From bugzilla-daemon at cs.uiuc.edu Tue Sep 1 01:31:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 01:31:36 -0500 Subject: [LLVMbugs] [Bug 4029] Function Passes can leave dangling pointers in CallGraph In-Reply-To: Message-ID: <200909010631.n816VauT009024@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4029 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #4 from Chris Lattner 2009-09-01 01:31:35 --- *** This bug has been marked as a duplicate of bug 3601 *** -- 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 1 01:34:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 01:34:05 -0500 Subject: [LLVMbugs] [Bug 3601] function passes vs call graph In-Reply-To: Message-ID: <200909010634.n816Y5sq009378@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3601 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Loop Optimizer |Interprocedural Analyses Resolution| |FIXED Summary|loop passes vs call graph |function passes vs call | |graph --- Comment #4 from Chris Lattner 2009-09-01 01:34:04 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086071.html Testcase here: test/Transforms/Inline/callgraph-update.ll -- 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 1 02:48:08 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 02:48:08 -0500 Subject: [LLVMbugs] [Bug 4847] New: [2.6] clang/llc prerelease: wrong assembly generated with -mcmodel=kernel Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4847 Summary: [2.6] clang/llc prerelease: wrong assembly generated with -mcmodel=kernel 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: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu clang -MD -MF arch/x86/kernel/.traps.o.d -nostdinc -isystem /usr/local/bin/../lib/^Cang/1.0/include -Iinclude -I/home/edwin/builds/linux-2.6/arch/x86/include -include include/linux/autoconf.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Werror-implicit-function-declaration -Wno-format-security -fno-delete-null-pointer-checks -Wno-unused-value -O2 -m64 -march=core2 -mno-red-zone -mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -pipe -Wno-sign-compare -fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -Wframe-larger-than=2048 -fno-stack-protector -fno-omit-frame-pointer -fno-optimize-sibling-calls -Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow -fno-dwarf2-cfi-asm -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(traps)" -D"KBUILD_MODNAME=KBUILD_STR(traps)" -c -emit-llvm arch/x86/kernel/traps.c /tmp/cc-O9QBCM.s:1717: Error: suffix or operands invalid for `movq' If I remove -mcmodel=kernel, then there is no error. To reproduce, this fails: $ clang -O1 -mcmodel=kernel -c traps.i This works: $ clang -O1 -c traps.i -- 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 1 02:54:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 02:54:59 -0500 Subject: [LLVMbugs] [Bug 4847] [2.6] clang/llc prerelease: wrong assembly generated with -mcmodel=kernel In-Reply-To: Message-ID: <200909010754.n817sxnY028193@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4847 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #5 from Anton Korobeynikov 2009-09-01 02:54:58 --- *** This bug has been marked as a duplicate of bug 4689 *** -- 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 1 03:05:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 03:05:23 -0500 Subject: [LLVMbugs] [Bug 4848] New: [2.6] instcombine runs out of memory (infinite loop?) when compiling kernel/hrtimer.c Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4848 Summary: [2.6] instcombine runs out of memory (infinite loop?) when compiling kernel/hrtimer.c Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Blocks: 4068 Created an attachment (id=3408) --> (http://llvm.org/bugs/attachment.cgi?id=3408) bugpoint-reduced-simplified.bc clang runs out of memory when compiling hrtimer.c, bugpoint reduced a testcase for instcombine. The testcase contains only 3 instructions. terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc 0 clang-cc 0x000000000128595f 1 clang-cc 0x0000000001287629 2 libpthread.so.0 0x0000003672e0e720 3 libc.so.6 0x0000003672231d25 gsignal + 53 4 libc.so.6 0x0000003672234de1 abort + 385 5 libstdc++.so.6 0x00000034cfcc6dc5 __gnu_cxx::__verbose_terminate_handler() + 277 6 libstdc++.so.6 0x00000034cfcc51f6 7 libstdc++.so.6 0x00000034cfcc5223 8 libstdc++.so.6 0x00000034cfcc531e 9 libstdc++.so.6 0x00000034cfcc57ad operator new(unsigned long) + 125 10 clang-cc 0x000000000113f1dd 11 clang-cc 0x000000000113f729 12 clang-cc 0x000000000113fc45 13 clang-cc 0x000000000105e3ba 14 clang-cc 0x0000000001095229 15 clang-cc 0x000000000109667e 16 clang-cc 0x000000000121336b 17 clang-cc 0x0000000001213d36 18 clang-cc 0x0000000001213fa0 19 clang-cc 0x0000000000427558 20 clang-cc 0x0000000000427ee7 21 clang-cc 0x000000000061f19b 22 clang-cc 0x000000000040dce9 23 clang-cc 0x00000000004167da main + 5066 24 libc.so.6 0x000000367221e5c6 __libc_start_main + 230 25 clang-cc 0x000000000040b869 Stack dump: 0. Program arguments: /usr/local/bin/../libexec/clang-cc -triple x86_64-unknown-linux-gnu -S -disable-free -main-file-name hrtimer.c --relocation-model static --disable-fp-elim --unwind-tables=1 --disable-red-zone -mcpu core2 -code-model kernel -target-feature -sse -target-feature -mmx -target-feature -sse2 -target-feature -3dnow --fmath-errno=1 -nostdinc -dependency-file kernel/.hrtimer.o.d -MT kernel/.tmp_hrtimer.o -sys-header-deps -isystem /usr/local/bin/../lib/clang/1.0/include -include include/linux/autoconf.h -D__KERNEL__ -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -DKBUILD_STR(s)=#s -DKBUILD_BASENAME=KBUILD_STR(hrtimer) -DKBUILD_MODNAME=KBUILD_STR(hrtimer) -Iinclude -I/home/edwin/builds/linux-2.6/arch/x86/include -O2 -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -Werror-implicit-function-declaration -Wno-format-security -Wno-unused-value -Wno-sign-compare -Wframe-larger-than=2048 -Wdeclaration-after-statement -Wno-pointer-sign --stack-protector=0 -fno-common -fdiagnostics-show-option -o /tmp/cc-Rz1pJ9.s -x c kernel/hrtimer.c 1. parser at end of file 2. Per-function optimization 3. Running pass 'Combine redundant instructions' on function '@init_hrtimers_cpu' clang: error: compiler command failed due to signal 6 (use -v to see invocation) make[1]: *** [kernel/hrtimer.o] Error 250 opt bugpoint-reduced-simplified.bc -instcombine WARNING: You're attempting to print out a bitcode file. This is inadvisable as it may cause display problems. If you REALLY want to taste LLVM bitcode first-hand, you can force output with the `-f' option. opt: Unexpected unknown exception occurred. -- 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 1 05:41:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 05:41:47 -0500 Subject: [LLVMbugs] [Bug 4849] New: Test failure in llvm2. 6-prelease 1 when compiled with llvm-gcc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4849 Summary: Test failure in llvm2.6-prelease 1 when compiled with llvm-gcc Product: Test Suite Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: c.r1 at gmx.de CC: llvmbugs at cs.uiuc.edu Hi, in the following circumstances, one test of the DejaGNU suite fails. build env: corni at h1425567:~/llvm-2.6/build$ gcc --version gcc (GCC) 4.2.4 (Ubuntu 4.2.4-1ubuntu4) Copyright (C) 2007 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. 1) built llvm-2.6 with the above gcc 2) run make check - everything fine 3) bootstrapped llvm-gcc-2.6 against the above llvm-2.6 4) compiled llvm-2.6 with the bootstrapped llvm-gcc 5) run make check - one failure: FAIL: /home/corni/llvm-2.6/llvm-2.6/test/Feature/load_module.ll for PR1318 Failed with exit(1) at line 1 while running: llvm-as < /home/corni/llvm-2.6/llvm-2.6/test/Feature/load_module.ll | opt -load=/home/corni/llvm-2.6/build/Release/lib/LLVMHello.so -hello -disable-output - |& /bin/grep Hello Error opening '/home/corni/llvm-2.6/build/Release/lib/LLVMHello.so': /home/corni/llvm-2.6/build/Release/lib/LLVMHello.so: undefined symbol: _ZNK4llvm4Pass11getPassNameEv child process exited abnormally The system is i686, Ubuntu 8.04 server on a Strato V-Power Server. -- 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 1 06:35:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 06:35:57 -0500 Subject: [LLVMbugs] [Bug 4850] New: The execution of code generated for MIPS is failure Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4850 Summary: The execution of code generated for MIPS is failure Product: new-bugs Version: 2.6 Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: miyo at wasamon.net CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3410) --> (http://llvm.org/bugs/attachment.cgi?id=3410) tried source code, generated code by llvm, and the dump of execution I compiled the program (swap0.c) included the attached archive by using clang-cc and native gcc for MIPS32. which is a simple program to swap arguments But the execution was failure. The process of compilation is the following: clang-cc swap0.c llvm-as swap0.ll llc -march=mips swap0.bc gcc -g swap0.s I checked a dump result of execution on MIPS (dump.txt). In the result, it seems that callee take the arguments by register although caller passed the arguments by stack. At the instruction at line 20 in caller, sw $4, 0($sp) callee stores one of the passed arguments into 0($sp) under assuming that the values was passed by $4. But, caller does not set the value to $4. So, when the program load a value from0($sp), the execution is failure. It is succeeded executing of codes for x86 and Cell SPU generated by same process. Do I have to set some options for a compilation of MIPS program ? Or, did I have mistakes in my compilation process? Target processor is MIPS and Linux is running on it. -- 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 1 07:32:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 07:32:35 -0500 Subject: [LLVMbugs] [Bug 4851] New: [MIPS] main() causes an assert Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4851 Summary: [MIPS] main() causes an assert Product: new-bugs Version: trunk Platform: PC OS/Version: Linux 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 main() is defined as a varargs function by my parser. The attached test case fails during code generation: [~/ellcc/test/source] main% llvm-as main.o.ll Error opening output file 'main.o.bc' Use -f command line argument to force output [~/ellcc/test/source] main% llvm-as -f main.o.ll [~/ellcc/test/source] main% llc main.o.bc llc: /home/rich/llvm-trunk-new/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp:5872: void llvm::SelectionDAGISel::LowerArguments(llvm::BasicBlock*): Assertion `InVals.size() == Ins.size() && "LowerFormalArguments didn't emit the correct number of values!"' failed. 0 llc 0x08fbf4d8 1 llc 0x08fbfaa4 2 0x00814400 __kernel_sigreturn + 0 3 libc.so.6 0x00bcce28 abort + 392 4 libc.so.6 0x00bc440e __assert_fail + 238 5 llc 0x08b9ee7b llvm::SelectionDAGISel::LowerArguments(llvm::BasicBlock*) + 1737 6 llc 0x08bd3767 llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 361 7 llc 0x08bd4e30 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 886 8 llc 0x08cc52de llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 86 9 llc 0x08f38669 llvm::FPPassManager::runOnFunction(llvm::Function&) + 299 10 llc 0x08f3916c llvm::FunctionPassManagerImpl::run(llvm::Function&) + 124 11 llc 0x08f39339 llvm::FunctionPassManager::run(llvm::Function&) + 135 12 llc 0x0845adea main + 3636 13 libc.so.6 0x00bb66e5 __libc_start_main + 229 14 llc 0x08458fd1 Stack dump: 0. Program arguments: llc main.o.bc 1. Running pass 'MIPS DAG->DAG Pattern Instruction Selection' on function '@main' Abort [~/ellcc/test/source] main% -- 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 1 08:25:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 08:25:47 -0500 Subject: [LLVMbugs] [Bug 4852] New: [Sparc] The code generator uses ".bss" instead of ". section .bss" Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4852 Summary: [Sparc] The code generator uses ".bss" instead of ".section .bss" Product: new-bugs Version: trunk Platform: PC OS/Version: Linux 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 Created an attachment (id=3412) --> (http://llvm.org/bugs/attachment.cgi?id=3412) A small example that generates .bss In the attached file, there is a static initialized variable that is initialized to zero. It is placed in the .bss section by using a ".bss" directive rather than a .section .bss" directive. -- 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 1 11:14:31 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 11:14:31 -0500 Subject: [LLVMbugs] [Bug 4061] clang breaks strict-aliasing rules according to g++4.4 In-Reply-To: Message-ID: <200909011614.n81GEVjp004127@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4061 Douglas Gregor changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Douglas Gregor 2009-09-01 11:14:30 --- Should be fixed here: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090831/020877.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 1 13:39:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 13:39:55 -0500 Subject: [LLVMbugs] [Bug 4415] crash with -indvars In-Reply-To: Message-ID: <200909011839.n81Idtg3024757@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4415 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #7 from Chris Lattner 2009-09-01 13:39:55 --- This apparently can't be reproduced. -- 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 1 17:43:01 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 17:43:01 -0500 Subject: [LLVMbugs] [Bug 4855] New: Missed optimization on pointer known equal to a constant global Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4855 Summary: Missed optimization on pointer known equal to a constant global Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: jyasskin at google.com CC: nicholas at mxc.ca, llvmbugs at cs.uiuc.edu Blocks: 1772 The following module is a fixed point under "opt -O3", while the optimizers ought to replace "load i8* %a" with "load i8* @gv" and then realize that loading it produces a constant value. "opt -predsimplify -O3" successfully optimizes it. ; ModuleID = '' @gv = internal constant i8 7 ; [#uses=1] define i8 @f(i8* %a) nounwind { entry: %cond = icmp eq i8* %a, @gv ; [#uses=1] br i1 %cond, label %bb2, label %bb bb: ; preds = %entry ret i8 0 bb2: ; preds = %entry %should_be_const = load i8* %a ; [#uses=1] ret i8 %should_be_const } This seems related to PR1768, but Nick assures me they're different. I'm interested because this is the form I was planning to use for Unladen Swallow's type guards. -- 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 1 19:19:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 19:19:44 -0500 Subject: [LLVMbugs] [Bug 4845] [2.6] crashing inside JITed code when running on x86_64 In-Reply-To: Message-ID: <200909020019.n820Ji9x006772@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4845 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Evan Cheng 2009-09-01 19:19:43 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086156.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 1 19:43:25 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 19:43:25 -0500 Subject: [LLVMbugs] [Bug 4857] New: Check/fix sys::getHostTriple on subtargets. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4857 Summary: Check/fix sys::getHostTriple on subtargets. Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu We should check getHostTriple to make sure it is returning an appropriate value for subtargets. Bug 4845 seems to indicate that on Darwin at least it is returning i386-apple-darwin... even when running in 64-bit mode. This should be fixed. -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Tue Sep 1 22:11:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 22:11:45 -0500 Subject: [LLVMbugs] [Bug 4860] New: Clang doesn' t honour -Wno-attributes when applying __attribute__(( __warn_unused_result__)) to variables. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4860 Summary: Clang doesn't honour -Wno-attributes when applying __attribute__((__warn_unused_result__)) to variables. Product: new-bugs Version: unspecified Platform: PC OS/Version: Solaris Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jpihlaja at cc.helsinki.fi CC: llvmbugs at cs.uiuc.edu, eocallaghan at auroraux.org The following program solicits a warning when the __warn_unused_result__ attribute is applied to a variable. gcc >= 4.1 does not emit a warning with -Wno-attributes. $ cat >wnoattr-var.c __attribute__((__warn_unused_result__)) int x; $ /opt/clang/bin/clang -c -Wall -Wno-attributes wnoattr-var.c wnoattr-var.c:1:16: warning: '__warn_unused_result__' attribute only applies to function types __attribute__((__warn_unused_result__)) int x; ^ 1 diagnostic generated. $ /opt/clang/bin/clang -v clang version 1.0 (http://llvm.org/svn/llvm-project/cfe/trunk 80665) Target: i386-pc-solaris2.11 Thread model: posix $ uname -a SunOS hydra 5.11 snv_110 i86pc i386 i86pc -- 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 1 23:32:46 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 1 Sep 2009 23:32:46 -0500 Subject: [LLVMbugs] [Bug 4061] clang breaks strict-aliasing rules according to g++4.4 In-Reply-To: Message-ID: <200909020432.n824WklC007466@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4061 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #7 from Duncan Sands 2009-09-01 23:32:45 --- Reopening because gcc-4.4 still reports a gazillion issues. Here are the first two: clang/lib/Basic/../../include/clang/Basic/SourceLocation.h: In member function ???bool clang::SourceManager::isBeforeInTranslationUnit(clang::SourceLocation, clang::SourceLocation) const???: clang/lib/Basic/../../include/clang/Basic/SourceLocation.h:43: warning: dereferencing pointer ?????? does break strict-aliasing rules /usr/include/c++/4.4/bits/stl_tree.h:175: note: initialized from here ... clang/lib/Parse/../../include/clang/Parse/DeclSpec.h: In member function ???clang::CXXScopeSpec& clang::DeclaratorChunk::MemberPointerTypeInfo::Scope()???: clang/lib/Parse/../../include/clang/Parse/DeclSpec.h:637: warning: dereferencing type-punned pointer will break strict-aliasing rules -- 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 2 00:35:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 00:35:55 -0500 Subject: [LLVMbugs] [Bug 4848] [2.6] instcombine runs out of memory (infinite loop?) when compiling kernel/hrtimer.c In-Reply-To: Message-ID: <200909020535.n825Ztww015707@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4848 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-02 00:35:55 --- Fixed in r80762 -- 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 2 01:11:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 01:11:05 -0500 Subject: [LLVMbugs] [Bug 4861] New: Eliminate VISIBILITY_HIDDEN from the code base Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4861 Summary: Eliminate VISIBILITY_HIDDEN from the code base Product: new-bugs Version: unspecified Platform: PC OS/Version: All Status: NEW Keywords: code-cleanup Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: clattner at apple.com CC: llvmbugs at cs.uiuc.edu VISIBILITY_HIDDEN is an optimization hint for compilers prior to GCC 4.2. By the time 2.7 ships, gcc 4.1 and earlier will be really really really old. This hint is cluttering up the codebase and should be removed. -- 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 2 01:31:38 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 01:31:38 -0500 Subject: [LLVMbugs] [Bug 4815] DSE invalidates iterator In-Reply-To: Message-ID: <200909020631.n826Vc76023057@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4815 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Summary|Assertion `NodePtr && "--'d |DSE invalidates iterator |off the beginning of an | |ilist!"' failed. | --- Comment #3 from Chris Lattner 2009-09-02 01:31:38 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086181.html Thanks John! -- 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 2 02:29:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 02:29:50 -0500 Subject: [LLVMbugs] [Bug 4845] [2.6] crashing inside JITed code when running on x86_64 In-Reply-To: Message-ID: <200909020729.n827To3v030695@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4845 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #7 from Evan Cheng 2009-09-02 02:29:50 --- Ok, JIT is fixed on x86_64-apple-darwin as far as I can tell. But the example is still failing for some reason. -- 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 gausszhch at gmail.com Wed Sep 2 06:42:17 2009 From: gausszhch at gmail.com (gauss) Date: Wed, 2 Sep 2009 19:42:17 +0800 Subject: [LLVMbugs] link-error: different visibilities Message-ID: <834cd97f0909020442l5260524ag2de4a4628c644c40@mail.gmail.com> When I use llvm-2.5 to compile gnash which is a GNU flash movie player, some errors appeared as follow: *llvm-ld: error: Cannot link in module '../libcore/.libs/libgnashcore.a(movie_root.o)': Linking globals named '_ZNKSt6vectorIN5gnash8geometry7Range2dIfEESaIS3_EE4sizeEv': symbols have different visibilities!* I demangle the name "_ZNKSt6vectorIN5gnash8geometry7Range2dIfEESaIS3_EE4sizeEv" with tool c++filt, and its corresponding source code is: *std::vector, std::allocator > >::size() const.* It is a template method "Vector<>::size()" inserted by llvm-g++ at compile time. LLVM-g++ compiled all the cpp file correctly, but llvm-ld failed to link them because of the different visibilities. Is there some wrong with LLVM-g++? How to deal with it? Any one will help me? Thanks! Gauss, 09-02 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvmbugs/attachments/20090902/abd5bada/attachment.html From bugzilla-daemon at cs.uiuc.edu Wed Sep 2 12:05:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 12:05:22 -0500 Subject: [LLVMbugs] [Bug 4850] The execution of code generated for MIPS is failure In-Reply-To: Message-ID: <200909021705.n82H5MVT022028@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4850 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |INVALID --- Comment #4 from Daniel Dunbar 2009-09-02 12:05:20 --- Clang is not yet supported for targetting MIPS. -- 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 2 13:55:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 13:55:41 -0500 Subject: [LLVMbugs] [Bug 4865] New: JIT dwarf exception handling seems broken Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4865 Summary: JIT dwarf exception handling seems broken Product: libraries Version: 2.6 Platform: PC OS/Version: All Status: NEW Severity: major Priority: P2 Component: Target-Independent JIT AssignedTo: unassignedbugs at nondot.org ReportedBy: lsansonetti at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3416) --> (http://llvm.org/bugs/attachment.cgi?id=3416) test source Tested on: Mac OS 10.6, Core 2 Duo, on both i386 and x86_64 modes. It looks like the dwarf exception tables emitted at runtime for libgcc causes a crash inside the personality function. This used to work at least until r72741 on the exact same environment (which has the new libgcc-compatible runtime), so I suspect a regression in LLVM. Attached is a simple program (both the source and the bitcode) and the crash is reproducible using: $ lli -enable-eh /tmp/t.bc 0 lli 0x0000000100756962 std::_Rb_tree, std::less, std::allocator >::_M_erase(std::_Rb_tree_node*) + 8866 1 lli 0x0000000100757426 std::_Rb_tree, std::less, std::allocator >::_M_erase(std::_Rb_tree_node*) + 11622 2 libSystem.B.dylib 0x00007fff8016a14a _sigtramp + 26 3 libSystem.B.dylib 000000000000000000 _sigtramp + 2146000592 4 libstdc++.6.dylib 0x00007fff84b61299 read_encoded_value_with_base(unsigned char, unsigned long, unsigned char const*, unsigned long*) + 267 5 libstdc++.6.dylib 0x00007fff84b613a1 parse_lsda_header(_Unwind_Context*, unsigned char const*, lsda_header_info*) + 68 6 libstdc++.6.dylib 0x00007fff84b616e2 __gxx_personality_v0 + 217 7 libSystem.B.dylib 0x00007fff8016ecb1 _Unwind_RaiseException + 145 8 libstdc++.6.dylib 0x00007fff84b61bef __cxa_throw + 91 9 libstdc++.6.dylib 0x0000000100b0010c __cxa_throw + 2079974776 10 lli 0x0000000100454bb2 std::vector >::_M_insert_aux(__gnu_cxx::__normal_iterator > >, llvm::sys::MemoryBlock const&) + 30722 11 lli 0x000000010001177d llvm::cl::ValuesClass llvm::cl::values(char const*, int, char const*, ...) + 7421 12 lli 0x0000000100011248 llvm::cl::ValuesClass llvm::cl::values(char const*, int, char const*, ...) + 6088 13 lli 0x0000000000000003 llvm::cl::ValuesClass llvm::cl::values(char const*, int, char const*, ...) + 4294903171 Stack dump: 0. Program arguments: lli -enable-eh /tmp/t.bc Abort trap I looked at the changes in JITDwarfEmitter.cpp since and tried to revert r79674, without success. -- 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 2 15:28:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 15:28:51 -0500 Subject: [LLVMbugs] [Bug 4866] New: Exciting new crash with -inline -dse -domtree Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4866 Summary: Exciting new crash with -inline -dse -domtree 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: baldrick at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3418) --> (http://llvm.org/bugs/attachment.cgi?id=3418) testcase .ll This problem didn't exist a few days ago. Reduced from Ada ACATS test c974001. $ opt -inline -dse -domtree dom_crash.bc 0 opt 0x0000000000990b57 1 opt 0x000000000099098a 2 libc.so.6 0x00007faad8e17530 3 opt 0x0000000000515006 __gnu_cxx::__normal_iterator**, std::__norm::vector*, std::allocator*> > >::__normal_iterator(llvm::DomTreeNodeBase** const&) + 16 4 opt 0x0000000000513f3b std::__norm::vector*, std::allocator*> >::begin() + 31 5 opt 0x0000000000512b8b std::__debug::vector*, std::allocator*> >::begin() + 33 6 opt 0x0000000000877a86 llvm::DomTreeNodeBase::begin() + 38 7 opt 0x00000000005116fc llvm::GraphTraits*>::child_begin(llvm::DomTreeNodeBase*) + 34 8 opt 0x0000000000514675 llvm::df_iterator*, llvm::SmallPtrSet*, 8u>, false, llvm::GraphTraits*> >::df_iterator(llvm::DomTreeNodeBase*) + 125 9 opt 0x0000000000513652 llvm::df_iterator*, llvm::SmallPtrSet*, 8u>, false, llvm::GraphTraits*> >::begin(llvm::DomTreeNodeBase* const&) + 47 10 opt 0x00000000005124b7 llvm::df_iterator*, llvm::SmallPtrSet*>::NodeType*, 8u>, false, llvm::GraphTraits*> > llvm::df_begin*>(llvm::DomTreeNodeBase* const&) + 34 11 opt 0x000000000050e19d 12 opt 0x00000000008d5e37 llvm::FPPassManager::runOnFunction(llvm::Function&) + 395 13 opt 0x0000000000708e8c 14 opt 0x0000000000709bac 15 opt 0x00000000008d63ce llvm::MPPassManager::runOnModule(llvm::Module&) + 504 16 opt 0x00000000008d6985 llvm::PassManagerImpl::run(llvm::Module&) + 125 17 opt 0x00000000008d6c8d llvm::PassManager::run(llvm::Module&) + 39 18 opt 0x00000000004203f8 main + 3595 19 libc.so.6 0x00007faad8e02abd __libc_start_main + 253 20 opt 0x00000000004002a9 Stack dump: 0. Running pass 'CallGraph Pass Manager' on module 'dom_crash.bc'. 1. Running pass 'Dead Store Elimination' on function '@c974001__lengthy_calculation.1736' Segmentation fault (core dumped) -- 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 2 19:03:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 19:03:16 -0500 Subject: [LLVMbugs] [Bug 4869] New: clang x86-32 doesn't implement ABI on FreeBSD Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4869 Summary: clang x86-32 doesn't implement ABI on FreeBSD Product: clang Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: csdavec at swan.ac.uk CC: llvmbugs at cs.uiuc.edu Running the GCC ABI test with clang as the other C compiler, I get these failures: FAIL: gcc.dg/compat/struct-by-value-22 c_compat_x_alt.o compile FAIL: gcc.dg/compat/struct-by-value-22 c_compat_y_alt.o compile FAIL: gcc.dg/compat/struct-complex-1 c_compat_x_alt.o compile FAIL: gcc.dg/compat/struct-complex-1 c_compat_y_alt.o compile FAIL: gcc.dg/compat/struct-complex-2 c_compat_x_alt.o compile FAIL: gcc.dg/compat/struct-complex-2 c_compat_y_alt.o compile FAIL: gcc.dg/compat/struct-layout-1 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/struct-layout-1 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/struct-return-19 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/struct-return-19 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/struct-return-21 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/struct-return-21 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/struct-return-2 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/struct-return-2 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/struct-return-3 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/struct-return-3 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/union-return-1 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/union-return-1 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/vector-1 c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/vector-1 c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/vector-1a c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/vector-1a c_compat_x_alt.o-c_compat_y_tst.o execute FAIL: gcc.dg/compat/vector-2 c_compat_x_alt.o compile FAIL: gcc.dg/compat/vector-2 c_compat_y_alt.o compile FAIL: gcc.dg/compat/vector-2a c_compat_x_tst.o-c_compat_y_alt.o execute FAIL: gcc.dg/compat/vector-2a c_compat_x_alt.o-c_compat_y_tst.o execute -- 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 2 21:13:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 21:13:43 -0500 Subject: [LLVMbugs] [Bug 4852] [Sparc] The code generator uses ".bss" instead of ".section .bss " In-Reply-To: Message-ID: <200909030213.n832Dh5r029822@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4852 Richard Pennington changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Richard Pennington 2009-09-02 21:13:43 --- This was fixed somewhere between the code freeze and 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 Wed Sep 2 22:22:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 2 Sep 2009 22:22:56 -0500 Subject: [LLVMbugs] [Bug 4870] New: Memory corruption in metadata something something Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4870 Summary: Memory corruption in metadata something something Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu My i386-apple-darwin10 LLVM build fails because opt crashes stripping the debug info in libprofile. This appears to be some kind of memory corruption, and it works for x86_64-apple-darwin10. -- ddunbar at giles:tmp$ /Users/ddunbar/llvm.obj.32/Debug/bin/opt -std-compile-opts -strip-debug -o /dev/null -f /tmp/t.bc ddunbar at giles:tmp$ /Users/../Users/ddunbar/llvm.obj.32/Debug/bin/opt -std-compile-opts -strip-debug -o /dev/null -f /tmp/t.bc ddunbar at giles:tmp$ /Users/../Users/../Users/ddunbar/llvm.obj.32/Debug/bin/opt -std-compile-opts -strip-debug -o /dev/null -f /tmp/t.bc While deleting: metadata % Use still stuck around after Def is destroyed:!0 = metadata !{i32 458798, i32 0, metadata !1, metadata !"__inline_memcpy_chk", metadata !"__inline_memcpy_chk", metadata !"__inline_memcpy_chk", metadata !2, i32 57, null, i1 true, i1 true} !1 = metadata !{i32 458769, i32 0, i32 1, metadata !"CommonProfiling.c", metadata !"/Volumes/Data/Users/ddunbar/llvm/runtime/libprofile", metadata !"4.2.1 (Based on Apple Inc. build 5649) (LLVM build)", i1 true, i1 false, metadata !"", i32 0} !2 = metadata !{i32 458769, i32 0, i32 1, metadata !"_string.h", metadata !"/usr/include/secure", metadata !"4.2.1 (Based on Apple Inc. build 5649) (LLVM build)", i1 false, i1 false, metadata !"", i32 0} Assertion failed: (use_empty() && "Uses remain when a value is destroyed!"), function ~Value, file /Volumes/Data/Users/ddunbar/llvm/lib/VMCore/Value.cpp, line 81. 0 opt 0x003d0270 PrintStackTrace(void*) + 45 1 opt 0x003d0878 SignalHandler(int) + 410 2 libSystem.B.dylib 0x93442bfb _sigtramp + 43 3 libSystem.B.dylib 0xffffffff _sigtramp + 1824248879 4 libSystem.B.dylib 0x934d0ba5 raise + 26 5 libSystem.B.dylib 0x934e6c5c abort + 93 6 libSystem.B.dylib 0x934d3804 __pthread_markcancel + 0 7 opt 0x0038048f llvm::Value::~Value() + 443 8 opt 0x002e357b llvm::User::~User() + 103 9 opt 0x00356514 llvm::MetadataBase::~MetadataBase() + 40 10 opt 0x0035520a llvm::MDNode::~MDNode() + 106 11 opt 0x00349e12 llvm::LLVMContext::RemoveDeadMetadata() + 258 12 opt 0x000a7c5a StripDebugInfo(llvm::Module&) + 719 13 opt 0x000a82f4 (anonymous namespace)::StripSymbols::runOnModule(llvm::Module&) + 26 14 opt 0x0035ea08 llvm::MPPassManager::runOnModule(llvm::Module&) + 378 15 opt 0x00360c52 llvm::PassManagerImpl::run(llvm::Module&) + 112 16 opt 0x00360ca9 llvm::PassManager::run(llvm::Module&) + 27 17 opt 0x0005adca main + 4469 18 opt 0x0004f0c9 start + 53 Stack dump: 0. Running pass 'Strip all symbols from a module' on module '/tmp/t.bc'. Abort trap -- (Yes, thats running the same binary three times with a longer argv[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 Thu Sep 3 03:47:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 03:47:05 -0500 Subject: [LLVMbugs] [Bug 4871] New: Assertion failed: (Bits != 0 && " Cannot print this instruction."), function printInstruction, file /data/ home/rdivacky/llvm/lib/Target/X86/AsmPrinter/../X86GenAsmWriter.inc, line 2251. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4871 Summary: Assertion failed: (Bits != 0 && "Cannot print this instruction."), function printInstruction, file /data/home/rdivacky/llvm/lib/Target/X86/AsmPrinter/../X8 6GenAsmWriter.inc, line 2251. Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu I am getting this assertion when compiling the attached lev /tmp$ clang -c divxc3.c Assertion failed: (Bits != 0 && "Cannot print this instruction."), function printInstruction, file /data/home/rdivacky/llvm/lib/Target/X86/AsmPrinter/../X86GenAsmWriter.inc, line 2251. Stack dump: 0. Program arguments: /usr/local/bin/../libexec/clang-cc -triple x86_64-unknown-freebsd7.2 -S -disable-free -main-file-name divxc3.c --relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64 --fmath-errno=1 -fdiagnostics-show-option -o /tmp/cc-3IzTHM.s -x c divxc3.c 1. parser at end of file 2. Code generation 3. Running pass 'X86 AT&T-Style Assembly Printer' on function '@__divxc3' clang: error: compiler command failed due to signal 6 (use -v to see invocation) this is with clang/llvm r80908 -- 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 3 04:08:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 04:08:17 -0500 Subject: [LLVMbugs] [Bug 4866] Exciting new crash with -inline -dse -domtree In-Reply-To: Message-ID: <200909030908.n8398HjZ032075@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4866 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #3 from Duncan Sands 2009-09-03 04:08:15 --- It turns out that this was due to a patch I had applied locally. Sorry for the noise. -- 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 3 06:15:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 06:15:48 -0500 Subject: [LLVMbugs] [Bug 4872] New: Bitcode linking of debug information fails Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4872 Summary: Bitcode linking of debug information fails Product: new-bugs Version: trunk Platform: PC OS/Version: Linux 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 Created an attachment (id=3422) --> (http://llvm.org/bugs/attachment.cgi?id=3422) First file. If I link two files with debugging metadata, the link failes. [~/ellcc/test/source] main% llvm-as a.o.ll [~/ellcc/test/source] main% llvm-as b.o.ll [~/ellcc/test/source] main% llvm-ld a.o.bc b.o.bc 0 llvm-ld 0x085986f8 1 llvm-ld 0x08598cc4 2 0x00d88400 __kernel_sigreturn + 0 3 llvm-ld 0x082a6c78 bool llvm::isa_impl(llvm::Value const&) + 29 4 llvm-ld 0x082aae3c llvm::isa_impl_wrap::doit(llvm::Value const&) + 29 5 llvm-ld 0x082aae5f bool llvm::isa_impl_cl::isa(llvm::Value const&) + 29 6 llvm-ld 0x082aae82 bool llvm::isa_impl_cl::isa(llvm::Value const&) + 29 7 llvm-ld 0x082aaea5 bool llvm::isa_impl_cl::isa(llvm::Value const*) + 29 8 llvm-ld 0x082aaeca bool llvm::isa(llvm::Value const* const&) + 31 9 llvm-ld 0x082be4ba llvm::cast_retty::ret_type llvm::dyn_cast(llvm::Value const* const&) + 29 10 llvm-ld 0x082c3bb8 11 llvm-ld 0x082c4184 12 llvm-ld 0x082c45e9 13 llvm-ld 0x082c47b6 14 llvm-ld 0x082c811d llvm::Linker::LinkModules(llvm::Module*, llvm::Module*, std::string*) + 1979 15 llvm-ld 0x082c24e2 llvm::Linker::LinkInModule(llvm::Module*, std::string*) + 46 16 llvm-ld 0x082c19b1 llvm::Linker::LinkInFile(llvm::sys::Path const&, bool&) + 1505 17 llvm-ld 0x082c20e9 llvm::Linker::LinkInItems(std::vector, std::allocator > > const&, std::vector, std::allocator > >&) + 259 18 llvm-ld 0x082980c0 main + 1589 19 libc.so.6 0x00bb66e5 __libc_start_main + 229 20 llvm-ld 0x0828e831 Stack dump: 0. Program arguments: llvm-ld a.o.bc b.o.bc Segmentation fault -- 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 3 07:45:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 07:45:32 -0500 Subject: [LLVMbugs] [Bug 4873] New: Infinite loop in APFloat::convertFromDecimalString() around APFloat.cpp:2378 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4873 Summary: Infinite loop in APFloat::convertFromDecimalString() around APFloat.cpp:2378 Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: d.g.gorbachev at gmail.com CC: llvmbugs at cs.uiuc.edu LLVM r80583, compiled with LLVM-GCC with -O3. Test program: int main(void) { llvm::APFloat af(0.0); af.convertFromString(llvm::StringRef("2.71828", 7), llvm::APFloat::rmNearestTiesToEven); return 0; } "p" is incremented, then assigned back to the beginning of the string. -- 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 3 07:49:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 07:49:03 -0500 Subject: [LLVMbugs] [Bug 4874] New: llvm-gcc build on Sloaris SPARC Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4874 Summary: llvm-gcc build on Sloaris SPARC Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: fedor.sakharov at gmail.com CC: llvmbugs at cs.uiuc.edu Building llvm-gcc-4.2 ends up with an error. llvm-gcc-4.2-src/gcc -I../../llvm-gcc-4.2-src/gcc/. -I../../llvm-gcc-4.2-src/gcc/../include -I../../llvm-gcc-4.2-src/gcc/../libcpp/include -I../../llvm-gcc-4.2-src/gcc/../libdecnumber -I../libdecnumber -I/export/home/sakharov/llvm-objects/include -I/export/home/sakharov/llvm-src/include \ -c ../../llvm-gcc-4.2-src/gcc/config/sparc/gmon-sol2.c -o gmon.o ../../llvm-gcc-4.2-src/gcc/config/sparc/gmon-sol2.c:424: internal compiler error: Segmentation Fault Please submit a full bug report, with preprocessed source if appropriate. See for instructions. make[2]: *** [gmon.o] Error 1 make[2]: Leaving directory `/export/home/sakharov/llvm-gcc-4.2-objects/gcc' make[1]: *** [all-gcc] Error 2 make[1]: Leaving directory `/export/home/sakharov/llvm-gcc-4.2-objects' make: *** [all] Error 2 -- 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 3 09:11:31 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 09:11:31 -0500 Subject: [LLVMbugs] [Bug 4875] New: llmv-ld is not able to link libprofile in debug mode. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4875 Summary: llmv-ld is not able to link libprofile in debug mode. Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: astifter-llvm at gmx.at CC: llvmbugs at cs.uiuc.edu, astifter-llvm at gmx.at When llvm is compiled with debugging enabled, then I get the following error during linking the libprofile: make[1]: Entering directory `/localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile' llvm[1]: Building Debug Bytecode Archive libprofile_rt.bca (internalize) 0 llvm-ld 0x0000000000974854 1 llvm-ld 0x0000000000974db3 2 libpthread.so.0 0x00002aaaaacd6a80 3 llvm-ld 0x00000000006b68de llvm::Value::getValueID() const + 12 4 llvm-ld 0x00000000006b68fc bool llvm::isa_impl(llvm::Value const&) + 21 5 llvm-ld 0x00000000006ba616 llvm::isa_impl_wrap::doit(llvm::Value const&) + 21 6 llvm-ld 0x00000000006ba62d bool llvm::isa_impl_cl::isa(llvm::Value const&) + 21 7 llvm-ld 0x00000000006ba644 bool llvm::isa_impl_cl::isa(llvm::Value const&) + 21 8 llvm-ld 0x00000000006ba65b bool llvm::isa_impl_cl::isa(llvm::Value const*) + 21 9 llvm-ld 0x00000000006ba675 bool llvm::isa(llvm::Value const* const&) + 24 10 llvm-ld 0x00000000006cb39f llvm::cast_retty::ret_type llvm::dyn_cast(llvm::Value const* const&) + 21 11 llvm-ld 0x00000000006d067a 12 llvm-ld 0x00000000006d0bd5 13 llvm-ld 0x00000000006d0bd5 14 llvm-ld 0x00000000006d1056 15 llvm-ld 0x00000000006d1210 16 llvm-ld 0x00000000006d4c51 llvm::Linker::LinkModules(llvm::Module*, llvm::Module*, std::string*) + 1897 17 llvm-ld 0x00000000006ceff5 llvm::Linker::LinkInModule(llvm::Module*, std::string*) + 41 18 llvm-ld 0x00000000006ce56b llvm::Linker::LinkInFile(llvm::sys::Path const&, bool&) + 1431 19 llvm-ld 0x00000000006cebf6 llvm::Linker::LinkInItems(std::vector, std::allocator > > const&, std::vector, std::allocator > >&) + 258 20 llvm-ld 0x00000000006a95c0 main + 1349 21 libc.so.6 0x00002aaaabaaf1a6 __libc_start_main + 230 22 llvm-ld 0x00000000006a05a9 __gxx_personality_v0 + 681 Stack dump: 0. Program arguments: /localtmp/astifter/llvm/llvm-svn-debug-obj/Debug/bin/llvm-ld -L/localtmp/astifter/llvm/llvm-van-install/bin/../lib -internalize-public-api-file=/localtmp/astifter/llvm/llvm-svn/runtime/libprofile/exported_symbols.lst -o /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/profile_rt.internalize /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/BasicBlockTracing.bc /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/BlockProfiling.bc /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/CommonProfiling.bc /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/EdgeProfiling.bc /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/FunctionProfiling.bc /localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile/Debug/OptimalEdgeProfiling.bc make[1]: *** [/localtmp/astifter/llvm/llvm-svn-debug-obj/Debug/lib/libprofile_rt.bca] Segmentation fault make[1]: Leaving directory `/localtmp/astifter/llvm/llvm-svn-debug-obj/runtime/libprofile' The system is "Linux 2.6.25-2-amd64 #1 SMP Mon Jul 14 11:05:23 UTC 2008 x86_64 GNU/Linux" with (almost) latest trunk llvm and llvm-gcc. Andi -- 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 3 12:04:26 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 12:04:26 -0500 Subject: [LLVMbugs] [Bug 4870] Memory corruption in metadata something something In-Reply-To: Message-ID: <200909031704.n83H4Qxj029478@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4870 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |devang.patel at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from devang.patel 2009-09-03 12:04:26 --- Fixed. rev. 80928. -- 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 3 15:20:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 15:20:13 -0500 Subject: [LLVMbugs] [Bug 4875] llmv-ld is not able to link libprofile in debug mode. In-Reply-To: Message-ID: <200909032020.n83KKDZ2023393@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4875 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |devang.patel at gmail.com Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #11 from devang.patel 2009-09-03 15:20:12 --- *** This bug has been marked as a duplicate of bug 4872 *** -- 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 3 15:25:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 15:25:40 -0500 Subject: [LLVMbugs] [Bug 1415] Refactor debug/eh emission code. In-Reply-To: Message-ID: <200909032025.n83KPeff024239@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1415 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |devang.patel at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from devang.patel 2009-09-03 15:25:39 --- Bill did refactoring in this area some 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 Thu Sep 3 15:36:28 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 15:36:28 -0500 Subject: [LLVMbugs] [Bug 4872] Bitcode linking of debug information fails In-Reply-To: Message-ID: <200909032036.n83KaScY025685@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4872 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |devang.patel at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from devang.patel 2009-09-03 15:36:28 --- Fixed. Revision 80941. -- 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 3 18:47:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 18:47:11 -0500 Subject: [LLVMbugs] [Bug 4877] New: Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/ Analysis/DebugInfo.cpp, line 258. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4877 Summary: Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/Analysis/DebugInfo. cpp, line 258. Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: dpatel at apple.com, llvmbugs at cs.uiuc.edu Created an attachment (id=3426) --> (http://llvm.org/bugs/attachment.cgi?id=3426) It goes boom. The attached bc asserts in llc: -- ddunbar at giles:tmp$ llc t2.bc Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/Analysis/DebugInfo.cpp, line 258. 0 llc 0x0000000100b85d65 PrintStackTrace(void*) + 38 1 llc 0x0000000100b862f3 SignalHandler(int) + 336 2 libSystem.B.dylib 0x00007fff8726214a _sigtramp + 26 3 libSystem.B.dylib 0x00007fff5fbfe1c0 _sigtramp + 3633954960 4 libSystem.B.dylib 0x00007fff872de104 __pthread_markcancel + 0 5 llc 0x0000000100a065ee llvm::DIArray::getNumElements() const + 64 6 llc 0x000000010083ab0f llvm::DwarfDebug::ConstructTypeDIE(llvm::CompileUnit*, llvm::DIE&, llvm::DICompositeType) + 565 7 llc 0x000000010083838f llvm::DwarfDebug::AddType(llvm::CompileUnit*, llvm::DIE*, llvm::DIType) + 405 8 llc 0x00000001008393c2 llvm::DwarfDebug::CreateSubprogramDIE(llvm::CompileUnit*, llvm::DISubprogram const&, bool, bool) + 548 9 llc 0x0000000100839bd0 llvm::DwarfDebug::ConstructSubprogram(llvm::MDNode*) + 126 10 llc 0x000000010083a4d6 llvm::DwarfDebug::BeginModule(llvm::Module*, llvm::MachineModuleInfo*) + 576 11 llc 0x0000000100856493 llvm::DwarfWriter::BeginModule(llvm::Module*, llvm::MachineModuleInfo*, llvm::raw_ostream&, llvm::AsmPrinter*, llvm::MCAsmInfo const*) + 157 12 llc 0x000000010082dafc llvm::AsmPrinter::doInitialization(llvm::Module&) + 806 13 llc 0x0000000100b02153 llvm::FPPassManager::doInitialization(llvm::Module&) + 65 14 llc 0x0000000100b02332 llvm::FunctionPassManagerImpl::doInitialization(llvm::Module&) + 58 15 llc 0x0000000100b0239b llvm::FunctionPassManager::doInitialization() + 41 16 llc 0x000000010002e834 main + 3045 17 llc 0x000000010002d540 start + 52 18 llc 0x0000000000000002 start + 4294781686 Stack dump: 0. Program arguments: llc t2.bc Abort trap -- Derived from an llvm-gcc ICE on x86_64-pc-linux-gnu. -- 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 3 20:56:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 20:56:29 -0500 Subject: [LLVMbugs] [Bug 3570] Mac OS X: Linking to llvm modifies behaviour of abort() In-Reply-To: Message-ID: <200909040156.n841uTXR002010@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3570 Eric Christopher changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |echristo at gmail.com Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #3 from Eric Christopher 2009-09-03 20:56:29 --- I can't reproduce this with either of my machines (both 10.6.0). Can anyone cc'd still reproduce? Pull it out of worksforme if so, 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 3 23:29:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 3 Sep 2009 23:29:47 -0500 Subject: [LLVMbugs] [Bug 4879] New: strange failure with VLAs and debug info Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4879 Summary: strange failure with VLAs and debug info Product: clang Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: daniel at zuster.org CC: dpatel at apple.com, llvmbugs at cs.uiuc.edu >From a user on IRC: -- ddunbar at giles:tmp$ cat t.c #include #include static void f0(const char * s, size_t len) { char buf2[len]; memmove(buf2, s, len); fprintf(stderr, "%s\n", buf2); } int main() { const char a[] = "Hello, World!"; f0(a, sizeof a); return 0; } ddunbar at giles:tmp$ clang -m64 t.c && ./a.out Hello, World! ddunbar at giles:tmp$ clang -g -m64 t.c && ./a.out ??.o"&???? ????_?? -- -- 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 4 03:00:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 03:00:23 -0500 Subject: [LLVMbugs] [Bug 4880] New: Assertion in instruction selector Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4880 Summary: Assertion in instruction selector Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: gohman at apple.com, llvmbugs at cs.uiuc.edu Created an attachment (id=3430) --> (http://llvm.org/bugs/attachment.cgi?id=3430) Testcase Consider the attached testcase. Running "./llc -combiner-alias-analysis -combiner-global-alias-analysis bugpoint-reduced-simplified.bc" yields llc: /home/asl/proj/llvm/src/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:4925: unsigned int llvm::SelectionDAG::AssignTopologicalOrder(): Assertion `SortedPos == AllNodes.end() && "Topological sort incomplete!"' 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 Fri Sep 4 03:16:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 03:16:13 -0500 Subject: [LLVMbugs] [Bug 4881] New: LLVM does not compile libruntime, opt problem. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4881 Summary: LLVM does not compile libruntime, opt problem. Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: astifter-llvm at gmx.at CC: llvmbugs at cs.uiuc.edu, astifter-llvm at gmx.at The libruntime gets not build with the problem being that not even the compilation from CommonProfiling.ll to CommonProfiling.bc does not work. I have reduced it to $> .../Debug/bin/llvm-as .../runtime/libprofile/Debug/CommonProfiling.ll -o test.bc $> .../Debug/bin/opt -std-compile-opts -o .../runtime/libprofile/Debug/CommonProfiling.bc < test.bc 0 opt 0x0000000000b2df54 1 opt 0x0000000000b2e4b3 2 libpthread.so.0 0x00007ffff7bd4a80 3 opt 0x000000000074f3bc llvm::Value::getValueID() const + 12 4 opt 0x00000000008ed86d llvm::MDNode::classof(llvm::Value const*) + 21 5 opt 0x00000000008ee050 bool llvm::isa_impl(llvm::Value const&) + 21 6 opt 0x00000000008ee067 llvm::isa_impl_wrap::doit(llvm::Value const&) + 21 7 opt 0x00000000008ee07e bool llvm::isa_impl_cl::isa(llvm::Value const&) + 21 8 opt 0x00000000008ee095 bool llvm::isa_impl_cl::isa(llvm::Value*) + 21 9 opt 0x0000000000a9d1f8 bool llvm::isa_impl_cl::isa(llvm::Value* const&) + 24 10 opt 0x0000000000a9d21c llvm::isa_impl_wrap::doit(llvm::WeakVH const&) + 34 11 opt 0x0000000000a9d233 bool llvm::isa_impl_cl::isa(llvm::WeakVH const&) + 21 12 opt 0x0000000000a9d24a bool llvm::isa(llvm::WeakVH const&) + 21 13 opt 0x0000000000a9d3a3 llvm::cast_retty::ret_type llvm::cast(llvm::WeakVH const&) + 21 14 opt 0x0000000000a9ad9e llvm::LLVMContext::RemoveDeadMetadata() + 222 15 opt 0x0000000000789686 16 opt 0x0000000000ab10ee llvm::MPPassManager::runOnModule(llvm::Module&) + 382 17 opt 0x0000000000ab336b llvm::PassManagerImpl::run(llvm::Module&) + 111 18 opt 0x0000000000ab33cd llvm::PassManager::run(llvm::Module&) + 33 19 opt 0x000000000075e678 main + 3674 20 libc.so.6 0x00007ffff6ce61a6 __libc_start_main + 230 21 opt 0x000000000074ecb9 __gxx_personality_v0 + 697 Stack dump: 0. Running pass 'Dead Global Elimination' on module ''. Segmentation fault Interesstingly this is not happening when opt gets the file on the command line, not from stdin. -- 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 4 04:25:31 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 04:25:31 -0500 Subject: [LLVMbugs] [Bug 4882] New: bus error in MemCpyOpt Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4882 Summary: bus error in MemCpyOpt Product: new-bugs Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: critical Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3431) --> (http://llvm.org/bugs/attachment.cgi?id=3431) reduction The attached crashes with opt -memcpyopt. -- 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 4 06:43:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 