From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 11:57:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 11:57:27 -0500 Subject: [LLVMbugs] [Bug 1531] NEW: type alignment non forwarded to malloc instruction in C and C++ frontends Message-ID: <200707021657.l62GvRZD030966@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1531 Summary: type alignment non forwarded to malloc instruction in C and C++ frontends Product: new-bugs Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: loufoque at gmail.com In the following C code (with GCC extensions): struct test { int a; int b __attribute__((aligned(16))); }; void do_something(struct test*); int main() { struct test* t = malloc(sizeof *t); do_something(t); return 0; } The type "struct test" has an associated alignment of 16. However, LLVM will do a normal malloc instruction, without saying it wants an alignment of 16. While it is true that GCC won't be able to do that anyway because to it malloc is a regular function that only allocates with an alignment of 8, the malloc instruction can be smart enough here to allocate with the appropriate alignment. Actual output: ; ModuleID = '/tmp/webcompile/_30695_0.bc' 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" target triple = "i686-pc-linux-gnu" %struct.test = type { i32, [12 x i8], i32, [12 x i8] } define i32 @main() { entry: %tmp16 = malloc %struct.test ; <%struct.test*> [#uses=1] tail call void @do_something( %struct.test* %tmp16 ) ret i32 0 } declare void @do_something(%struct.test*) Expected output: ; ModuleID = '/tmp/webcompile/_30695_0.bc' 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" target triple = "i686-pc-linux-gnu" %struct.test = type { i32, [12 x i8], i32, [12 x i8] } define i32 @main() { entry: %tmp16 = malloc %struct.test, align 16 ; <%struct.test*> [#uses=1] tail call void @do_something( %struct.test* %tmp16 ) ret i32 0 } declare void @do_something(%struct.test*) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 12:16:40 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 12:16:40 -0500 Subject: [LLVMbugs] [Bug 1532] NEW: Alignment too big with alloca in C/C++ frontend Message-ID: <200707021716.l62HGeuS031708@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1532 Summary: Alignment too big with alloca in C/C++ frontend Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: loufoque at gmail.com The following C code struct test { int a; char b; }; void do_something(struct test*); int main() { struct test t; do_something(&t); } generates an alloca instruction with an alignment of 8 on the web demo. However, the type "struct test" only requires an alignment of 4 for the x86 target. I think it shouldn't even be specified here, because the type definition is enough to know the required alignment. Actual output: ; ModuleID = '/tmp/webcompile/_31500_0.bc' 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" target triple = "i686-pc-linux-gnu" %struct.test = type { i32, i8 } define i32 @main() { entry: %t = alloca %struct.test, align 8 ; <%struct.test*> [#uses=1] call void @do_something( %struct.test* %t ) ret i32 undef } declare void @do_something(%struct.test*) Expected output: ; ModuleID = '/tmp/webcompile/_31500_0.bc' 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" target triple = "i686-pc-linux-gnu" %struct.test = type { i32, i8 } define i32 @main() { entry: %t = alloca %struct.test ; <%struct.test*> [#uses=1] call void @do_something( %struct.test* %t ) ret i32 undef } declare void @do_something(%struct.test*) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 12:21:43 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 12:21:43 -0500 Subject: [LLVMbugs] [Bug 1532] Alignment too big with alloca in C/C++ frontend Message-ID: <200707021721.l62HLhnh031934@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1532 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID ------- Additional Comments From sabre at nondot.org 2007-07-02 12:21 ------- This is a decision made by the GCC front-end. The extra alignment can sometimes help the optimizer, by allowing extra aligned load/stores to be used. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 12:47:56 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 12:47:56 -0500 Subject: [LLVMbugs] [Bug 1531] type alignment non forwarded to malloc instruction in C and C++ frontends Message-ID: <200707021747.l62HluWQ000637@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1531 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From sabre at nondot.org 2007-07-02 12:47 ------- *** This bug has been marked as a duplicate of 1362 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 13:44:00 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 13:44:00 -0500 Subject: [LLVMbugs] [Bug 1533] NEW: -indvars pass fails to rewirte the exit condition Message-ID: <200707021844.l62Ii0hB002575@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1533 Summary: -indvars pass fails to rewirte the exit condition Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: major Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: zhiruz at gmail.com //// C code //// #define N 128 // try 32 instead int array[N]; void loop(int x) { int i; for (i = N - 1; i >= 0; --i) array[i] = x; } ////////////// "llvm-gcc -c -emit-llvm -O3 loop.c -o loop.bc" followed by "opt -indvars loop.bc -o new.bc -f" will yield this: define void @loop(i32 %x) { entry: br label %bb bb: ; preds = %bb, %entry %indvar = phi i32 [ 0, %entry ], [ %indvar.next2, %bb ] ; [#uses=3] %i.010.0 = sub i32 127, %indvar ; [#uses=1] %tmp1 = getelementptr [128 x i32]* @array, i32 0, i32 %i.010.0 ; [#uses=1] store i32 %x, i32* %tmp1 %tmp4 = sub i32 126, %indvar ; [#uses=1] %tmp7 = icmp sgt i32 %tmp4, -1 ; [#uses=1] %indvar.next2 = add i32 %indvar, 1 ; [#uses=1] br i1 %tmp7, label %bb, label %return return: ; preds = %bb ret void } Interestingly, if I define N to be 32 instead (actually any integer number < 102 is good), everything will be just fine. See below: define void @loop(i32 %x) { entry: br label %bb bb: ; preds = %bb, %entry %indvar = phi i32 [ 0, %entry ], [ %indvar.next2, %bb ] ; [#uses=2] %i.010.0 = sub i32 31, %indvar ; [#uses=1] %tmp1 = getelementptr [32 x i32]* @array, i32 0, i32 %i.010.0 ; [#uses=1] store i32 %x, i32* %tmp1 %indvar.next2 = add i32 %indvar, 1 ; [#uses=2] %exitcond3 = icmp eq i32 %indvar.next2, 32 ; [#uses=1] br i1 %exitcond3, label %return, label %bb return: ; preds = %bb ret void } ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 19:55:52 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 19:55:52 -0500 Subject: [LLVMbugs] [Bug 1534] NEW: Testcase MultiSource/Application/oggenc fails (llc) with -relocation-model=pic -mcpu=i486. Message-ID: <200707030055.l630tqOT013645@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1534 Summary: Testcase MultiSource/Application/oggenc fails (llc) with -relocation-model=pic -mcpu=i486. Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: dalej at apple.com Testcase MultiSource/Application/oggenc fails (llc) with -relocation-model=pic -mcpu=i486. (This failure precedes rev 37847, an obvious suspect). ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 2 19:59:01 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Jul 2007 19:59:01 -0500 Subject: [LLVMbugs] [Bug 1505] X87 float arg passing bug Message-ID: <200707030059.l630x1Nx013772@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1505 dalej at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From dalej at apple.com 2007-07-02 19:59 ------- This should be fixed now. The two small testcases above are in the repository. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 3 13:05:44 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Jul 2007 13:05:44 -0500 Subject: [LLVMbugs] [Bug 1535] NEW: SVN Conversion Related Bugs Message-ID: <200707031805.l63I5iT0018215@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1535 Summary: SVN Conversion Related Bugs Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org Migrating to SVN is not complete. There are many things left to do and bugs related to the conversion. This is a generic bug to capture all todo items. Current Todo List: - CommandGuide doesn't exist. We should have the website behave like the old one and check out the docs and update/build. - Python is throwing an exception when viewing "Path" links for changes. (ie cpp file) - cvs is mentioned in tons of places still (ie. most docs still do) - doxygen is still using cvs - llvmgrep should ignore svn related directories - llvm-test should be renamed to whatever we have it name in svn. This impacts the nightlytester (maybe accept scripts too) and docs. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 3 13:11:40 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Jul 2007 13:11:40 -0500 Subject: [LLVMbugs] [Bug 1536] NEW: BFtoLLVM needs to be updated for LLVM 2.0 syntax changes Message-ID: <200707031811.l63IBecW018427@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1536 Summary: BFtoLLVM needs to be updated for LLVM 2.0 syntax changes Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: trivial Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: resistor at mac.com BFtoLLVM, the example Brainfuck frontend for LLVM, needs to be updated to emit LLVM 2.0-compatible .ll files. