From bugzilla-daemon at cs.uiuc.edu Mon Sep 1 07:49:53 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 07:49:53 -0500 Subject: [LLVMbugs] [Bug 2741] New: PP output loses line count with -C Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2741 Summary: PP output loses line count with -C Product: clang Version: unspecified Platform: PC OS/Version: NetBSD Status: NEW Severity: minor Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: neil at daikokuya.co.uk CC: llvmbugs at cs.uiuc.edu For example, preprocessing the following with -C puts y on the wrong line. /* a boring comment */ y Although a problem with -C, the issue is more generic, for example if the raw string literals patch is accepted it would have the same bug. -- 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 Sep 1 10:21:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 10:21:55 -0500 Subject: [LLVMbugs] [Bug 2742] New: endless trouble due to functions with weak linkage Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2742 Summary: endless trouble due to functions with weak linkage 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 CC: llvmbugs at cs.uiuc.edu Blocks: 1678,2398,2411 All kinds of passes, eg: IP constant propagation, Andersen's, GlobalsModRef, the inliner, and probably many other IP passes, are all wrong for functions with weak linkage. The reason is that such functions have a definition, but the definition may be wrong - it can be replaced with something completely different at link time. Rather than fixing up all passes, here is a suggestion for how to solve all problems in one fell swoop: (0) make it illegal for a function to have weak linkage. (1) when creating a function F which should have weak linkage, instead output it with a different name ("F2"), and create a weak alias F which links to F2. Calls to the function will be calls to F, the alias. The optimizers won't see the body for F, since there is none, so won't make mistakes. (2) at codegen time, undo this transform: output F2 with name F and with weak linkage. -- 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 Sep 1 11:32:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 11:32:11 -0500 Subject: [LLVMbugs] [Bug 2743] New: referencing a static function that isn' t defined crashes codegen Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2743 Summary: referencing a static function that isn't defined crashes codegen Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu referencing a static function that isn't defined crashes codegen. GCC promotes the static symbol to an external reference, but I think it would be better to just reject this. e.g.: static void foo(); void bar() { foo(); } -- 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 Sep 1 16:40:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 16:40:11 -0500 Subject: [LLVMbugs] [Bug 2744] New: 150x slowdown with llvm-gcc -O2 over -O1 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2744 Summary: 150x slowdown with llvm-gcc -O2 over -O1 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: nicholas at mxc.ca CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1974) --> (http://llvm.org/bugs/attachment.cgi?id=1974) testcase Attached is part of the preprocessed source for bash 3.2: $ time llvm-gcc foo.c -O2 -c real 3m33.351s user 1m52.482s sys 0m0.592s $ time llvm-gcc foo.c -O1 -c real 0m1.377s user 0m0.652s sys 0m0.014s I'm not sure which pass is responsible. The problem doesn't occur with opt -std-compile-opts. -- 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 Sep 1 17:32:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 17:32:11 -0500 Subject: [LLVMbugs] [Bug 2746] New: handling casts to union transparently Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2746 Summary: handling casts to union transparently Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu transparently casting to union is not handled by clang. Note that this is not the same as PR2015. test case (gcc accepts it, but produces a warning with -pedantic): union rqinfou { struct bio *bio; }; int vinumstart(struct bio *bio) { logrq((union rqinfou) bio, bio); } -- 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 Sep 1 17:56:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 17:56:03 -0500 Subject: [LLVMbugs] [Bug 2747] New: ptr & 0 not considered as constant Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2747 Summary: ptr & 0 not considered as constant Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu the following code is rejected by clang (with "initializer element is not a compile-time constant"): struct sym_reg { char nc_gpreg; }; int sym_fw1a_scr[] = { ((int)(&((struct sym_reg *)0)->nc_gpreg)) & 0 }; -- 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 Sep 1 18:32:46 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 18:32:46 -0500 Subject: [LLVMbugs] [Bug 2748] New: ice: Invalid extra_subreg instruction! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2748 Summary: ice: Invalid extra_subreg instruction! 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen on r55609 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp20$ llvm-gcc -Os small.c small.c:30: warning: overflow in implicit constant conversion cc1: SimpleRegisterCoalescing.cpp:1032: bool llvm::SimpleRegisterCoalescing::JoinCopy(llvm::CopyRec&, bool&): Assertion `RealDstReg && "Invalid extra_subreg instruction!"' failed. small.c:50: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp20$ cat small.c typedef short int int16_t; typedef int int32_t; typedef unsigned char uint8_t; typedef unsigned short int uint16_t; __extension__ typedef unsigned long long int uint64_t; static inline int rshift_s_s (int left, int right) { return left >> right; } static inline uint16_t safe_mod_uint8_t_u_u (uint8_t ui1, uint8_t ui2) { if (ui2) return ui1; return ui1 % ui2; } static inline uint16_t safe_sub_uint16_t_u_u (uint16_t ui1, uint16_t ui2) { } static inline uint64_t safe_sub_uint64_t_u_u (uint64_t ui1, uint64_t ui2) { return ui1 - ui2; } int32_t g_5; uint16_t g_24; int32_t g_73 = 0x2F47D66A4358B9E0LL; int32_t func_2 (int32_t p_4) { return g_5; } int32_t func_44 (int16_t p_46) { int32_t l_47; func_48 ((0xE1705295536CB472LL && (safe_mod_uint8_t_u_u ((safe_sub_uint64_t_u_u ((safe_sub_uint16_t_u_u (l_47, 1)), p_46)), g_73))), (rshift_s_s (func_2 (1), 1)) > 1, (safe_div_int32_t_s_s (p_46, (safe_sub_uint8_t_u_u ((safe_mod_uint8_t_u_u (g_24, 1)), 1))))); } -- 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 Sep 1 18:43:13 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 18:43:13 -0500 Subject: [LLVMbugs] [Bug 2749] New: C99-style struct initializers reject sizeof() Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2749 Summary: C99-style struct initializers reject sizeof() Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu The code below is rejected by clang ("error: initializer for aggregate with no elements requires explicit braces"), but accepted by gcc: struct disklabel_ops { struct {} type; int labelsize; }; struct disklabel_ops disklabel64_ops = { .labelsize = sizeof(struct disklabel_ops) }; -- 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 Sep 1 23:00:12 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 1 Sep 2008 23:00:12 -0500 Subject: [LLVMbugs] [Bug 2750] New: Extremely strange "exponent has no digits" error Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2750 Summary: Extremely strange "exponent has no digits" error Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: sharparrow1 at yahoo.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1976) --> (http://llvm.org/bugs/attachment.cgi?id=1976) Testcase I see the following behavior with the attached testcase: $ clang yy.c yy.c:10:1: error: exponent has no digits XRECORD (XRECORD (1, cons), cons); ^ This error makes absolutely no sense, and goes away if I instead run clang -E | clang. I think it's a preprocessor bug. Testcase reduced from eldap.c in xemacs (see also bug 2745). -- 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 Sep 2 04:38:13 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 04:38:13 -0500 Subject: [LLVMbugs] [Bug 2747] ptr & 0 not considered as constant In-Reply-To: Message-ID: <200809020938.m829cDTo025978@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2747 Eli Friedman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Eli Friedman 2008-09-02 04:38:13 --- Fixed in r55627. -- 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 Sep 2 06:47:25 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 06:47:25 -0500 Subject: [LLVMbugs] [Bug 2751] New: -relocation-model=static failed with -march=mips Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2751 Summary: -relocation-model=static failed with -march=mips Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: patrice.gerin at imag.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1978) --> (http://llvm.org/bugs/attachment.cgi?id=1978) test file Hi, llc failed to generate Non-relocatable code for mips. llc -march=mips -relocation-model=static test.bc -f -o test.s Cannot yet select: 0x8afc4a4: i32,ch = load 0x8afb7d8, 0x8afc30c, 0x8afba04 alignment=4 llc[0x893cf88] /lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d4ba01] llc[0x81d7515] Aborted Please find a simple test bitcode in attachment. Thanks, -- 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 Sep 2 06:50:05 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 06:50:05 -0500 Subject: [LLVMbugs] [Bug 2752] New: Ocaml bindings for parameter attributes Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2752 Summary: Ocaml bindings for parameter attributes 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: gkorf at softlab.ntua.gr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1979) --> (http://llvm.org/bugs/attachment.cgi?id=1979) Ocaml bindings for parameter attributes I attach a patch for adding ocaml bindings for parameter attributes. I think there is a bit of asymmetry in the respective APIs for function declarations and call instructions. In particular, I couldn't find how to add an attribute to the return type in a function declaration. I made a new function for that, LLVMAddFunctionAttr. I have also updated the test file, but didn't add any grep comments, because I couldn't manage the run the tests on my system. -- 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 Sep 2 09:29:47 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 09:29:47 -0500 Subject: [LLVMbugs] [Bug 2753] New: repeated structure declaration crashes clang Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2753 Summary: repeated structure declaration crashes clang Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu The following code crashes clang: void foo() { struct xpto; struct xpto; } The crash happens because both stmts point to the same decl. When destroying the stmts, both will free the same decl, and thus crashing. This may get fixed with Ted's proposal for struct stuff, but here's one more test case. -- 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 Sep 2 09:31:57 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 09:31:57 -0500 Subject: [LLVMbugs] [Bug 2724] problem with endianness of bitfields In-Reply-To: Message-ID: <200809021431.m82EVvjP003435@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2724 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #6 from Duncan Sands 2008-09-02 09:31:56 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080901/066603.html -- 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 Sep 2 09:37:50 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 09:37:50 -0500 Subject: [LLVMbugs] [Bug 2417] clang emits wrong bitcode with initialization of a global variable with a static function ptr In-Reply-To: Message-ID: <200809021437.m82EbooQ003696@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2417 Nuno Lopes changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #4 from Nuno Lopes 2008-09-02 09:37:49 --- *** This bug has been marked as a duplicate of bug 2743 *** -- 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 Sep 2 19:34:28 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 19:34:28 -0500 Subject: [LLVMbugs] [Bug 2756] New: infinite loop in instcombine after mem2reg Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2756 Summary: infinite loop in instcombine after mem2reg Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: paul at rapidmind.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1984) --> (http://llvm.org/bugs/attachment.cgi?id=1984) ll file to reproduce the infinite loop the attached ll file can be used to generate an infinite loop in instruction combining when called after a single mem2reg pass. reproduce by: 1) llvm-as inf_loop_instcombine.ll 2) opt inf_loop_instcombine.ll -mem2reg -instcombine -o=foo.bc Note: without -mem2reg or by using the default optimizations the pass runs to completion. -- 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 Sep 2 22:05:04 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 2 Sep 2008 22:05:04 -0500 Subject: [LLVMbugs] [Bug 2757] New: llc aborting Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2757 Summary: llc aborting 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen on Ubuntu Hardy on ia32. clang small.c --emit-llvm -o - 2>/dev/null | llvm-as | opt -std-compile-opts | llc llc[0x8938988] /lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d16a01] llc(_ZNK4llvm18X86GenRegisterInfo9getSubRegEjj+0x18)[0x851a768] Aborted regehr at john-home:~/volatile/tmp24$ cat small.c typedef signed char int8_t; typedef short int int16_t; typedef unsigned char uint8_t; typedef unsigned int uint32_t; __extension__ typedef unsigned long long int uint64_t; static inline int rshift_s_s (int left, int right) { return left; } static inline uint32_t safe_add_int8_t_s_s (int8_t si1, uint32_t si2) { if ((si1 > 0) && (si2 > 0) && (si1 > (127 - si2)) || (si1 < 0) && (si2 < 0) && (si1 < (-128 - si2))) return si1; return si1 + si2; } static inline int16_t safe_mod_int16_t_s_s (int16_t si1, uint32_t si2) { if ((si2) || (si1 == (-32767 - 1)) && (si2)) return si1; return si1 % si2; } static inline uint32_t safe_sub_uint8_t_u_u (uint8_t ui1, uint32_t ui2) { return ui1 - ui2; } uint32_t g_3; uint32_t g_112 = 0x9819055545B60A5FLL; uint32_t func_85 (uint32_t p_86, uint32_t p_87, uint32_t p_88) { for (1; p_86; p_86 = safe_sub_uint8_t_u_u (p_86, 1)) { for (0; p_87; p_87 = safe_add_int8_t_s_s (p_87, 1)) if (0) { } } } uint32_t func_106 (uint64_t p_107, uint32_t p_109, uint32_t p_111) { func_85 (g_112, g_112, 1); return g_3; } uint32_t func_125 (uint32_t p_126, uint32_t p_128, uint32_t p_129) { uint32_t l_138 = 0x1A8DCBC2B75F9727LL; func_104 (safe_mod_int16_t_s_s (func_106 (1, 0x96L, 1), 1 != (rshift_s_s (func_106 (1, l_138, 1), 1)))); } -- 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 Wed Sep 3 01:26:25 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 01:26:25 -0500 Subject: [LLVMbugs] [Bug 2756] infinite loop in instcombine after mem2reg In-Reply-To: Message-ID: <200809030626.m836QPIM003799@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2756 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Nick Lewycky 2008-09-03 01:26:24 --- Fixed in r55690 by disabling that transform on vectors. Let me know if you think this optz'n (X % -Y --> X % Y) is important for your vector code, let me know and I'll add the appropriate code. Apply optz'n when # of negative elements > # of strictly-positive elements. -- 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 Wed Sep 3 02:26:04 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 02:26:04 -0500 Subject: [LLVMbugs] [Bug 1678] wrong weak alias optimization with llvm-gcc-4.2 In-Reply-To: Message-ID: <200809030726.m837Q4KY005729@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1678 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Duncan Sands 2008-09-03 02:26:03 --- The #linkage section doesn't seem to exist; instead there is a section on visibility styles that's not mentioned in the contents! Anyway, I'm closing this bug because it has been fixed! The output is now: @A = alias weak void ()* @B ; [#uses=1] define i32 @active() nounwind { entry: ret i32 zext (i1 icmp ne (void ()* @A, void ()* null) to i32) } declare extern_weak void @B() -- 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 Wed Sep 3 16:06:17 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 16:06:17 -0500 Subject: [LLVMbugs] [Bug 2758] New: Objc for ... in loop wrong code generation Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2758 Summary: Objc for ... in loop wrong code generation Product: clang Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedbugs at nondot.org ReportedBy: thomascl at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1986) --> (http://llvm.org/bugs/attachment.cgi?id=1986) patch There is a small error in the fast enumeration code generation which produces an infinite loop in generated code. The counter is set to 0 at each iteration. Attached is the suggested patch. Thomas -- 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 Wed Sep 3 16:54:36 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 16:54:36 -0500 Subject: [LLVMbugs] [Bug 2730] Address of register missed In-Reply-To: Message-ID: <200809032154.m83LsaY8016453@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2730 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Daniel Dunbar 2008-09-03 16:54:35 --- Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=55739 -- 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 Wed Sep 3 17:03:38 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:03:38 -0500 Subject: [LLVMbugs] [Bug 2016] clang doesn't support gcc's computed goto tables In-Reply-To: Message-ID: <200809032203.m83M3caK016758@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2016 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #5 from Daniel Dunbar 2008-09-03 17:03:38 --- This was fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=54837 -- 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 Wed Sep 3 17:10:18 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:10:18 -0500 Subject: [LLVMbugs] [Bug 2320] clang crashes on nested enum In-Reply-To: Message-ID: <200809032210.m83MAI0t016989@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2320 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Daniel Dunbar 2008-09-03 17:10:17 --- This was fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=54463 -- 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 Wed Sep 3 17:11:18 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:11:18 -0500 Subject: [LLVMbugs] [Bug 2321] clang crashes redeclaring union tag as enum In-Reply-To: Message-ID: <200809032211.m83MBILe017048@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2321 