06:43:36 -0500 Subject: [LLVMbugs] [Bug 4883] New: Codegen/X86-32: missed optimization on first arg passing, unneccessary stack moves Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4883 Summary: Codegen/X86-32: missed optimization on first arg passing, unneccessary stack moves 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: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu For intra-class method calls LLVM generates suboptimal code. Consider this code: class A { public: __attribute__((noinline)) int bar(int b) { return b+4; } __attribute__((noinline)) int foo(int b) { return bar(b*4); } __attribute__((noinline)) int foo() { return bar(8); } }; A a; int main() { return a.foo(4); } LLVM-G++ generates this at -O3 for foo: ZN1A3fooEi: pushl %ebp movl %esp, %ebp subl $8, %esp movl 12(%ebp), %eax shll $2, %eax movl %eax, 4(%esp) movl 8(%ebp), %eax movl %eax, (%esp) call _ZN1A3barEi addl $8, %esp popl %ebp ret _ZN1A4foo2Ev: pushl %ebp movl %esp, %ebp subl $8, %esp movl 8(%ebp), %eax movl %eax, (%esp) movl $8, 4(%esp) call _ZN1A3barEi addl $8, %esp popl %ebp ret _ZN1A4foo3Eii: pushl %ebp movl %esp, %ebp subl $8, %esp movl 16(%ebp), %eax addl 12(%ebp), %eax movl %eax, 4(%esp) movl 8(%ebp), %eax movl %eax, (%esp) call _ZN1A3barEi addl $8, %esp popl %ebp ret G++-4.3 generates this code: _ZN1A4foo3Eii: .LFB5: pushl %ebp .LCFI7: movl %esp, %ebp .LCFI8: movl 16(%ebp), %eax addl %eax, 12(%ebp) popl %ebp jmp _ZN1A3barEi _ZN1A4foo2Ev: .LFB4: pushl %ebp .LCFI4: movl %esp, %ebp .LCFI5: subl $8, %esp .LCFI6: movl 8(%ebp), %eax movl $8, 4(%esp) movl %eax, (%esp) call _ZN1A3barEi leave ret _ZN1A4foo3Eii: .LFB5: pushl %ebp .LCFI7: movl %esp, %ebp .LCFI8: movl 16(%ebp), %eax addl %eax, 12(%ebp) popl %ebp jmp _ZN1A3barEi With LLVM there are a lot of unnecessary stack moves. This doesn't occur on x86-64 because args are passed in registers. -- 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 4 07:35:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 07:35:43 -0500 Subject: [LLVMbugs] [Bug 4884] New: function attributes of template methods of template classes are ignored Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4884 Summary: function attributes of template methods of template classes are ignored Product: tools Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-g++ AssignedTo: unassignedbugs at nondot.org ReportedBy: torvald at se.inf.tu-dresden.de CC: llvmbugs at cs.uiuc.edu test case: template X bar(X x) __attribute__((noinline)); template X bar(X x) { return x + 1; } template class Bar { public: S s; Bar() : s(1) {} template X bar(X x) __attribute__((noinline)); }; template template X Bar::bar(X x) { return x + 1 + s; } int test() { int a = bar((int)1); Bar bar; int b = bar.bar((int)1); return a + b; } llvm-g++ -emit-llvm -o foo.bc foo.cpp -O3 -c yields: efine i32 @_Z4testv() nounwind { entry: %0 = tail call i32 @_Z3barIiET_S0_(i32 1) nounwind ; [#uses=1] %1 = add nsw i32 %0, 3 ; [#uses=1] ret i32 %1 } define linkonce_odr i32 @_Z3barIiET_S0_(i32 %x) nounwind readnone noinline { entry: %0 = add nsw i32 %x, 1 ; [#uses=1] ret i32 %0 } Bar::bar isn't marked as noinline. Likewise, there is no compiler warning when using non-existing function attributes. It does seem to work for the template function, though. -- 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 4 09:50:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 09:50:50 -0500 Subject: [LLVMbugs] [Bug 4885] New: Warnings from compiling llvm with icc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4885 Summary: Warnings from compiling llvm with icc Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: idadesub at users.sourceforge.net CC: llvmbugs at cs.uiuc.edu This bug is to track fixing the icc warnings. I'll be putting together a patch to fix these. Here are the current ones: ----------------------- /scratch/llvm/llvm/lib/Support/regcomp.c(259): warning #592: variable "prevback" is used before its value is set sopno prevback = prevback; ^ ----------------------- /scratch/llvm/llvm/lib/Target/ARM/ARMJITInfo.cpp(202): warning #69: integer conversion resulted in truncation JCE.emitWordLE((intptr_t)ARMCompilationCallback); ----------------------- http://software.intel.com/en-us/articles/cdiag111/ /scratch/llvm/llvm/lib/System/Mutex.cpp(119): remark #111: statement is unreachable return false; ^ Not sure what to do about this one, it looks like icc is inlining the "pthread_enabled", and then seeing that there's no possible way to reach the "return false" in: bool MutexImpl::release() { if (pthread_enabled) { pthread_mutex_t* mutex = static_cast(data_); assert(mutex != 0); int errorcode = pthread_mutex_unlock(mutex); return errorcode == 0; } return false; } ----------------------- /scratch/llvm/llvm/include/llvm/Support/Registry.h(80): warning #1170: invalid redeclaration of nested class class iterator; ^ This one's a simple fix, "class iterator" is listed twice. ----------------------- http://software.intel.com/en-us/articles/cdiag1418/ /scratch/llvm/llvm/include/llvm/Support/CommandLine.h(307): remark #1418: external function definition with no prior declaration LocationClass location(Ty &L) { return LocationClass(L); } this one is happening because the warning level is high enough to give stylistic suggestions. This one is saying that you should declare functions before implementing them, even if they're in a .h file. ----------------------- http://software.intel.com/en-us/articles/cdiag1419/ /scratch/llvm/llvm/lib/Support/Timer.cpp(26): remark #1419: external declaration in primary source file namespace llvm { extern raw_ostream *GetLibSupportInfoOutputFile(); } ^ ----------------------- http://software.intel.com/en-us/articles/cdiag1572/ /scratch/llvm/llvm/lib/Support/Timer.cpp(291): remark #1572: floating-point equality and inequality comparisons are unreliable if (Total.UserTime) ^ For Timer.cpp, I could replace these with "if (Total.UserTime > 0)", but that reads weird to me. There are a lot of these warnings though. ----------------------- http://software.intel.com/en-us/articles/cdiag177/ /scratch/llvm/llvm/lib/System/DynamicLibrary.cpp(33): remark #177: variable "Dummy" was declared but never referenced } Dummy; ^ This is an easy fix, since this is a c++ file and doesn't need the trailing name for the struct. http://software.intel.com/en-us/articles/cdiag177/ /scratch/llvm/llvm/lib/Support/APInt.cpp(1484): remark #177: variable "allOnes" was declared but never referenced APInt allOnes = APInt::getAllOnesValue(d.getBitWidth()); ^ /scratch/llvm/llvm/lib/Support/APInt.cpp(1486): remark #177: variable "signedMax" was declared but never referenced APInt signedMax = APInt::getSignedMaxValue(d.getBitWidth()); /scratch/llvm/llvm/utils/TableGen/FastISelEmitter.cpp(219): warning #177: function "::FastISelMap::PrintClass" was declared but never referenced void PrintClass(raw_ostream &OS); ^ /scratch/llvm/llvm/lib/VMCore/AsmWriter.cpp(1285): warning #177: function "::AssemblyWriter::getModule" was declared but never referenced const Module* getModule() { return TheModule; } ^ Also simple fixes. ----------------------- http://software.intel.com/en-us/articles/cdiag1599/ /scratch/llvm/llvm/lib/Transforms/IPO/PruneEH.cpp(111): remark #1599: declaration hides variable "E" (declared at line 100) for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I) We're getting a lot of these errors. ----------------------- http://software.intel.com/en-us/articles/cdiag1128/ /scratch/llvm/llvm/lib/System/DynamicLibrary.cpp(28): warning #1128: a storage class may not be specified here static struct ExplicitSymbolsDeleter { ^ Does this do anything? If not, I'll remove it. ----------------------- http://software.intel.com/en-us/articles/cdiag1128/ /scratch/llvm/llvm/lib/System/Unix/Signals.inc(229): remark #181: argument is incompatible with corresponding format string conversion fprintf(stderr, " + %tu",(char*)StackTrace[i]-(char*)dlinfo.dli_saddr); ----------------------- I'll post more in a bit. -- 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 4 10:45:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 10:45:06 -0500 Subject: [LLVMbugs] [Bug 4886] New: [2.6] Master 2.6 bug Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4886 Summary: [2.6] Master 2.6 bug Product: new-bugs Version: 2.6 Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu This bug is to track all regressions for 2.6. -- 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 4 11:55:25 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 11:55:25 -0500 Subject: [LLVMbugs] [Bug 4887] New: [2.6] test/Driver/hello.c fails Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4887 Summary: [2.6] test/Driver/hello.c fails Product: new-bugs Version: 2.6 Platform: Other OS/Version: OpenBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jsg at openbsd.org CC: llvmbugs at cs.uiuc.edu "/usr/users/jsg/src/llvm-2.6/Release/bin/clang-cc" -triple x86_64-unknown-openbsd4.6 -S -disable-free -main-file-name hello.c --relocation-model static --di sable-fp-elim --unwind-tables=1 --mcpu=x86-64 --fmath-errno=1 -fdiagnostics-show-option -o /tmp/cc-gRHCxp.s -x c /usr/users/jsg/src/llvm-2.6/tools/clang/test /Driver/hello.c "/usr/bin/as" -o /tmp/cc-ODqStn.o /tmp/cc-gRHCxp.s "/usr/bin/ld" -e __start --eh-frame-hdr -Bdynamic -dynamic-linker /usr/libexec/ld.so -o /usr/users/jsg/src/llvm-2.6/tools/clang/test/Output/Driver/hello.c.t mp /usr/lib/crt0.o /usr/lib/crtbegin.o -L/usr/lib/gcc-lib/x86_64-unknown-openbsd4.6/3.3.5 /tmp/cc-ODqStn.o -lgcc -lc -lgcc /usr/lib/crtend.o /usr/bin/ld: cannot find -lgcc clang: error: linker command failed with exit code 1 (use -v to see invocation) The path /usr/lib/gcc-lib/x86_64-unknown-openbsd4.6/3.3.5 is wrong and does not exist /usr/lib/gcc-lib/amd64-unknown-openbsd4.6/3.3.5 exists and should be used -- 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 4 12:06:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 12:06:05 -0500 Subject: [LLVMbugs] [Bug 4888] New: [2.6] test/CodeGen/pragma-weak.c fails Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4888 Summary: [2.6] test/CodeGen/pragma-weak.c fails Product: new-bugs Version: 2.6 Platform: Other OS/Version: OpenBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jsg at openbsd.org CC: llvmbugs at cs.uiuc.edu 2.6 pre-release 1, compiled with gcc 4.2.4 on OpenBSD/amd64. Exit Code: 1 Command Output (stdout): -- count: expected 1 lines and got 0. -- Command Output (stderr): -- /usr/users/jsg/src/llvm-2.6/tools/clang/test/CodeGen/pragma-weak.c:40:14: warning: weak identifier '__unused_alias' never declared #pragma weak unused_alias = __unused_alias // expected-warning {{weak identifier '__unused_alias' never declared}} ^ /usr/users/jsg/src/llvm-2.6/tools/clang/test/CodeGen/pragma-weak.c:39:14: warning: weak identifier 'unused' never declared #pragma weak unused // expected-warning {{weak identifier 'unused' never declared}} ^ /usr/users/jsg/src/llvm-2.6/tools/clang/test/CodeGen/pragma-weak.c:42:14: warning: weak identifier 'td' never declared #pragma weak td // expected-warning {{weak identifier 'td' never declared}} ^ /usr/users/jsg/src/llvm-2.6/tools/clang/test/CodeGen/pragma-weak.c:11:14: warning: weak identifier 'param' never declared #pragma weak param // expected-warning {{weak identifier 'param' never declared}} ^ /usr/users/jsg/src/llvm-2.6/tools/clang/test/CodeGen/pragma-weak.c:45:14: warning: weak identifier '__td2' never declared #pragma weak td2 = __td2 // expected-warning {{weak identifier '__td2' never declared}} ^ 5 diagnostics generated. -- 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 4 12:31:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 12:31:43 -0500 Subject: [LLVMbugs] [Bug 4889] New: loop-unswitch assertion fails: " Cannot change null node pointers!" Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4889 Summary: loop-unswitch assertion fails: "Cannot change null node pointers!" Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: bob.wilson at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3432) --> (http://llvm.org/bugs/attachment.cgi?id=3432) testcase There were 3 regressions in the armv6 nightly tester last night, and all of them appear to be caused by this same problem. I have verified that it still occurs in llvm trunk and reduced a testcase with bugpoint. (The bugpoint-reduced-simplified output exposed a different failure, so the testcase is the bugpoint-reduced-function version.) To reproduce: $ opt bugpoint-reduced-function.bc -loop-unswitch > out.bc Assertion failed: (N && NewIDom && "Cannot change null node pointers!"), function changeImmediateDominator, file /Users/bwilson/local/llvm/llvm/include/llvm/Analysis/Dominators.h, line 490. 0 opt 0x003f8ef4 PrintStackTrace(void*) + 45 1 opt 0x003f94fc SignalHandler(int) + 410 2 libSystem.B.dylib 0x98115bfb _sigtramp + 43 3 libSystem.B.dylib 0xffffffff _sigtramp + 1743692847 4 libSystem.B.dylib 0x981a3ba5 raise + 26 5 libSystem.B.dylib 0x981b9c5c abort + 93 6 libSystem.B.dylib 0x981a6804 __pthread_markcancel + 0 7 opt 0x00335369 llvm::DominatorTreeBase::changeImmediateDominator(llvm::DomTreeNodeBase*, llvm::DomTreeNodeBase*) + 85 8 opt 0x001d8a4f void llvm::DominatorTreeBase::Split >(llvm::DominatorTreeBase::NodeType>&, llvm::GraphTraits::NodeType*) + 897 9 opt 0x00335fc1 llvm::DominatorTreeBase::splitBlock(llvm::BasicBlock*) + 69 10 opt 0x001bc033 llvm::DominatorTree::splitBlock(llvm::BasicBlock*) + 27 11 opt 0x001bac2b llvm::SplitBlockPredecessors(llvm::BasicBlock*, llvm::BasicBlock* const*, unsigned int, char const*, llvm::Pass*) + 544 12 opt 0x0015d6c7 (anonymous namespace)::LoopUnswitch::SplitExitEdges(llvm::Loop*, llvm::SmallVector const&) + 229 13 opt 0x0015df30 (anonymous namespace)::LoopUnswitch::UnswitchNontrivialCondition(llvm::Value*, llvm::Constant*, llvm::Loop*) + 692 14 opt 0x0015eec8 (anonymous namespace)::LoopUnswitch::UnswitchIfProfitable(llvm::Value*, llvm::Constant*) + 538 15 opt 0x0015f0dd (anonymous namespace)::LoopUnswitch::processCurrentLoop() + 229 16 opt 0x0015f3e4 (anonymous namespace)::LoopUnswitch::runOnLoop(llvm::Loop*, llvm::LPPassManager&) + 206 17 opt 0x002584b2 llvm::LPPassManager::runOnFunction(llvm::Function&) + 1026 18 opt 0x003891a7 llvm::FPPassManager::runOnFunction(llvm::Function&) + 297 19 opt 0x00389356 llvm::FPPassManager::runOnModule(llvm::Module&) + 84 20 opt 0x00387390 llvm::MPPassManager::runOnModule(llvm::Module&) + 378 21 opt 0x003895da llvm::PassManagerImpl::run(llvm::Module&) + 112 22 opt 0x00389631 llvm::PassManager::run(llvm::Module&) + 27 23 opt 0x0005cdb6 main + 4337 24 opt 0x00050ff1 start + 53 -- 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 4 15:11:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 15:11:13 -0500 Subject: [LLVMbugs] [Bug 4890] New: Assert when generating debug info for C++ destructor Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4890 Summary: Assert when generating debug info for C++ destructor Product: clang Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: dgregor at apple.com CC: llvmbugs at cs.uiuc.edu Given dtor.cpp: struct X { ~X(); }; X::~X() { } clang -c dtor.cpp -g produces: Assertion failed: (!isa(D) && "Use mangleCXXDtor for destructor decls!"), function mangleName, file /Volumes/Data/dgregor/Projects/llvm/tools/clang/lib/CodeGen/Mangle.cpp, line 847. Stack dump: 0. Program arguments: /Volumes/Data/dgregor/Projects/llvm-build-xcode/bin/Debug/clang-cc -triple i386-apple-darwin10 -S -disable-free -main-file-name dtor.cpp --relocation-model pic -pic-level=1 --disable-fp-elim --unwind-tables=0 --mcpu=yonah --fmath-errno=0 -mmacosx-version-min=10.6.0 -g -fdiagnostics-show-option -o /var/folders/4o/4oX0yyc9FZ0tdikPS5odWE+++TI/-Tmp-/cc-lyMxzB.s -x c++ dtor.cpp 1. parser at end of file 2. dtor.cpp:5:4: LLVM IR generation of declaration 'X::~X' clang: error: compiler command failed due to signal 6 (use -v to see invocation) -- 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 4 15:29:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 15:29:50 -0500 Subject: [LLVMbugs] [Bug 4891] New: Backend producing incorrect code for aliased buffers Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4891 Summary: Backend producing incorrect code for aliased buffers Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: benedict.gaster at amd.com CC: llvmbugs at cs.uiuc.edu When compiling the following program it is producing incorrect x86 code. //foo.c #include #include size_t global_id[4]; size_t get_index(unsigned int dimidx) { return global_id[dimidx]; } typedef short short2 __attribute__ ((vector_size (4))); union _X { short2 s; int i; }; typedef union _X X; inline short2 as_short2(int x) { X result; result.i = x; return result.s; } inline int as_int(short2 x) { X result; result.s = x; return result.i; } void short2_int_swap( short2* b, int* c) { int gidx = get_index(0); short2 bval = b[gidx]; int cval = c[gidx]; b[gidx] = as_short2(cval); c[gidx] = as_int(bval); } Compiling with TOT or 2.6 with: llvm-gcc -O3 -S foo.c Gives the assembler for short2_int_swap: .align 16 .globl short2_int_swap .type short2_int_swap, at function short2_int_swap: .Leh_func_begin4: movslq global_id(%rip), %rax movl (%rsi,%rax,4), %ecx movl %ecx, (%rdi,%rax,4) movl (%rdi,%rax,4), %ecx movl %ecx, (%rsi,%rax,4) ret .size short2_int_swap, .