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 3 16:37:50 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Jul 2007 16:37:50 -0500 Subject: [LLVMbugs] [Bug 1536] BFtoLLVM needs to be removed Message-ID: <200707032137.l63LbotL024133@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1536 resistor at mac.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From resistor at mac.com 2007-07-03 16:37 ------- This is done as of: http://llvm.org/viewvc/llvm-project?view=revision&revision=37858 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 3 19:59:28 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Jul 2007 19:59:28 -0500 Subject: [LLVMbugs] [Bug 1537] NEW: Suspicious code in Dominators.h Message-ID: <200707040059.l640xSe8029829@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1537 Summary: Suspicious code in Dominators.h Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Global Analyses AssignedTo: unassignedbugs at nondot.org ReportedBy: zhiruz at gmail.com CC: dpatel at apple.com DominatorTreeBase::getIDomBlock 00152 /// getIDomBlock - return basic block BB's immediate dominator basic block. 00153 /// 00154 BasicBlock *getIDomBlock(BasicBlock *BB) { 00155 DomTreeNode *N = getNode(BB); 00156 assert (N && "Missing dominator tree node"); 00157 DomTreeNode *I = N->getIDom(); 00158 assert (N && "Missing immediate dominator"); 00159 return I->getBlock(); 00160 } Line 158: seems that 'I' should be asserted instead of 'N'. But what if 'N' is the root? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 3 20:06:21 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Jul 2007 20:06:21 -0500 Subject: [LLVMbugs] [Bug 1537] Suspicious code in Dominators.h Message-ID: <200707040106.l6416LiG030061@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1537 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-03 20:06 ------- Nice catch. It is expected that BB is not root. Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070702/050973.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 4 14:09:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Jul 2007 14:09:46 -0500 Subject: [LLVMbugs] [Bug 1442] Transforms/LICM/scalar_promote.ll randomly fails Message-ID: <200707041909.l64J9kI9004269@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1442 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-07-04 14:09 ------- I haven't seen this lately. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 4 17:39:31 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Jul 2007 17:39:31 -0500 Subject: [LLVMbugs] [Bug 1538] NEW: llvm2cpp should be made a target of llc Message-ID: <200707042239.l64MdVsH009869@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1538 Summary: llvm2cpp should be made a target of llc Product: tools Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: llvm2cpp AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org There is no reason for llvm2cpp to be a stand-alone tool. Its functionality should be moved into llc, as a new "cpp" target or something. In other words, use: llc -march=cpp x.bc -o x.cpp instead of: llvm2cpp x.bc -o x.cpp This makes it work more like the CBE and MSIL backends. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 4 18:06:15 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Jul 2007 18:06:15 -0500 Subject: [LLVMbugs] [Bug 1539] NEW: passmanger assertion on -analyze -indvars Message-ID: <200707042306.l64N6FoZ010575@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1539 Summary: passmanger assertion on -analyze -indvars Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca This works on any bytecode: $ opt -debug-pass=Structure -analyze -indvars b.bc -disable-output Pass Arguments: -domtree -loops -loopsimplify -lcssa -scalar-evolution -indvars -domtree -verify Target Data Layout ModulePass Manager FunctionPass Manager Dominator Tree Construction Natural Loop Construction Canonicalize natural loops Loop-Closed SSA Form Pass Scalar Evolution Analysis Loop Pass Manager Canonicalize Induction Variables -- Canonicalize Induction Variables -- Loop-Closed SSA Form Pass -- Canonicalize natural loops -- Natural Loop Construction -- Dominator Tree Construction -- Scalar Evolution Analysis 'Pass' Printer N4llvm23FunctionPassManagerImplE -- 'Pass' Printer FunctionPass Manager Dominator Tree Construction Module Verifier -- Module Verifier -- Dominator Tree Construction Printing analysis 'Canonicalize Induction Variables': opt: /home/nicholas/llvm-commit/include/llvm/PassAnalysisSupport.h:193: AnalysisType& llvm::Pass::getAnalysisID(const llvm::PassInfo*) const [with AnalysisType = llvm::Pass]: Assertion `ResultPass && "getAnalysis*() called on an analysis that was not " "'required' by pass!"' failed. llvm-commit/Debug/bin/opt((anonymous namespace)::PrintStackTrace()+0x1a)[0x86599c6] llvm-commit/Debug/bin/opt((anonymous namespace)::SignalHandler(int)+0x112)[0x8659c8c] [0xffffe500] Aborted The backtrace is this: Program received signal SIGABRT, Aborted. [Switching to Thread -137570608 (LWP 14170)] 0xffffe405 in __kernel_vsyscall () (gdb) bt #0 0xffffe405 in __kernel_vsyscall () #1 0xf7cf8d60 in raise () from /lib/i686/cmov/libc.so.6 #2 0xf7cfa5b1 in abort () from /lib/i686/cmov/libc.so.6 #3 0xf7cf242b in __assert_fail () from /lib/i686/cmov/libc.so.6 #4 0x083820aa in llvm::Pass::getAnalysisID (this=0x872b8a0, PI=0x8724c20) at /home/nicholas/llvm-commit/include/llvm/PassAnalysisSupport.h:191 #5 0x08384c91 in (anonymous namespace)::ModulePassPrinter::runOnModule ( this=0x872b8a0, M=@0x8729798) at opt.cpp:138 #6 0x085f350f in llvm::MPPassManager::runOnModule (this=0x87290f8, M=@0x8729798) at PassManager.cpp:1208 #7 0x085f36dc in llvm::PassManagerImpl::run (this=0x8728f08, M=@0x8729798) at PassManager.cpp:1281 #8 0x085f372e in llvm::PassManager::run (this=0xff9bd108, M=@0x8729798) at PassManager.cpp:1313 #9 0x0837e141 in main (argc=5, argv=0xff9bd304) at opt.cpp:397 This is the same assertion as PR1503 but an entirely different backtrace. They may or may not be the same problem. Also, "N4llvm23FunctionPassManagerImplE" should be fixed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 5 02:45:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Jul 2007 02:45:16 -0500 Subject: [LLVMbugs] [Bug 1540] NEW: LiveIntervalAnalysis assertion failure for SPARC back-end Message-ID: <200707050745.l657jGqd000440@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1540 Summary: LiveIntervalAnalysis assertion failure for SPARC back- end Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: hvdieren at elis.ugent.be CC: hvdieren at elis.ugent.be I have tried to compile LLVM bytecode for the SPEC CPU2000 benchmark bzip2 to SPARC assembly in a cross-compiler setting (LLVM running on x86_64). I am using llvm-2.0 as distributed in a tar-ball from llvm.org. Bugpoint reduced the error to this case: ; ModuleID = 'bugpoint-reduced-simplified.bc' target datalayout = "e-p:64:64:64-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" target triple = "x86_64-unknown-linux-gnu" @seedi = external global i64 ; [#uses=1] define i32 @spec_random_load(i32 %fd) { entry: br i1 false, label %bb, label %bb69 bb: ; preds = %entry br i1 false, label %cond_true12, label %bb56 cond_true12: ; preds = %bb %tmp21 = load i64* @seedi ; [#uses=1] %tmp24 = srem i64 %tmp21, 127773 ; [#uses=1] %tmp26 = mul i64 %tmp24, 16807 ; [#uses=1] %tmp29 = sub i64 %tmp26, 0 ; [#uses=1] %storemerge = add i64 0, %tmp29 ; [#uses=1] %tmp3940 = sitofp i64 %storemerge to float ; [#uses=1] %tmp41 = mul float %tmp3940, 0x3E00000000000000 ; [#uses=1] %tmp4142 = fpext float %tmp41 to double ; [#uses=1] %tmp47 = mul double %tmp4142, 2.560000e+02 ; [#uses=1] %tmp4748 = fptosi double %tmp47 to i32 ; [#uses=1] %tmp474849 = trunc i32 %tmp4748 to i8 ; [#uses=1] store i8 %tmp474849, i8* null ret i32 0 bb56: ; preds = %bb ret i32 0 bb69: ; preds = %entry ret i32 0 } the command: llc -march=sparc -f -o bzip2.s bugpoint-reduced-function.bc crashes: llc: /.../llvm/llvm-2.0/include/llvm/CodeGen/LiveIntervalAnalysis.h:121: llvm::LiveInterval& llvm::LiveIntervals::getInterval(unsigned int): Assertion `I != r2iMap_.end() && "Interval does not exist for register"' failed. Some analysis shows that the error occurs in expanded floating-point bytecodes. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 5 07:32:53 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Jul 2007 07:32:53 -0500 Subject: [LLVMbugs] [Bug 1541] NEW: incorrect code generation from LLVM to x86 asm w/r to __alloca Message-ID: <200707051232.l65CWrRY025430@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1541 Summary: incorrect code generation from LLVM to x86 asm w/r to __alloca Product: new-bugs Version: unspecified Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: nadiasvertex at gmail.com I'm using LLVM as the backend compiler for a language compiler I am running. While trying to interface to the C runtime libraries on the platform, I consistently got garbage to the display when executing puts. Running through a debugger, I see that the memory being set aside by __alloca is being lost. I have included both the LLVM and the generated x86 code, as well a comment where I think the generated code is broken. The LLVM code: %r2 = select i1 true, %type.metal_string_t * %data, %type.metal_string_t * %data ;id_val ; Copy the metal string and make it into a c-string. %r4 = getelementptr %type.metal_string_t *%r2, i32 0, i32 0 %r5 = load i32 *%r4 %r6 = add i32 %r5, 1 %r3 = alloca i8, i32 %r6 call void @llvm.memset.i32(i8 *%r3, i8 0, i32 %r6, i32 0) %r7 = getelementptr %type.metal_string_t *%r2, i32 0, i32 1 %r8 = load i8 **%r7 call void @llvm.memcpy.i32(i8 *%r3, i8 *%r8, i32 %r5, i32 0) ; End copy into a c-string. call i32 @puts(i8 * %r3) The actual x86 assembler: movl (%esi), %edi movl %edi, %eax addl $16, %eax andl $4294967280, %eax call __alloca subl $16, %esp ; THIS IS THE PROBLEM: The return value from __alloca is being lost. movl %edi, %eax incl %eax movl %eax, 8(%esp) movl $0, 4(%esp) movl %esp, %ebx movl %ebx, (%esp) call _memset addl $16, %esp movl 4(%esi), %eax subl $16, %esp movl %edi, 8(%esp) movl %eax, 4(%esp) movl %ebx, (%esp) call _memcpy addl $16, %esp subl $16, %esp movl %ebx, (%esp) call _puts ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 5 10:04:09 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Jul 2007 10:04:09 -0500 Subject: [LLVMbugs] [Bug 1342] Using ETForest in LCSSA causes bootstrap failure Message-ID: <200707051504.l65F490K029223@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1342 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-05 10:04 ------- yes. ETForest is gone. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 5 10:32:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Jul 2007 10:32:27 -0500 Subject: [LLVMbugs] [Bug 1539] passmanger assertion on -analyze -indvars Message-ID: <200707051532.l65FWRq0030074@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1539 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-05 10:32 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070702/051035.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 5 15:37:18 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Jul 2007 15:37:18 -0500 Subject: [LLVMbugs] [Bug 1541] incorrect code generation from LLVM to x86 asm w/r to __alloca Message-ID: <200707052037.l65KbImN006286@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1541 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-07-05 15:37 ------- Yeah, you're right. Fixed in: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070702/051058.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 6 16:45:29 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Jul 2007 16:45:29 -0500 Subject: [LLVMbugs] [Bug 1320] Loop Rotate does not preserve dominator info Message-ID: <200707062145.l66LjTFi022286@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1320 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-06 16:45 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070702/051107.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 6 20:19:12 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Jul 2007 20:19:12 -0500 Subject: [LLVMbugs] [Bug 1542] NEW: loop variants hoisted out of loop Message-ID: <200707070119.l671JCdG030336@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1542 Summary: loop variants hoisted out of loop Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca The attached is a miscompile was reduced by bugpoint. Run "opt bugpoint-tooptimize.bc -loop-rotate -licm -lcssa" to reproduce. I strongly recommend that anyone trying to visualize what's going on use opt --print-cfg on the before and after. The problem is in the optimization of the bb38 and bb32 loop. This sequence in bb32 is clearly decrementing a value: %tmp36 = load i32* %tmp5 %tmp37 = sub i32 %tmp36, 1 store i32 %tmp37, i32* %tmp5 however in the optimized code, the load of %tmp5 is hoisted into bb.nph outside the loop (%tmp5.promoted). ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Jul 8 04:43:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Jul 2007 04:43:24 -0500 Subject: [LLVMbugs] [Bug 1320] Loop Rotate does not preserve dominator info Message-ID: <200707080943.l689hOZv006920@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1320 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- BugsThisDependsOn| |1542 Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From asl at math.spbu.ru 2007-07-08 04:43 ------- This should be reopened, because change was backed out. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Jul 8 05:00:15 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Jul 2007 05:00:15 -0500 Subject: [LLVMbugs] [Bug 1543] NEW: LoopRotate results to broken module Message-ID: <200707081000.l68A0FxB007515@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1543 Summary: LoopRotate results to broken module Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: dpatel at apple.com ReportedBy: asl at math.spbu.ru I don't know, whether this bug is similar to PR1320 & PR1524. I'm having on TOT (with patch to PR1320 backed out): ./opt -loop-rotate -licm inflate-opt.bc Instruction does not dominate all uses! %tmp1859 = add i32 %tmp1852.promoted, 1 ; [#uses=2] %tmp1852.tmp.0 = phi i32 [ %tmp1852.tmp.11, %bb1871.preheader ], [ %tmp1859, %bb1871.loopexit ] ; [#uses=5] Everything is ok without looprotate. I'm attached bugpoint reduction & original bytecode. This prevents Qt compilation. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 9 02:06:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Jul 2007 02:06:46 -0500 Subject: [LLVMbugs] [Bug 1544] NEW: Bugpoint failed to reduce the bitcode Message-ID: <200707090706.l6976kEA021679@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1544 Summary: Bugpoint failed to reduce the bitcode Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: bugpoint AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Consider the attached bitcode. Running ./bugpoint -run-llc nsPrintEngine.bc -tool-args -mcpu=athlon -relocation-model=pic results: Read input file : 'nsPrintEngine.bc' *** All input ok Found gcc: /usr/bin/gcc Initializing execution environment: Running the code generator to test for a crash: Error running tool: ./llc -mcpu=athlon -relocation-model=pic -o bugpoint-test-program.bc-H78I0s.llc.s -f bugpoint-test-program.bc-H78I0s llc: /home/asl/proj/llvm/src/lib/Target/X86/X86FloatingPoint.cpp:768: void ::FPS::handleTwoArgFP(llvm::ilist_iterator&): Assertion `UpdatedSlot < StackTop && Dest < 7' failed. ./llc((anonymous namespace)::PrintStackTrace()+0x1f)[0x86ca4bf] /lib/libc.so.6(abort+0xec)[0xb7ca6c0c] /lib/libc.so.6(__assert_fail+0xeb)[0xb7c9e95b] ./llc((anonymous namespace)::FPS::handleTwoArgFP(llvm::ilist_iterator&)+0xd49)[0x82d0069] *** Debugging code generator crash! bugpoint: /home/asl/proj/llvm/src/lib/Bitcode/Writer/ValueEnumerator.h:68: unsigned int llvm::ValueEnumerator::getValueID(const llvm::Value*) const: Assertion `I != ValueMap.end() && "Value not in slotcalculator!"' failed. Checking to see if we can delete global inits: ./bugpoint((anonymous namespace)::PrintStackTrace()+0x1f)[0x83f898f] /lib/libc.so.6(abort+0xec)[0xb7d87c0c] /lib/libc.so.6(__assert_fail+0xeb)[0xb7d7f95b] ./bugpoint[0x81436ca] Is bugpoint broken? Or, maybe some bitcode writing machinery? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 9 02:58:58 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Jul 2007 02:58:58 -0500 Subject: [LLVMbugs] [Bug 1545] NEW: Assertion failed during codegeneration of the easy FP code Message-ID: <200707090758.l697ww56023402@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1545 Summary: Assertion failed during codegeneration of the easy FP code Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Consider the attahced testcase (reduced from Mozilla). Running ./llc -mcpu=athlon -relocation-model=pic bugpoint-reduced-simplified.bc results: llc: /home/asl/proj/llvm/src/lib/Target/X86/X86FloatingPoint.cpp:768: void ::FPS::handleTwoArgFP(llvm::ilist_iterator&): Assertion `UpdatedSlot < StackTop && Dest < 7' failed. ./llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x88e48fe] ./llc((anonymous namespace)::SignalHandler(int)+0x102)[0x88e4ba6] [0xb7fa1420] /lib/libc.so.6(abort+0xec)[0xb7d4ec0c] /lib/libc.so.6(__assert_fail+0xeb)[0xb7d4695b] ./llc((anonymous namespace)::FPS::handleTwoArgFP(llvm::ilist_iterator&)+0x5bb)[0x8540613] ./llc((anonymous namespace)::FPS::processBasicBlock(llvm::MachineFunction&, llvm::MachineBasicBlock&)+0x275)[0x853f3ef] ./llc((anonymous namespace)::FPS::runOnMachineFunction(llvm::MachineFunction&)+0x130)[0x853f130] ./llc(llvm::MachineFunctionPass::runOnFunction(llvm::Function&)+0x29)[0x8445ff9] ./llc(llvm::FPPassManager::runOnFunction(llvm::Function&)+0x158)[0x8883bcc] ./llc(llvm::FunctionPassManagerImpl::run(llvm::Function&)+0x86)[0x8883970] ./llc(llvm::FunctionPassManager::run(llvm::Function&)+0x88)[0x8883880] I never seen this assertion before (on the same code and on the same build flags), so I think it's definitely a regression. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 9 09:08:13 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Jul 2007 09:08:13 -0500 Subject: [LLVMbugs] [Bug 1546] NEW: PowerPC tests failing on Solaris Message-ID: <200707091408.l69E8DIP005276@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1546 Summary: PowerPC tests failing on Solaris Product: Test Suite Version: trunk Platform: Sun OS/Version: SunOS Status: NEW Severity: normal Priority: P2 Component: Programs Tests AssignedTo: unassignedbugs at nondot.org ReportedBy: ggreif at gmail.com (Other non-Solaris targets also fail) Here is a typical error message: FAIL: /home/ggreif/llvm/test/CodeGen/PowerPC/compare-duplicate.ll Failed with exit(1) at line 1 while running: llvm-as < /home/ggreif/llvm/test/CodeGen/PowerPC/compare-duplicate.ll | llc | not grep slwi llc: error auto-selecting target for module 'Cannot choose between targets "c" and "msil"'. Please use the -march option to explicitly pick a target. And another: FAIL: /home/ggreif/llvm/test/CodeGen/PowerPC/darwin-labels.ll Failed with exit(1) at line 1 while running: llvm-upgrade < /home/ggreif/llvm/test/CodeGen/PowerPC/darwin-labels.ll | llvm-as | llc | grep {foo bar":} llc: error auto-selecting target for module 'Cannot choose between targets "sparc" and "ppc32"'. Please use the -march option to explicitly pick a target. My impression is that explicit target triples must be added. If somebody describes the system how to do it, I can try to do this. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 9 12:26:20 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Jul 2007 12:26:20 -0500 Subject: [LLVMbugs] [Bug 1546] PowerPC tests failing on Solaris Message-ID: <200707091726.l69HQKS9013636@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1546 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-07-09 12:26 ------- Fixed, patches here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051654.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051655.html Please verify that the failures are fixed for you, thanks! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 9 14:18:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Jul 2007 14:18:33 -0500 Subject: [LLVMbugs] [Bug 1547] NEW: Documentation Requirements Message-ID: <200707091918.l69JIXep019333@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1547 Summary: Documentation Requirements Product: Website Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Documentation AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at reidspencer.com With the new website coming online soonish, we need to think about how llvm modules' documentation will be presented. Right now, everything's in one directory (llvm/trunk/docs) but with the need to integrate llvm-gcc, hlvm, etc., we need a new plan. This PR is for tracking the issues and developing a plan. Here are some of the requirements: 1. We want to retain a mapping of http://llvm.org/docs to the top level project (community) documentation, probably located the website module, or possibly the llvm-top module. This would have the stuff that applies to all modules (DeveloperPolicy, License, etc.) 2. Each module should put its module specific documentation into a directory named "docs". (i.e. .../trunk/docs). 3. Each module's documentation should appear under http://llvm.org/docs/{mod} where {mod} is the module name. These become the "sections" of the overall documentation, one per module. 4. The top level documentation should have a menu of the various modules with a link to each. 5. We want to allow users to review the documentation offline as well as online. 6. Auto-generated documentation should be as easy to set up offline as well as online. Some of the issues are: 1. With Chris' planned usage of SSI on the new web site, offline viewing could get rather ugly. The doc pages wouldn't be stand-alone any more and would not be properly formatted without the web server. One way to do it is to use XHTML frameset/frame elements on the website so that the modules' documentation is incorporated into the web site. This allows the modules' documentation to be their own complete pages and not fragments with SSI. This allows us to retain the web site "shell" (top and left bars) while also making the documentation stand alone in each module. 2. Many of the documents use keywords for things like Date, Author, Revision, etc. For example, the timestamp at the bottom of the LLVM documentation. Unfortunately, when Subversion serves these pages directly, it doesn't expand the keywords (e.g. $Date$). Consequently, we need to check out the documentation for presentation on the web site. This is not how it is currently arranged. I'm sure there's much more to discuss. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 10 14:13:18 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Jul 2007 14:13:18 -0500 Subject: [LLVMbugs] [Bug 1544] Bugpoint failed to reduce the bitcode Message-ID: <200707101913.l6AJDIOq011282@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1544 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-07-10 14:11 ------- Fixed in http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051715.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 03:57:26 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 03:57:26 -0500 Subject: [LLVMbugs] [Bug 1545] Assertion failed during codegeneration of the easy FP code Message-ID: <200707110857.l6B8vQtQ003376@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1545 evan.cheng at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From evan.cheng at apple.com 2007-07-11 03:57 ------- http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051736.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 11:40:59 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 11:40:59 -0500 Subject: [LLVMbugs] [Bug 1545] Assertion failed during codegeneration of the easy FP code Message-ID: <200707111640.l6BGex9A026080@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1545 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From asl at math.spbu.ru 2007-07-11 11:40 ------- Look into http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051742.html Testcase is attached here ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 14:30:06 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 14:30:06 -0500 Subject: [LLVMbugs] [Bug 1545] Assertion failed during codegeneration of the easy FP code Message-ID: <200707111930.l6BJU6r7005653@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1545 evan.cheng at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From evan.cheng at apple.com 2007-07-11 14:30 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051758.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 16:45:37 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 16:45:37 -0500 Subject: [LLVMbugs] [Bug 1548] NEW: Problem with packed structs on targets that doesn't support unaligned memory access Message-ID: <200707112145.l6BLjbeL010892@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1548 Summary: Problem with packed structs on targets that doesn't support unaligned memory access Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: lauro.venancio at gmail.com test case: #include struct t { char a; int b; } __attribute__ ((packed)); struct t s = { 1, 10 }; int main(){ printf("It must print 10!\n"); printf("%d\n", s.b); return 0; } emitted code (ARM): main: stmfd sp!, {r11, lr} ldr r0, .LCPI1_0 bl puts ldr r3, .LCPI1_1 ldr r1, [r3, #+1] <======== unaligned load (invalid) ldr r0, .LCPI1_2 bl printf mov r0, #0 ldmfd sp!, {r11, pc} If a value is unaligned, we must build it using some byte loads. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 18:49:08 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 18:49:08 -0500 Subject: [LLVMbugs] [Bug 1320] Loop Rotate does not preserve dominator info Message-ID: <200707112349.l6BNn8lb015336@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1320 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-11 18:49 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051774.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 18:49:38 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 18:49:38 -0500 Subject: [LLVMbugs] [Bug 1542] loop-rotate doesn't correctly update analysis info Message-ID: <200707112349.l6BNncIb015384@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1542 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-11 18:49 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051774.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 11 18:52:49 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Jul 2007 18:52:49 -0500 Subject: [LLVMbugs] [Bug 1543] LoopRotate results to broken module Message-ID: <200707112352.l6BNqnGj015560@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1543 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-11 18:52 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051774.