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Daniel Dunbar 2008-09-03 17:11:17 --- I'm closing as 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 Wed Sep 3 17:19:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:19:55 -0500 Subject: [LLVMbugs] [Bug 2707] Wrong Objective-C diagnostic (incompatible pointer types) In-Reply-To: Message-ID: <200809032219.m83MJtJe017379@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2707 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #7 from Daniel Dunbar 2008-09-03 17:19:55 --- This has been fixed (although I lost track of revision). -- 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 Wed Sep 3 17:50:18 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:50:18 -0500 Subject: [LLVMbugs] [Bug 2759] New: C99 array declarators accepted in C90 mode Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2759 Summary: C99 array declarators accepted in C90 mode Product: clang Version: unspecified Platform: PC OS/Version: NetBSD Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: neil at daikokuya.co.uk CC: llvmbugs at cs.uiuc.edu Without any diagnosis as a syntax error. -- 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 Wed Sep 3 17:53:21 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 3 Sep 2008 17:53:21 -0500 Subject: [LLVMbugs] [Bug 2760] New: Clang rejects valid redeclaration Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2760 Summary: Clang rejects valid redeclaration Product: clang Version: unspecified Platform: PC OS/Version: NetBSD Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: neil at daikokuya.co.uk CC: llvmbugs at cs.uiuc.edu void f (void) { extern double *p; extern double *p; } -- 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 Thu Sep 4 04:19:48 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 04:19:48 -0500 Subject: [LLVMbugs] [Bug 2761] New: Analyser: command line option for checking GC code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2761 Summary: Analyser: command line option for checking GC code 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: roddi at mac.com CC: llvmbugs at cs.uiuc.edu The clang static analyser reports a lot of false positives for a GC-only project. There should be either a command line switch for skipping unnecessary tests, or a way for the analyser to gather that from the xcode project file -- 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 Thu Sep 4 13:45:50 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 13:45:50 -0500 Subject: [LLVMbugs] [Bug 2762] New: LegalizeTypes abort on vector cast Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2762 Summary: LegalizeTypes abort on vector cast Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: gohman at apple.com CC: llvmbugs at cs.uiuc.edu Running the following on x86 with legalize-types enabled: define void @foo(<4 x i32>* %p, <4 x double>* %q) { %n = load <4 x i32>* %p %z = sitofp <4 x i32> %n to <4 x double> store <4 x double> %z, <4 x double>* %q ret void } results in this assertion failure: Assertion failed: (Entry.first.getNode() && "Operand isn't split"), function GetSplitVector, file lib/CodeGen/SelectionDAG/LegalizeTypes.cpp, line 532. The <4 x i32> is Legal, so it isn't split, but then the <4 x double> is not Legal and needs to be split, and LegalizeTypes assumes that the operands have already been split. -- 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 Thu Sep 4 14:27:44 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 14:27:44 -0500 Subject: [LLVMbugs] [Bug 2765] New: CloneFunction Fails to correctly set Attributes Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2765 Summary: CloneFunction Fails to correctly set Attributes Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Interprocedural Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: alenhar2 at uiuc.edu CC: llvmbugs at cs.uiuc.edu CloneFunction allows the ValueMap to remove arguments from the resultant function during the clone and replace those arguments with other values. However, CloneFunction.cpp:84 copies attributes based on argument position, not taking into account that arguments may have been removed in the destination function. While on this topic, why are the ParamAttributes not attached directly to each Argument? -- 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 Thu Sep 4 16:20:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 16:20:03 -0500 Subject: [LLVMbugs] [Bug 2756] infinite loop in instcombine after mem2reg In-Reply-To: Message-ID: <200809042120.m84LK3xR013505@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2756 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #5 from Nick Lewycky 2008-09-04 16:20:03 --- Sure, I realized this morning that I had the vector version of this transform wrong (in my head) and that's why I didn't implement it. The correction way (to look at the vector component-wise) is much simpler. I'll add that. -- 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 Thu Sep 4 16:55:21 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 16:55:21 -0500 Subject: [LLVMbugs] [Bug 2758] Objc for ... in loop wrong code generation In-Reply-To: Message-ID: <200809042155.m84LtLVK014805@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2758 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Daniel Dunbar 2008-09-04 16:55:21 --- Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=55804 Thanks! -- 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 Thu Sep 4 17:26:14 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 17:26:14 -0500 Subject: [LLVMbugs] [Bug 2767] New: -C output inconsistent with -E for trigraphs Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2767 Summary: -C output inconsistent with -E for trigraphs Product: clang Version: unspecified Platform: PC OS/Version: NetBSD Status: NEW Severity: minor Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: neil at daikokuya.co.uk CC: llvmbugs at cs.uiuc.edu For the following two-line file: ??< ??> /* ??< ??> */ Clang gives: $ clang /tmp/test.c -E -C -trigraphs 2>/dev/null # 1 "/tmp/test.c" 1 # 0 "/tmp/test.c" # 1 "" 1 # 1 "/tmp/test.c" 2 { } /* ??< ??> */ GCC gives: $ cpp /tmp/test.c -E -C -trigraphs 2>/dev/null # 1 "/tmp/test.c" # 1 "" # 1 "" # 1 "/tmp/test.c" { } /* { } */ -- 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 Thu Sep 4 17:47:29 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 4 Sep 2008 17:47:29 -0500 Subject: [LLVMbugs] [Bug 2721] -indvars and -loop-unroll do not agree on names ( preventing -O2 option) In-Reply-To: Message-ID: <200809042247.m84MlTUA016695@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2721 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #4 from Devang Patel 2008-09-04 17:47:29 --- I am unable to trigger this bug using latest llvm. If you have a bitcode file that triggers this bug then please attach it with this PR. thanks. -- 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 Fri Sep 5 08:14:38 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 08:14:38 -0500 Subject: [LLVMbugs] [Bug 2770] New: crash with -relocation-model=pic Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2770 Summary: crash with -relocation-model=pic Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: rafael.espindola at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=1994) --> (http://llvm.org/bugs/attachment.cgi?id=1994) test llvm-as test.ll -o - | llc -relocation-model=pic llc: /usr/local/espindola/llvm/llvm/lib/CodeGen/LiveVariables.cpp:129: void llvm::LiveVariables::HandleVirtRegUse(unsigned int, llvm::MachineBasicBlock*, llvm::MachineInstr*): Assertion `MRI->getVRegDef(reg) && "Register use before def!"' failed. llc[0xf8f7a8] llc[0xf8f8ca] /lib/libc.so.6[0x7f3bfb339100] /lib/libc.so.6(gsignal+0x35)[0x7f3bfb339095] /lib/libc.so.6(abort+0x110)[0x7f3bfb33aaf0] /lib/libc.so.6(__assert_fail+0xef)[0x7f3bfb3322df] llc(_ZN4llvm13LiveVariables16HandleVirtRegUseEjPNS_17MachineBasicBlockEPNS_12MachineInstrE+0xd5)[0xd5e60f] llc(_ZN4llvm13LiveVariables20runOnMachineFunctionERNS_15MachineFunctionE+0x611)[0xd5fc85] llc(_ZN4llvm19MachineFunctionPass13runOnFunctionERNS_8FunctionE+0x31)[0x845b51] llc(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0xef)[0xf1e297] llc(_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE+0x74)[0xf1e816] llc(_ZN4llvm19FunctionPassManager3runERNS_8FunctionE+0x80)[0xf1e974] llc(main+0x90f)[0x7e2b1b] /lib/libc.so.6(__libc_start_main+0xf4)[0x7f3bfb3251c4] llc(atan+0xe9)[0x7e1269] Aborted -- 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 Fri Sep 5 09:41:06 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 09:41:06 -0500 Subject: [LLVMbugs] [Bug 2771] New: opt loads LLVMHello only when called from build directory, fails from installed Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2771 Summary: opt loads LLVMHello only when called from build directory, fails from installed Product: libraries Version: 2.3 Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Transformation Utilities AssignedTo: unassignedbugs at nondot.org ReportedBy: israfel33 at gmail.com CC: llvmbugs at cs.uiuc.edu Once llvm is build, opt can only be called from Release/bin/opt to load LLVMHello. If called from directory where llvm is installed, it fails. Works fine on Linux. make -C lib/Transforms/Hello Works: Release/bin/opt -load=Release/lib/LLVMHello.dylib -help|grep hello -hello - Hello World Pass -hello2 - Hello World Pass (with getAnalysisUsage implemented) Fails: /opt/llvm/bin/opt -load=Release/lib/LLVMHello.dylib -help|grep hello Error opening '/opt/llvm/lib/LLVMHello.dylib': dlopen(/opt/llvm/lib/LLVMHello.dylib, 9): Symbol not found: __ZN4llvm12FunctionPass11runOnModuleERNS_6ModuleE Referenced from: /opt/llvm/lib/LLVMHello.dylib Expected in: dynamic lookup -load request ignored. Also fails: /opt/llvm/bin/opt -load=/opt/llvm/lib/LLVMHello.dylib -help|grep hello Error opening '/opt/llvm/lib/LLVMHello.dylib': dlopen(/opt/llvm/lib/LLVMHello.dylib, 9): Symbol not found: __ZN4llvm12FunctionPass11runOnModuleERNS_6ModuleE Referenced from: /opt/llvm/lib/LLVMHello.dylib Expected in: dynamic lookup -load request ignored. -- 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 Fri Sep 5 11:49:34 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 11:49:34 -0500 Subject: [LLVMbugs] [Bug 2772] New: clang doesn't accept -c flag Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2772 Summary: clang doesn't accept -c flag Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Basic AssignedTo: unassignedbugs at nondot.org ReportedBy: nlewycky at google.com CC: llvmbugs at cs.uiuc.edu AnMaster on IRC points out that the clang driver doesn't seem to follow POSIX standard for a compiler's flags: SYNOPSIS c99 [-c][-D name[=value]]...[-E][-g][-I directory] ... [-L directory] ... [-o outfile][-Ooptlevel][-s][-U name]... operand ... notably that the -c flag is missing. -- 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 Fri Sep 5 11:53:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 11:53:11 -0500 Subject: [LLVMbugs] [Bug 2772] clang doesn't accept -c flag In-Reply-To: Message-ID: <200809051653.m85GrBqU002656@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2772 Ted Kremenek changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |kremenek at apple.com Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Ted Kremenek 2008-09-05 11:53:11 --- The executable 'clang' is not meant to be a drop in replacement for 'cc'. That's the purpose of 'ccc', which calls 'clang' with the proper arguments. 'ccc' definitely accepts -c. Eventually we will evolve the driver infrastructure to have a good cc replacement that is not implemented in python. Marking as invalid. -- 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 Fri Sep 5 13:07:24 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 13:07:24 -0500 Subject: [LLVMbugs] [Bug 2773] New: Define __USER_LABEL_PREFIX__ Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2773 Summary: Define __USER_LABEL_PREFIX__ Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu This is used by some glibc headers. From: http://gcc.gnu.org/onlinedocs/cpp/Common-Predefined-Macros.html#Common-Predefined-Macros -- __USER_LABEL_PREFIX__ This macro expands to a single token which is the prefix applied to user labels (symbols visible to C code) in assembly. For example, in the m68k-aout environment it expands to an `_', but in the m68k-coff environment it expands to nothing. This macro will have the correct definition even if -f(no-)underscores is in use, but it will not be correct if target-specific options that adjust this prefix are used (e.g. the OSF/rose -mno-underscores option). -- -- 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 Fri Sep 5 18:08:26 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 5 Sep 2008 18:08:26 -0500 Subject: [LLVMbugs] [Bug 2687] x86 Cannot yet select v2f64 = sint_to_fp In-Reply-To: Message-ID: <200809052308.m85N8Qrd015304@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2687 Eli Friedman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #4 from Eli Friedman 2008-09-05 18:08:26 --- Committed in r55864. -- 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 Sat Sep 6 10:32:31 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 6 Sep 2008 10:32:31 -0500 Subject: [LLVMbugs] [Bug 2774] New: Please update Doxygen 1.3 to the new 1.5 version Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2774 Summary: Please update Doxygen 1.3 to the new 1.5 version Product: Documentation Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Doxygen AssignedTo: unassignedbugs at nondot.org ReportedBy: nadav256 at gmail.com CC: llvmbugs at cs.uiuc.edu The current LLVM.org doxygen docs are created using Doxygen 1.3. Doxygen 1.5 is already available. Please create doxygen files using the new version. It looks much nicer and is easier to follow. -- 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 Sat Sep 6 10:52:19 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 6 Sep 2008 10:52:19 -0500 Subject: [LLVMbugs] [Bug 2775] New: ice: invalid register-register move instruction Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2775 Summary: ice: invalid register-register move instruction 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r55870 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp25$ llvm-gcc -Os small.c cc1: X86InstrInfo.cpp:684: virtual bool llvm::X86InstrInfo::isMoveInstr(const llvm::MachineInstr&, unsigned int&, unsigned int&) const: Assertion `MI.getNumOperands() >= 2 && MI.getOperand(0).isRegister() && MI.getOperand(1).isRegister() && "invalid register-register move instruction"' failed. small.c:47: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp25$ cat small.c typedef int int32_t; typedef unsigned char uint8_t; typedef unsigned int uint32_t; static inline unsigned int lshift_u_s (unsigned int left, int right) { return left << right; } static inline unsigned int rshift_u_u (unsigned int left, unsigned int right) { return left >> right; } static inline unsigned long int mod_rhs (long int rhs) { return rhs; } uint32_t g_3; uint8_t g_59; uint32_t g_75; uint32_t g_80; uint32_t g_159; int32_t func_93 (uint8_t p_94); uint32_t func_58 (uint8_t p_60, uint32_t p_61, uint8_t p_63) { uint32_t l_170; uint32_t l_180 = 7L; for (l_170 = 0; (l_170 == 0); l_170 -= 1) if ((1 % mod_rhs (((l_180 > (0 || g_159)) || g_3)))) l_180 = 1; func_93 ((mod_rhs (rshift_u_u (p_61, p_63)))); if (func_39 (1)) g_80 = (((p_63 & func_93 (1)) * (lshift_u_s (p_63, l_180)))); } int32_t func_93 (uint8_t p_94) { if (p_94) return g_59; for (0; 1; g_75--) { } } -- 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 Sat Sep 6 15:54:47 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 6 Sep 2008 15:54:47 -0500 Subject: [LLVMbugs] [Bug 2776] New: possible volatile bug -- too many loads Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2776 Summary: possible volatile bug -- too many loads 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu This is for r55873 on Ubuntu Hardy on ia32. Looks to me like the right answer is loading g_45 just one time. regehr at john-home:~/volatile/tmp26$ llvm-gcc -O1 -S --emit-llvm small.c -o foo.s ; llvm-as -f foo.s ; lli --force-interpreter --interpreter-print-volatile foo.s.bc Volatile load %0 = volatile load i32* @g_45, align 4 ; [#uses=0] Volatile load %5 = volatile load i32* @g_124, align 4 ; [#uses=1] checksum = 0 regehr at john-home:~/volatile/tmp26$ llvm-gcc -O3 -S --emit-llvm small.c -o foo.s ; llvm-as -f foo.s ; lli --force-interpreter --interpreter-print-volatile foo.s.bc Volatile load %0 = volatile load i32* @g_45, align 4 ; [#uses=0] Volatile load %1 = volatile load i32* @g_45, align 4 ; [#uses=0] Volatile load %2 = volatile load i32* @g_124, align 4 ; [#uses=0] Volatile load %1 = volatile load i32* @g_45, align 4 ; [#uses=0] checksum = 0 regehr at john-home:~/volatile/tmp26$ cat small.c #include typedef signed char int8_t; typedef short int int16_t; typedef int int32_t; typedef unsigned short int uint16_t; typedef unsigned int uint32_t; uint32_t g_20; volatile uint32_t g_45 ; int32_t g_83; volatile uint32_t g_124; volatile uint32_t g_208; int32_t func_100 (uint32_t p_103); int32_t func_100 (uint32_t p_103) { return 1; } int32_t func_119 (uint32_t p_120); int32_t func_119 (uint32_t p_120) { return 1; } int32_t func_96 (int32_t p_97); int32_t func_96 (int32_t p_97) { if ((g_45 > 1) > 1 > g_83) { func_100 (g_208); return 1; } func_119 (g_124); return 1; } void func_90 (int16_t p_92); void func_90 (int16_t p_92) { for (g_83 = 0; g_83 <= 0; g_83++) func_96 (1); } int main (void) { func_90 (1); printf ("checksum = 0\n"); return 0; } -- 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 Sat Sep 6 20:52:29 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 6 Sep 2008 20:52:29 -0500 Subject: [LLVMbugs] [Bug 2777] New: ice: this->replaceAllUsesWith(this) is NOT valid! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2777 Summary: ice: this->replaceAllUsesWith(this) is NOT valid! 