-short2_int_swap .Leh_func_end4: It looks like an edge is bring droped and thus leading to incorrect code. Compiling the same code with gcc 4.3.3: gcc -O3 -S foo.c gives globl short2_int_swap .type short2_int_swap, @function short2_int_swap: .LFB9: movslq global_id(%rip),%rax salq $2, %rax addq %rax, %rdi addq %rax, %rsi movl (%rdi), %edx movl (%rsi), %eax movl %edx, (%rsi) movl %eax, (%rdi) ret -- 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 4 16:32:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 16:32:23 -0500 Subject: [LLVMbugs] [Bug 4877] globaldce : Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/ Analysis/DebugInfo.cpp, line 258. In-Reply-To: Message-ID: <200909042132.n84LWNPp007189@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4877 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from devang.patel 2009-09-04 16:32:23 --- Fixed in rev. 81045. -- 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 4 17:04:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 17:04:03 -0500 Subject: [LLVMbugs] [Bug 4892] New: templates and friends cause false positive complaints Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4892 Summary: templates and friends cause false positive complaints Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jreiser at BitWagon.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3433) --> (http://llvm.org/bugs/attachment.cgi?id=3433) gzipped, pre-processed C++ source code I get many false positive complaints (hundreds per file) that appear to be caused by templates and friend classes in C++. This obscures any potential benefits of using the tools. The attached file is compressed with gzip after being generated by "g++ -E p_mach.cpp" with the correct -I options, etc. The original source is available from https://www.pysol.org:4443/hg/upx.hg . The incorrect complaints start with: ----- $ /usr/local/llvm/libexec/clang-cc -fsyntax-only foo.cxx p_mach.cpp:85:5: error: use of undeclared identifier 'bele' bele = N_BELE_CTP::getRTP((const BeLePolicy*) __null); ^ p_mach.cpp:168:5: error: no matching function for call to 'addLoader' addLoader("MACOS000", __null); ^~~~~~~~~ p_mach.cpp:171:13: error: use of undeclared identifier 'ph' ( ((ph.method) >= 8 && (ph.method) <= 10) ? "NRV_HEAD,NRV2E,NRV_TAIL" ^ p_mach.cpp:176:9: error: no matching function for call to 'hasLoaderSection' if (hasLoaderSection("CFLUSH")) ^~~~~~~~~~~~~~~~ p_mach.cpp:177:9: error: no matching function for call to 'addLoader' addLoader("CFLUSH"); ^~~~~~~~~ p_mach.cpp:178:5: error: no matching function for call to 'addLoader' addLoader("ELFMAINY,IDENTSTR,+40,ELFMAINZ,FOLDEXEC", __null); ^~~~~~~~~ p_mach.cpp:290:5: error: no matching function for call to 'initLoader' initLoader(proto, szproto); ^~~~~~~~~~ p_mach.cpp:296:34: error: use of undeclared identifier 'ph' h.b_method = (unsigned char) ph.method; ^ p_mach.cpp:297:32: error: use of undeclared identifier 'ph' h.b_ftid = (unsigned char) ph.filter; ^ p_mach.cpp:298:32: error: use of undeclared identifier 'ph' h.b_cto8 = (unsigned char) ph.filter_cto; ^ p_mach.cpp:306:17: error: use of undeclared identifier 'ph' __null, ph.method, 10, __null, __null ); ^ p_mach.cpp:314:5: error: use of undeclared identifier 'linker' linker->addSection("FOLDEXEC", cprLoader, sizeof(h) + h.sz_cpr, 0); ^ p_mach.cpp:324:5: error: no matching function for call to 'relocateLoader' relocateLoader(); ^~~~~~~~~~~~~~ p_mach.cpp:384:32: error: no matching function for call to 'getLoader' unsigned char *const ptr = getLoader(); ^~~~~~~~~ p_mach.cpp:389:36: error: use of undeclared identifier 'lsize' lp->l_lsize = (unsigned short) lsize; ^ ----- -- 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 4 19:34:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 19:34:48 -0500 Subject: [LLVMbugs] [Bug 4879] strange failure with VLAs and debug info In-Reply-To: Message-ID: <200909050034.n850YmDC030941@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4879 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |devang.patel at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from devang.patel 2009-09-04 19:34:48 --- Fixed in rev 81058. -- 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 4 20:09:25 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 20:09:25 -0500 Subject: [LLVMbugs] [Bug 4845] [2.6] crashing inside JITed code when running on x86_64 In-Reply-To: Message-ID: <200909050109.n8519PlP002994@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4845 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #17 from Daniel Dunbar 2009-09-04 20:09:25 --- This should be fixed 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 Fri Sep 4 20:35:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 20:35:57 -0500 Subject: [LLVMbugs] [Bug 4877] globaldce : Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/ Analysis/DebugInfo.cpp, line 258. In-Reply-To: Message-ID: <200909050135.n851Zv0S006782@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4877 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #8 from Daniel Dunbar 2009-09-04 20:35:57 --- The self host build on the public buildbot is still failing because of the same or a similar problem. Current build setup is: llvm (debug) -> llvm-gcc (using system gcc) -> llvm (release, using first llvm-gcc) -> llvm-gcc (using first llvm-gcc) and one test is failing in the second llvm build step, and second llvm-gcc buid is failing. Attaching failing .bc input, which crashes with opt -std-compile-opts on that builder (using either the first or second llvm opt binary). -- 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 4 20:58:49 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 20:58:49 -0500 Subject: [LLVMbugs] [Bug 4892] templates and friends cause false positive complaints In-Reply-To: Message-ID: <200909050158.n851wnhs009823@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4892 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |LATER --- Comment #1 from Daniel Dunbar 2009-09-04 20:58:48 --- Clang does not yet have officially supported C++ support. Bugs for small isolated features -- generally when the feature already seems to largely work -- are still useful, but we are not yet to the point of accepting bugs for general code compiled with 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 Fri Sep 4 21:31:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 4 Sep 2009 21:31:48 -0500 Subject: [LLVMbugs] [Bug 4894] New: Crash while emitting debug information for complex Objective-C classes Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4894 Summary: Crash while emitting debug information for complex Objective-C classes Product: new-bugs Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: csdavec at swan.ac.uk CC: llvmbugs at cs.uiuc.edu At some point during compilation of the attached preprocessed source file, the stack-allocated FwdDecl instance's DbgNode pointer becomes dangling, causing a crash when it is dereferenced, with the following stack trace: #0 0x0818f3d5 in llvm::Type::getForwardedType (this=0x5a5a5a5a) at Type.h:302 #1 0x0818f66f in llvm::PATypeHolder::get (this=0x29cdeb48) at Type.h:435 #2 0x0818f6bd in llvm::PATypeHolder::operator llvm::Type* (this=0x29cdeb48) at AbstractTypeUser.h:151 #3 0x0818f6f0 in llvm::Value::getType (this=0x29cdeb40) at Value.h:103 #4 0x0915df8d in llvm::Value::replaceAllUsesWith (this=0x29cdeb40, New=0x29ceabc0) at Value.cpp:326 #5 0x0903c81e in llvm::DIDerivedType::replaceAllUsesWith (this=0xbfbfbb80, D=@0xbfbfbb78) at DebugInfo.cpp:270 #6 0x081beee0 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29b9c1d0, Unit=@0xbfbfbd04) at CGDebugInfo.cpp:642 #7 0x081bcf61 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 700039632}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfbe20) at CGDebugInfo.cpp:783 #8 0x081c16a6 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29b9c228, Unit=@0xbfbfbeb4) at CGDebugInfo.cpp:205 #9 0x081bced4 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 700039720}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfc108) at CGDebugInfo.cpp:781 #10 0x081be9c3 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29a953e0, Unit=@0xbfbfc224) at CGDebugInfo.cpp:580 #11 0x081bcf61 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 698962912}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfc340) at CGDebugInfo.cpp:783 #12 0x081c16a6 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29a95438, Unit=@0xbfbfc3d4) at CGDebugInfo.cpp:205 #13 0x081bced4 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 698963000}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfc628) at CGDebugInfo.cpp:781 #14 0x081be9c3 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29a95008, Unit=@0xbfbfc744) at CGDebugInfo.cpp:580 #15 0x081bcf61 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 698961928}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfc960) at CGDebugInfo.cpp:783 #16 0x081be79c in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29a84328, Unit=@0xbfbfcaa4) at CGDebugInfo.cpp:566 #17 0x081bcf61 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 698893096}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfcbc0) at CGDebugInfo.cpp:783 #18 0x081c16a6 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29a84380, Unit=@0xbfbfcc54) at CGDebugInfo.cpp:205 #19 0x081bced4 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 698893184}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfcea8) at CGDebugInfo.cpp:781 #20 0x081be9c3 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29cd04a8, Unit=@0xbfbfcfc4) at CGDebugInfo.cpp:580 #21 0x081bcf61 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 701301928}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfd0e0) at CGDebugInfo.cpp:783 #22 0x081c16a6 in clang::CodeGen::CGDebugInfo::CreateType (this=0x29730780, Ty=0x29cd04e8, Unit=@0xbfbfd174) at CGDebugInfo.cpp:205 #23 0x081bced4 in clang::CodeGen::CGDebugInfo::getOrCreateType (this=0x29730780, Ty={Value = {Value = 701301992}, static MaxAddressSpace = 16777215}, Unit=@0xbfbfd2a4) at CGDebugInfo.cpp:781 #24 0x081bdc16 in clang::CodeGen::CGDebugInfo::EmitDeclare (this=0x29730780, Decl=0x29cd0540, Tag=257, Storage=0x29cd818c, Builder=@0xbfbfd694) at CGDebugInfo.cpp:904 #25 0x081bddea in clang::CodeGen::CGDebugInfo::EmitDeclareOfArgVariable (this=0x29730780, Decl=0x29cd0540, AI=0x29cd818c, Builder=@0xbfbfd694) at CGDebugInfo.cpp:934 #26 0x081c766d in clang::CodeGen::CodeGenFunction::EmitParmDecl (this=0xbfbfd610, D=@0x29cd0540, Arg=0x29cd91e0) at CGDecl.cpp:521 #27 0x081b7a4c in clang::CodeGen::CodeGenFunction::EmitFunctionProlog (this=0xbfbfd610, FI=@0x29c148f0, Fn=0x29cda1a0, Args=@0xbfbfd4e0) at CGCall.cpp:580 #28 0x08237823 in clang::CodeGen::CodeGenFunction::StartFunction (this=0xbfbfd610, D=0x29cd0460, RetTy={Value = {Value = 695829456}, static MaxAddressSpace = 16777215}, Fn=0x29cda1a0, Args=@0xbfbfd4e0, StartLoc={ID = 823520}) at CodeGenFunction.cpp:189 #29 0x081f7cf4 in clang::CodeGen::CodeGenFunction::StartObjCMethod (this=0xbfbfd610, OMD=0x29cd0460, CD=0x29ccc150) at CGObjC.cpp:122 #30 0x081f8ef5 in clang::CodeGen::CodeGenFunction::GenerateObjCMethod (this=0xbfbfd610, OMD=0x29cd0460) at CGObjC.cpp:131 #31 0x08181516 in clang::CodeGen::CodeGenModule::EmitTopLevelDecl (this=0x2970dc00, D=0x29cd0460) at CodeGenModule.cpp:1699 #32 0x0817b240 in HandleTopLevelDecl (this=0x297066c0, DG={D = 0x29cd0460}) at ModuleBuilder.cpp:67 #33 0x080973f0 in HandleTopLevelDecl (this=0x29786100, D={D = 0x29cd0460}) at Backend.cpp:130 #34 0x08336353 in clang::ParseAST (PP=@0x297581c0, Consumer=0x29786100, Ctx=@0x2978c400, PrintStats=false, CompleteTranslationUnit=true) at ParseAST.cpp:70 #35 0x0806fd8a in ProcessInputFile (PP=@0x297581c0, PPF=@0xbfbfe414, InFile=@0x297070e0, PA=EmitAssembly, Features=@0xbfbfe440, Context=@0x297070ac) at clang-cc.cpp:2084 #36 0x08071af0 in main (argc=20, argv=0xbfbfe540) at clang-cc.cpp:2355 This occurs while emitting the type information for the self implicit argument in the first method of the class. The class itself has a number of instance variables which refer to other classes, so the problem may be related different classes at different depths in this structure referring to each other, so that the debug node forward declaration is already replaced by something else by the time this is reached. -- 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 5 00:14:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 00:14:47 -0500 Subject: [LLVMbugs] [Bug 4895] New: Missed phi/phi/select folding opportunity Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4895 Summary: Missed phi/phi/select folding opportunity Product: new-bugs Version: unspecified Platform: Other OS/Version: other Status: NEW Severity: minor Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: me22.ca at gmail.com CC: llvmbugs at cs.uiuc.edu I compiled the following C++ code with bitter melon on 2009-09-05T05:00Z with LTO enabled: #include int main() { std::set s; if (s.find(0) != s.end()) return *s.find(0); } This resulted in this basic block (among others): _ZNSt8_Rb_treeIiiSt9_IdentityIiESt4lessIiESaIiEED1Ev.exit: ; preds = %invcont2, %invcont1 %.0 = phi i32 [ %29, %invcont2 ], [ undef, %invcont1 ] ; [#uses=1] %finally_tmp.74.0 = phi i1 [ true, %invcont2 ], [ false, %invcont1 ] ; [#uses=1] %.1 = select i1 %finally_tmp.74.0, i32 %.0, i32 0 ; [#uses=1] ret i32 %.1 Which, which names shortened, looks like this: %.0 = phi i32 [%29, %A], [undef, %B] %t = phi i1 [true, %A], [false, %B] %.1 = select i1 %t, i32 %.0, i32 0 ret i32 %.1 Selecting on the result of a phi[true][false] seems strange to me, as AFAIK it could have simply been written like this: %.0 = phi i32 [%29, %A], [0, %B] ret i32 %.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 5 01:11:20 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 01:11:20 -0500 Subject: [LLVMbugs] [Bug 4897] New: [2.6] clang doesn' t search CWD for include files when reading from stdin Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4897 Summary: [2.6] clang doesn't search CWD for include files when reading from stdin Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Perl's t/comp/cpp fails, when using clang: 1..3 ok 1 ok 2 :2:10: fatal error: 'Comp_cpp.inc' file not found #include "Comp_cpp.inc" ^ 1 diagnostic generated. not ok 3 This is because the following command gives an error: $ clang -E -C -I/usr/local/lib/perl5/5.10.0 - :2:10: fatal error: 'Comp_cpp.inc' file not found #include "Comp_cpp.inc" ^ # 1 "" # 1 "" 1 # 1 "" 1 # 1 "" 3 # 105 "" 3 # 105 "" 1 # 1 "" 2 $ok = "not ok 3\n"; print $ok; 1 diagnostic generated. Running this with llvm-gcc or gcc works: clang -E -C -I/usr/local/lib/perl5/5.10.0 - :2:10: fatal error: 'Comp_cpp.inc' file not found #include "Comp_cpp.inc" ^ # 1 "" # 1 "" 1 # 1 "" 1 # 1 "" 3 # 105 "" 3 # 105 "" 1 # 1 "" 2 $ok = "not ok 3\n"; print $ok; 1 diagnostic generated. Comp_cpp.inc is in the current directory, and contains this: #define OK "ok 3\n" This is not due to -I and -C, even clang -E - http://llvm.org/bugs/show_bug.cgi?id=4898 Summary: llvm-gcc: fails to link ARCH=um Linux kernel undefined reference to `__bad_size' Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu $ make CC=llvm-gcc HOSTCC=llvm-gcc ARCH=um defconfig $ make CC=llvm-gcc HOSTCC=llvm-gcc ARCH=um mm/built-in.o: In function `kmem_cache_init': /home/edwin/builds/linux-2.6/mm/slab.c:343: undefined reference to `__bad_size' /home/edwin/builds/linux-2.6/mm/slab.c:343: undefined reference to `__bad_size' /home/edwin/builds/linux-2.6/mm/slab.c:343: undefined reference to `__bad_size' /home/edwin/builds/linux-2.6/mm/slab.c:1461: undefined reference to `__bad_size' /home/edwin/builds/linux-2.6/mm/slab.c:1461: undefined reference to `__bad_size' mm/built-in.o:/home/edwin/builds/linux-2.6/mm/slab.c:1517: more undefined references to `__bad_size' follow collect2: ld returned 1 exit status This is usually due to not inlining something the kernel expected to inline, thus __builtin_constant_p() returning false. Will try to attach a 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 Sat Sep 5 02:10:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 02:10:11 -0500 Subject: [LLVMbugs] [Bug 4899] New: clang: __builtin_longjmp not supported Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4899 Summary: clang: __builtin_longjmp not supported Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu valgrind 3.4.1 doesn't compile: m_signals.c:1435:7: error: cannot compile this builtin function yet __builtin_longjmp(tst->sched_jmpbuf, True); ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -- 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 5 03:40:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 03:40:11 -0500 Subject: [LLVMbugs] [Bug 4900] New: clang: blender 2.49b renders wrong image Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4900 Summary: clang: blender 2.49b renders wrong image Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3449) --> (http://llvm.org/bugs/attachment.cgi?id=3449) compare-blender.sh I built blender using gcc, llvm-gcc and clang (in objgcc, obj, objclang dirs respectively), doing something like: export CC=llvm-gcc CXX=llvm-g++ cmake ../blender-2.49b CC=llvm-gcc CXX=llvm-g++ make -j8 Then I wrote a script to render .blend files with 2 blenders and compare outputs (compare-blender.sh attached). Running it on the blender 2.49 regression files from blender.org I got