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 01:27:02 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 01:27:02 -0500 Subject: [LLVMbugs] [Bug 1549] NEW: x86-64 backend missing some sse builtins Message-ID: <200707120627.l6C6R2fl030752@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1549 Summary: x86-64 backend missing some sse builtins Product: libraries Version: 2.0 Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org It looks like the x86-64-specific builtins are missing from emmintrin.h (and probably others). For example, this doesn't compile: #include typedef long long __m128i __attribute__ ((__vector_size__ (16))); long long foo_0(__m128i* val) { return _mm_cvtsi128_si64(*val); } long long foo_1(__m128i* val) { return __builtin_ia32_vec_ext_v2di(*val, 1); } When it is supported, we should verify that we produce decent code in these cases: http://gcc.gnu.org/ml/gcc-patches/2007-07/msg01077.html -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 08:19:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 08:19:17 -0500 Subject: [LLVMbugs] [Bug 1546] PowerPC tests failing on Solaris Message-ID: <200707121319.l6CDJHmC022270@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1546 ggreif at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 08:22:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 08:22:57 -0500 Subject: [LLVMbugs] [Bug 1546] PowerPC tests failing on Solaris Message-ID: <200707121322.l6CDMvZH022435@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1546 ggreif at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From ggreif at gmail.com 2007-07-12 08:22 ------- This resolves the remaining 3 problems: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051778.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 10:03:23 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 10:03:23 -0500 Subject: [LLVMbugs] [Bug 1550] NEW: test/CodeGen/Generic/intrinsics.ll fails on Sparc Message-ID: <200707121503.l6CF3NDj027262@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1550 Summary: test/CodeGen/Generic/intrinsics.ll fails on Sparc Product: libraries Version: trunk Platform: Sun OS/Version: SunOS Status: NEW Severity: normal Priority: P2 Component: Backend: SparcV8 AssignedTo: unassignedbugs at nondot.org ReportedBy: ggreif at gmail.com I get this error: FAIL: /home/ggreif/llvm/test/CodeGen/Generic/intrinsics.ll Failed with signal(SIGABRT) at line 1 while running: llvm-upgrade < /home/ggreif/llvm/test/CodeGen/Generic/intrinsics.ll | llvm-as | llc .text .align 16 .globl test_sqrt .type test_sqrt, #function test_sqrt: save -96, %o6, %o6 st %i0, [%i6+-4] ld [%i6+-4], %f0 fsqrts %f0, %f0 fstod %f0, %f0 fsqrtd %f0, %f0 restore %g0, %g0, %g0 retl nop Assertion failed: I != r2iMap_.end() && "Interval does not exist for register", file /home/ggreif/llvm/include/llvm/CodeGen/LiveIntervalAnalysis.h, line 102 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 10:05:44 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 10:05:44 -0500 Subject: [LLVMbugs] [Bug 1551] NEW: test/CodeGen/Generic/print-arith-fp.ll fails on Sparc Message-ID: <200707121505.l6CF5ioB027348@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1551 Summary: test/CodeGen/Generic/print-arith-fp.ll fails on Sparc Product: libraries Version: trunk Platform: Sun OS/Version: SunOS Status: NEW Severity: normal Priority: P2 Component: Backend: SparcV8 AssignedTo: unassignedbugs at nondot.org ReportedBy: ggreif at gmail.com Here is the error message: FAIL: /home/ggreif/llvm/test/CodeGen/Generic/print-arith-fp.ll Failed with signal(SIGABRT) at line 1 while running: llvm-upgrade < /home/ggreif/llvm/test/CodeGen/Generic/print-arith-fp.ll | llvm-as | llc Cannot yet select: 0x7e57a0: f64 = frem 0x7eb828, 0x7eb748 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 10:29:19 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 10:29:19 -0500 Subject: [LLVMbugs] [Bug 1552] NEW: Need to fix llvm-gcc-4.2 Message-ID: <200707121529.l6CFTJBa028255@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1552 Summary: Need to fix llvm-gcc-4.2 Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru >From Devang: Thing that require immediate attention : 1) /* FIXME FIXME */ markers. :) 2) In GCC-4.2 CONSTRUCTOR_ELTS are now represented as VEC instead of a tree node chain. 3) FILE_TYPE, CHAR_TYPE etc.. tree nodes are gone. 4) In llvm-gcc x86 world, SSE intrinsics are listed in header file, where as FSF GCC lists them in source file. When I removed list from .c file, I did not synchronize it with .h file content. I'll let x86 code gen folks worry about it :) 5) There are many GCC tree structure changes that I have not noticed yet. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 12 11:58:12 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Jul 2007 11:58:12 -0500 Subject: [LLVMbugs] [Bug 1553] NEW: syntax error on anonymous zext Message-ID: <200707121658.l6CGwC5u003365@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1553 Summary: syntax error on anonymous zext Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM assembly language parser AssignedTo: unassignedbugs at nondot.org ReportedBy: djg at cray.com For this test case: define void @hem() { %t = call i8 @hem( i8 10 ) bitcast i8 %t to i8 ret void } define void @bar() { %t = call i8 @foo( i8 10 ) zext i8 %t to i32 ret void } declare i8 @foo(i8) I get this error: llvm-as: test.ll:9,0: test.ll:9: error: syntax error, unexpected INTTYPE, expecting LOAD or STORE while reading token: 'i8' The anonymous bitcast instruction seems to be ok, which the anonymous zext instruction which is otherwise essentially syntactically the same gets an error. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 00:17:40 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 00:17:40 -0500 Subject: [LLVMbugs] [Bug 1554] NEW: llvm-extract hangs on certain files/functions Message-ID: <200707130517.l6D5He6Y012642@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1554 Summary: llvm-extract hangs on certain files/functions Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: resistor at mac.com llvm-extract seems to hang on certain bitcode file/function combination. I'm attach a testcase that hangs with: llvm-extract -func=_ZNSt5dequeIPN2kc14impl_argumentsESaIS2_EEC1ERKS4_ -f -o missedopt10.bc ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 11:25:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 11:25:03 -0500 Subject: [LLVMbugs] [Bug 1551] test/CodeGen/Generic/print-arith-fp.ll fails on Sparc Message-ID: <200707131625.l6DGP3Ng001340@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1551 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |compile-fail Resolution| |FIXED Target Milestone|--- |2.1 ------- Additional Comments From sabre at nondot.org 2007-07-13 11:25 ------- Fixed, thanks! http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051806.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 11:28:07 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 11:28:07 -0500 Subject: [LLVMbugs] [Bug 1550] test/CodeGen/Generic/intrinsics.ll fails on Sparc Message-ID: <200707131628.l6DGS7fr001447@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1550 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From sabre at nondot.org 2007-07-13 11:28 ------- *** This bug has been marked as a duplicate of 1540 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 12:01:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 12:01:57 -0500 Subject: [LLVMbugs] [Bug 1555] NEW: Upgrade all llvm assembly files Message-ID: <200707131701.l6DH1vRD002844@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1555 Summary: Upgrade all llvm assembly files Product: Test Suite Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org Instead of running llvm-upgrade on everything, we should convert all the test cases. This will speed up the dejagnu tester. We should keep the llvm-upgrade test cases though. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 12:19:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 12:19:17 -0500 Subject: [LLVMbugs] [Bug 1554] llvm-extract hangs on certain files/functions Message-ID: <200707131719.l6DHJHwU003361@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1554 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID ------- Additional Comments From sabre at nondot.org 2007-07-13 12:19 ------- Works for me. It's hanging reading stdin because you didn't specify a filename to read from ;-) -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 13:06:42 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 13:06:42 -0500 Subject: [LLVMbugs] [Bug 1556] NEW: scalarrepl miscompiles TargetData::setAlignment Message-ID: <200707131806.l6DI6g3B004926@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1556 Summary: scalarrepl miscompiles TargetData::setAlignment Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: critical Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: evan.cheng at apple.com llvm-gcc compiled llc cannot even compile hello world (x86 Mac OS X): TargetData.cpp:295: failed assertion `BestMatchIdx != -1 && "Didn't find alignment info for this datatype!"' Abort trap I tracked this down to scalarrepl miscompiling setAlignment. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 13 18:58:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Jul 2007 18:58:24 -0500 Subject: [LLVMbugs] [Bug 1540] LiveIntervalAnalysis assertion failure for SPARC back-end Message-ID: <200707132358.l6DNwOHO019260@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1540 evan.cheng at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From evan.cheng at apple.com 2007-07-13 18:58 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051824.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Jul 15 21:08:36 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Jul 2007 21:08:36 -0500 Subject: [LLVMbugs] [Bug 1533] -indvars pass fails to rewrite the exit condition Message-ID: <200707160208.l6G28aD6014107@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1533 nicholas at mxc.ca changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From nicholas at mxc.ca 2007-07-15 21:08 ------- Fixed. Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070709/051853.html Testcase: test/Analysis/ScalarEvolution/2007-07-15-NegativeStride.ll ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 16 09:02:31 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Jul 2007 09:02:31 -0500 Subject: [LLVMbugs] [Bug 1557] NEW: Inline assembly problems for generic tests on SPARC Message-ID: <200707161402.l6GE2V1J004740@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1557 Summary: Inline assembly problems for generic tests on SPARC Product: libraries Version: trunk Platform: Sun OS/Version: SunOS Status: NEW Severity: normal Priority: P2 Component: LLVM assembly language parser AssignedTo: unassignedbugs at nondot.org ReportedBy: ggreif at gmail.com I get these test 3 failures: FAIL: /home/ggreif/llvm/test/CodeGen/Generic/2007-04-08-MultipleFrameIndices.ll for PR1308 Failed with exit(1) at line 1 while running: llvm-as < /home/ggreif/llvm/test/CodeGen/Generic/2007-04-08-MultipleFrameIndices.ll | llc Couldn't allocate output reg for contraint 'r'! FAIL: /home/ggreif/llvm/test/CodeGen/Generic/2007-04-27-InlineAsm-X-Dest.ll Failed with exit(1) at line 1 while running: llvm-as < /home/ggreif/llvm/test/CodeGen/Generic/2007-04-27-InlineAsm-X-Dest.ll | llc Could not match memory address. Inline asm failure! FAIL: /home/ggreif/llvm/test/CodeGen/Generic/2007-04-27-LargeMemObject.ll Failed with exit(1) at line 1 while running: llvm-as < /home/ggreif/llvm/test/CodeGen/Generic/2007-04-27-LargeMemObject.ll | llc Could not match memory address. Inline asm failure! The first FAIL seems different in nature from #2 and #3. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 16 09:21:54 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Jul 2007 09:21:54 -0500 Subject: [LLVMbugs] [Bug 1558] NEW: 7 tests from test suit fail on Solaris when no llvm-gcc installed Message-ID: <200707161421.l6GELspD005447@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1558 Summary: 7 tests from test suit fail on Solaris when no llvm-gcc installed Product: Test Suite Version: trunk Platform: Sun OS/Version: SunOS Status: NEW Severity: normal Priority: P2 Component: QMTest AssignedTo: unassignedbugs at nondot.org ReportedBy: ggreif at gmail.com This is a low-priority bug IMHO. However, we usually avoid running tests when the corresponding binary is not accessible. But not in these cases: ggreif at my [!963] make check |& grep "running: -emit-llvm" while running: -emit-llvm -S /home/ggreif/llvm/test/CodeGen/CBackend/2007-01-06-Signless.c -o - | llvm-as | llc -march=c | grep {(unsigned short} while running: -emit-llvm -S /home/ggreif/llvm/test/CodeGen/Generic/BasicInstrs.c -o - | llvm-as | llc while running: -emit-llvm -S -O0 -g /home/ggreif/llvm/test/DebugInfo/2006-11-06-StackTrace.cpp -o - | llvm-as | llc --disable-fp-elim -o Output/2006-11-06-StackTrace.cpp.tmp.s -f while running: -emit-llvm -O0 -g -c /home/ggreif/llvm/test/DebugInfo/2006-11-20-GlobalSymbols.cpp while running: -emit-llvm -S -O0 -g /home/ggreif/llvm/test/DebugInfo/2006-11-30-NoCompileUnit.cpp -o - | llvm-as | llc --disable-fp-elim -o Output/NoCompileUnit.s -f while running: -emit-llvm -S -O0 -g /home/ggreif/llvm/test/DebugInfo/2006-11-30-Pubnames.cpp -o - | llvm-as | llc --disable-fp-elim -o Output/2006-11-30-Pubnames.cpp.tmp.s -f while running: -emit-llvm -O0 -c -g /home/ggreif/llvm/test/DebugInfo/2007-01-02-UnboundedArray.cpp The names of the tests should be obvious from the RUN: line. obviously the substitution of %llvmgcc goes awry which result in the bizarre error. What about failing the testcase when %llvmgcc is not available with a better error message? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 17 01:26:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Jul 2007 01:26:16 -0500 Subject: [LLVMbugs] [Bug 1558] 7 tests from test suit fail on Solaris when no llvm-gcc installed Message-ID: <200707170626.l6H6QGYH010846@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1558 rspencer at reidspencer.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at reidspencer.com 2007-07-17 01:26 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070716/051920.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 17 04:23:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Jul 2007 04:23:35 -0500 Subject: [LLVMbugs] [Bug 1559] NEW: Something is horrible broken in the loop optimizer Message-ID: <200707170923.l6H9NZro016291@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1559 Summary: Something is horrible broken in the loop optimizer Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru This stopped llvm-gcc build here. The process ate all memory, because LCSSA recursed infinitely. I suppose, that this wasn't caught on mainline, because I have LoopRotate disabled here (it was broken some time ago). And LoopRotate seems to hide the bug. So, now I have LoopUnswitch disabled also :) ./opt -loop-unswitch bugpoint-reduced-simplified.bc Dunno, whether it's the same bug, but anyway it segfaults :) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 18 12:11:41 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 18 Jul 2007 12:11:41 -0500 Subject: [LLVMbugs] [Bug 1560] NEW: --version should include svn revision number Message-ID: <200707181711.l6IHBf1r003686@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1560 Summary: --version should include svn revision number Product: tools Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com All llvm tools should include svn revision number in --version output. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Jul 18 18:49:05 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 18 Jul 2007 18:49:05 -0500 Subject: [LLVMbugs] [Bug 1559] loop unswitch does not preserve dom info for loop headers Message-ID: <200707182349.l6INn5Th016115@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1559 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-07-18 18:49 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070716/051996.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 01:23:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 01:23:57 -0500 Subject: [LLVMbugs] [Bug 1549] x86-64 backend missing some sse builtins Message-ID: <200707190623.l6J6Nvoo029460@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1549 isanbard at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From isanbard at gmail.com 2007-07-19 01:23 ------- With this patch: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070716/052008.html We get this: $ llvm-gcc -S -Os -msse -m64 m.c -o - .text .align 4 .globl _foo_0 _foo_0: pushq %rbp movq %rsp, %rbp movq (%rdi), %rax popq %rbp ret .align 4 .globl _foo_1 _foo_1: pushq %rbp movq %rsp, %rbp movq 8(%rdi), %rax popq %rbp ret .subsections_via_symbols $ llvm-gcc -S -Os -msse -m64 -fomit-frame-pointer m.c -o - .text .align 4 .globl _foo_0 _foo_0: movq (%rdi), %rax ret .align 4 .globl _foo_1 _foo_1: movq 8(%rdi), %rax ret .subsections_via_symbols ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 03:34:28 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 03:34:28 -0500 Subject: [LLVMbugs] [Bug 1561] NEW: g++-4.2 produces a bunch of warnings in lib/System/ltdl.c Message-ID: <200707190834.l6J8YSrU001414@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1561 Summary: g++-4.2 produces a bunch of warnings in lib/System/ltdl.c Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr lib/System/ltdl.c: In function ???try_dlopen???: lib/System/ltdl.c:3202: warning: the address of ???sys_search_path??? will always evaluate as ???true??? lib/System/ltdl.c: In function ???lt_dlopenext???: lib/System/ltdl.c:3494: warning: the address of ???archive_ext??? will always evaluate as ???true??? lib/System/ltdl.c:3515: warning: the address of ???shlib_ext??? will always evaluate as ???true??? lib/System/ltdl.c:3515: warning: the address of ???archive_ext??? will always evaluate as ???true??? lib/System/ltdl.c:3518: warning: the address of ???shlib_ext??? will always evaluate as ???true??? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 13:11:52 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 13:11:52 -0500 Subject: [LLVMbugs] [Bug 1562] NEW: ./configure ignores path settings Message-ID: <200707191811.l6JIBqgX019289@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1562 Summary: ./configure ignores path settings Product: Build scripts Version: 2.0 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: autoconf AssignedTo: unassignedbugs at nondot.org ReportedBy: jlh at gmx.ch Running ./configure --mandir=/usr/share/man --sysconfdir=/etc doesn't make it use those directories, as one would expect. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 16:18:37 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 16:18:37 -0500 Subject: [LLVMbugs] [Bug 1563] NEW: BitVector::resize does not initialize all new bits Message-ID: <200707192118.l6JLIb6Q024580@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1563 Summary: BitVector::resize does not initialize all new bits Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: resistor at mac.com CC: evan.cheng at apple.com When resizing a BitVector that does not cause new allocations (say, resizing from 16 to 20 bits), BitVector::resize fails to initial the four bits that are now being used. This can lead to inappropriate behavior, and is preventing GVNPRE from using BitVectors are aggressively as would be possible. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 18:13:19 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 18:13:19 -0500 Subject: [LLVMbugs] [Bug 1553] syntax error on anonymous zext Message-ID: <200707192313.l6JNDJbs027137@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1553 rspencer at reidspencer.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at reidspencer.com 2007-07-19 18:13 ------- Resolved with r40069 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Jul 19 20:57:01 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Jul 2007 20:57:01 -0500 Subject: [LLVMbugs] [Bug 1564] NEW: opt -adce produces incorrect code Message-ID: <200707200157.l6K1v1GC032328@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1564 Summary: opt -adce produces incorrect code Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: nick.smallbone at gmail.com I have the following module: define fastcc void @out() { start: br label %loop unreachable: unreachable loop: switch i32 0, label %unreachable [ i32 0, label %loop ] } As far as I can see, this program should be an infinite loop. However, with LLVM 2.0, running opt -adce on the module produces define fastcc void @out() { br label %unreachable unreachable: unreachable } which is an incorrect program, since the "unreachable" statement is obviously reachable. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 20 13:17:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 20 Jul 2007 13:17:46 -0500 Subject: [LLVMbugs] [Bug 1564] opt -adce produces incorrect code Message-ID: <200707201817.l6KIHk4w031171@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1564 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Jul 20 21:43:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 20 Jul 2007 21:43:22 -0500 Subject: [LLVMbugs] [Bug 1564] opt -adce produces incorrect code Message-ID: <200707210243.l6L2hMRq016865@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1564 nicholas at mxc.ca changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|INVALID | ------- Additional Comments From nicholas at mxc.ca 2007-07-20 21:43 ------- Indeed. Devang, I think your analysis was wrong on a few points: - unreachable doesn't mean that the block has no predecessors - further, the unreachable in this case is entirely correct; the default case *is* unreachable (i32 0 can't != i32 0) - if the switch was malformed, why didn't the verifier complain? - the switch doesn't pass control to two blocks, it passes it to the %loop block because that's the case that applies I'm reopening this bug. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Jul 21 03:58:20 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 21 Jul 2007 03:58:20 -0500 Subject: [LLVMbugs] [Bug 1549] x86-64 backend missing some sse builtins Message-ID: <200707210858.l6L8wK4Y028457@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1549 isanbard at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From isanbard at gmail.com 2007-07-21 03:58 ------- Need to add more x86-64 builtins. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Jul 21 16:51:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 21 Jul 2007 16:51:24 -0500 Subject: [LLVMbugs] [Bug 1556] scalarrepl miscompiles TargetData::setAlignment Message-ID: <200707212151.l6LLpOxH020681@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1556 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |miscompilation Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-07-21 16:51 ------- fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070716/052100.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Jul 22 22:10:49 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 22 Jul 2007 22:10:49 -0500 Subject: [LLVMbugs] [Bug 1549] x86-64 backend missing some sse builtins Message-ID: <200707230310.l6N3AnsI028579@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1549 isanbard at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From isanbard at gmail.com 2007-07-22 22:10 ------- Added the final SSE builtins: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070716/052141.html -bw ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 23 01:25:54 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Jul 2007 01:25:54 -0500 Subject: [LLVMbugs] [Bug 1565] NEW: Vector program fails to compile. Message-ID: <200707230625.l6N6Ps7E023618@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1565 Summary: Vector program fails to compile. Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: PowerPC AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com Reporting this as a PPC backend problem, but it also fails to compile on X86. Attaching the .c and .ll files. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 23 16:48:53 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Jul 2007 16:48:53 -0500 Subject: [LLVMbugs] [Bug 1566] NEW: self hosted llvm fails Transforms/SimplifyLibCalls/floor.ll Message-ID: <200707232148.l6NLmrsf016357@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1566 Summary: self hosted llvm fails Transforms/SimplifyLibCalls/floor.ll Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com SimplifyLibCalls.cpp is miscompiled. This is related to frame pointer elimination optimization. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 23 17:29:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Jul 2007 17:29:46 -0500 Subject: [LLVMbugs] [Bug 1567] NEW: Testcase 'CFrontend/exact-div-expr.c' Fails (sometimes) on 64bit Message-ID: <200707232229.l6NMTkV0018052@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1567 Summary: Testcase 'CFrontend/exact-div-expr.c' Fails (sometimes) on 64bit Product: Test Suite Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: chandlerc at gmail.com ReportedBy: chandlerc at gmail.com Sometimes, but not always, the CFrontend/exact-div-expr test fails. This is due to a very bizarre confluence of 64bit words and optimization passes enabled by default in llvm-gcc. The test case is ensuring that a divide by a constant power of 2 is reduced to an arithmetic shift right. To test it, pointer arithmetic is used, as this is the most common place it is used. The test case: ---- int test(int *A, int *B) { return A-B; } ---- The 32bit .ll: ---- ; ModuleID = 'CFrontend/exact-div-expr.c' target datalayout = "e-p:64:64:64-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" target triple = "x86_64-pc-linux-gnu" define i32 @test(i32* %A, i32* %B) { entry: %tmp12 = ptrtoint i32* %A to i32 ; [#uses=1] %tmp34 = ptrtoint i32* %B to i32 ; [#uses=1] %tmp5 = sub i32 %tmp12, %tmp34 ; [#uses=1] %tmp6 = ashr i32 %tmp5, 2 ; [#uses=1] ret i32 %tmp6 } ---- The 64-bit .ll: ---- ; ModuleID = 'CFrontend/exact-div-expr.c' target datalayout = "e-p:64:64:64-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" target triple = "x86_64-pc-linux-gnu" define i32 @test(i32* %A, i32* %B) { entry: %tmp12 = ptrtoint i32* %A to i64 ; [#uses=1] %tmp34 = ptrtoint i32* %B to i64 ; [#uses=1] %tmp5 = sub i64 %tmp12, %tmp34 ; [#uses=1] %tmp610 = lshr i64 %tmp5, 2 ; [#uses=1] %tmp67 = trunc i64 %tmp610 to i32 ; [#uses=1] ret i32 %tmp67 } ---- Note that the "ashr" has turned into an "lshr". This is valid because of the following "trunc" which removes any non-signed bits shifted into the left 32bits. This is a classic "int is not a word, pointers are words" problem with C. I Think the following test case will compile correctly and generate the desired code on any architecture: proposed new test case: ---- int *test(int *A, int *B) { return A-B; } ---- This should keep the same bitwidth on all operations on any architecture, and not suffer from any loss of precisions that allow unexpected optimizations to mess up the pattern matching on the code. -Chandler ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 23 17:43:58 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Jul 2007 17:43:58 -0500 Subject: [LLVMbugs] [Bug 1567] Testcase 'CFrontend/exact-div-expr.c' Fails (sometimes) on 64bit Message-ID: <200707232243.l6NMhwIq018627@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1567 chandlerc at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From chandlerc at gmail.com 2007-07-23 17:43 ------- Fixed using 'long long' in revision #40451. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 23 19:09:37 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Jul 2007 19:09:37 -0500 Subject: [LLVMbugs] [Bug 1379] Need support for eh-related builtins Message-ID: <200707240009.l6O09b3F021276@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1379 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-07-23 19:09 ------- It seems, that this was really fixed :) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 24 08:57:49 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Jul 2007 08:57:49 -0500 Subject: [LLVMbugs] [Bug 1568] NEW: segfault during domtree construction on invalid bytecode Message-ID: <200707241357.l6ODvngH007358@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1568 Summary: segfault during domtree construction on invalid bytecode Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr The verifier uses the result of domtree construction. Thus domtree construction must not crash on invalid bytecode. I hit the following crash while working on gcc-4.2, caused by a branch to a basic block which lacks a terminator. Here's the backtrace: #0 0x0881f563 in llvm::TerminatorInst::getNumSuccessors (this=0x0) at InstrTypes.h:56 #1 0x08be6c0f in llvm::DominatorTree::DFSPass (this=0x93c8e60, V=0x93d98e0, VInfo=@0x9454134, N=3) at /home/duncan/LLVM/llvm.master/lib/VMCore/Dominators.cpp:190 #2 0x08be7c66 in llvm::DominatorTree::calculate (this=0x93c8e60, F=@0x9417a70) at /home/duncan/LLVM/llvm.master/lib/VMCore/Dominators.cpp:329 #3 0x08be817a in llvm::DominatorTree::runOnFunction (this=0x93c8e60, F=@0x9417a70) at /home/duncan/LLVM/llvm.master/lib/VMCore/Dominators.cpp:594 The segfault occurs here 190 if (NextSucc == BB->getTerminator()->getNumSuccessors()) { because BB has no terminator: (gdb) call BB->dump() bb: ; Error: Block without parent! (gdb) p BB->getTerminator() $2 = (class llvm::TerminatorInst *) 0x0 I would love to attach a testcase .ll, but I failed to construct one: my examples were rejected by llvm-as. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 24 22:20:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Jul 2007 22:20:16 -0500 Subject: [LLVMbugs] [Bug 1569] NEW: PPC hazard not detected Message-ID: <200707250320.l6P3KGSo028555@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1569 Summary: PPC hazard not detected Product: libraries Version: trunk Platform: Macintosh OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: PowerPC AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca There is code in the PPCHazardRecognizer to detect mtclr / bctrl hazards, however, this sequence is generated by an indirect call and is put into one SUnit. Only the first member of the SUnit (mtclr) is passed to the hazard recognizer by the DAG scheduler and so it doesn't detect the hazard. This breaks crtbegin.o on PPC/Linux. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 24 23:16:54 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Jul 2007 23:16:54 -0500 Subject: [LLVMbugs] [Bug 1570] NEW: instcombine doesn't collapse zext/xor/zext Message-ID: <200707250416.l6P4GsE4030047@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1570 Summary: instcombine doesn't collapse zext/xor/zext Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca This function: define i32 @test2(float %X, float %Y) { entry: %tmp3 = fcmp uno float %X, %Y ; [#uses=1] %tmp34 = zext i1 %tmp3 to i8 ; [#uses=1] %tmp = xor i8 %tmp34, 1 ; [#uses=1] %toBoolnot5 = zext i8 %tmp to i32 ; [#uses=1] ret i32 %toBoolnot5 } could be optimized further. Instcombine should use its bitwise analysis to collapse the zext/xor/zext structure to an xor/zext and then remove the xor by reversing the fcmp. Desired output: define i32 @test2(float %X, float %Y) { entry: %tmp3 = fcmp ord float %X, %Y ; [#uses=1] %tmp34 = zext i1 %tmp3 to i32 ; [#uses=1] ret i32 %tmp34 } ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Jul 24 23:21:01 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Jul 2007 23:21:01 -0500 Subject: [LLVMbugs] [Bug 1571] NEW: signext attribute test ABI dependent Message-ID: <200707250421.l6P4L1uN030164@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1571 Summary: signext attribute test ABI dependent Product: Test Suite Version: trunk Platform: Macintosh OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca 2007-06-18-SextAttrAggregate.c appears to depend on the ABI of the system it's running on. On PPC/Linux, it produces a zeroext for %C instead. Is there a better fix for this than XFAILing it on architectures where it doesn't work as desired? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Jul 30 13:37:42 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Jul 2007 13:37:42 -0500 Subject: [LLVMbugs] [Bug 1585] New: Test auto email Message-ID: http://llvm.org/bugs/show_bug.cgi?id=1585 Summary: Test auto email Product: Bugzilla Admin Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Mail AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu Estimated Hours: 0.0 Testing that the auto email for new bugs works for this component. -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Mon Jul 30 13:38:26 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Jul 2007 13:38:26 -0500 Subject: [LLVMbugs] [Bug 1585] Test auto email In-Reply-To: Message-ID: <200707301838.l6UIcQRg013530@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1585 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Tue Jul 31 10:23:11 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 31 Jul 2007 10:23:11 -0500 Subject: [LLVMbugs] [Bug 1586] New: llvm.used should be in section llvm.metadata Message-ID: http://llvm.org/bugs/show_bug.cgi?id=1586 Summary: llvm.used should be in section llvm.metadata Product: cfe Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: enhancement Priority: P2 Component: Basic AssignedTo: unassignedbugs at nondot.org ReportedBy: alenhar2 at uiuc.edu CC: llvmbugs at cs.uiuc.edu All metadata used by llvm should be in the llvm.metadata section. This way a pass wanting to ignore metadata related globals, such as llvm.used, can do so in a simple and future proof way. -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Tue Jul 31 11:01:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 31 Jul 2007 11:01:16 -0500 Subject: [LLVMbugs] [Bug 1587] New: Include svn revision number in nightly tester web page report Message-ID: http://llvm.org/bugs/show_bug.cgi?id=1587 Summary: Include svn revision number in nightly tester web page report Product: Test Suite Version: trunk Platform: Macintosh OS/Version: Mac System 9.x Status: NEW Severity: enhancement Priority: P2 Component: Nightly Tester AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com CC: llvmbugs at cs.uiuc.edu Please include svn revision number in www.llvm.org/nighltlytest table. This will make it easier to understand whether particular test run includes selected patch or not. -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Tue Jul 31 16:35:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 31 Jul 2007 16:35:33 -0500 Subject: [LLVMbugs] [Bug 1588] New: Freelist Broken Assertion (JITEmitter.cpp:111) -- llvm 2. 1cvs Message-ID: http://llvm.org/bugs/show_bug.cgi?id=1588 Summary: Freelist Broken Assertion (JITEmitter.cpp:111) -- llvm 2.1cvs Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: dgu at stanford.edu CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1059) --> (http://llvm.org/bugs/attachment.cgi?id=1059) source code This bug only when running bytecode with lli version 2.1 from cvs (and not with version 2.0). The following error message is emitted: lli: JITEmitter.cpp:111: ::FreeRangeHeader*::FreeRangeHeader::RemoveFromFreeList(): Assertion `Next->Prev == this && Prev->Next == this && "Freelist broken!"' failed. lli((anonymous namespace)::PrintStackTrace()+0x19)[0x84d3559] /lib/libc.so.6(abort+0x103)[0xb7ccdea3] /lib/libc.so.6(__assert_fail+0xfb)[0xb7cc601b] lli((anonymous namespace)::JITEmitter::startFunction(llvm::MachineFunction&)+0x70)[0x820f580] Aborted When I compile and run the following C code (llvm-gcc -c main.c --emit-llvm -o main.bc && lli main.bc): void bug(); int main() { bug(); } void bug() { int zero = 0; while( zero < 1 ) if( zero == 0 ) break; } Note: Very strangely, if bug is *defined* before main, the above error message is not emitted. -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Tue Jul 31 19:54:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 31 Jul 2007 19:54:34 -0500 Subject: [LLVMbugs] [Bug 1589] New: Crash in loop optimizer Message-ID: http://llvm.org/bugs/show_bug.cgi?id=1589 Summary: Crash in loop optimizer Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: enhancement Priority: P2 Component: Loop Optimizer AssignedTo: dpatel at apple.com ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1061) --> (http://llvm.org/bugs/attachment.cgi?id=1061) bugpoint-reduced-simplified.bc It's segfaulting again... opt bugpoint-reduced-simplified.bc -licm -loop-unswitch -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.