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu This is seen using r55873 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp27$ llvm-gcc -O2 small.c cc1: Value.cpp:317: void llvm::Value::replaceAllUsesWith(llvm::Value*): Assertion `New != this && "this->replaceAllUsesWith(this) is NOT valid!"' failed. small.c:43: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp27$ cat small.c typedef int int32_t; typedef unsigned char uint8_t; typedef unsigned int uint32_t; static inline unsigned long int mod_rhs (long int rhs) { } static inline unsigned long int div_rhs (long int rhs) { if (rhs) return 1; } uint32_t g_103; int32_t func_135 (uint8_t p_137, uint32_t p_138, int32_t p_140); uint32_t func_127 (int32_t p_129) { uint32_t l_181; if (g_103) { return 1; } for (l_181 = 1; l_181; ++l_181) { if (p_129) { if (g_103) { } else for (1; p_129; ++p_129) { } func_135 (func_135 (1, (mod_rhs (div_rhs (p_129))), 1), 1, 1); } } } int32_t func_135 (uint8_t p_137, uint32_t p_138, int32_t p_140) { } -- 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 Sep 8 00:15:23 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 00:15:23 -0500 Subject: [LLVMbugs] [Bug 2779] New: likely loop bug Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2779 Summary: likely loop bug 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r55890 on Ubuntu Hardy on ia32. I think the -O1 output is wrong, the loop looks like it has no choice but to execute nearly 2^16 times. regehr at john-home:~/volatile/tmp28$ llvm-gcc -O0 small.c -o small regehr at john-home:~/volatile/tmp28$ ./small 65508 regehr at john-home:~/volatile/tmp28$ llvm-gcc -O1 small.c -o small regehr at john-home:~/volatile/tmp28$ ./small 36 regehr at john-home:~/volatile/tmp28$ cat small.c #include #include #include static inline int lshift_s_s(int left, int right) { if ((left < 0) || (right < 0) || (right >= sizeof(int)*CHAR_BIT) || (left > (INT_MAX >> right))) { /* Avoid undefined behavior. */ return left; } return left << right; } static inline unsigned int lshift_u_u(unsigned int left, unsigned int right) { if ((right >= sizeof(unsigned int)*CHAR_BIT) || (left > (UINT_MAX >> right))) { /* Avoid undefined behavior. */ return left; } return left << right; } int32_t func_8 (uint8_t p_9); int32_t func_8 (uint8_t p_9) { return 1; } int32_t func_3 (int8_t p_5); int32_t func_3 (int8_t p_5) { return 1; } uint32_t g_19; void func_1 (void); void func_1 (void) { uint16_t l_2; for (l_2 = -28; l_2; l_2--) { func_3 ( lshift_s_s (l_2, lshift_u_u ( func_8 (g_19++), 1) ) ); } } int main (void) { func_1 (); printf ("%d\n", g_19); return 0; } -- 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 Sep 8 04:29:46 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 04:29:46 -0500 Subject: [LLVMbugs] [Bug 2780] New: llvm-g++ prints (bad?) warning not present in g++ 4.2 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2780 Summary: llvm-g++ prints (bad?) warning not present in g++ 4.2 Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-g++ AssignedTo: unassignedbugs at nondot.org ReportedBy: rafael.espindola at gmail.com CC: llvmbugs at cs.uiuc.edu, nlewycky at google.com Given the code ---------------------------------------------------------- extern int printf (__const char *__restrict __format, ...); void g(const char *s) { printf(s); } ----------------------------------------------------------- llvm-g++ prints -------------------------------------------------------------- testcase.ii: In function 'void g(const char*)': testcase.ii:3: warning: format not a string literal and no format arguments ----------------------------------------------------------- g++ 4.2 prints no warnings with and without -Wall -- 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 Sep 8 07:28:44 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 07:28:44 -0500 Subject: [LLVMbugs] [Bug 2770] crash with -relocation-model=pic In-Reply-To: Message-ID: <200809081228.m88CSiYd001584@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2770 Rafael ??vila de Esp??ndola 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 Mon Sep 8 10:22:04 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 10:22:04 -0500 Subject: [LLVMbugs] [Bug 1565] Legalizer can't legalize invalid vector arguments yet In-Reply-To: Message-ID: <200809081522.m88FM4HS007606@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1565 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #4 from Dan Gohman 2008-09-08 10:22:02 --- The attached testcase currently aborts like this: "Call result #1 has unhandled type f32" This is being tracked with PR2660. *** This bug has been marked as a duplicate of bug 2660 *** -- 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 Sep 8 10:28:42 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 10:28:42 -0500 Subject: [LLVMbugs] [Bug 1672] legalizer doesn' t scalarize illegal vector return correctly in some cases In-Reply-To: Message-ID: <200809081528.m88FSg3t007925@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1672 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #4 from Dan Gohman 2008-09-08 10:28:41 --- This is being tracked with PR2660. *** This bug has been marked as a duplicate of bug 2660 *** -- 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 Sep 8 10:35:04 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 10:35:04 -0500 Subject: [LLVMbugs] [Bug 2127] Build Problem on Open Suse x86_64 In-Reply-To: Message-ID: <200809081535.m88FZ4G4008162@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2127 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Dan Gohman 2008-09-08 10:35:02 --- I don't have an OpenSUSE box to test on, but I'm pretty sure this specific bug was fixed, and builds on other similar configurations succeed. Please reopen this, or submit a new PR, if you see any problems building on OpenSUSE. -- 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 Sep 8 10:37:09 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 10:37:09 -0500 Subject: [LLVMbugs] [Bug 2781] New: can't add 8 to a void * pointer Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2781 Summary: can't add 8 to a void * pointer Product: tools Version: 2.2 Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: jay.foad at antixlabs.com CC: llvmbugs at cs.uiuc.edu I get this: $ cat x.c void *f(void *p) { return (void *)((int)p + 8); } $ ~/llvm/llvm-gcc4.2-2.2-x86-linux-RHEL4/libexec/gcc/i686-pc-linux-gnu/4.2.1/cc1 -quiet -O1 -emit-llvm x.c && cat x.s ; ModuleID = 'x.c' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i686-pc-linux-gnu" define i8* @f(i8* %p) nounwind { entry: %tmp34 = getelementptr i8* %p, i32 1 ; [#uses=1] ret i8* %tmp34 } The generated code is adding 1 to the pointer, not 8! -- 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 Sep 8 10:45:59 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 10:45:59 -0500 Subject: [LLVMbugs] [Bug 2197] llvm-test/MultiSource/Benchmarks/Olden/health fails on x86-64 In-Reply-To: Message-ID: <200809081545.m88Fjxgn008688@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2197 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #14 from Dan Gohman 2008-09-08 10:45:58 --- This was fixed a while ago, and Olden/health now passes on x86-64. -- 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 Sep 8 18:46:56 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 18:46:56 -0500 Subject: [LLVMbugs] [Bug 2743] referencing a static function that isn't defined crashes irgen In-Reply-To: Message-ID: <200809082346.m88NkuwK006998@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2743 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #2 from Daniel Dunbar 2008-09-08 18:46:55 --- Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=55966 -- 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 Sep 8 20:07:35 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 8 Sep 2008 20:07:35 -0500 Subject: [LLVMbugs] [Bug 2777] ice: this->replaceAllUsesWith(this) is NOT valid! In-Reply-To: Message-ID: <200809090107.m8917ZSa009803@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2777 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Devang Patel 2008-09-08 20:07:34 --- Fixed. rev. 55971 -- 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 Sep 9 07:32:24 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 07:32:24 -0500 Subject: [LLVMbugs] [Bug 2782] New: Aliases should be forwarded if possible Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2782 Summary: Aliases should be forwarded if possible 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 CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu The following testcase @A = alias void ()* @F ; [#uses=1] declare void @F() define void @C() { call void @A( ) ret void } could (and should) be turned into @A = alias void ()* @F ; [#uses=1] declare void @F() define void @C() { call void @F( ) ret void } where the call is now a direct call to F. (This is only correct if the alias does not have weak linkage or similar). Turning indirect calls into direct calls opens up new optimization opportunities. -- 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 Sep 9 12:57:51 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 12:57:51 -0500 Subject: [LLVMbugs] [Bug 2781] can't add 8 to a void * pointer In-Reply-To: Message-ID: <200809091757.m89Hvpdn025597@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2781 Bill Wendling changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |wendling at apple.com Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #4 from Bill Wendling 2008-09-09 12:57:49 --- (In reply to comment #3) > Yes. This was reduced from a more complicated example which, given an arbitrary > pointer p, did something like: > > printf("%p %p\n", p, f(p)); > > and the numbers printed differed by 1, not by 8. > When I compile this code: #include void *f(void *p) __attribute__((noinline)); void *f(void *p) { return (void*)((int)p + 8); } int main(int argc, char **argv) { printf("%p %p\n", argv, f(argv)); } I get this: define i8* @f(i8* %p) nounwind notes(inline=never) { entry: %0 = getelementptr i8* %p, i32 8 ; [#uses=1] ret i8* %0 } define i32 @main(i32 %argc, i8** %argv) nounwind { ... } with top of tree. Please use top-of-tree. -- 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 Sep 9 13:20:05 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 13:20:05 -0500 Subject: [LLVMbugs] [Bug 2645] Vector splat within loop not optimized In-Reply-To: Message-ID: <200809091820.m89IK5wj026313@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2645 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Dan Gohman 2008-09-09 13:20:04 --- I fixed a few bugs in Eli's patch and tidied it a little, and applied it, here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/066984.html Thanks for analyzing this bug and preparing the initial patch Eli! -- 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 Sep 9 14:06:22 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 14:06:22 -0500 Subject: [LLVMbugs] [Bug 2782] Aliases should be forwarded if possible In-Reply-To: Message-ID: <200809091906.m89J6M2K027937@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2782 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Anton Korobeynikov 2008-09-09 14:06:21 --- Implemented in http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/066992.html -- 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 Sep 9 14:08:05 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 14:08:05 -0500 Subject: [LLVMbugs] [Bug 2500] interpreter broken? In-Reply-To: Message-ID: <200809091908.m89J85u1028001@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2500 John Regehr changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from John Regehr 2008-09-09 14:08:05 --- Chris fixed this a while ago. -- 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 Sep 9 15:24:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 15:24:55 -0500 Subject: [LLVMbugs] [Bug 2757] llc aborting In-Reply-To: Message-ID: <200809092024.m89KOtOn030635@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2757 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |evan.cheng at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Evan Cheng 2008-09-09 15:24:54 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067001.html -- 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 Sep 9 15:54:54 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 15:54:54 -0500 Subject: [LLVMbugs] [Bug 2779] likely loop bug In-Reply-To: Message-ID: <200809092054.m89KssQV031719@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2779 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Devang Patel 2008-09-09 15:54:54 --- Fixed overflow check. revision 56011. -- 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 Sep 9 16:33:07 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 16:33:07 -0500 Subject: [LLVMbugs] [Bug 2552] FunctionPassManager crashed with function inling pass In-Reply-To: Message-ID: <200809092133.m89LX7Pl000580@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2552 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |dpatel at apple.com Status|NEW |RESOLVED Resolution| |WONTFIX --- Comment #3 from Devang Patel 2008-09-09 16:33:07 --- Added assertion check to catch this. At the moment, the pass manager does not provide any mechanism to check whether it is able to handle given pass or not. I am closing this PR for now. -- 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 Sep 9 22:48:16 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 9 Sep 2008 22:48:16 -0500 Subject: [LLVMbugs] [Bug 2783] New: likely regression causing ice: Invalid vreg! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2783 Summary: likely regression causing ice: Invalid vreg! 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu I never saw this error until tonight, and now it is happening quire frequently, meaning that it most likely is being triggered by a commit that happened during the last 24 hours. This is on Ubuntu Hardy on ia32 using r56034. LLVM is valgrind-clean on this run. regehr at john-home:~/volatile/tmp30$ llvm-gcc -O1 small.c cc1: /home/regehr/llvm/include/llvm/CodeGen/MachineRegisterInfo.h:135: const llvm::TargetRegisterClass* llvm::MachineRegisterInfo::getRegClass(unsigned int) const: Assertion `Reg < VRegInfo.size() && "Invalid vreg!"' failed. small.c:29: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp30$ cat small.c typedef unsigned short int uint16_t; typedef unsigned int uint32_t; static inline unsigned int lshift_u_s (unsigned int left, int right) { return left; } static inline int rshift_s_s (int left, int right) { if ((left < 0) || (right < 0) || (right >= sizeof (int))) return left; return left >> right; } uint16_t g_15; uint32_t func_3 (uint32_t p_5) { uint32_t l_16; uint32_t l_19; uint32_t l_22; uint32_t l_304; func_7 (rshift_s_s (((lshift_u_s (((1 & g_15) >= ((rshift_u_s (1)) <= 1)) - ((g_15 < 1) >= func_20 (1)), 1)) <= 1), (1 % mod_rhs (1)))); } -- 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 Wed Sep 10 10:04:52 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 10:04:52 -0500 Subject: [LLVMbugs] [Bug 2645] Vector splat within loop not optimized In-Reply-To: Message-ID: <200809101504.m8AF4q5S017562@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2645 Nicolas Capens changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #5 from Nicolas Capens 2008-09-10 10:04:50 --- Thanks Dan. Unfortunately the patch breaks things, in the same way Eli's original patch did. So I fear there's some incorrect transformation being made. I don't know what and where exactly, but my JIT vector application is definitely giving wrong output with this patch applied (55995) while the revision before it is fine. So I'm reopening this and will further analyze where things go bad. -- 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 Wed Sep 10 12:29:26 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 12:29:26 -0500 Subject: [LLVMbugs] [Bug 2784] New: Assert violation when generating debug info for pointer to opaque type Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2784 Summary: Assert violation when generating debug info for pointer to opaque type Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedbugs at nondot.org ReportedBy: shackan at gmail.com CC: llvmbugs at cs.uiuc.edu $ echo "struct OPAQUE; typedef struct OPAQUE *PTR; PTR p;" | clang -g -emit-llvm clang: (...)/llvm/tools/clang/lib/AST/ASTContext.cpp:487: const clang::ASTRecordLayout& clang::ASTContext::getASTRecordLayout(const clang::RecordDecl*): Assertion `D && "Cannot get layout of forward declarations!"' failed. -- 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 Wed Sep 10 13:50:57 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 13:50:57 -0500 Subject: [LLVMbugs] [Bug 2785] New: Incorrect C code produce by llc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2785 Summary: Incorrect C code produce by llc 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: akasim at swri.org CC: llvmbugs at cs.uiuc.edu When I convert llvm bytecode to c using the following command: llc -march=c file.bc file.c I'm getting some c statements which are not compilable (and/or does not make sense). Here is one such statement: __asm__ volatile ("btl %1,%2 ; setcb %b0" :"=q"(llvm_cbe_asmtmp) :"r"(llvm_cbe_tmp5)"m"(*(llvm_cbe_tmp5))"cc");; note that llvm_cbe_asmtmp is unsigned char and llvm_cbe_tmp5 is unsigned int. Shouldn't the code more be along the lines of : __asm__ volatile ("btl %1,%2 ; setcb %b0" :"=q"(llvm_cbe_asmtmp) :"r"(llvm_cbe_tmp5), "m"((llvm_cbe_tmp5)) :"cc"); -- 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 Wed Sep 10 13:51:22 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 13:51:22 -0500 Subject: [LLVMbugs] [Bug 2785] New: Incorrect C code produced by llc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2785 Summary: Incorrect C code produced by llc 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: akasim at swri.org CC: llvmbugs at cs.uiuc.edu Arif Kasim changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|Incorrect C code produce by |Incorrect C code produced by |llc |llc When I convert llvm bytecode to c using the following command: llc -march=c file.bc file.c I'm getting some c statements which are not compilable (and/or does not make sense). Here is one such statement: __asm__ volatile ("btl %1,%2 ; setcb %b0" :"=q"(llvm_cbe_asmtmp) :"r"(llvm_cbe_tmp5)"m"(*(llvm_cbe_tmp5))"cc");; note that llvm_cbe_asmtmp is unsigned char and llvm_cbe_tmp5 is unsigned int. Shouldn't the code more be along the lines of : __asm__ volatile ("btl %1,%2 ; setcb %b0" :"=q"(llvm_cbe_asmtmp) :"r"(llvm_cbe_tmp5), "m"((llvm_cbe_tmp5)) :"cc"); -- 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 Wed Sep 10 13:55:58 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 13:55:58 -0500 Subject: [LLVMbugs] [Bug 2785] Incorrect C code produced by llc In-Reply-To: Message-ID: <200809101855.m8AItwd4025056@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2785 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Anton Korobeynikov 2008-09-10 13:53:32 --- 1. Please attach bytecode in question as well 2. Inline assembler support inside CBE is really limited --- Comment #2 from Anton Korobeynikov 2008-09-10 13:55:58 --- *** This bug has been marked as a duplicate of bug 802 *** -- 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 Wed Sep 10 15:10:27 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 15:10:27 -0500 Subject: [LLVMbugs] [Bug 2664] ice In-Reply-To: Message-ID: <200809102010.m8AKARob027670@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2664 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Evan Cheng 2008-09-10 15:10:23 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067051.html -- 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 Wed Sep 10 18:04:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 18:04:55 -0500 Subject: [LLVMbugs] [Bug 2786] New: in correct branch code produced by code generator, under certain circumstances Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2786 Summary: in correct branch code produced by code generator, under certain circumstances Product: libraries Version: 2.3 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: lefever at crhc.uiuc.edu CC: llvmbugs at cs.uiuc.edu The the machine code generator produces incorrect machine code when processing the attached bitcode file. The bitcode has a main function that sets up a few variables and calls strtol___efw______augfun___(). strtol___efw______augfun___() peforms roughly as such: --------------------------------- 1: ... 