some failures: Render of ./teapot_soft.blend is not exact Image Difference (MeanSquaredError): Normalized Absolute ============ ========== Red: 0.0000081281 0.0 Green: 0.0000084478 0.0 Blue: 0.0000151673 0.0 Total: 0.0000105811 0.0 Checking whether reference render is deterministic ...Yes When the script finds a different render, it reruns the render with the reference blender again, and checks whether the output is deterministic (for some renders it is not). For teapot_soft.blend the render with reference blender is deterministic, but differs from the clang compiled blender. When building it with llvm-gcc everything is OK for this particular render. Attached are also the 2 images, and the output of: gm compare -highlight-style Assign -file diff.miff file_ref0001.png file_new0001.png -- 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 5 03:46:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 03:46:23 -0500 Subject: [LLVMbugs] [Bug 4901] New: llvm-ld: Segmentation fault running pass ' Dead Global Elimination' Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4901 Summary: llvm-ld: Segmentation fault running pass 'Dead Global Elimination' Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: sanxiyn at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3453) --> (http://llvm.org/bugs/attachment.cgi?id=3453) TRE agrep build script llvm-ld segfaults while running the attached build script. -- 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 5 06:27:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 06:27:45 -0500 Subject: [LLVMbugs] [Bug 3801] non- PIC-aware assembler code in X86 jit In-Reply-To: Message-ID: <200909051127.n85BRjdg030484@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3801 Corrado Zoccolo changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #15 from Corrado Zoccolo 2009-09-05 06:27:44 --- (In reply to comment #9) > What is the status of this patch? Is in TOT? > > If this is safe for 2.6, I'd like to know ASAP. > It was confirmed from many sources that the patch works correctly for both 32 and 64 bit linux, and fixes the issue. I think the problem reported by Anton was just due an unclean build, so please include it in 2.6 -- 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 5 11:12:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 11:12:03 -0500 Subject: [LLVMbugs] [Bug 4902] New: clang tries to free a global variable Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4902 Summary: clang tries to free a global variable Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: baldrick at free.fr CC: llvmbugs at cs.uiuc.edu Valgrind detected the following problem in several clang/test/CodeGenObjC tests, for example objc2-assign-global.m. It is trying to free something that wasn't malloced, the data symbol _ZNSs4_Rep20_S_empty_rep_storageE. ==750== Invalid free() / delete / delete[] ==750== at 0x4C24AEA: operator delete(void*) (vg_replace_malloc.c:346) ==750== by 0x41C52F: __gnu_cxx::new_allocator::deallocate(char*, unsigned long) (new_allocator.h:95) ==750== by 0x418484: std::string::_Rep::_M_destroy(std::allocator const&) (basic_string.tcc:427) ==750== by 0x414011: std::string::_Rep::_M_dispose(std::allocator const&) (basic_string.h:231) ==750== by 0x40CE37: std::basic_string, std::allocator >::~basic_string() (basic_string.h:498) ==750== by 0x5EF5EA: (anonymous namespace)::CGObjCMac::FinishModule() (CGObjCMac.cpp:3512) ==750== by 0x5E9EBD: (anonymous namespace)::CGObjCMac::ModuleInitFunction() (CGObjCMac.cpp:2340) ==750== by 0x54E7F8: clang::CodeGen::CodeGenModule::Release() (CodeGenModule.cpp:70) ==750== by 0x54D66E: (anonymous namespace)::CodeGeneratorImpl::HandleTranslationUnit(clang::ASTContext&) (ModuleBuilder.cpp:85) ==750== by 0x4361FB: (anonymous namespace)::BackendConsumer::HandleTranslationUnit(clang::ASTContext&) (Backend.cpp:142) ==750== by 0x732285: clang::ParseAST(clang::Preprocessor&, clang::ASTConsumer*, clang::ASTContext&, bool, bool) (ParseAST.cpp:79) ==750== by 0x4052CB: ProcessInputFile(clang::Preprocessor&, clang::PreprocessorFactory&, std::string const&, ProgActions, llvm::StringMap const&, llvm::LLVMContext&) (clang-cc.cpp:2084) ==750== Address 0x532b160 is 0 bytes inside data symbol "_ZNSs4_Rep20_S_empty_rep_storageE" -- 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 5 15:52:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 15:52:27 -0500 Subject: [LLVMbugs] [Bug 3801] non- PIC-aware assembler code in X86 jit In-Reply-To: Message-ID: <200909052052.n85KqREF003571@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3801 Corrado Zoccolo changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED 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 Sat Sep 5 16:05:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 16:05:59 -0500 Subject: [LLVMbugs] [Bug 4810] [2.6] ExecutionEngine test failures on mixed 64/32 system In-Reply-To: Message-ID: <200909052105.n85L5xeJ005391@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4810 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Nick Lewycky 2009-09-05 16:05:58 --- Humm, this problem has gone away on HEAD. -- 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 5 16:35:21 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 16:35:21 -0500 Subject: [LLVMbugs] [Bug 4906] New: printf(3C) format argument not analized correctly Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4906 Summary: printf(3C) format argument not analized correctly Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: bkorb at gnu.org CC: llvmbugs at cs.uiuc.edu This fails: printf("%1$d %1$d %2$d\n", 1, 2); -- 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 5 18:41:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 18:41:27 -0500 Subject: [LLVMbugs] [Bug 4907] New: llvm fails to build, error: 'WeakVH' was not declared Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4907 Summary: llvm fails to build, error: 'WeakVH' was not declared Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: blocker Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: moritzwarning at web.de CC: llvmbugs at cs.uiuc.edu LLVM fails to build on Debian linux (x86_64). (FreeBSD 7.2 32&64bit as well). release_26 branch. llvm[3]: Compiling BasicBlockPlacement.cpp for Release build llvm[3]: Compiling CodeGenPrepare.cpp for Release build llvm[3]: Compiling CondPropagate.cpp for Release build llvm[3]: Compiling ConstantProp.cpp for Release build llvm[3]: Compiling DCE.cpp for Release build llvm[3]: Compiling DeadStoreElimination.cpp for Release build /home/mwarning/llvm-trunk/lib/Transforms/Scalar/DeadStoreElimination.cpp: In member function ???bool::DSE::runOnBasicBlock(llvm::BasicBlock&)???: /home/mwarning/llvm-trunk/lib/Transforms/Scalar/DeadStoreElimination.cpp:144: error: ???WeakVH??? was not declared in this scope /home/mwarning/llvm-trunk/lib/Transforms/Scalar/DeadStoreElimination.cpp:144: error: expected `;' before ???NextInst??? /home/mwarning/llvm-trunk/lib/Transforms/Scalar/DeadStoreElimination.cpp:148: error: ???NextInst??? was not declared in this scope make[3]: *** [/home/mwarning/llvm-trunk-obj/lib/Transforms/Scalar/Release/DeadStoreElimination.o] Error 1 make[3]: Leaving directory `/home/mwarning/llvm-trunk-obj/lib/Transforms/Scalar' make[2]: *** [Scalar/.makeall] Error 2 make[2]: Leaving directory `/home/mwarning/llvm-trunk-obj/lib/Transforms' make[1]: *** [Transforms/.makeall] Error 2 make[1]: Leaving directory `/home/mwarning/llvm-trunk-obj/lib' make: *** [all] Error 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 Sat Sep 5 21:18:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 5 Sep 2009 21:18:40 -0500 Subject: [LLVMbugs] [Bug 4908] New: infinite loop in instcombine Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4908 Summary: infinite loop in instcombine Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu Instcombine loops on: -- ; ModuleID = '' target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" target triple = "x86_64-apple-darwin10.0" define void @short2_int_swap(<1 x i16>* nocapture %b, i32* nocapture %c) nounwind ssp { entry: %arrayidx = getelementptr inbounds <1 x i16>* %b, i64 undef ; <<1 x i16>*> [#uses=1] %tmp2 = load <1 x i16>* %arrayidx ; <<1 x i16>> [#uses=1] %tmp6 = bitcast <1 x i16> %tmp2 to i16 ; [#uses=1] %tmp7 = zext i16 %tmp6 to i32 ; [#uses=1] %ins = or i32 0, %tmp7 ; [#uses=1] %arrayidx20 = getelementptr inbounds i32* %c, i64 undef ; [#uses=1] store i32 %ins, i32* %arrayidx20 ret void } -- -- 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 6 02:41:28 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 02:41:28 -0500 Subject: [LLVMbugs] [Bug 4907] llvm fails to build, error: 'WeakVH' was not declared In-Reply-To: Message-ID: <200909060741.n867fSY9023307@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4907 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Duncan Sands 2009-09-06 02:41:27 --- Please test the 2.6 candidates released by Tanya (http://llvm.org/prereleases/2.6/), and not the current state of the release_26 repository. -- 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 6 14:56:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 14:56:51 -0500 Subject: [LLVMbugs] [Bug 4909] New: Crash in opt -partial-inliner Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4909 Summary: Crash in opt -partial-inliner Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: d.g.gorbachev at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3458) --> (http://llvm.org/bugs/attachment.cgi?id=3458) Reduced BC file opt -disable-output -partial-inliner bugpoint-reduced-simplified.bc 0x080fa657 in llvm::cast (Val=@0x88978c4) at /tmp/llvm/include/llvm/Support/Casting.h:199 199 assert(isa(Val) && "cast() argument of incompatible type!"); (gdb) bt #0 0x080fa657 in llvm::cast (Val=@0x88978c4) at /tmp/llvm/include/llvm/Support/Casting.h:199 #1 0x0812ee8d in runOnModule (this=0x88903f0, M=@0x88906d0) at /tmp/llvm/lib/Transforms/IPO/PartialInlining.cpp:75 #2 0x084028aa in llvm::MPPassManager::runOnModule (this=0x88952c0, M=@0x88906d0) at /tmp/llvm/lib/VMCore/PassManager.cpp:1447 #3 0x08402c87 in llvm::PassManagerImpl::run (this=0x8891178, M=@0x88906d0) at /tmp/llvm/lib/VMCore/PassManager.cpp:1528 #4 0x08402d37 in llvm::PassManager::run (this=0xbfd1939c, M=@0x88906d0) at /tmp/llvm/lib/VMCore/PassManager.cpp:1557 #5 0x080f052e in main (argc=Cannot access memory at address 0x9) at /tmp/llvm/tools/opt/opt.cpp:512 (gdb) print Val $1 = (class llvm::Value * const&) @0x9fe68c4: 0x0 LLVM rev. 81123, compiled with GCC 4.3.3 with -O2. -- 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 6 15:22:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 15:22:29 -0500 Subject: [LLVMbugs] [Bug 3801] non- PIC-aware assembler code in X86 jit In-Reply-To: Message-ID: <200909062022.n86KMTWh022292@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3801 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #19 from Anton Korobeynikov 2009-09-06 15:22:29 --- (In reply to comment #18) > (In reply to comment #17) > > (In reply to comment #15) > > > It was confirmed from many sources that the patch works correctly for both 32 > > > and 64 bit linux, and fixes the issue. I think the problem reported by Anton > > > was just due an unclean build, so please include it in 2.6 > > No, it is not. I think you haven't tried to build libLTO.so, have you? > > > > The patch works correctly on 32-bit Linux. Right, the problem is only on 64-bit systems, because 32 bit pcrel relocation is generated which is not suitable for PIC. I've committed different version of this patch, which seems to work: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090831/086505.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 Sun Sep 6 15:33:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 15:33:39 -0500 Subject: [LLVMbugs] [Bug 4910] New: fatal error: error in backend: Ran out of registers during register allocation! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4910 Summary: fatal error: error in backend: Ran out of registers during register allocation! Product: libraries Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu when compiling attached test case on i386 with clang/llvm (r81131) I am getting: witten /tmp# clang -c -O2 mp_machdep.c ... warnings ... fatal error: error in backend: Ran out of registers during register allocation! this might be related to bug 4733 -- 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 6 16:44:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 16:44:17 -0500 Subject: [LLVMbugs] [Bug 4911] New: Tcl detection should look for tcl.h, not tclsh Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4911 Summary: Tcl detection should look for tcl.h, not tclsh Product: Build scripts Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: autoconf AssignedTo: unassignedbugs at nondot.org ReportedBy: salimma at fedoraproject.org CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3461) --> (http://llvm.org/bugs/attachment.cgi?id=3461) Look for tcl.h, not tclsh LLVM currently validates the with_tclinclude setting by checking if the tclsh binary exists. This binary is, however, installed in different places on some Linux distributions: Fedora puts it in /usr/bin . To enable dejagnu unit tests on Fedora, the configure script has to be slightly modified to search for tcl.h instead -- the tests seem to run fine. The patch attached is for 2.5, but it applies cleanly to 2.6 prerelease-1. If it could be applied before the 2.6 release, we could avoid having to carry the patch throughout the 2.6 release cycle. Thanks, -- Michel / Fedora Project -- 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 6 20:23:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 20:23:14 -0500 Subject: [LLVMbugs] [Bug 4912] New: "opt -insert-rs-profiling-framework" seems stuck in an infinite loop Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4912 Summary: "opt -insert-rs-profiling-framework" seems stuck in an infinite loop Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: d.g.gorbachev at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3462) --> (http://llvm.org/bugs/attachment.cgi?id=3462) BC file opt -insert-rs-profiling-framework -disable-output xxx.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 Sun Sep 6 23:27:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 6 Sep 2009 23:27:51 -0500 Subject: [LLVMbugs] [Bug 4913] New: unittests should build with build and test with make check Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4913 Summary: unittests should build with build and test with make check Product: Test Suite Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu The unittests currently are not well integrated with the makefiles and its too easy to forget to keep them up to date. The unittests should be build during a standard build, so that code changes force the unittests to be updated. The unittests should be run during 'make check', so that unittest failures are detect like normal test failures. -- 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 7 04:05:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 04:05:36 -0500 Subject: [LLVMbugs] [Bug 4902] clang tries to free a global variable In-Reply-To: Message-ID: <200909070905.n8795aw1003420@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4902 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #7 from Duncan Sands 2009-09-07 04:05:35 --- This now seems to be working, but the original issue may not have been fixed (Daniel thinks some changes he made may have masked the underlying problem). -- 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 7 14:00:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:00:50 -0500 Subject: [LLVMbugs] [Bug 4914] New: llvm-gcc doesn't do range analysis on induction variable Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4914 Summary: llvm-gcc doesn't do range analysis on induction variable Product: libraries Version: trunk Platform: PC URL: http://codepad.org/n6dGCqje OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: rjmccall at apple.com CC: llvmbugs at cs.uiuc.edu This is a forwarded complaint by "bearophile" on LLVM IRC, who unfortunately never files bugs. I am willing to act as a stand-in in this case. The code in the URL produces worse code on (an unspecified version of) llvm-gcc than (an unspecified version of) gcc does; assembly output from both is inline in the test. The difference seems to involve three things: (1) gcc knows that i is never < 0, and so elides the first check when computing sat(i). (2) gcc knows that -i is always <= 0, and so infers that sat(-i) is always 0. (3) gcc hoists a constant load outside of the loop. -- 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 7 14:07:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:07:09 -0500 Subject: [LLVMbugs] [Bug 4915] New: error on postdomfrontier for a function without exit block Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4915 Summary: error on postdomfrontier for a function without exit block Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Global Analyses AssignedTo: unassignedbugs at nondot.org ReportedBy: kuba at gcc.gnu.org CC: llvmbugs at cs.uiuc.edu opt pdf_crash.bc -postdomfrontier -debug -disable-output causes error because PDT.getRootNode() returns NULL for a function without exit block. -- 