @@ -461,9 +479,27 @@ /* string -> lltype -> llmodule -> llvalue */ CAMLprim LLVMValueRef llvm_declare_function(value Name, LLVMTypeRef Ty, LLVMModuleRef M) { + LLVMValueRef Fn; + if ((Fn = LLVMGetNamedFunction(M, String_val(Name)))) { + if (LLVMGetElementType(LLVMTypeOf(Fn)) != Ty) + return LLVMConstBitCast(Fn, LLVMPointerType(Ty)); + return Fn; + } return LLVMAddFunction(M, String_val(Name), Ty); } +/* string -> llmodule -> llvalue option */ +CAMLprim value llvm_lookup_function(value Name, LLVMModuleRef M) { + CAMLparam1(Name); + LLVMValueRef Fn; + if ((Fn = LLVMGetNamedFunction(M, String_val(Name)))) { + value Option = caml_alloc(1, 1); + Field(Option, 0) = (value) Fn; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + /* string -> lltype -> llmodule -> llvalue */ CAMLprim LLVMValueRef llvm_define_function(value Name, LLVMTypeRef Ty, LLVMModuleRef M) { Modified: llvm/trunk/include/llvm-c/Core.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm-c/Core.h?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/include/llvm-c/Core.h (original) +++ llvm/trunk/include/llvm-c/Core.h Sun Oct 7 22:45:09 2007 @@ -318,6 +318,7 @@ /* Operations on global variables */ LLVMValueRef LLVMAddGlobal(LLVMModuleRef M, LLVMTypeRef Ty, const char *Name); +LLVMValueRef LLVMGetNamedGlobal(LLVMModuleRef M, const char *Name); void LLVMDeleteGlobal(LLVMValueRef GlobalVar); int LLVMHasInitializer(LLVMValueRef GlobalVar); LLVMValueRef LLVMGetInitializer(LLVMValueRef GlobalVar); @@ -330,6 +331,7 @@ /* Operations on functions */ LLVMValueRef LLVMAddFunction(LLVMModuleRef M, const char *Name, LLVMTypeRef FunctionTy); +LLVMValueRef LLVMGetNamedFunction(LLVMModuleRef M, const char *Name); void LLVMDeleteFunction(LLVMValueRef Fn); unsigned LLVMCountParams(LLVMValueRef Fn); void LLVMGetParams(LLVMValueRef Fn, LLVMValueRef *Params); Modified: llvm/trunk/lib/VMCore/Core.cpp URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/Core.cpp?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/lib/VMCore/Core.cpp (original) +++ llvm/trunk/lib/VMCore/Core.cpp Sun Oct 7 22:45:09 2007 @@ -532,6 +532,10 @@ GlobalValue::ExternalLinkage, 0, Name, unwrap(M))); } +LLVMValueRef LLVMGetNamedGlobal(LLVMModuleRef M, const char *Name) { + return wrap(unwrap(M)->getNamedGlobal(Name)); +} + void LLVMDeleteGlobal(LLVMValueRef GlobalVar) { unwrap(GlobalVar)->eraseFromParent(); } @@ -576,6 +580,10 @@ GlobalValue::ExternalLinkage, Name, unwrap(M))); } +LLVMValueRef LLVMGetNamedFunction(LLVMModuleRef M, const char *Name) { + return wrap(unwrap(M)->getFunction(Name)); +} + void LLVMDeleteFunction(LLVMValueRef Fn) { unwrap(Fn)->eraseFromParent(); } Modified: llvm/trunk/test/Bindings/Ocaml/vmcore.ml URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Bindings/Ocaml/vmcore.ml?rev=42740&r1=42739&r2=42740&view=diff ============================================================================== --- llvm/trunk/test/Bindings/Ocaml/vmcore.ml (original) +++ llvm/trunk/test/Bindings/Ocaml/vmcore.ml Sun Oct 7 22:45:09 2007 @@ -393,8 +393,14 @@ (* RUN: grep {GVar01.