2: int tmp = (memcmp(...) == 0); 3: ... 4: printf(stderr, "%d\n", tmp); 5: ... 6: if(tmp != 0){ 7: printf(stderr, "%d\n", tmp); 8: } 9: ... 10: // later code causes a seg fault ----------------------------- The problem is that lines 4 and 7 print tmp to be 1; however, we should never execute line 7, if tmp is 1. This bug is confirmed on version 2.3 of LLVM. I tried to test with the latest SVN version of LLVM, but I could not get the gcc frontend to compile on my x86-64 Linux machine. I will attach appropriate bugpoint files shortly. -- 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 Wed Sep 10 18:31:56 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 18:31:56 -0500 Subject: [LLVMbugs] [Bug 2740] Lo is not <= Hi in range emission code! In-Reply-To: Message-ID: <200809102331.m8ANVukK001548@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2740 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Dan Gohman 2008-09-10 18:31:56 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067061.html -- 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 Wed Sep 10 20:03:35 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 10 Sep 2008 20:03:35 -0500 Subject: [LLVMbugs] [Bug 2681] bad asm generated In-Reply-To: Message-ID: <200809110103.m8B13Zda004526@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2681 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Evan Cheng 2008-09-10 20:03:34 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067063.html -- 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 Thu Sep 11 00:57:49 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 11 Sep 2008 00:57:49 -0500 Subject: [LLVMbugs] [Bug 2787] New: Add demangling to PrintStackTrace Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2787 Summary: Add demangling to PrintStackTrace Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: viridia at gmail.com CC: llvmbugs at cs.uiuc.edu Here's the code snippet which does demangling on OS X. To make this work for other unix-like platforms, you'll need to change the code which extracts the symbol name from the line of text returned by backtrace symbols. So essentially, you will need to: 1) Add a test for in the configure script. 2) Optionally, you can add a test for __cxa_demangle. There are autoconfig tests for this out on the net that can be used, but personally I think this step is not necessary. 3) Write a platform-specific function to extract the name of the mangled symbol from the line of text produced by backtrace_symbols. ---------------------------------------------------------------- #if HAVE_CXXABI_H #include #endif /// .... static void PrintStackTrace(int skipFrames) { #ifdef HAVE_BACKTRACE // Use backtrace() to output a backtrace on Linux systems with glibc. int depth = backtrace(stackTrace, static_cast(arrayLengthOf(StackTrace))); #ifdef HAVE_CXXABI_H if (char ** symbols = backtrace_symbols(StackTrace, depth)) { // Name buffer used to contain demangling result. size_t sz = 256; char * buffer = (char *)malloc(sz); for (int i = 0; i < depth; ++i) { // Skip this frame, and possibly the assert machinery as well. if (i >= skipFrames) { char * symbol = symbols[i]; // TODO: This is a very cheesy way to extract the symbol name, // need to come up with something that will work on various platforms. // fprintf(outstream, "%s\n", symbol); char * begin = strchr(symbol, '_'); char * demangled_name = NULL; if (begin) { char * end = strchr(begin, ' '); if (end) { *end = 0; int status; demangled_name = abi::__cxa_demangle(begin, buffer, &sz, &status); } } if (demangled_name != NULL) { fprintf(outstream, "%s\n", demangled_name); // Result may be a realloc of input buffer. buffer = demangled_name; } } } free(symbols); free(buffer); } #else backtrace_symbols_fd(StackTrace, depth, STDERR_FILENO); #endif #endif } ---------------------------------------------------------------- -- 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 Thu Sep 11 13:41:22 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 11 Sep 2008 13:41:22 -0500 Subject: [LLVMbugs] [Bug 2783] likely regression causing ice: Invalid vreg! In-Reply-To: Message-ID: <200809111841.m8BIfMDq025888@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2783 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from Evan Cheng 2008-09-11 13:41:20 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067087.html -- 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 Thu Sep 11 15:07:53 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 11 Sep 2008 15:07:53 -0500 Subject: [LLVMbugs] [Bug 2748] ice: Invalid extra_subreg instruction! In-Reply-To: Message-ID: <200809112007.m8BK7rCF028627@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2748 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Evan Cheng 2008-09-11 15:07:52 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067093.html -- 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 Thu Sep 11 15:11:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 11 Sep 2008 15:11:11 -0500 Subject: [LLVMbugs] [Bug 2788] New: Support #line and # [0-9]+ directives Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2788 Summary: Support #line and # [0-9]+ directives Product: clang Version: unspecified Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu This is a placeholder for implementing support for preprocessor line directives. This is apparently important for Parrot as well as for correctly reparsing our own preprocessor output. -- 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 Thu Sep 11 17:51:15 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 11 Sep 2008 17:51:15 -0500 Subject: [LLVMbugs] [Bug 2645] Vector splat within loop not optimized In-Reply-To: Message-ID: <200809112251.m8BMpFBr001044@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2645 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #11 from Dan Gohman 2008-09-11 17:51:14 --- It turns out the bug affected one of our internal tests as well. I modified your attached testcase slightly and verified that it still demonstrates the bug. The fix is committed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067102.html Thanks! -- 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 Fri Sep 12 17:15:46 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 12 Sep 2008 17:15:46 -0500 Subject: [LLVMbugs] [Bug 2789] New: module does not pass 'verifierPass' after 'loop unroll' Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2789 Summary: module does not pass 'verifierPass' after 'loop unroll' Product: libraries Version: trunk Platform: PC OS/Version: Windows NT Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: asitgoes at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2013) --> (http://llvm.org/bugs/attachment.cgi?id=2013) bitcode after loop unroll pass Hi, When I'm playing with the loop unroll pass to unroll a loop whose iteration count is unknown with 'count=2', I found that after the loop unroll pass, the module has invalid phi instructions as below PHINode should have one entry for each predecessor of its parent basic block! %r6.x.0.lcssa32 = phi float [ %r6.x.0.1, %__while_52_entry.1 ] ; [#uses=1] Broken module found, compilation aborted! It seems like that at line 332 in 'Transforms/Utils/UnrollLoop.cpp', a branch is added from each necessary conditionals to the exit block, but the PHI nodes in the exit block are not updated to have incomings for the branch. I attached the bitcode after loop-unroll. Please let me know if I'm guessing wrong. Thanks. - Ji Young -- 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 Fri Sep 12 20:02:02 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 12 Sep 2008 20:02:02 -0500 Subject: [LLVMbugs] [Bug 2790] New: missing warning on return value in void function Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2790 Summary: missing warning on return value in void function Product: clang Version: unspecified Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu clang misses a warning on: -- void f2(int x) { return x; } -- -- 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 Fri Sep 12 21:45:30 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 12 Sep 2008 21:45:30 -0500 Subject: [LLVMbugs] [Bug 2791] New: possible induction variable bug Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2791 Summary: possible induction variable bug 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r56177 on Ubuntu Hardy on ia32. The -O3 result is clearly wrong. Notice that g_5 is not assigned to anywhere except in the header of the for loop. llvm-gcc is valgrind-clean on the -O3 run. regehr at john-home:~/volatile/tmp31$ llvm-gcc -O1 -w small.c -o small regehr at john-home:~/volatile/tmp31$ ./small 2 regehr at john-home:~/volatile/tmp31$ llvm-gcc -O3 -w small.c -o small regehr at john-home:~/volatile/tmp31$ ./small 1 regehr at john-home:~/volatile/tmp31$ cat small.c #include #include #include static inline unsigned long int mod_rhs(const long int rhs) { if (rhs == 0) return 1; return rhs; } static inline int rshift_s_s(int left, int right) { if ((left < 0) || (right < 0) || (right >= sizeof(int)*CHAR_BIT)) { /* Avoid implementation-defined and undefined behavior. */ return left; } return left >> right; } static inline unsigned int lshift_u_u(unsigned int left, unsigned int right) { if ((right >= sizeof(unsigned int)*CHAR_BIT) || (left > (UINT_MAX >> right))) { /* Avoid undefined behavior. */ return left; } return left << right; } static inline int lshift_s_s(int left, int right) { if ((left < 0) || (right < 0) || (right >= sizeof(int)*CHAR_BIT) || (left > (INT_MAX >> right))) { /* Avoid undefined behavior. */ return left; } return left << right; } uint32_t g_5; uint32_t g_234; uint32_t g_248; volatile uint32_t g_283; uint32_t g_286; int32_t func_12 (uint32_t p_13, uint32_t p_14, uint32_t p_15); int32_t func_12 (uint32_t p_13, uint32_t p_14, uint32_t p_15) { return 1; } int32_t func_50 (uint32_t p_51, int8_t p_53); int32_t func_50 (uint32_t p_51, int8_t p_53) { return 1; } int32_t func_55 (uint16_t p_57); int32_t func_55 (uint16_t p_57) { if (1 % mod_rhs (1)) return 1; return 1; } int32_t func_95 (int8_t p_96); int32_t func_95 (int8_t p_96) { return 1; } int32_t func_103 (uint32_t p_105); int32_t func_103 (uint32_t p_105) { return 1; } void func_8 (void); void func_8 (void) { int32_t p_9 = 1; uint32_t l_11 = 0; uint32_t l_281; if (g_5) return; uint32_t l_280; uint32_t l_282 = 1; if (((rshift_s_s (g_5 && 1, func_12 (1 & (l_280 * g_234), func_50 ((1 | l_11), l_281), (l_11 > (g_283 && p_9))))) + (lshift_u_u (g_5, 0)) == 1) + (func_55 (func_103 (&l_282) & g_248 + (lshift_s_s ((lshift_u_u (g_286, 1)), func_95 (p_9)))) > ((lshift_u_u (p_9, 1)) % mod_rhs (1)))) { } else for (1; p_9; p_9--) { } } void func_1 (void); void func_1 (void) { for (g_5 = 0; g_5 <= 1; g_5++) { func_8 (); } } int main (void) { func_1 (); printf ("%d\n", g_5); return 0; } -- 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 Fri Sep 12 22:28:10 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 12 Sep 2008 22:28:10 -0500 Subject: [LLVMbugs] [Bug 2792] New: volatile regression today Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2792 Summary: volatile regression today 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu This is seen using r56184 on Ubuntu Hardy on ia32. The bad behavior is new as of 9/12/08 and causes a very high volatile error rate in random programs. func_1() is translated properly but the copy of it that is inlined into main is incorrectly optimized away. regehr at john-home:~/volatile/tmp32$ llvm-gcc -O1 --emit-llvm -S -o - small.c ; ModuleID = 'small.c' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i386-pc-linux-gnu" @g_102 = common global i32 0 ; [#uses=1] define i32 @func_1() nounwind readonly { entry: %0 = volatile load i32* @g_102, align 4 ; [#uses=1] ret i32 %0 } define i32 @main() nounwind readonly { entry: ret i32 0 } regehr at john-home:~/volatile/tmp32$ cat small.c volatile int g_102; int func_1 (void); int func_1 (void) { return g_102; } int main (void) { func_1 (); return 0; } -- 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 Sat Sep 13 07:46:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 13 Sep 2008 07:46:55 -0500 Subject: [LLVMbugs] [Bug 2792] volatile regression today In-Reply-To: Message-ID: <200809131246.m8DCktCm013134@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2792 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Duncan Sands 2008-09-13 07:46:54 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080908/067135.html -- 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 Sat Sep 13 19:05:58 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 13 Sep 2008 19:05:58 -0500 Subject: [LLVMbugs] [Bug 2793] New: --docdir=DIR listed but not supported Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2793 Summary: --docdir=DIR listed but not supported Product: Build scripts Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: autoconf AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca CC: llvmbugs at cs.uiuc.edu LLVM configure --help lists: --docdir=DIR documentation root [DATAROOTDIR/doc/-llvm-] but trying to set it is futile: $ ./configure --docdir=/usr/share/doc/llvm-svn [...] === configuring in projects/sample (/home/nicholas/llvm-commit/projects/sample) configure: running /bin/sh ./configure --prefix=/usr/local '--docdir=/usr/share/doc/llvm-svn' --cache-file=/dev/null --srcdir=. configure: error: unrecognized option: --docdir=/usr/share/doc/llvm-svn This was originally spotted on https://trac.macports.org/ticket/15724 . -- 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 Sun Sep 14 08:29:24 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 14 Sep 2008 08:29:24 -0500 Subject: [LLVMbugs] [Bug 1820] Macro expansion bug In-Reply-To: Message-ID: <200809141329.m8EDTOeL003274@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1820 Neil Booth changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #2 from Neil Booth 2008-09-14 08:29:22 --- I noticed that clang treats keywords specially and differently to identifiers. This bug was fixed for non-keyword identifiers, but is unfixed for keywords. There may be other "tok::identifier" bugs in clang's preprocessor; it's tricky to tell at-a-glance. -- 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 Sun Sep 14 11:01:25 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 14 Sep 2008 11:01:25 -0500 Subject: [LLVMbugs] [Bug 2794] New: ISD:: SIGN_EXTEND_INREG added during Legalization and not Legalized Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2794 Summary: ISD::SIGN_EXTEND_INREG added during Legalization and not Legalized 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: rich at pennware.com CC: llvmbugs at cs.uiuc.edu LegalizeDAG.cpp adds a ISD::SIGN_EXTEND_INREG node during promotion but doesn't make sure to verify that the resulting DAG is legal for the target. The fix is: Index: LegalizeDAG.cpp =================================================================== --- LegalizeDAG.cpp (revision 56194) +++ LegalizeDAG.cpp (working copy) @@ -4670,6 +4670,8 @@ DAG.getValueType(VT)); Tmp2 = DAG.getNode(ISD::SIGN_EXTEND_INREG, NVT, Tmp2, DAG.getValueType(VT)); + Tmp1 = LegalizeOp(Tmp1); + Tmp2 = LegalizeOp(Tmp2); break; } } -- 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 Sun Sep 14 12:01:20 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 14 Sep 2008 12:01:20 -0500 Subject: [LLVMbugs] [Bug 2795] New: llc: assertion failed in Support/Recycler.h:89 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2795 Summary: llc: assertion failed in Support/Recycler.h:89 Product: tools Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: sothman at web.de CC: llvmbugs at cs.uiuc.edu Assertion failed: AlignOf::Alignment <= Align && "Recycler allocation alignment is less than object alignment!", file k:/sources/llvm.git/include/llvm/Support/Recycler.h, line 89 llvm rev 56195, win32, compiled via msys/mingw used to work in rev 55777 fails for most simple testcases: --------- ; ModuleID = 'test' 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-f80:32:32-v64:64:64-v128:128:128-a0:0:64" target triple = "i686-pc-mingw32" define i32 @main(i32 %argc, i8** %argv) nounwind { ret i32 undef } --------- -- 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 Sep 15 16:13:37 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 15 Sep 2008 16:13:37 -0500 Subject: [LLVMbugs] [Bug 2751] -relocation-model=static failed with -march=mips In-Reply-To: Message-ID: <200809152113.m8FLDbpK014212@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2751 Bruno Cardoso Lopes changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Bruno Cardoso Lopes 2008-09-15 16:13:33 --- Hi Patrice, Fixed : http://llvm.org/viewvc/llvm-project?view=rev&revision=56228 I suggest that you generate Mips bytecode instead of x86, this can be accomplished using cc1 generated by the llvm-gcc mips cross-compilation, even when the compilation breaks. If you configured llvm-gcc obj dir as /p/llvm-gcc/mips/obj/, you'll find cc1 in /p/llvm-gcc/mips/obj/gcc/cc1 after the compilation breaks. So you must use it to generate the mips bytecode. Using the x86 bytecode as input for the mips back-end (see the target-triple of the bytecode you attached) can lead to strange behaviors. Thanks. :) -- 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 Sep 15 19:08:40 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 15 Sep 2008 19:08:40 -0500 Subject: [LLVMbugs] [Bug 2797] New: Internal error: Segmentation fault (program cc1) Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2797 Summary: Internal error: Segmentation fault (program cc1) 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r56232 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp34$ llvm-gcc -O0 small.c llvm-gcc: Internal error: Segmentation fault (program cc1) Please submit a full bug report. See for instructions. regehr at john-home:~/volatile/tmp34$ cat small.c unsigned int func_48 (signed char p_49) { signed char l_340; func_44 (&((1 ^ 1 == (lshift_u_s (1)) != (l_340 < 1)) & 1L)); } -- 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 Sep 15 19:10:26 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 15 Sep 2008 19:10:26 -0500 Subject: [LLVMbugs] [Bug 2797] New: Internal error: Segmentation fault (program cc1) Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2797 Summary: Internal error: Segmentation fault (program cc1) 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: regehr at cs.utah.edu CC: nicholas at mxc.ca, llvmbugs at cs.uiuc.edu Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |nicholas at mxc.ca Seen using r56232 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp34$ llvm-gcc -O0 small.c llvm-gcc: Internal error: Segmentation fault (program cc1) Please submit a full bug report. See for instructions. regehr at john-home:~/volatile/tmp34$ cat small.c unsigned int func_48 (signed char p_49) { signed char l_340; func_44 (&((1 ^ 1 == (lshift_u_s (1)) != (l_340 < 1)) & 1L)); } -- 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 Sep 16 07:16:56 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 16 Sep 2008 07:16:56 -0500 Subject: [LLVMbugs] [Bug 2799] New: LLC fails to emit MIPS assembly code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2799 Summary: LLC fails to emit MIPS assembly code Product: new-bugs Version: unspecified Platform: PC OS/Version: Windows XP Status: NEW Keywords: compile-fail Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: ihusar at fit.vutbr.cz CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2019) --> (http://llvm.org/bugs/attachment.cgi?id=2019) failing bc file I compiled a simple "Hello world" example i C language to LLVM IR. Now when I use the LLC to emit assembly for MIPS architecture, the LLC fails. 