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 7 14:19:54 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:19:54 -0500 Subject: [LLVMbugs] [Bug 4916] New: Optimize add and unsigned compare to add-with-overflow Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4916 Summary: Optimize add and unsigned compare to add-with-overflow Product: libraries Version: trunk Platform: PC URL: http://codepad.org/BqJOSpv6 OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rjmccall at apple.com CC: llvmbugs at cs.uiuc.edu This is a forwarded complaint by "bearophile" on LLVM IRC, who unfortunately never files bugs. The URL contains the original test case. llvm-gcc: gcc version 4.2.1 (Based on Apple Inc. build 5636) (LLVM build) The following should be optimized to an add-with-overflow: %c = add iN %a, $b %d = icmp ult iN %c, %a -- 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 7 14:28:25 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:28:25 -0500 Subject: [LLVMbugs] [Bug 4917] New: Optimize mul and unsigned compare to mul-with-overflow Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4917 Summary: Optimize mul and unsigned compare to mul-with-overflow Product: libraries Version: trunk Platform: PC URL: http://codepad.org/qIPeNgVm OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rjmccall at apple.com CC: llvmbugs at cs.uiuc.edu This is a forwarded complaint by "bearophile" on LLVM IRC, who unfortunately never files bugs. The URL contains an unreduced test case. The following should be optimized to a mul-with-overflow: %L_ext = zext i32 %L to i64 %R_ext = zext i32 %R to i64 %mul64 = mul i64 %L_ext, %R_ext # no uses except the following: %overflow = icmp ugt i64 %mul64, (i64 0xffffffff) br i1 %overflow, label %elsewhere, label %cont ... %mul32 = mul i32 %L, %R -- 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 7 14:52:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:52:00 -0500 Subject: [LLVMbugs] [Bug 4918] New: Optimize mul and equality test to mul-with-overflow Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4918 Summary: Optimize mul and equality test to mul-with-overflow Product: libraries Version: trunk Platform: PC URL: http://codepad.org/ugwQ37sh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rjmccall at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3468) --> (http://llvm.org/bugs/attachment.cgi?id=3468) Pastebin This is a forwarded complaint by "bearophile" on LLVM IRC, who unfortunately never files bugs. The following should be optimized to a mul-with-overflow: %L_ext = zext i32 %L to i64 %R_ext = zext i32 %R to i64 %mul64 = mul i64 %L_ext, %R_ext # no uses except the following: %trunc = trunc i64 %mul64 to i32 %zext = zext i32 %trunc to i64 %overflow = icmp eq i64 %mul64, %zext This pattern should create either a smul-with-overflow or a umul-with-overflow depending on whether a sext or a zext is used. -- 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 7 14:58:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 14:58:40 -0500 Subject: [LLVMbugs] [Bug 4919] New: llvm-gcc doesn't optimize simple factorial into a loop Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4919 Summary: llvm-gcc doesn't optimize simple factorial into a loop Product: libraries Version: trunk Platform: PC URL: http://codepad.org/2K5gNj3e OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Interprocedural Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rjmccall at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3469) --> (http://llvm.org/bugs/attachment.cgi?id=3469) Original pastebin This is a forwarded complaint by "bearophile" on LLVM IRC, who unfortunately never files bugs. The following naive factorial function does not get optimized into a loop, even at -O3. long fact(long x) { if (x <= 0) return 1; else return x * fact(x - 1); } gcc-4.3.3 apparently can do 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 Mon Sep 7 15:31:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 15:31:29 -0500 Subject: [LLVMbugs] [Bug 4810] [2.6] ExecutionEngine test failures on mixed 64/32 system In-Reply-To: Message-ID: <200909072031.n87KVTSC027415@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4810 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #5 from Tanya Lattner 2009-09-07 15:31:29 --- Can someone see if this repo's with the 2.6 branch? I'm reopening this because it occurred with pre-release1 and we need to make sure those patches get merged in. Please make sure to note which patches in the bug so I can track everything. -- 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 7 16:00:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 16:00:48 -0500 Subject: [LLVMbugs] [Bug 4920] New: Document undef semantics Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4920 Summary: Document undef semantics Product: Documentation Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: General docs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu LangRef should document undef better, some thoughts from a discussion on IRC: x = undef a = xor x, x What is the value of x? x = undef i1 a = icmp x, 4 i2 b = icmp x,5 Both a and b can be true here select undef,a,b -> either a or b br undef, L1, L2 -> choose either br L1 or br L2, not unreachable undef is a value in the range of the type, but it can change its value during its lifetime. Does using an 'undef' value destroy SSA properties of the program when you RAUW a value with undef? (previously it was a single value, and optimizers could rely on this, after the replace it can become different values at different points). -- 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 7 17:53:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 17:53:03 -0500 Subject: [LLVMbugs] [Bug 4920] Document undef semantics In-Reply-To: Message-ID: <200909072253.n87Mr3BT013229@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4920 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-07 17:53:02 --- implemented in r81167 -- 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 7 18:02:46 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 18:02:46 -0500 Subject: [LLVMbugs] [Bug 4921] New: No warning for control reaching end of non-void function Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4921 Summary: No warning for control reaching end of non-void function Product: clang Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: self at mattmower.com CC: llvmbugs at cs.uiuc.edu I'm using the version of Clang/LLVM that comes with Mac OSX 10.6.0. When compiling the following Objective-C code: - (int)foo { } I notice that neither the compiler, nor the analyzer, notice that the code does not return. Compiling with GCC I get the expected "control reaches end of non-void function". I'm compiling with -Wall so I'd expect to get something. -- 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 7 18:52:30 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 18:52:30 -0500 Subject: [LLVMbugs] [Bug 4921] No warning for control reaching end of non-void function In-Reply-To: Message-ID: <200909072352.n87NqU08020726@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4921 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Chris Lattner 2009-09-07 18:52:29 --- This is fixed on mainline, 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 7 19:27:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 19:27:34 -0500 Subject: [LLVMbugs] [Bug 4882] bus error in MemCpyOpt In-Reply-To: Message-ID: <200909080027.n880RY8B025303@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4882 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Chris Lattner 2009-09-07 19:27:33 --- Fixed by Jakub here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086551.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 7 20:23:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 20:23:15 -0500 Subject: [LLVMbugs] [Bug 4915] error on postdomfrontier for a function without exit block In-Reply-To: Message-ID: <200909080123.n881NFcP032606@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4915 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-07 20:23:14 --- Fixed, thanks: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086552.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 7 20:29:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 20:29:22 -0500 Subject: [LLVMbugs] [Bug 4918] Optimize mul and equality test to mul-with-overflow In-Reply-To: Message-ID: <200909080129.n881TM7D000957@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4918 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Chris Lattner 2009-09-07 20:29:22 --- This is very related to 4917. Although not a strict dupe, it seems reasonable that anyone tackling 4917 would fix this too. *** This bug has been marked as a duplicate of bug 4917 *** -- 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 7 22:45:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 22:45:16 -0500 Subject: [LLVMbugs] [Bug 4908] infinite loop in instcombine In-Reply-To: Message-ID: <200909080345.n883jGZP017908@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4908 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-07 22:45:14 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086562.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 7 23:15:38 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 23:15:38 -0500 Subject: [LLVMbugs] [Bug 4922] New: Clang's Sema:: ActOnEndOfTranslationUnit completes tentative decls in nondeterminstic order Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4922 Summary: Clang's Sema::ActOnEndOfTranslationUnit completes tentative decls in nondeterminstic order Product: clang Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: clattner at apple.com CC: llvmbugs at cs.uiuc.edu While investigating PR4888, I discovered that Sema::ActOnEndOfTranslationUnit is completing tentative definitions in DenseMap iteration order, leading to nondeterminstic generated code. This is badness. One possible fix is to have Sema put tentative definitions in a vector as well as the densemap, and then emit in vector order. -- 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 7 23:56:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 7 Sep 2009 23:56:03 -0500 Subject: [LLVMbugs] [Bug 4767] Cannot print this instruction. In-Reply-To: Message-ID: <200909080456.n884u33h026999@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4767 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #8 from Chris Lattner 2009-09-07 23:56:03 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086567.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 8 00:18:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 00:18:51 -0500 Subject: [LLVMbugs] [Bug 4871] Assertion failed: (Bits != 0 && "Cannot print this instruction." ), function printInstruction, file /data/home/rdivacky/llvm/lib/Target/X86/ AsmPrinter/../X86GenAsmWriter.inc, line 2251. In-Reply-To: Message-ID: <200909080518.n885IplZ029930@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4871 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #2 from Chris Lattner 2009-09-08 00:18:50 --- *** This bug has been marked as a duplicate of bug 4767 *** -- 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 8 01:23:52 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 01:23:52 -0500 Subject: [LLVMbugs] [Bug 3484] test-suite should support building only C/ObjC programs In-Reply-To: Message-ID: <200909080623.n886Nqx3005793@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3484 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Daniel Dunbar 2009-09-08 01:23:52 --- I done did this. 'make DISABLE_CXX=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 Tue Sep 8 01:39:30 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 01:39:30 -0500 Subject: [LLVMbugs] [Bug 4689] Regalloc problems due to subregs? In-Reply-To: Message-ID: <200909080639.n886dUnj007903@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4689 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Evan Cheng 2009-09-08 01:39:29 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086586.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 8 03:26:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 03:26:22 -0500 Subject: [LLVMbugs] [Bug 4924] New: Assertion failed: (I->Codes.size() == 1 && " Too many asm constraint codes to handle") Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4924 Summary: Assertion failed: (I->Codes.size() == 1 && "Too many asm constraint codes to handle") Product: libraries Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: Backend: C AssignedTo: unassignedbugs at nondot.org ReportedBy: pawel.worach at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3471) --> (http://llvm.org/bugs/attachment.cgi?id=3471) bugpoint-reduced-simplified.bc While bugpointing bug 4910 llc -march=c dies on the code attached. # llc -march=c bugpoint-reduced-simplified.bc Assertion failed: (I->Codes.size() == 1 && "Too many asm constraint codes to handle"), function visitInlineAsm, file CBackend.cpp, line 3365. Stack dump: 0. Program arguments: llc -march=c bugpoint-reduced-simplified.bc 1. Running pass 'Function Pass Manager' on module 'bugpoint-reduced-simplified.bc'. 2. Running pass 'C backend' on function '@cpustop_handler' Abort (core dumped) -- 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 8 03:33:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 03:33:27 -0500 Subject: [LLVMbugs] [Bug 4924] Assertion failed: (I->Codes.size() == 1 && " Too many asm constraint codes to handle") In-Reply-To: Message-ID: <200909080833.n888XR5o003146@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4924 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Anton Korobeynikov 2009-09-08 03:33:27 --- *** 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 Tue Sep 8 08:14:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 08:14:45 -0500 Subject: [LLVMbugs] [Bug 4925] New: llc crash when a const struct includes a zero-length member Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4925 Summary: llc crash when a const struct includes a zero-length member Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Keywords: compile-fail Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: kennethuil at gmail.com CC: llvmbugs at cs.uiuc.edu When I run llc on the (assembled version of) the following code on Linux x86: %testType = type {i32, [0 x i32], i32} define i32 @main() { %1 = alloca %testType store %testType {i32 1, [0 x i32] zeroinitializer, i32 2}, %testType* %1 ret i32 0 } llc crashes with a segmentation fault. It happens in the 2.5 release version and in the version I pulled from svn trunk about two days ago. The trouble starts in SelectionDAGLowering::getValue(const Value* V), within the if (isa(C) || isa(C)) check, where the zero-length member (which can be either a zero-length array or an empty struct) causes getValue(*OI).getNode() to return NULL. The segfault occurs when this null pointer is used as the instance pointer for SDNode::getNumValues(). It doesn't matter whether the zero-length member appears at the beginning, the end, or somewhere in the middle. A simple null check on the result of getValue(*OI).getNode() clears up the segfault and makes my larger test program run properly after llc'ing and assembling (not surprising, since leaving out the zero-length member leaves the result of the store unchanged). -- 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 8 10:35:01 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 10:35:01 -0500 Subject: [LLVMbugs] [Bug 4925] llc crash when a const struct includes a zero-length member In-Reply-To: Message-ID: <200909081535.n88FZ1kk026366@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4925 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Chris Lattner 2009-09-08 10:35:00 --- This was recently fixed on mainline. -- 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 8 11:15:10 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 11:15:10 -0500 Subject: [LLVMbugs] [Bug 4926] New: Assertion in rewriteARMFrameIndex() Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4926 Summary: Assertion in rewriteARMFrameIndex() Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3472) --> (http://llvm.org/bugs/attachment.cgi?id=3472) Testcase Consider the attached bytecode. It causes the following assertion: llc: /home/asl/proj/llvm/src_arm/lib/Target/ARM/ARMBaseInstrInfo.cpp:999: bool llvm::rewriteARMFrameIndex(llvm::MachineInstr&, unsigned int, unsigned int, int&, const llvm::ARMBaseInstrInfo&): Assertion `(Offset & (Scale-1)) == 0 && "Can't encode this offset!"' 