*external} < %t.ll *) group "declarations"; + insist (None == lookup_global "GVar01" m); let g = declare_global i32_type "GVar01" m in insist (is_declaration g); + insist (pointer_type float_type == + type_of (declare_global float_type "GVar01" m)); + insist (g == declare_global i32_type "GVar01" m); + insist (match lookup_global "GVar01" m with Some x -> x = g + | None -> false); (* RUN: grep {GVar02.*42} < %t.ll * RUN: grep {GVar03.*42} < %t.ll @@ -433,15 +439,21 @@ let test_functions () = let ty = function_type i32_type [| i32_type; i64_type |] in - let pty = pointer_type ty in + let ty2 = function_type i8_type [| i8_type; i64_type |] in (* RUN: grep {declare i32 @Fn1\(i32, i64\)} < %t.ll *) group "declare"; + insist (None = lookup_function "Fn1" m); let fn = declare_function "Fn1" ty m in - insist (pty = type_of fn); + insist (pointer_type ty = type_of fn); insist (is_declaration fn); insist (0 = Array.length (basic_blocks fn)); + insist (pointer_type ty2 == type_of (declare_function "Fn1" ty2 m)); + insist (fn == declare_function "Fn1" ty m); + insist (None <> lookup_function "Fn1" m); + insist (match lookup_function "Fn1" m with Some x -> x = fn + | None -> false); (* RUN: grep -v {Fn2} < %t.ll *) From sabre at nondot.org Sun Oct 7 22:45:44 2007 From: sabre at nondot.org (Chris Lattner) Date: Mon, 08 Oct 2007 03:45:44 -0000 Subject: [llvm-commits] [llvm] r42741 - /llvm/trunk/include/llvm/Pass.h Message-ID: <200710080345.l983jiuN026791@zion.cs.uiuc.edu> Author: lattner Date: Sun Oct 7 22:45:44 2007 New Revision: 42741 URL: http://llvm.org/viewvc/llvm-project?rev=42741&view=rev Log: Allow passes to just pass up "FunctionPass(&ID)" instead of "FunctionPass((intptr_t)&ID)" Modified: llvm/trunk/include/llvm/Pass.h Modified: llvm/trunk/include/llvm/Pass.h URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Pass.h?rev=42741&r1=42740&r2=42741&view=diff ============================================================================== --- llvm/trunk/include/llvm/Pass.h (original) +++ llvm/trunk/include/llvm/Pass.h Sun Oct 7 22:45:44 2007 @@ -86,6 +86,7 @@ Pass(const Pass &); // DO NOT IMPLEMENT public: explicit Pass(intptr_t pid) : Resolver(0), PassID(pid) {} + explicit Pass(const void *pid) : Resolver(0), PassID((intptr_t)pid) {} virtual ~Pass(); /// getPassName - Return a nice clean name for a pass. This usually @@ -241,6 +242,7 @@ } explicit ModulePass(intptr_t pid) : Pass(pid) {} + explicit ModulePass(const void *pid) : Pass(pid) {} // Force out-of-line virtual method. virtual ~ModulePass(); }; @@ -266,6 +268,8 @@ bool runOnModule(Module &M) { return false; } explicit ImmutablePass(intptr_t pid) : ModulePass(pid) {} + explicit ImmutablePass(const void *pid) : ModulePass(pid) {} + // Force out-of-line virtual method. virtual ~ImmutablePass(); }; @@ -282,6 +286,7 @@ class FunctionPass : public Pass { public: explicit FunctionPass(intptr_t pid) : Pass(pid) {} + explicit FunctionPass(const void *pid) : Pass(pid) {} /// doInitialization - Virtual method overridden by subclasses to do /// any necessary per-module initialization. @@ -333,6 +338,7 @@ class BasicBlockPass : public Pass { public: explicit BasicBlockPass(intptr_t pid) : Pass(pid) {} + explicit BasicBlockPass(const void *pid) : Pass(pid) {} /// doInitialization - Virtual method overridden by subclasses to do /// any necessary per-module initialization.