1) $cat test.c #include int main() { printf("Ahoj"); return 0; } 2) $llvm-gcc test.c -emit-llvm -c -o test.bc 3) $llc -march=mips test.bc terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc ./llc[0x882472f] Aborted All the other supported architectures (x86, x86-4, alpha, sparc, ppc32, ppc64, arm, thumb, cellspu) work fine and a correct assembly file is generated. ----------------------------------------------------------------------- Here are listed versions of tools/OS I used: $llvm-gcc --version llvm-gcc (GCC) 4.2.1 (Based on Apple Inc. build 5555) (LLVM build 2.3) $llc --version Low Level Virtual Machine (http://llvm.org/): llvm version 2.3 Optimized build with assertions. $cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 65 model name : Dual-Core AMD Opteron(tm) Processor 2216 ... $cat /etc/issue CentOS release 5 (Final) -- 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 Sep 16 09:01:02 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 16 Sep 2008 09:01:02 -0500 Subject: [LLVMbugs] [Bug 2800] New: simplifycfg turns IR invalid Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2800 Summary: simplifycfg turns IR invalid 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: tomas.l.olsen at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2020) --> (http://llvm.org/bugs/attachment.cgi?id=2020) bitcode from just before the simplifycfg pass that turns it invalid I have a module which seems to be working fine, it verifies, compiles with llc etc. However a specific optimization route turns the IR invalid on the simplifycfg pass. Steps to reproduce: opt -o=foo.bc -simplifycfg SharedLib.bc -- 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 Sep 16 11:12:42 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 16 Sep 2008 11:12:42 -0500 Subject: [LLVMbugs] [Bug 2799] LLC fails to emit MIPS assembly code In-Reply-To: Message-ID: <200809161612.m8GGCgEY001605@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2799 Bruno Cardoso Lopes changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Bruno Cardoso Lopes 2008-09-16 11:12:42 --- Hi, This bug is solved in trunk. http://llvm.org/viewvc/llvm-project?view=rev&revision=51899 Thanks. *** This bug has been marked as a duplicate of bug 2543 *** -- 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 Wed Sep 17 05:33:14 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 05:33:14 -0500 Subject: [LLVMbugs] [Bug 2801] New: x86-64 doesn't save nonvolatile XMM6-XMM15 registers Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2801 Summary: x86-64 doesn't save nonvolatile XMM6-XMM15 registers 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: nicolas at capens.net CC: llvmbugs at cs.uiuc.edu The x86-64 ABI specifies that XMM6 to XMM15 are non-volatile, and should be preserved by the callee as needed. It appears that LLVM currently doesn't do this, causing unpredictable behavior with floating-point calculations. -- 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 Wed Sep 17 06:59:15 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 06:59:15 -0500 Subject: [LLVMbugs] [Bug 2802] New: online demo bug Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2802 Summary: online demo bug Product: new-bugs Version: unspecified Platform: All URL: http://llvm.org/demo/ OS/Version: All Status: NEW Keywords: quality-of-implementation Severity: trivial Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: philcolbourn at gmail.com CC: llvmbugs at cs.uiuc.edu argv[0] should be argv[1] in the demo source code. -- 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 Wed Sep 17 09:08:31 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 09:08:31 -0500 Subject: [LLVMbugs] [Bug 2760] Clang rejects valid redeclaration In-Reply-To: Message-ID: <200809171408.m8HE8Vms030228@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2760 snaroff at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from snaroff at apple.com 2008-09-17 09:08:30 --- Fixed in http://llvm.org/viewvc/llvm-project?view=rev&revision=56280. Thanks Neil. -- 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 Wed Sep 17 12:54:13 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 12:54:13 -0500 Subject: [LLVMbugs] [Bug 2791] possible induction variable bug In-Reply-To: Message-ID: <200809171754.m8HHsDfd006555@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2791 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from Devang Patel 2008-09-17 12:54:12 --- Fixed. Committed revision 56284. -- 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 Wed Sep 17 13:22:09 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 13:22:09 -0500 Subject: [LLVMbugs] [Bug 2800] simplifycfg turns IR invalid In-Reply-To: Message-ID: <200809171822.m8HIM9bj007578@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2800 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Devang Patel 2008-09-17 13:22:09 --- Fixed. Committed revision 56286. -- 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 Wed Sep 17 13:23:43 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 13:23:43 -0500 Subject: [LLVMbugs] [Bug 2803] New: using x86.mmx. emms intrinsic can clobber results returned on the fp stack Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2803 Summary: using x86.mmx.emms intrinsic can clobber results returned on the fp stack Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: paul at rapidmind.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2023) --> (http://llvm.org/bugs/attachment.cgi?id=2023) ll file to reproduce bug To work around bug 1838 (x86 backend doesn't insert emms when it is needed) I've manually insert calls to x86.mmx.emms before calling functions that use the floating point stack. e.g., ... %45 = extractelement <2 x float> %39, i32 1 ; [#uses=1] %46 = extractelement <2 x float> %40, i32 1 ; [#uses=1] call void @llvm.x86.mmx.emms( ) %47 = call float @fmaxf( float %45, float %46 ) ; [#uses=1] %48 = insertelement <2 x float> %44, float %47, i32 1 ; <<2 x float>> [#uses=2] ... and here is a snippet of generated code: call fmaxf fstpt 88(%esp) movss 160(%esp), %xmm0 movss %xmm0, 4(%esp) movss 144(%esp), %xmm0 movss %xmm0, (%esp) call fminf emms fstps 212(%esp) movss 212(%esp), %xmm0 movss %xmm0, 128(%esp) movss %xmm0, 4(%esp) fldt 100(%esp) fstps 232(%esp) movss 232(%esp), %xmm0 movss %xmm0, 112(%esp) movss %xmm0, (%esp) call fminf I'm not exactly sure what's going on but either 1) emms is moving after the call to fminf 2) emms is moving before the result of a previous fminf is popped from the floating point stack and clobbering it 3) both? Sorry for providing such a large test program but in the generated code sometimes the emms is in the right spot and sometimes 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 Wed Sep 17 13:56:54 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 13:56:54 -0500 Subject: [LLVMbugs] [Bug 2804] New: Cannot select bitcast of <2 x i32> to double on x86 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2804 Summary: Cannot select bitcast of <2 x i32> to double on x86 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: sharparrow1 at yahoo.com CC: llvmbugs at cs.uiuc.edu Testcase: define double @z(<2 x i32> %a) nounwind { entry: %cast = bitcast <2 x i32> %a to double ret double %cast } Run through llc on x86, gives "Cannot yet select: 0x8b3aa64: f64 = bit_convert 0x8b3ac84". -- 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 Wed Sep 17 17:53:07 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 17 Sep 2008 17:53:07 -0500 Subject: [LLVMbugs] [Bug 2802] online demo bug In-Reply-To: Message-ID: <200809172253.m8HMr7rZ018273@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2802 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Tanya Lattner 2008-09-17 17:53:06 --- Fixed: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080915/067219.html -- 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 Thu Sep 18 08:40:23 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 08:40:23 -0500 Subject: [LLVMbugs] [Bug 2805] New: ice: Unexpected split condition predicate Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2805 Summary: ice: Unexpected split condition predicate 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r56295 on Ubuntu Hardy in ia32. regehr at john-home:~/volatile/tmp39$ llvm-gcc -O1 small.c cc1: LoopIndexSplit.cpp:1303: void::LoopIndexSplit::calculateLoopBounds(::LoopIndexSplit::SplitInfo&): Assertion `0 && "Unexpected split condition predicate"' failed. small.c:16: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp39$ cat small.c typedef unsigned int uint32_t; uint32_t g_330; uint32_t func_45 (uint32_t p_47) { for (p_47 = 0; (p_47 < 22); ++p_47) { if (p_47) { func_70 (1); } else g_330 = 1; } func_56 (func_95 (p_47)); } -- 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 Thu Sep 18 17:45:52 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 17:45:52 -0500 Subject: [LLVMbugs] [Bug 2806] New: assertion failure in DCE. cpp when executed in debug mode on win32 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2806 Summary: assertion failure in DCE.cpp when executed in debug mode on win32 Product: libraries Version: trunk Platform: PC OS/Version: Windows NT Status: NEW Severity: minor Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: asitgoes at gmail.com CC: llvmbugs at cs.uiuc.edu Hi, The following code in DCE.cpp(line 113 to 118) causes an assertion failure in 'std::vector' operator--(). for (std::vector::iterator WI = WorkList.begin(); WI != WorkList.end(); ++WI) if (*WI == I) { WI = WorkList.erase(WI); --WI; } When 'WorkList' becomes empty after WorkList.erase, 'WI' will point to the beginning of the internal buffer of the std::vector. Thus, '--WI' will make 'WI' point to an invalid address. I believe that this will function correctly since 'WI' will be then incremented by '++WI' of the for loop, however, the 'std::vector' implementation of the visual studio detects that 'WI' will points to an invalid address after '--' operation, and raises an assertion failure. I'd suggest to change this code into for (std::vector::iterator WI = WorkList.begin(); WI != WorkList.end(); ) if (*WI == I) { WI = WorkList.erase(WI); } else { ++WI; } Please let me know if you don't think this need to be fixed. Thanks. -- 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 Thu Sep 18 18:04:46 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 18:04:46 -0500 Subject: [LLVMbugs] [Bug 2806] assertion failure in DCE. cpp when executed in debug mode on win32 In-Reply-To: Message-ID: <200809182304.m8IN4kOW015059@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2806 Bill Wendling changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |wendling at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Bill Wendling 2008-09-18 18:04:46 --- Patch applied here: http://llvm.org/viewvc/llvm-project?rev=56316&view=rev Thanks! -- 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 Thu Sep 18 18:45:39 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 18:45:39 -0500 Subject: [LLVMbugs] [Bug 2805] ice: Unexpected split condition predicate In-Reply-To: Message-ID: <200809182345.m8INjdg2016724@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2805 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Devang Patel 2008-09-18 18:45:38 --- Fixed. Committed revision 56321. -- 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 Thu Sep 18 21:54:46 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 21:54:46 -0500 Subject: [LLVMbugs] [Bug 2807] New: CellSPU cannot select fdiv Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2807 Summary: CellSPU cannot select fdiv 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: ph.silva at gmail.com CC: llvmbugs at cs.uiuc.edu This code: ; ModuleID = '' define double @div(double, double) nounwind { entry: %2 = fdiv double %0, %1 ; [#uses=1] ret double %2 } fails to compile with: $ llvm-as < div.ll | llc -march=cellspu Cannot yet select: 0x8a20ab4: f64 = fdiv 0x8a20cd4, 0x8a20de4 llc[0x88b1775] Aborted While writing this report, I found a comment on SPUInstrInfo.td asking for info about how to map fdiv and other to Cell. GCC folks use some algorithms "taken from STI's gcc" (see spu.md on GCC tree), but I don't know if GPL code is welcome on a BSD-like licensed project. The mentioned commented make this bug a "current feature", anyway would be good to work on this. -- 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 Thu Sep 18 23:56:55 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 18 Sep 2008 23:56:55 -0500 Subject: [LLVMbugs] [Bug 2808] New: ice: attempt to spill already spilled interval! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2808 Summary: ice: attempt to spill already spilled interval! 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r56321 on Ubuntu Hardy on ia32. regehr at john-home:~/volatile/tmp41$ llvm-gcc -O2 small.c small.c: In function ???func_4???: small.c:19: warning: passing argument 2 of ???rshift_u_s??? makes integer from pointer without a cast cc1: LiveIntervalAnalysis.cpp:1803: std::vector > llvm::LiveIntervals::addIntervalsForSpills(const llvm::LiveInterval&, const llvm::MachineLoopInfo*, llvm::VirtRegMap&, float&): Assertion `li.weight != (__builtin_huge_valf()) && "attempt to spill already spilled interval!"' failed. small.c: At top level: small.c:20: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. regehr at john-home:~/volatile/tmp41$ cat small.c int rshift_u_s (int left, int right) { if (right < 0 || right >= 4) return left; return left >> right; } unsigned int mod_rhs (int rhs) { return rhs; } int g_3; int func_4 (void) { char l_2 = 1 - g_3; (g_3 % mod_rhs (rshift_u_s (l_2, func_4))) && lshift_u_s (1); } -- 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 Fri Sep 19 10:29:27 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 19 Sep 2008 10:29:27 -0500 Subject: [LLVMbugs] [Bug 2809] New: Avoid execution domain bypass Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2809 Summary: Avoid execution domain bypass Product: new-bugs Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: nicolas at capens.net CC: llvmbugs at cs.uiuc.edu On x86 a logical operation on 128-bit vectors can be lowered to three different instrucitons: andps, andpd, pand (in the case of logical and). According to the intel optimization guide, andps and andpd execute in the 'floating-point domain' and pand executes in the 'integer domain'. Data bypass between domains cost an additional clock cycle. So the use of a pand instruction in the middle of floating-point calculations costs two extra clock cycles. I've seen LLVM generate both andps and pand. But since LLVM's IR only allows and operations between integer vectors this behavior seems rather uncontrolled. So for best performance instruction selection should ensure that no unnecessary execution domain bypasses are made. -- 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 Fri Sep 19 13:10:04 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 19 Sep 2008 13:10:04 -0500 Subject: [LLVMbugs] [Bug 2795] llc: assertion failed in Support/Recycler.h:89 In-Reply-To: Message-ID: <200809191810.m8JIA4Q5030634@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2795 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from Dan Gohman 2008-09-19 13:10:03 --- Thanks! Applied here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080915/067276.html -- 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 Fri Sep 19 20:28:22 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 19 Sep 2008 20:28:22 -0500 Subject: [LLVMbugs] [Bug 2808] ice: attempt to spill already spilled interval! In-Reply-To: Message-ID: <200809200128.m8K1SMic012178@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2808 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Evan Cheng 2008-09-19 20:28:22 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080915/067298.html -- 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 Fri Sep 19 23:55:57 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 19 Sep 2008 23:55:57 -0500 Subject: [LLVMbugs] [Bug 2774] Please update Doxygen 1.3 to the new 1.5 version In-Reply-To: Message-ID: <200809200455.m8K4tvRL019082@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2774 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Chris Lattner 2008-09-19 23:55:57 --- Tanya did this. -- 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 Sat Sep 20 06:33:45 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 20 Sep 2008 06:33:45 -0500 Subject: [LLVMbugs] [Bug 2812] New: Segfault on big endian processor in LegalizeDAG.cpp Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2812 Summary: Segfault on big endian processor in LegalizeDAG.cpp 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: rich at pennware.com CC: llvmbugs at cs.uiuc.edu The following patch fixes a segfault that occurs in LegalizeDAG.cpp when the target processor is big endian: Index: LegalizeDAG.cpp =================================================================== --- LegalizeDAG.cpp (revision 847) +++ LegalizeDAG.cpp (working copy) @@ -2480,7 +2480,7 @@ ExpandOp(ST->getValue(), Lo, Hi); IncrementSize = Hi.getNode() ? Hi.getValueType().getSizeInBits()/8 : 0; - if (TLI.isBigEndian()) + if (Hi.getNode() && TLI.isBigEndian()) std::swap(Lo, Hi); } Unfortunately I don't have a test case for this, because it only fails in my code generator for the Nios2. All of the other code generator either don't go down this path or fail for other reasons. The code that triggered this in my code generator was an i32 to f32 conversion. It is clear from looking at the code that Hi.getNode() is expected to be NULL in the case of i32 <-> f32 conversions. After I applied this patch, the test results: === Summary === # of expected passes 2736 # of expected failures 7 -- 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 Sat Sep 20 13:58:21 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 20 Sep 2008 13:58:21 -0500 Subject: [LLVMbugs] [Bug 1820] Macro expansion bug In-Reply-To: Message-ID: <200809201858.m8KIwLcl028365@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1820 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #3 from Chris Lattner 2008-09-20 13:58:16 --- Please include a testcase. Also, please file a new bug for new issues. This bug adequately covers the testcase you provided, I'd prefer a new bugzilla for new (even if related) issues. Thanks Neil! -- 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 Sun Sep 21 06:05:44 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 21 Sep 2008 06:05:44 -0500 Subject: [LLVMbugs] [Bug 2814] New: request for adding function note to force emission of stack frame pointer Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2814 Summary: request for adding function note to force emission of stack frame pointer Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tomas.l.olsen at gmail.com CC: llvmbugs at cs.uiuc.edu I'd like to request that a new function note be added, which forces the output of a stack frame in the "noted" function. Currently there are only hacks to accomplish this, like calling llvm.unwind.init somewhere. I need to guarantee that the stack frame is emitted (ie. EBP on x86), for functions using inline asm, and currently the unwind.init hack is working well, however, It'd be much nicer if we could just put a note. I asked about this in the IRC channel a while ago but was told to basically wait a bit as there was going to be a proposal for this stuff (function notes) soon, since it's implemented now, now feels like a good time to suggest its addition. Would be nice to get it into the 2.4 release. -- 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 Sun Sep 21 16:03:40 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 21 Sep 2008 16:03:40 -0500 Subject: [LLVMbugs] [Bug 1523] trivial functions not inlined at -O2 In-Reply-To: Message-ID: <200809212103.m8LL3e1Z001854@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1523 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Chris Lattner 2008-09-21 16:03:40 --- This got fixed a long time ago. -- 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 Sun Sep 21 16:18:45 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 21 Sep 2008 16:18:45 -0500 Subject: [LLVMbugs] [Bug 1162] MachineModuleInfo should migrate away from Uniquevector. In-Reply-To: Message-ID: <200809212118.m8LLIjKQ005502@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1162 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #1 from Chris Lattner 2008-09-21 16:18:45 --- This is only used in MachineModuleInfo for a couple of high level things, it isn't a real perf issue. -- 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 Sun Sep 21 18:48:16 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 21 Sep 2008 18:48:16 -0500 Subject: [LLVMbugs] [Bug 2815] New: escaped chars in string literal cause incorrect caret position in diagnostic messages Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2815 Summary: escaped chars in string literal cause incorrect caret position in diagnostic messages Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: trivial Priority: P2 Component: Basic AssignedTo: unassignedbugs at nondot.org ReportedBy: richard at godbee.net CC: llvmbugs at cs.uiuc.edu Given the following code... main() { printf("\0"); printf("\n\0"); printf("\n\n\0"); printf("\n\n\n\0"); return 0; } ...clang (r56415) produces the following diagnostics... c.c:2:9: warning: format string contains '\0' within the string body printf("\0"); ~^~~ c.c:3:10: warning: format string contains '\0' within the string body printf("\n\0"); ~~^~~~ c.c:4:11: warning: format string contains '\0' within the string body printf("\n\n\0"); ~~~^~~~~ c.c:5:12: warning: format string contains '\0' within the string body printf("\n\n\n\0"); ~~~~^~~~~~ 4 diagnostics generated. Note that the carets are pointing to the wrong positions in all cases except the first. Whatever code determines the caret's position (and length?) seems to be treating escaped characters as only taking up one char in the source code. The same problem is seen with "\xNN" (hex) and "\NNN" (octal) escapes. -- 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 Sun Sep 21 21:54:21 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 21 Sep 2008 21:54:21 -0500 Subject: [LLVMbugs] [Bug 1656] Failed to run Datastructure analysis(dsa-local) on Linux kernel file In-Reply-To: Message-ID: <200809220254.m8M2sLj5022013@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1656 Andrew Lenharth changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |WONTFIX --- Comment #6 from Andrew Lenharth 2008-09-21 21:54:20 --- Code base has moved too much. If still a problem, please refile. -- 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 Sep 22 01:14:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 01:14:03 -0500 Subject: [LLVMbugs] [Bug 2816] New: Assertion failure when parsing "id const" method argument Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2816 Summary: Assertion failure when parsing "id const" method argument Product: clang Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: major Priority: P2 Component: parser AssignedTo: unassignedbugs at nondot.org ReportedBy: nikita at zhuk.fi CC: llvmbugs at cs.uiuc.edu Clang parser throws an assertion failure when parsing an ObjC method with argument of type "id const". GCC compiles the test case without warnings (tested with GCC 4.0 on Mac OS X 10.5.5). Clang revision: 56386 Test case: @protocol MyProtocol - (void)doSomething; @end @interface MyClass - (void)m1:(id const)arg1; @end @implementation MyClass - (void)m1:(id const)arg1 { } @end Output: MyClass.m:15:29: error: expected ')' - (void)m1:(id const)arg1; ^ MyClass.m:15:12: error: to match this '(' - (void)m1:(id const)arg1; ^ Assertion failed: (isTokenParen() && "wrong consume method"), function ConsumeParen, file /Volumes/DocsHD/Ohjelmointi/llvm/tools/clang/lib/Parse/../../include/clang/Parse/Parser.h, line 145. 0 clang 0x003250c1 _ZN4llvm3sys7Program17FindProgramByNameERKSs + 611 1 clang 0x0032538b _ZN4llvm3sys28PrintStackTraceOnErrorSignalEv + 593 2 libSystem.B.dylib 0x966f509b _sigtramp + 43 3 ??? 0xffffffff 0x0 + 4294967295 4 libSystem.B.dylib 0x9676dec2 raise + 26 5 libSystem.B.dylib 0x9677d47f abort + 73 6 libSystem.B.dylib 0x9676f063 __assert_rtn + 101 7 clang 0x001f9f3d _ZN5clang6Parser12ConsumeParenEv + 91 8 clang 0x001edb6d _ZN5clang6Parser17ParseObjCTypeNameERNS_12ObjCDeclSpecE + 409 9 clang 0x001f1294 _ZN5clang6Parser19ParseObjCMethodDeclENS_14SourceLocationENS_3tok9TokenKindEPvNS2_15ObjCKeywordKindE + 788 10 clang 0x001f178a _ZN5clang6Parser24ParseObjCMethodPrototypeEPvNS_3tok15ObjCKeywordKindE + 204 11 clang 0x001f1e45 _ZN5clang6Parser26ParseObjCInterfaceDeclListEPvNS_3tok15ObjCKeywordKindE + 1211 12 clang 0x001f2b8d _ZN5clang6Parser31ParseObjCAtInterfaceDeclarationENS_14SourceLocationEPNS_13AttributeListE + 1655 13 clang 0x001f2d00 _ZN5clang6Parser21ParseObjCAtDirectivesEv + 224 14 clang 0x001f9450 _ZN5clang6Parser24ParseExternalDeclarationEv + 414 15 clang 0x001f9549 _ZN5clang6Parser17ParseTopLevelDeclERPv + 85 16 clang 0x0012b604 _ZN5clang8ParseASTERNS_12PreprocessorEPNS_11ASTConsumerEb + 404 17 clang 0x0002fbcb _Z22InitializeIncludePathsPKcRN5clang12HeaderSearchERNS1_11FileManagerERKNS1_11LangOptionsE + 2971 18 clang 0x00030d4b main + 1643 19 clang 0x0000208a start + 54 -- 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 Sep 22 02:16:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 02:16:11 -0500 Subject: [LLVMbugs] [Bug 2043] llvm-gcc4.2 boostrap fails: Assertion 'castIsValid(getOpcode(), S, Ty) && "Illegal BitCast"' In-Reply-To: Message-ID: <200809220716.m8M7GBB1031764@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2043 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Duncan Sands 2008-09-22 02:16:10 --- I remember this problem, though not the details - it was fixed long ago. -- 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 Sep 22 02:20:41 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 02:20:41 -0500 Subject: [LLVMbugs] [Bug 2221] Assertion failure when inlining function. In-Reply-To: Message-ID: <200809220720.m8M7KffE031922@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2221 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Duncan Sands 2008-09-22 02:20:41 --- It looks like I fixed this when I corrected callgraph updating in a bunch of passes (including the inliner). -- 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 Sep 22 05:14:01 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 05:14:01 -0500 Subject: [LLVMbugs] [Bug 2817] New: clang can't build autoconfed applications: checking size of short... configure: error: cannot determine a size for short Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2817 Summary: clang can't build autoconfed applications: checking size of short... configure: error: cannot determine a size for short Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu The problem is that 'switch (0) case 0: case sizeof(short) == 2):;' fails too: Switch constants must all be same type as switch value! switch i32 0, label %sw.epilog [ i32 0, label %sw.bb i64 1, label %sw.bb1 ] Broken module found, compilation aborted! clang[0x73923f] /lib/libc.so.6[0x3e93231f80] /lib/libc.so.6(gsignal+0x35)[0x3e93231ef5] /lib/libc.so.6(abort+0x183)[0x3e93233413] clang[0x6f6710] clang(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0x14c)[0x6d073c] clang(_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE+0x7b)[0x6d0b8b] clang(_ZN4llvm19FunctionPassManager3runERNS_8FunctionE+0x3e)[0x6d0cfe] clang(_ZN4llvm14verifyFunctionERKNS_8FunctionENS_21VerifierFailureActionE+0x264)[0x6ea874] clang(_ZN5clang7CodeGen15CodeGenFunction14FinishFunctionENS_14SourceLocationE+0x80)[0x518860] clang(_ZN5clang7CodeGen15CodeGenFunction12GenerateCodeEPKNS_12FunctionDeclEPN4llvm8FunctionE+0x3ce)[0x518ebe] clang(_ZN5clang7CodeGen13CodeGenModule28EmitGlobalFunctionDefinitionEPKNS_12FunctionDeclE+0x39b)[0x4ea78b] clang(_ZN5clang7CodeGen13CodeGenModule20EmitGlobalDefinitionEPKNS_9ValueDeclE+0x65)[0x4ea9a5] clang(_ZN5clang7CodeGen13CodeGenModule10EmitGlobalEPKNS_9ValueDeclE+0xcc)[0x4eaa7c] clang(_ZN5clang7CodeGen13CodeGenModule16EmitTopLevelDeclEPNS_4DeclE+0x1c5)[0x4eacb5] clang[0x4e652c] clang(_ZN5clang8ParseASTERNS_12PreprocessorEPNS_11ASTConsumerEb+0x15b)[0x5996eb] clang[0x4deb62] clang(main+0x7dd)[0x4e273d] /lib/libc.so.6(__libc_start_main+0xe6)[0x3e9321e1a6] clang(_ZNSt8ios_base4InitD1Ev+0x41)[0x4beaf9] configure:12964: checking size of short configure:12994: /home/edwin/llvm-svn/llvm/tools/clang/utils/ccc -c -U __GNUC__ -m32 conftest.c >&5 clang -emit-llvm-bc -x c -o conftest.o conftest.c -U__GNUC__ conftest.c:52:25: error: duplicate case value '0' switch (0) case 0: case (sizeof (short) == 4):; ^ conftest.c:52:17: error: previous case value occurrence defined here switch (0) case 0: case (sizeof (short) == 4):; ^ 2 diagnostics generated. configure:13000: $? = 1 configure: failed program was: | /* confdefs.h. */ | #define PACKAGE_NAME "clamav" | #define PACKAGE_TARNAME "clamav" | #define PACKAGE_VERSION "devel" | #define PACKAGE_STRING "clamav devel" | #define PACKAGE_BUGREPORT "http://bugs.clamav.net/" | #define PACKAGE PACKAGE_NAME | #define VERSION "devel-20080922" | #define HAVE_SYS_TYPES_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRING_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define LT_OBJDIR ".libs/" | #define SCANBUFF 131072 | #define FILEBUFF 8192 | #define WORDS_BIGENDIAN 0 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_STRING_H 1 | #define HAVE_SYS_MMAN_H 1 | #define HAVE_SYS_PARAM_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_SYS_TYPES_H 1 | #define HAVE_MALLOC_H 1 | #define HAVE_POLL_H 1 | #define HAVE_LIMITS_H 1 | #define HAVE_SYS_UIO_H 1 | #define HAVE_TERMIOS_H 1 | #define HAVE_STDBOOL_H 1 | #define HAVE_PWD_H 1 | #define HAVE_GRP_H 1 | #define USE_SYSLOG 1 | /* end confdefs.h. */ | #include "confdefs.h" | #include | | | int | main () | { | switch (0) case 0: case (sizeof (short) == 4):; | ; | return 0; | } configure:12994: /home/edwin/llvm-svn/llvm/tools/clang/utils/ccc -c -U __GNUC__ -m32 conftest.c >&5 clang -emit-llvm-bc -x c -o conftest.o conftest.c -U__GNUC__ conftest.c:52:25: error: duplicate case value '0' switch (0) case 0: case (sizeof (short) == 8):; ^ conftest.c:52:17: error: previous case value occurrence defined here switch (0) case 0: case (sizeof (short) == 8):; ^ 2 diagnostics generated. configure:13000: $? = 1 configure: failed program was: | /* confdefs.h. */ | #define PACKAGE_NAME "clamav" | #define PACKAGE_TARNAME "clamav" | #define PACKAGE_VERSION "devel" | #define PACKAGE_STRING "clamav devel" | #define PACKAGE_BUGREPORT "http://bugs.clamav.net/" | #define PACKAGE PACKAGE_NAME | #define VERSION "devel-20080922" | #define HAVE_SYS_TYPES_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRING_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define LT_OBJDIR ".libs/" | #define SCANBUFF 131072 | #define FILEBUFF 8192 | #define WORDS_BIGENDIAN 0 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_STRING_H 1 | #define HAVE_SYS_MMAN_H 1 | #define HAVE_SYS_PARAM_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_SYS_TYPES_H 1 | #define HAVE_MALLOC_H 1 | #define HAVE_POLL_H 1 | #define HAVE_LIMITS_H 1 | #define HAVE_SYS_UIO_H 1 | #define HAVE_TERMIOS_H 1 | #define HAVE_STDBOOL_H 1 | #define HAVE_PWD_H 1 | #define HAVE_GRP_H 1 | #define USE_SYSLOG 1 | /* end confdefs.h. */ | #include "confdefs.h" | #include | | | int | main () | { | switch (0) case 0: case (sizeof (short) == 8):; | ; | return 0; | } configure:12994: /home/edwin/llvm-svn/llvm/tools/clang/utils/ccc -c -U __GNUC__ -m32 conftest.c >&5 clang -emit-llvm-bc -x c -o conftest.o conftest.c -U__GNUC__ conftest.c:52:25: error: duplicate case value '0' switch (0) case 0: case (sizeof (short) == 1):; ^ conftest.c:52:17: error: previous case value occurrence defined here switch (0) case 0: case (sizeof (short) == 1):; ^ 2 diagnostics generated. configure:13000: $? = 1 configure: failed program was: | /* confdefs.h. */ | #define PACKAGE_NAME "clamav" | #define PACKAGE_TARNAME "clamav" | #define PACKAGE_VERSION "devel" | #define PACKAGE_STRING "clamav devel" | #define PACKAGE_BUGREPORT "http://bugs.clamav.net/" | #define PACKAGE PACKAGE_NAME | #define VERSION "devel-20080922" | #define HAVE_SYS_TYPES_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRING_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define LT_OBJDIR ".libs/" | #define SCANBUFF 131072 | #define FILEBUFF 8192 | #define WORDS_BIGENDIAN 0 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_STRING_H 1 | #define HAVE_SYS_MMAN_H 1 | #define HAVE_SYS_PARAM_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_SYS_TYPES_H 1 | #define HAVE_MALLOC_H 1 | #define HAVE_POLL_H 1 | #define HAVE_LIMITS_H 1 | #define HAVE_SYS_UIO_H 1 | #define HAVE_TERMIOS_H 1 | #define HAVE_STDBOOL_H 1 | #define HAVE_PWD_H 1 | #define HAVE_GRP_H 1 | #define USE_SYSLOG 1 | /* end confdefs.h. */ | #include "confdefs.h" | #include | | | int | main () | { | switch (0) case 0: case (sizeof (short) == 1):; | ; | return 0; | } configure:12994: /home/edwin/llvm-svn/llvm/tools/clang/utils/ccc -c -U __GNUC__ -m32 conftest.c >&5 clang -emit-llvm-bc -x c -o conftest.o conftest.c -U__GNUC__ Switch constants must all be same type as switch value! switch i32 0, label %sw.epilog [ i32 0, label %sw.bb i64 1, label %sw.bb1 ] Broken module found, compilation aborted! clang[0x73923f] /lib/libc.so.6[0x3e93231f80] /lib/libc.so.6(gsignal+0x35)[0x3e93231ef5] /lib/libc.so.6(abort+0x183)[0x3e93233413] clang[0x6f6710] clang(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0x14c)[0x6d073c] clang(_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE+0x7b)[0x6d0b8b] clang(_ZN4llvm19FunctionPassManager3runERNS_8FunctionE+0x3e)[0x6d0cfe] clang(_ZN4llvm14verifyFunctionERKNS_8FunctionENS_21VerifierFailureActionE+0x264)[0x6ea874] clang(_ZN5clang7CodeGen15CodeGenFunction14FinishFunctionENS_14SourceLocationE+0x80)[0x518860] clang(_ZN5clang7CodeGen15CodeGenFunction12GenerateCodeEPKNS_12FunctionDeclEPN4llvm8FunctionE+0x3ce)[0x518ebe] clang(_ZN5clang7CodeGen13CodeGenModule28EmitGlobalFunctionDefinitionEPKNS_12FunctionDeclE+0x39b)[0x4ea78b] clang(_ZN5clang7CodeGen13CodeGenModule20EmitGlobalDefinitionEPKNS_9ValueDeclE+0x65)[0x4ea9a5] clang(_ZN5clang7CodeGen13CodeGenModule10EmitGlobalEPKNS_9ValueDeclE+0xcc)[0x4eaa7c] clang(_ZN5clang7CodeGen13CodeGenModule16EmitTopLevelDeclEPNS_4DeclE+0x1c5)[0x4eacb5] clang[0x4e652c] clang(_ZN5clang8ParseASTERNS_12PreprocessorEPNS_11ASTConsumerEb+0x15b)[0x5996eb] clang[0x4deb62] clang(main+0x7dd)[0x4e273d] /lib/libc.so.6(__libc_start_main+0xe6)[0x3e9321e1a6] clang(_ZNSt8ios_base4InitD1Ev+0x41)[0x4beaf9] configure:13000: $? = 250 configure: failed program was: | /* confdefs.h. */ | #define PACKAGE_NAME "clamav" | #define PACKAGE_TARNAME "clamav" | #define PACKAGE_VERSION "devel" | #define PACKAGE_STRING "clamav devel" | #define PACKAGE_BUGREPORT "http://bugs.clamav.net/" | #define PACKAGE PACKAGE_NAME | #define VERSION "devel-20080922" | #define HAVE_SYS_TYPES_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRING_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define LT_OBJDIR ".libs/" | #define SCANBUFF 131072 | #define FILEBUFF 8192 | #define WORDS_BIGENDIAN 0 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_STRING_H 1 | #define HAVE_SYS_MMAN_H 1 | #define HAVE_SYS_PARAM_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_SYS_TYPES_H 1 | #define HAVE_MALLOC_H 1 | #define HAVE_POLL_H 1 | #define HAVE_LIMITS_H 1 | #define HAVE_SYS_UIO_H 1 | #define HAVE_TERMIOS_H 1 | #define HAVE_STDBOOL_H 1 | #define HAVE_PWD_H 1 | #define HAVE_GRP_H 1 | #define USE_SYSLOG 1 | /* end confdefs.h. */ | #include "confdefs.h" | #include | | | int | main () | { | switch (0) case 0: case (sizeof (short) == 2):; | ; | return 0; | } configure:12994: /home/edwin/llvm-svn/llvm/tools/clang/utils/ccc -c -U __GNUC__ -m32 conftest.c >&5 clang -emit-llvm-bc -x c -o conftest.o conftest.c -U__GNUC__ conftest.c:52:25: error: duplicate case value '0' switch (0) case 0: case (sizeof (short) == 16):; ^ conftest.c:52:17: error: previous case value occurrence defined here switch (0) case 0: case (sizeof (short) == 16):; ^ 2 diagnostics generated. configure:13000: $? = 1 configure: failed program was: | /* confdefs.h. */ | #define PACKAGE_NAME "clamav" | #define PACKAGE_TARNAME "clamav" | #define PACKAGE_VERSION "devel" | #define PACKAGE_STRING "clamav devel" | #define PACKAGE_BUGREPORT "http://bugs.clamav.net/" | #define PACKAGE PACKAGE_NAME | #define VERSION "devel-20080922" | #define HAVE_SYS_TYPES_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRING_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define LT_OBJDIR ".libs/" | #define SCANBUFF 131072 | #define FILEBUFF 8192 | #define WORDS_BIGENDIAN 0 | #define HAVE_STDINT_H 1 | #define HAVE_UNISTD_H 1 | #define HAVE_DLFCN_H 1 | #define HAVE_INTTYPES_H 1 | #define HAVE_MEMORY_H 1 | #define HAVE_STDLIB_H 1 | #define HAVE_STRINGS_H 1 | #define HAVE_STRING_H 1 | #define HAVE_SYS_MMAN_H 1 | #define HAVE_SYS_PARAM_H 1 | #define HAVE_SYS_STAT_H 1 | #define HAVE_SYS_TYPES_H 1 | #define HAVE_MALLOC_H 1 | #define HAVE_POLL_H 1 | #define HAVE_LIMITS_H 1 | #define HAVE_SYS_UIO_H 1 | #define HAVE_TERMIOS_H 1 | #define HAVE_STDBOOL_H 1 | #define HAVE_PWD_H 1 | #define HAVE_GRP_H 1 | #define USE_SYSLOG 1 | /* end confdefs.h. */ | #include "confdefs.h" | #include | | | int | main () | { | switch (0) case 0: case (sizeof (short) == 16):; | ; | return 0; | } configure:13020: error: cannot determine a size for short -- 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 Sep 22 05:30:25 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 05:30:25 -0500 Subject: [LLVMbugs] [Bug 2816] Assertion failure when parsing "id const" method argument In-Reply-To: Message-ID: <200809221030.m8MAUP4D015362@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2816 snaroff at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from snaroff at apple.com 2008-09-22 05:30:24 --- Fixed in http://llvm.org/viewvc/llvm-project?view=rev&revision=56433. -- 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 Sep 22 09:38:36 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 09:38:36 -0500 Subject: [LLVMbugs] [Bug 2818] New: Segmentation fault on PPC64 generated code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2818 Summary: Segmentation fault on PPC64 generated 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: ph.silva at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2032) --> (http://llvm.org/bugs/attachment.cgi?id=2032) LLVM .ll I tried to compile the following code using llvm-gcc 2.3 and it compiles right, but the PPC64 generated code (from llc) crashs with segfault. == sum.c == int SumOrig_sum_ARI_I (int* A) { A[8] = 42; A[9876] = 876; return A[0]+A[1]+1; } I used the following main function to test this: == main.c === int main (void) { int a[1000000]; a[0] = 10; a[1] = 20; printf("%d\n", SumOrig_sum_ARI_I(a)); printf("%d\n", a[9876]); return 0; } On x86 the result is right: henrique at r2d2:~$ ./main 31 876 Follow attached LLVM .ll and the PPC64 code ( llc -march=ppc64 -mcpu=ppc64) and just for comparison the GCC code for the same function. The crash was on a PPC64 running Fedora 9. -- 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 Sep 22 10:13:16 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 10:13:16 -0500 Subject: [LLVMbugs] [Bug 1812] ExpandUnalignedLoad doesn't handle vector loads In-Reply-To: Message-ID: <200809221513.m8MFDGNp006286@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1812 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #2 from Chris Lattner 2008-09-22 10:13:15 --- ok, must have already been fixed :) Thanks Duncan -- 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 Sep 22 17:42:32 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 17:42:32 -0500 Subject: [LLVMbugs] [Bug 2821] New: clang rejects ANSI C style prototype + K&R style definition Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2821 Summary: clang rejects ANSI C style prototype + K&R style definition Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedbugs at nondot.org ReportedBy: richard at godbee.net CC: llvmbugs at cs.uiuc.edu % cat t9.c void f(char c); void f(c) char c; { } % gcc -Wall -c t9.c % gcc -Wall -pedantic -c t9.c t9.c: In function ???f???: t9.c:2: warning: promoted argument ???c??? doesn???t match prototype t9.c:1: warning: prototype declaration % clang -x c t9.c t9.c:2:6: error: conflicting types for 'f' void f(c) char c; { } ^ t9.c:1:6: error: previous declaration is here void f(char c); ^ 2 diagnostics generated. zsh: exit 1 clang -x c t9.c Summary: - gcc: compiles w/o warnings - gcc -pedantic: compiles w/ warnings - clang: error llvm r56425 / clang r56427 on Mac OS X 10.5.5 (PPC) -- 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 Sep 22 18:54:40 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 22 Sep 2008 18:54:40 -0500 Subject: [LLVMbugs] [Bug 2817] clang can't build autoconfed applications: checking size of short... configure: error: cannot determine a size for short In-Reply-To: Message-ID: <200809222354.m8MNse1t007733@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2817 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Daniel Dunbar 2008-09-22 18:54:40 --- Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=56482 Thanks! -- 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 Sep 23 05:08:53 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 23 Sep 2008 05:08:53 -0500 Subject: [LLVMbugs] [Bug 2823] New: llvm-gcc does not respect ABI alignment for structs-with-vectors Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2823 Summary: llvm-gcc does not respect ABI alignment for structs- with-vectors 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: matthijs at stdin.nl CC: llvmbugs at cs.uiuc.edu The below program contains a struct that contains a single vector of 4 i32's. When llvm-gcc generates alloca's for this struct, the alignment gets set to 4, while the ABI alignment of the contained vector is 16. Since the program passes the contained vector to another function, this violates alignment requirements. The same goes for a union containing a vector, as illustrated in the main2 function. Since the div function assumes that it's arguments are ABI aligned, llc will generate aligned SSE instructions (movaps) operating on the function arguments. Since the caller does not respect the required alignment, this segfaults. $ llvm-gcc align.c -S -emit-llvm -o - | llvm-as | llc -o align.S $ gcc align.S $ ./a.out Segmentation fault (core dumped) In most cases, this behaviour is not observed, since instcombine will change the type of an alloca when it is bitcasted to something with a higher alignment. Since we're not running instcombine (this example would require inlining as well), the bug is not masked off. This example is mostly based on the test-suite/SingleSource/UnitTests/Vector/divide.c file, since that crashed with a local modification to make instcombine not change alloca instructions in some cases. Note that when changing the below example to not use a seperate div function, but doing the dividing inline, the vector loads will be generated with a conservative alignment of 1, llc will use unaligned SSE instructions (movups) and everything works fine. On a related note, clang does do the right thing, and produces the alloca's with the correct alignment right away. So, here's the code: typedef unsigned uvec __attribute__ ((__vector_size__ (16))); static void div(uvec *A, uvec *B, uvec *R) { *R = *A / *B; } typedef union { uvec V; unsigned A[4]; } UV; typedef struct { uvec V; } SV; int main(int argc, char**argv) { SV V; div (&V.V, &V.V, &V.V); return (unsigned)&V.V; } int main2(int argc, char**argv) { UV V; div (&V.V, &V.V, &V.V); return (unsigned)&V.V; } Which produces (main2 function left out): ; ModuleID = 'align.c' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i386-pc-linux-gnu" %struct.SV = type <{ <4 x i32> }> %struct.UV = type <{ <4 x i32> }> define internal void @div(<4 x i32>* %A, <4 x i32>* %B, <4 x i32>* %R) nounwind { entry: %A_addr = alloca <4 x i32>* ; <<4 x i32>**> [#uses=2] %B_addr = alloca <4 x i32>* ; <<4 x i32>**> [#uses=2] %R_addr = alloca <4 x i32>* ; <<4 x i32>**> [#uses=2] %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] store <4 x i32>* %A, <4 x i32>** %A_addr store <4 x i32>* %B, <4 x i32>** %B_addr store <4 x i32>* %R, <4 x i32>** %R_addr %0 = load <4 x i32>** %A_addr, align 4 ; <<4 x i32>*> [#uses=1] %1 = load <4 x i32>* %0, align 16 ; <<4 x i32>> [#uses=1] %2 = load <4 x i32>** %B_addr, align 4 ; <<4 x i32>*> [#uses=1] %3 = load <4 x i32>* %2, align 16 ; <<4 x i32>> [#uses=1] %4 = udiv <4 x i32> %1, %3 ; <<4 x i32>> [#uses=1] %5 = load <4 x i32>** %R_addr, align 4 ; <<4 x i32>*> [#uses=1] store <4 x i32> %4, <4 x i32>* %5, align 16 br label %return return: ; preds = %entry ret void } define i32 @main(i32 %argc, i8** %argv) nounwind { entry: %argc_addr = alloca i32 ; [#uses=1] %argv_addr = alloca i8** ; [#uses=1] %retval = alloca i32 ; [#uses=2] %V = alloca %struct.SV, align 4 ; <%struct.SV*> [#uses=4] %0 = alloca i32 ; [#uses=2] %"alloca point" = bitcast i32 0 to i32 ; [#uses=0] store i32 %argc, i32* %argc_addr store i8** %argv, i8*** %argv_addr %1 = getelementptr %struct.SV* %V, i32 0, i32 0 ; <<4 x i32>*> [#uses=1] %2 = getelementptr %struct.SV* %V, i32 0, i32 0 ; <<4 x i32>*> [#uses=1] %3 = getelementptr %struct.SV* %V, i32 0, i32 0 ; <<4 x i32>*> [#uses=1] call void @div(<4 x i32>* %1, <4 x i32>* %2, <4 x i32>* %3) nounwind %4 = getelementptr %struct.SV* %V, i32 0, i32 0 ; <<4 x i32>*> [#uses=1] %5 = ptrtoint <4 x i32>* %4 to i32 ; [#uses=1] store i32 %5, i32* %0, align 4 %6 = load i32* %0, align 4 ; [#uses=1] store i32 %6, i32* %retval, align 4 br label %return return: ; preds = %entry %retval1 = load i32* %retval ; [#uses=1] ret i32 %retval1 } -- 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 Sep 23 05:12:15 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 23 Sep 2008 05:12:15 -0500 Subject: [LLVMbugs] [Bug 2824] New: __USER_LABEL_PREFIX__ incompabitibility Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2824 Summary: __USER_LABEL_PREFIX__ incompabitibility Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: preprocessor AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu While compiling a program I got this linker error: (.text+0x3244): undefined reference to `__USER_LABEL_PREFIX____isoc99_sscanf' Indeed the object file contains this symbol: $ llvm-nm client.o|grep sscanf U __USER_LABEL_PREFIX____isoc99_sscanf This is the definition of sscanf (wrong): extern int sscanf (__const char *__restrict __s, __const char *__restrict __format, ...) __asm__ ("__USER_LABEL_PREFIX__" "__isoc99_sscanf") __attribute__ ((__nothrow__)); This is the definition of sscanf when using gcc (or llvm-gcc) -std=c99: extern int sscanf (__const char *__restrict __s, __const char *__restrict __format, ...) __asm__ ("" "__isoc99_sscanf") __attribute__ ((__nothrow__)); This is the definition of sscanf if I -U __GNUC__: extern int sscanf (const char * __s, const char * __format, ...) ; Testcase: $ cat x.c __USER_LABEL_PREFIX__ $ gcc -E x.c # 1 "x.c" # 1 "" # 1 "" # 1 "x.c" $ clang -E x.c # 1 "x.c" # 1 "x.c" 1 # 0 "x.c" # 1 "" 1 # 1 "x.c" 2 __USER_LABEL_PREFIX__ -- 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 Sep 23 05:16:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 23 Sep 2008 05:16:03 -0500 Subject: [LLVMbugs] [Bug 2825] New: cannot codegen __builtin_va_arg Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2825 Summary: cannot codegen __builtin_va_arg Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Using clang SVN r56486: $ clang -emit-llvm-bc -x c -o x.o x.c x.c:4:28: error: cannot codegen this l-value expression yet (void) __builtin_va_arg(ap, void*); ^~ 1 diagnostic generated. $ cat x.c void foo(void *f, ...) { __builtin_va_list ap; __builtin_va_start(ap, f); (void) __builtin_va_arg(ap, void*); __builtin_va_end(ap); } -- 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 Sep 23 12:38:27 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 23 Sep 2008 12:38:27 -0500 Subject: [LLVMbugs] [Bug 2824] __USER_LABEL_PREFIX__ incompabitibility In-Reply-To: Message-ID: <200809231738.m8NHcRV5020851@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2824 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #4 from Daniel Dunbar 2008-09-23 12:38:26 --- Patched here: http://llvm.org/viewvc/llvm-project?view=rev&revision=56491 Closing as we already have PR2773 for the larger issue. -- 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 Wed Sep 24 01:42:29 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 24 Sep 2008 01:42:29 -0500 Subject: [LLVMbugs] [Bug 2826] New: google-perftools GetStackTrace fails under LLVM Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2826 Summary: google-perftools GetStackTrace fails under LLVM 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: nicholas at mxc.ca CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2038) --> (http://llvm.org/bugs/attachment.cgi?id=2038) GetStackTrace extracted testcase google-perftools includes a heap checker which works by a malloc replacement. The attached testcase is an extraction from the larger code, which tries to retrieve a stack trace. For some reason, this program doesn't work under LLVM: $ g++ -Wall -pedantic reduced.cpp -o reduced -g -O0 $ ./reduced 0x80483f1 0xf7cea455 0x80485be $ llvm-g++ -Wall -pedantic reduced.cpp -o reduced -g -O0 $ ./reduced $ I realize that 'stack' isn't entirely defined in the C standard, but I don't think this program is invoking undefined behaviour in a way that LLVM is allowed to optimize away. Changing optz'n levels and debugging on/off has no effect. The program is valgrind clean, no matter how it's compiled. I'm seeing this problem on i386. I haven't tried x86_64. -- 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 Wed Sep 24 04:14:11 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 24 Sep 2008 04:14:11 -0500 Subject: [LLVMbugs] [Bug 2827] New: CodeEmitterGen produce wrong values in generated function getBinaryCodeForInstr (...) Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2827 Summary: CodeEmitterGen produce wrong values in generated function getBinaryCodeForInstr(...) Product: tools Version: 2.3 Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: TableGen AssignedTo: unassignedbugs at nondot.org ReportedBy: waldemar.knorr at eckcellent-it.de CC: llvmbugs at cs.uiuc.edu In generated XXXCodeEmitter::getBinaryCodeForInstr(MachineInstr &MI) tablegen produce wrong values, if a 32 bit instruction operand is used. So the generated code looks like this: case sharc::PUSHGAddr: { // op: registerINumber op = getMachineOpValue(MI, MI.getOperand(0)); Value |= (op & 7U) << 41; // op: registerMNumber op = getMachineOpValue(MI, MI.getOperand(1)); Value |= (op & 7U) << 38; // op: immediateData op = getMachineOpValue(MI, MI.getOperand(2)); Value |= op & 0U; // <<<<<<<<<<<<<<<<<<<<<<<<<<<< BUG. break; } The reason for that is that in CodeGenEmitterGen.cpp line 190: unsigned opMask = (1 << N) - 1; If N=32, than the opMask is zero, instead of 0xffffffff Solution: CodeGenEmitterGen.cpp line 190: unsigned opMask = (1ULL << N) - 1; -- 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 Wed Sep 24 17:14:51 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 24 Sep 2008 17:14:51 -0500 Subject: [LLVMbugs] [Bug 2801] x86-64 doesn't save nonvolatile XMM6-XMM15 registers In-Reply-To: Message-ID: <200809242214.m8OMEpgR024982@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2801 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #10 from Anton Korobeynikov 2008-09-24 17:14:50 --- Nicolas, I applied the callee-saved register part of your patch. See r56568. I believe, that alignment of stack variables is separate issue. Please check, if there are any PRs opened already on this problem, and if not - fill new PR with testcase. -- 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 Thu Sep 25 11:16:15 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 25 Sep 2008 11:16:15 -0500 Subject: [LLVMbugs] [Bug 2812] Segfault on big endian processor in LegalizeDAG.cpp In-Reply-To: Message-ID: <200809251616.m8PGGFlA008312@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2812 Richard Pennington changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Richard Pennington 2008-09-25 11:16:14 --- Committed in revision 56609. -- 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 Thu Sep 25 19:37:00 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 25 Sep 2008 19:37:00 -0500 Subject: [LLVMbugs] [Bug 2829] New: llvm.memory. barrier fails to produce write-before-read fence (mfence) Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2829 Summary: llvm.memory.barrier fails to produce write-before-read fence (mfence) 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: luked at cs.rochester.edu CC: llvmbugs at cs.uiuc.edu The x86 allows load bypassing as per http://www.intel.com/products/processor/manuals/318147.pdf, and provides the sse2 "mfence" instruction as a write-before-read fence. When the "ls" parameter to llvm.memory.barrier is true, llc should emit the mfence (or some equivalent smarter thing, like an xchg or other locked instruction if it's already in the instruction sequence). Currently building the supplied bitcode with "llc -mattr=+sse2 -f -o test.s test.bc" fails to insert that fence. This appears to be caused by the 5th parameter to llvm.memory.barrier. Any time it is false it suppresses the fence. The fifth parameter is very poorly documented, so I'm not 100% sure if the problem is a documentation bug, or a bug in the code. The llvm.memory.barrier is generated by a __sync_synchronize() intrinsic in llvm-gcc. It could be that the intrinsic should emit the barrier with the 5th parameter as true, and the documentation needs to be changed. This also may be related to the gcc bug http://gcc.gnu.org/ml/gcc-bugs/2008-07/msg01046.html. Luke -- 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 Fri Sep 26 00:43:00 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 26 Sep 2008 00:43:00 -0500 Subject: [LLVMbugs] [Bug 2826] google-perftools GetStackTrace fails under LLVM In-Reply-To: Message-ID: <200809260543.m8Q5h0Kc002640@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2826 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #10 from Nick Lewycky 2008-09-26 00:42:59 --- Thanks everyone. I'll take it up with upstream. -- 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 Fri Sep 26 04:19:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 26 Sep 2008 04:19:03 -0500 Subject: [LLVMbugs] [Bug 2830] New: Cpp Backend doesn' t generate correct code for Switch instructions Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2830 Summary: Cpp Backend doesn't generate correct code for Switch instructions Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Keywords: miscompilation Severity: normal Priority: P2 Component: Backend: C AssignedTo: unassignedbugs at nondot.org ReportedBy: bobby at sharedrealm.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2039) --> (http://llvm.org/bugs/attachment.cgi?id=2039) bitcode containing switch insturction with 'i32 0' case statement. I am trying to compile some bitcode to LLVM API calls using the Cpp backend: llc -march=cpp --cppgen=program -o test.cpp test.bc Compiling the generated C++ code fails with this error: test.cpp: In function 'llvm::Module* makeLLVMModule()': test.cpp:148: error: invalid conversion from 'llvm::Constant*' to 'llvm::ConstantInt*' test.cpp:148: error: initializing argument 1 of 'void llvm::SwitchInst::addCase(llvm::ConstantInt*, llvm::BasicBlock*)' The problem is with this LLVM IR code: switch i32 %1, label %bb5 [ i32 0, label %bb6 i32 1, label %bb2 ] The 'i32' constants '0' & '1' are defined with this c++ code: // Constant Definitions Constant* const_int32_7 = Constant::getNullValue(IntegerType::get(32)); ConstantInt* const_int32_8 = ConstantInt::get(APInt(32, "1", 1, 10)); and the switch instruction is created with this code: SwitchInst* void_11 = SwitchInst::Create(int32_10, label_bb5, 3, label_entry); void_11->addCase(const_int32_7, label_bb6); void_11->addCase(const_int32_8, label_bb2); The 'addCase' method accepts a llvm::ConstantInt* as it's first argument, but the 'i32 0' constant was created as just an 'llvm::Constant*' P.S. Can someone add a "Backend: Cpp" Component to Bugzilla. -- 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 Fri Sep 26 08:51:34 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 26 Sep 2008 08:51:34 -0500 Subject: [LLVMbugs] [Bug 2831] New: instcombine complicates expression, and also gets it wrong! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2831 Summary: instcombine complicates expression, and also gets it wrong! 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 CC: llvmbugs at cs.uiuc.edu Running "opt -instcombine" on this testcase: 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-f80:32:32" target triple = "i386-pc-linux-gnu" define i32 @main(i32 %argc, i8** %argv) nounwind { entry: %0 = ptrtoint i8** %argv to i32 ; [#uses=1] %1 = add i32 %0, 1 ; [#uses=1] ret i32 %1 } Turns it into: define i32 @main(i32 %argc, i8** %argv) nounwind { entry: %argv1 = bitcast i8** %argv to i8* ; [#uses=1] %ctg2 = getelementptr i8* %argv1, i32 1 ; [#uses=1] %0 = ptrtoint i8* %ctg2 to i32 ; [#uses=1] ret i32 %0 } Not only is this more complicated, but overflow behaviour is different: i32 overflow is well-defined, while gep overflow is not (at least, I don't think it is...). So this transform is wrong as well as silly. -- 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 Fri Sep 26 14:07:49 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 26 Sep 2008 14:07:49 -0500 Subject: [LLVMbugs] [Bug 2275] assertion failure with C++ namespaces In-Reply-To: Message-ID: <200809261907.m8QJ7nKK010031@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2275 Ted Kremenek changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |kremenek at apple.com 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 Fri Sep 26 17:19:54 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 26 Sep 2008 17:19:54 -0500 Subject: [LLVMbugs] [Bug 2832] New: ICE doesn't consider addrof as constant Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2832 Summary: ICE doesn't consider addrof as constant Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu isConstantExpression() doesn't consider adresses as constant. This can be seen by running 'clang -emit-llvm' with the following code: struct s { void *p; }; void foo(void) { static struct s var = {((void*)&((char*)0)[0])}; } -- 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 Sat Sep 27 15:29:37 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 27 Sep 2008 15:29:37 -0500 Subject: [LLVMbugs] [Bug 2833] New: AtomicSwap64 not supported by x86 backend Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2833 Summary: AtomicSwap64 not supported by x86 backend Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: minor Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca CC: llvmbugs at cs.uiuc.edu This program fails to compile: long long NoBarrier_AtomicExchange(volatile long long* ptr, long long new_value) { return __sync_lock_test_and_set(ptr, new_value); } results in: NODE: 0x8ef333c: i64,ch = AtomicSwap64 0x8ef3198, 0x8ef377c, 0x8ef3b34 <0x8eafa20:0> alignment=4 cc1: LegalizeDAG.cpp:5885: void::SelectionDAGLegalize::ExpandOp(llvm::SDValue, llvm::SDValue&, llvm::SDValue&): Assertion `0 && "Do not know how to expand this operator!"' failed. My proposed expansion uses a cmpxchg and a spin loop: movl (%edi), %ebx ; something to move the 64-bit value to ecx:ebx movl 4(%edi), %ecx 0: movl (%esi), %eax movl 4(%esi), %edx lock; cmpxchg8b (%esi) jnz 0b -- 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 Sun Sep 28 22:49:08 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 28 Sep 2008 22:49:08 -0500 Subject: [LLVMbugs] [Bug 2835] New: likely volatile error in x86 backend Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2835 Summary: likely volatile error in x86 backend 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu Seen using r56775 on Ubuntu Hardy on ia32. The translation below to LLVM is correct: the volatile load from g_407 is i32. However, the x86 looks wrong: movzbl is used, which loads only the lowest 8 bits. regehr at john-home:~/volatile/tmp47$ cat small.c #include #include static inline int lshift_s_s(int left, int right) { if ((left < 0) || (right < 0) || (right >= sizeof(int)*CHAR_BIT) || (left > (INT_MAX >> right))) { /* Avoid undefined behavior. */ return left; } return left << right; } uint32_t g_114; volatile uint32_t g_407 ; int32_t func_45 (uint8_t p_48); int32_t func_45 (uint8_t p_48) { for (0; g_114; g_114--) { } return 1; } int main (void) { func_45 (lshift_s_s (1, 1) <= 1 | g_407); return 0; } regehr at john-home:~/volatile/tmp47$ llvm-gcc -O -I/home/regehr/randprog-classic/src small.c -S --emit-llvm -o - ; ModuleID = 'small.c' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i386-pc-linux-gnu" @g_114 = common global i32 0 ; [#uses=2] @g_407 = common global i32 0 ; [#uses=1] define i32 @func_45(i8 zeroext %p_48) nounwind { entry: %0 = load i32* @g_114, align 4 ; [#uses=1] %1 = icmp eq i32 %0, 0 ; [#uses=1] br i1 %1, label %bb2, label %bb.nph bb.nph: ; preds = %entry store i32 0, i32* @g_114 ret i32 1 bb2: ; preds = %entry ret i32 1 } define i32 @main() nounwind { entry: %0 = volatile load i32* @g_407, align 4 ; [#uses=1] %1 = trunc i32 %0 to i8 ; [#uses=1] %2 = tail call i32 @func_45(i8 zeroext %1) nounwind ; [#uses=0] ret i32 0 } regehr at john-home:~/volatile/tmp47$ llvm-gcc -O -I/home/regehr/randprog-classic/src small.c -S -o - .ident "GCC: (GNU) 4.2.1 (Based on Apple Inc. build 5623) (LLVM build)" .text .align 16 .globl func_45 .type func_45, at function func_45: pushl %ebp movl %esp, %ebp cmpl $0, g_114 je .LBB1_2 # bb2 .LBB1_1: # bb.nph movl $0, g_114 .LBB1_2: # bb2 movl $1, %eax popl %ebp ret .size func_45, .-func_45 .align 16 .globl main .type main, at function main: pushl %ebp movl %esp, %ebp subl $8, %esp movzbl g_407, %eax movl %eax, (%esp) call func_45 xorl %eax, %eax addl $8, %esp popl %ebp ret .size main, .-main .type g_114, at object .section .gnu.linkonce.b.g_114,"aw", at nobits .comm g_114,4,4 # g_114 .type g_407, at object .section .gnu.linkonce.b.g_407,"aw", at nobits .comm g_407,4,4 # g_407 .section .note.GNU-stack,"", at progbits regehr at john-home:~/volatile/tmp47$ -- 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 Sep 29 06:45:10 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 06:45:10 -0500 Subject: [LLVMbugs] [Bug 2836] New: Objc: Extended setter property not synthesized Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2836 Summary: Objc: Extended setter property not synthesized Product: clang Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedbugs at nondot.org ReportedBy: thomascl at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2041) --> (http://llvm.org/bugs/attachment.cgi?id=2041) test case The extended setter property is not synthesized by the code generation. The attached test case is actually taken from the 'Objective-C 2 Programming language' guide. Thanks, Thomas -- 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 Sep 29 06:48:09 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 06:48:09 -0500 Subject: [LLVMbugs] [Bug 2837] New: Objc fsyntax-only synthesized properties and missing implementation warnings Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2837 Summary: Objc fsyntax-only synthesized properties and missing implementation warnings Product: clang Version: unspecified Platform: Macintosh OS/Version: Mac System 9.x Status: NEW Severity: normal Priority: P2 Component: parser AssignedTo: unassignedbugs at nondot.org ReportedBy: thomascl at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2042) --> (http://llvm.org/bugs/attachment.cgi?id=2042) test case clang -fsyntax-only emits irrelevant warnings about missing implementation methods while these methods are actually synthesized by the compiler. See the attached test case. Thanks, Thomas -- 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 Sep 29 12:27:00 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 12:27:00 -0500 Subject: [LLVMbugs] [Bug 2835] likely volatile error in x86 backend In-Reply-To: Message-ID: <200809291727.m8THR08u025684@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2835 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |evan.cheng at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Evan Cheng 2008-09-29 12:26:59 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080929/067752.html -- 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 Sep 29 14:16:51 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 14:16:51 -0500 Subject: [LLVMbugs] [Bug 2839] New: C Backend does not handle inline assembly correctly. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2839 Summary: C Backend does not handle inline assembly correctly. Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: C AssignedTo: unassignedbugs at nondot.org ReportedBy: haohui.mai at gmail.com CC: llvmbugs at cs.uiuc.edu The LLVM IR is the following: define void @test() nounwind { tail call void asm sideeffect "", "~{dirflag},~{fpsr},~{flags},~{memory}"() nounwind ret void } Here is the generated C code: void test() { __asm__ volatile ("" : :"memory"); return; } It should be __asm__ volative("":::"memory"). The generated code lacks a colon that make the C code failed to compile. -- 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 Sep 29 14:43:38 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 14:43:38 -0500 Subject: [LLVMbugs] [Bug 2839] C Backend does not handle inline assembly correctly. In-Reply-To: Message-ID: <200809291943.m8TJhc5M009799@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2839 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Anton Korobeynikov 2008-09-29 14:43:37 --- *** This bug has been marked as a duplicate of bug 802 *** -- 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 Sep 29 15:24:12 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 15:24:12 -0500 Subject: [LLVMbugs] [Bug 2842] New: instcombine should generate bswap Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2842 Summary: instcombine should generate bswap Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: alenhar2 at uiuc.edu CC: llvmbugs at cs.uiuc.edu Created an attachment (id=2045) --> (http://llvm.org/bugs/attachment.cgi?id=2045) two hton, equal in stature in the following code, instcombine should generate bswap. (amusingly enough, the inline asm to intrinsic promoter recognizes the inline asm and promotes it to a bswap in the first case). C source attached, little endian .ll file also attached. -- 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 Sep 29 15:47:01 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 15:47:01 -0500 Subject: [LLVMbugs] [Bug 2843] New: llvm-gcc has -Wformat-nonliteral on by default, can' t switch it off Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2843 Summary: llvm-gcc has -Wformat-nonliteral on by default, can't switch it off Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: nlewycky at google.com CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu $ cat foo.cpp #include extern const char *foo(); int main(void) { const char *the_string = foo(); printf(the_string); } $ llvm-g++ -c foo.cpp -Wno-format-nonliteral foo.cpp: In function ???int main()???: foo.cpp:7: warning: format not a string literal and no format arguments $ g++ -c foo.cpp $ GCC will pick it up with -Wformat -Wformat-nonliteral, or -Wall -Wformat-nonliteral, but llvm-gcc picks it up always. In my case, this is worsened by being combined with -Werror ... -- 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 Sep 29 18:13:31 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 18:13:31 -0500 Subject: [LLVMbugs] [Bug 2750] Extremely strange "exponent has no digits" error In-Reply-To: Message-ID: <200809292313.m8TNDVkM000359@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2750 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Chris Lattner 2008-09-29 18:13:30 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080929/007823.html Thanks for tackling this Eli, please let me know if you think the approach is wrong. -Chris -- 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 Sep 29 21:58:03 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 29 Sep 2008 21:58:03 -0500 Subject: [LLVMbugs] [Bug 2844] New: likely integer code miscompilation Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2844 Summary: likely integer code miscompilation 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: regehr at cs.utah.edu CC: llvmbugs at cs.uiuc.edu This is seen on r56829 on Ubuntu Hardy on ia32. llvm-gcc is valgrind-clean at both optimization levels. 0 is the right answer I believe. regehr at john-home:~/volatile/tmp49$ llvm-gcc -O1 small.c -o small regehr at john-home:~/volatile/tmp49$ ./small 0 regehr at john-home:~/volatile/tmp49$ llvm-gcc -O2 small.c -o small regehr at john-home:~/volatile/tmp49$ ./small 1 regehr at john-home:~/volatile/tmp49$ cat small.c #include static inline unsigned long int mod_rhs(const long int rhs) { if (rhs == 0) return 1; return rhs; } static inline unsigned long int div_rhs(const long int rhs) { if (rhs == 0) return 1; return rhs; } static inline unsigned int rshift_u_u(unsigned int left, unsigned int right) { if (right >= sizeof(unsigned int)*8) { return left; } return left >> right; } unsigned int g_52; unsigned int g_241; unsigned int g_247; unsigned int g_253; int func_8 (unsigned int p_10); int func_8 (unsigned int p_10) { unsigned int l_11 = 0; for (1; l_11; l_11 -= 1) { } return 1; } int func_65 (unsigned short p_67, int p_69); int func_65 (unsigned short p_67, int p_69) { unsigned int l_70 = 0xD9F5B60AL; return l_70; } int func_76 (signed char p_78); int func_76 (signed char p_78) { p_78 = 1 % mod_rhs (1); return 1; } int func_73 (int p_74); int func_73 (int p_74) { unsigned int l_75 = 1; unsigned int l_77 = 1; unsigned int l_264 = 0; func_65 (((rshift_u_u (l_75, (func_76 (p_74) || l_75 + func_8 (1)) <= 1 & 1 < (g_241 ^ (1 == p_74)) * (func_65 ((g_247 * l_77), 1) && (g_241 & (1 >= l_77))))) - 1), p_74 && ((g_52 | l_77) && l_75 && (1 / div_rhs (1))) * ((p_74 ^ (func_8 (1) != (1 % mod_rhs (1)))) <= 1) - (1 >= l_264)); g_253 = (p_74 && (p_74 > func_65 (0x76L, 1))); return 1; } int main (void) { func_73 (0); printf ("%d\n", g_253); return 0; } -- 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 Sep 30 01:09:00 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 30 Sep 2008 01:09:00 -0500 Subject: [LLVMbugs] [Bug 2844] likely integer code miscompilation In-Reply-To: Message-ID: <200809300609.m8U690s9000604@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2844 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #4 from Nick Lewycky 2008-09-30 01:09:00 --- Fixed in r56834. -- 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 Sep 30 10:44:18 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 30 Sep 2008 10:44:18 -0500 Subject: [LLVMbugs] [Bug 2845] New: ICE doesn't like cast to non-integer types Message-ID: http://llvm.org/bugs/show_bug.cgi?id=2845 Summary: ICE doesn't like cast to non-integer types Product: clang Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu clang chokes on the following code because ICE doesn't like the cast to char*: int foo() { switch (0) { case (int)(char*)0: break; } } -- 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 Sep 30 20:27:31 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 30 Sep 2008 20:27:31 -0500 Subject: [LLVMbugs] [Bug 2843] llvm-gcc has -Wformat-nonliteral on by default, can' t switch it off In-Reply-To: Message-ID: <200810010127.m911RVXJ014517@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2843 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Devang Patel 2008-09-30 20:27:30 --- When -Wformat is used it seems -Wno-format-nonliteral is not effective. Anyway, now llvm-gcc will not enforce -Wformat by default on non-darwin platforms after revision 56905. -- 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 Sep 30 20:28:31 2008 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 30 Sep 2008 20:28:31 -0500 Subject: [LLVMbugs] [Bug 2780] llvm-g++ prints (bad?) warning not present in g++ 4.2 In-Reply-To: Message-ID: <200810010128.m911SVMd014585@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=2780 Devang Patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |dpatel at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Devang Patel 2008-09-30 20:28:30 --- llvm-gcc accidently enabled -Wformat for all targets by default, instead of just darwin. After revision 56905, Wformat is enabled by default on darwin only. -- 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.