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 Tue Sep 8 13:20:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 13:20:09 -0500 Subject: [LLVMbugs] [Bug 4922] Clang's Sema:: ActOnEndOfTranslationUnit completes tentative decls in nondeterminstic order In-Reply-To: Message-ID: <200909081820.n88IK9jo016564@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4922 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Chris Lattner 2009-09-08 13:20:08 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090907/021105.html Sema still has other problems though. -- 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 8 13:40:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 13:40:45 -0500 Subject: [LLVMbugs] [Bug 4889] loop-unswitch assertion fails: " Cannot change null node pointers!" In-Reply-To: Message-ID: <200909081840.n88IejWB019390@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4889 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Dan Gohman 2009-09-08 13:40:45 --- Both problems are now fixed in llvm tot r81221. -- 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 8 13:44:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 13:44:17 -0500 Subject: [LLVMbugs] [Bug 4888] [2.6] test/CodeGen/pragma-weak.c fails In-Reply-To: Message-ID: <200909081844.n88IiHe8019844@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4888 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-08 13:44:17 --- Hopefully the combination of the fix for 4922 and this patch: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090907/021107.html will fix this. If not, please reopen. -- 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 8 14:54:02 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 14:54:02 -0500 Subject: [LLVMbugs] [Bug 4909] Crash in opt -partial-inliner In-Reply-To: Message-ID: <200909081954.n88Js2Ek029367@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4909 Owen Anderson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Owen Anderson 2009-09-08 14:54:02 --- Approved and committed in r81250. Thanks Jakub! -- 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 8 14:55:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 14:55:29 -0500 Subject: [LLVMbugs] [Bug 4927] New: assertion "no allocatable register" breaks Apple-style llvm-gcc build Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4927 Summary: assertion "no allocatable register" breaks Apple-style llvm-gcc build Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: bob.wilson at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3473) --> (http://llvm.org/bugs/attachment.cgi?id=3473) testcase Apple-style builds of llvm-gcc are broken on llvm trunk if you include ARM support. I'm getting an assertion failure when using the x86 llvm-gcc to build the arm llvm-gcc. It seems to happen only with fast isel. To reproduce, compile the attached testcase with "llc -O0": Assertion failed: (I != E && "No allocatable register in this register class!"), function getFreePhysReg, file /Volumes/LocalHD/bwilson/llvm/build/llvmCore.roots/llvmCore~obj/src/lib/CodeGen/RegAllocLinearScan.cpp, line 1362. 0 llc 0x00900fa4 PrintStackTrace(void*) + 45 1 llc 0x009015ac SignalHandler(int) + 410 2 libSystem.B.dylib 0x98115bfb _sigtramp + 43 3 libSystem.B.dylib 0xffffffff _sigtramp + 1743692847 4 libSystem.B.dylib 0x981a3ba5 raise + 26 5 libSystem.B.dylib 0x981b9c5c abort + 93 6 libSystem.B.dylib 0x981a6804 __pthread_markcancel + 0 7 llc 0x0067d725 (anonymous namespace)::RALinScan::getFreePhysReg(llvm::LiveInterval*, llvm::TargetRegisterClass const*, unsigned int, llvm::SmallVector&, bool) + 335 8 llc 0x0067e80d (anonymous namespace)::RALinScan::getFreePhysReg(llvm::LiveInterval*) + 1073 9 llc 0x00680860 (anonymous namespace)::RALinScan::assignRegOrStackSlotAtInterval(llvm::LiveInterval*) + 1928 10 llc 0x006834b1 (anonymous namespace)::RALinScan::linearScan() + 667 11 llc 0x00683d99 (anonymous namespace)::RALinScan::runOnMachineFunction(llvm::MachineFunction&) + 523 12 llc 0x00642737 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 77 13 llc 0x0087f8d7 llvm::FPPassManager::runOnFunction(llvm::Function&) + 297 14 llc 0x0087fe0a llvm::FunctionPassManagerImpl::run(llvm::Function&) + 112 15 llc 0x0087ffd6 llvm::FunctionPassManager::run(llvm::Function&) + 130 16 llc 0x0002b134 main + 3395 17 llc 0x00029c85 start + 53 -- 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 8 16:49:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 16:49:35 -0500 Subject: [LLVMbugs] [Bug 4929] New: [2.6] MultiSource/Applications/lemon JIT regression Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4929 Summary: [2.6] MultiSource/Applications/lemon JIT regression Product: new-bugs Version: 2.6 Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: critical Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu Blocks: 4886 This is 2.6 pre-release1. This is a regression from 2.5. Will verify with current 2.6 release branch. If someone could verify with TOT, that would be great. -- 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 8 19:11:52 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 19:11:52 -0500 Subject: [LLVMbugs] [Bug 4865] [2.6] JIT dwarf exception handling seems broken In-Reply-To: Message-ID: <200909090011.n890Bq2K002630@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4865 Bill Wendling changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #15 from Bill Wendling 2009-09-08 19:11:52 --- Fixed: http://llvm.org/viewvc/llvm-project?rev=81295&view=rev Laurent, Please let me know if this fixes your MacRuby problems. -- 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 8 20:19:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 20:19:44 -0500 Subject: [LLVMbugs] [Bug 4930] New: Simple code breaks with llc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4930 Summary: Simple code breaks with llc Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: wendling at apple.com CC: llvmbugs at cs.uiuc.edu $ cat t.ll ; ModuleID = 'bugpoint-reduced-simplified.bc' define i32 @main() ssp { entry: invoke void @_ZL3foov() ssp to label %bb6 unwind label %lpad bb6: ; preds = %entry ret i32 undef lpad: ; preds = %entry %eh_ptr = call i8* @llvm.eh.exception() ; [#uses=5] %eh_select9 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; [#uses=0] unreachable } declare void @_ZL3foov() ssp declare i8* @llvm.eh.exception() declare i32 @llvm.eh.selector.i32(i8*, i8*, ...) nounwind declare i32 @__gxx_personality_v0(...) $ llvm-as < t.ll | /Developer/usr/local/bin/llc -o - -march=x86-64 Assertion failed: (BestRC && "Couldn't find the register class"), function getPhysicalRegisterRegClass, file /Volumes/Sandbox/llvm/Roots/llvmCore-2204.1.roots/llvmCore-2204.1~obj/src/lib/Target/TargetRegisterInfo.cpp, line 59. 0 llc 0x0000000100812732 std::vector >::_M_insert_aux(__gnu_cxx::__normal_iterator > >, llvm::sys::Path const&) + 7042 1 llc 0x0000000100812dc6 std::vector >::_M_insert_aux(__gnu_cxx::__normal_iterator > >, llvm::sys::Path const&) + 8726 2 libSystem.B.dylib 0x00007fff8289014a _sigtramp + 26 3 libSystem.B.dylib 0x00007fff82834caa tiny_malloc_from_free_list + 1196 4 libSystem.B.dylib 0x00007fff8290c104 __pthread_markcancel + 0 5 llc 0x0000000100716cae void std::__introsort_loop<__gnu_cxx::__normal_iterator*, std::vector, std::allocator > > >, long>(__gnu_cxx::__normal_iterator*, std::vector, std::allocator > > >, __gnu_cxx::__normal_iterator*, std::vector, std::allocator > > >, long) + 1966 6 llc 0x000000010041f5fd std::vector >::reserve(unsigned long) + 7069 7 llc 0x0000000100421674 std::vector >::reserve(unsigned long) + 15380 8 llc 0x000000010042237a std::vector >::reserve(unsigned long) + 18714 9 llc 0x00000001004c02ea llvm::FunctionLoweringInfo::InitializeRegForValue(llvm::Value const*) + 21466 10 llc 0x00000001004c25d7 llvm::FunctionLoweringInfo::InitializeRegForValue(llvm::Value const*) + 30407 11 llc 0x00000001004c34d9 llvm::FunctionLoweringInfo::InitializeRegForValue(llvm::Value const*) + 34249 12 llc 0x00000001004c4727 llvm::FunctionLoweringInfo::InitializeRegForValue(llvm::Value const*) + 38935 13 llc 0x00000001007a1c82 llvm::FunctionPass::~FunctionPass() + 36690 14 llc 0x00000001007a2aa0 llvm::FunctionPass::~FunctionPass() + 40304 15 llc 0x00000001007a2d77 llvm::FunctionPass::~FunctionPass() + 41031 16 llc 0x0000000100017868 std::_Rb_tree, std::_Select1st >, std::less, std::allocator > >::_M_insert_unique(std::pair const&) + 18072 17 llc 0x00000001000160f8 std::_Rb_tree, std::_Select1st >, std::less, std::allocator > >::_M_insert_unique(std::pair const&) + 12072 Stack dump: 0. Program arguments: /Developer/usr/local/bin/llc -o - -march=x86-64 1. Running pass 'X86 DAG->DAG Instruction Selection' on function '@main' Abort trap [bitters:llvm] /Developer/usr/local/bin/llc --version Low Level Virtual Machine (http://llvm.org/): llvm version 2.6svn Apple Build #2204-01 Optimized build with assertions. Built Aug 21 2009(15:46:27). -- 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 8 20:24:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 8 Sep 2009 20:24:12 -0500 Subject: [LLVMbugs] [Bug 4931] New: [meta] Improve inlining effectiveness Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4931 Summary: [meta] Improve inlining effectiveness 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 Blocks: 2353,2973 Limitations in inlining in LLVM (from llvmdev discussion]: - Inlining doesn't take into account whether a load from a constant pointer is constant. - Inlining is purely bottom-up: we inline large functions into small functions when it would be better to inline the small function. - The cost metrics for constant-propagating into a branch or switch are extremely imprecise. - Cost metrics can easily end up being negative (i.e. we always inline) in cases where inlining will lead to arbitrarily large increases in code size. This might be okay at -O3, but it isn't for -Os. - We only use one inlining pass normally; IIRC, the gcc devs found that an additional early inlining pass is beneficial. Partial inlining also probably falls into here somewhere. Filing per request from 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 Wed Sep 9 02:40:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 02:40:41 -0500 Subject: [LLVMbugs] [Bug 4932] New: clang: doesn't honor -fno-asynchronous-unwind-tables Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4932 Summary: clang: doesn't honor -fno-asynchronous-unwind-tables 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: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Blocks: 4068 llvm-gcc and gcc do not emit .eh_frame with that flag, clang does. Testcase: $ echo 'int main(){}'|clang -fno-asynchronous-unwind-tables -x c - -S -o -|grep eh_frame clang: warning: argument unused during compilation: '-fno-asynchronous-unwind-tables' .section .eh_frame,"aw", at progbits .Lsection_eh_frame: .Leh_frame_common: .long .Leh_frame_common_end-.Leh_frame_common_begin .Leh_frame_common_begin: .Leh_frame_common_end: .long .Leh_frame_end1-.Leh_frame_begin1 .Leh_frame_begin1: .long . -- 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 9 03:47:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 03:47:43 -0500 Subject: [LLVMbugs] [Bug 4933] New: clang/llvm-gcc/ llc wrong section flags output nobits vs progbits Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4933 Summary: clang/llvm-gcc/llc wrong section flags output nobits vs progbits 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: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu $ llvm-gcc -Wno-unused-value -S ioremap.i -S -o -|grep bss .bss .bss .bss .section .bss.page_aligned,"aw", at progbits .bss $ clang -Wno-unused-value -S ioremap.i -S -o -|grep bss section .bss.page_aligned,"aw", at progbits .bss .bss .bss $ gcc -Wno-unused-value -S ioremap.i -S -o -|grep bss .section .bss.page_aligned,"aw", at nobits So it outputs the wrong flags for bss.page_aligned, and also puts some symbols into .bss, while gcc doesn't put any symbols there. -- 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 9 04:08:01 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 04:08:01 -0500 Subject: [LLVMbugs] [Bug 4933] clang/llvm-gcc/llc wrong section flags output nobits vs progbits In-Reply-To: Message-ID: <200909090908.n89981NZ019726@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4933 T??r??k Edwin changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from T??r??k Edwin 2009-09-09 04:08:01 --- This works now, thanks for the quick fix. -- 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 9 05:03:42 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 05:03:42 -0500 Subject: [LLVMbugs] [Bug 4934] New: -loop-unswitch: cLoopUnswitch.cpp:204: virtual bool::LoopUnswitch::runOnLoop(llvm::Loop*, llvm::LPPassManager&): Assertion `currentLoop->isLCSSAForm()' failed. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4934 Summary: -loop-unswitch: cLoopUnswitch.cpp:204: virtual bool::LoopUnswitch::runOnLoop(llvm::Loop*, llvm::LPPassManager&): Assertion `currentLoop- >isLCSSAForm()' failed. Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: gohman at apple.com, llvmbugs at cs.uiuc.edu Created an attachment (id=3481) --> (http://llvm.org/bugs/attachment.cgi?id=3481) bugpoint-reduced-simplified.bc With ToT I get this: $ opt: /home/edwin/llvm-git/llvm/lib/Transforms/Scalar/LoopUnswitch.cpp:204: virtual bool::LoopUnswitch::runOnLoop(llvm::Loop*, llvm::LPPassManager&): Assertion `currentLoop->isLCSSAForm()' failed. 0 opt 0x000000000081281f 1 opt 0x00000000008144e9 2 libpthread.so.0 0x000000318de0e720 3 libc.so.6 0x000000318d231d25 gsignal + 53 4 libc.so.6 0x000000318d234de1 abort + 385 5 libc.so.6 0x000000318d22af99 __assert_fail + 233 6 opt 0x00000000005a95e7 7 opt 0x000000000068fd70 llvm::LPPassManager::runOnFunction(llvm::Function&) + 1136 8 opt 0x00000000007a371b llvm::FPPassManager::runOnFunction(llvm::Function&) + 571 9 opt 0x00000000007a384b llvm::FPPassManager::runOnModule(llvm::Module&) + 75 10 opt 0x00000000007a1bfc llvm::MPPassManager::runOnModule(llvm::Module&) + 460 11 opt 0x00000000007a3f00 llvm::PassManagerImpl::run(llvm::Module&) + 160 12 opt 0x00000000004b9363 main + 1171 13 libc.so.6 0x000000318d21e5c6 __libc_start_main + 230 14 opt 0x00000000004ae5f9 Stack dump: 0. Running pass 'Function Pass Manager' on module 'bugpoint-reduced-simplified.bc'. 1. Running pass 'Loop Pass Manager' on function '@pnp_check_irq' 2. Running pass 'Unswitch loops' on basic block '%while.cond.i' 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 Wed Sep 9 05:29:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 05:29:04 -0500 Subject: [LLVMbugs] [Bug 4592] ARM JIT 1+1=0 In-Reply-To: Message-ID: <200909091029.n89AT4D3032707@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4592 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Xerxes R??nby 2009-09-09 05:29:03 --- This bug got fixed during Evan Chengs ARM JIT bughunt after rev 81308, 81310, 81311! Thanks! xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ ./Kaleidoscope ready> 0; ready> Evaluated to 0.000000 ready> 1; ready> Evaluated to 1.000000 ready> 2; ready> Evaluated to 2.000000 ready> 3; ready> Evaluated to 3.000000 ready> 4; ready> Evaluated to 4.000000 ready> 5; ready> Evaluated to 5.000000 ready> 6; ready> Evaluated to 6.000000 ready> def test(x) 1+x; ready> Read function definition: define double @test(double %x) { entry: %addtmp = fadd double %x, 1.000000e+00 ; [#uses=1] ret double %addtmp } ready> test(2); ready> Evaluated to 3.000000 ready> -- 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 9 05:40:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 05:40:16 -0500 Subject: [LLVMbugs] [Bug 4601] assertion failure: Don't have operand info for this instruction! In-Reply-To: Message-ID: <200909091040.n89AeGjI002117@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4601 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |xerxes at zafena.se Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Xerxes R??nby 2009-09-09 05:40:15 --- looks OK when testing using svnr81322 xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ ./llc -O3 -relocation-model=pi c -mcpu=arm1136jf-s -f /home/xerxes/4601.bc xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ cat /home/xerxes/4601.s .section __TEXT,__text,regular,pure_instructions .globl _cli_ac_addsig .align 2 _cli_ac_addsig: cmp r0, #0 bne LBB1_14 @ %bb126 LBB1_1: @ %bb1 cmp r0, #0 beq LBB1_6 @ %cli_calloc.exit.thread LBB1_2: @ %cli_calloc.exit cmp r0, #0 bne LBB1_14 @ %bb126 LBB1_3: @ %bb4 cmp r0, #0 beq LBB1_6 @ %cli_calloc.exit.thread LBB1_4: @ %bb1.i3 cmp r0, #0 LBB1_5: @ %cli_strdup.exit cmp r0, #0 bne LBB1_13 @ %bb45 LBB1_6: @ %cli_calloc.exit.thread mvn r0, #113 bx lr LBB1_7: @ %bb9 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop cmp r0, #0 cmpeq r0, #0 bne LBB1_14 @ %bb126 LBB1_8: @ %bb13 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop cmp r0, #0 LBB1_9: @ %cli_calloc.exit70 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop cmp r0, #0 LBB1_10: @ %bb1.i84 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop cmp r0, #0 LBB1_11: @ %bb17 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop cmp r0, #0 bne LBB1_14 @ %bb126 LBB1_12: @ %bb24 @ Loop Depth 1 @ Loop Header is BB1_13 @ Inner Loop mov r0, #10 ldrh r0, [r0] add r0, r0, #1 mov r1, #10 strh r0, [r1] uxth r0, r0 add r0, r0, r0, lsl #1 sub r0, r0, #1 cmp r0, r0 cmpne r0, #0 LBB1_13: @ %bb45 @ Loop Depth 1 @ Loop Header @ Inner Loop cmp r0, #0 bne LBB1_7 @ %bb9 LBB1_14: @ %bb126 mvn r0, #116 bx lr .subsections_via_symbols xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ -- 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 9 05:45:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 05:45:17 -0500 Subject: [LLVMbugs] [Bug 4821] Vector splat not selected In-Reply-To: Message-ID: <200909091045.n89AjHGN002980@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4821 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |xerxes at zafena.se Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Xerxes R??nby 2009-09-09 05:45:17 --- testcase looks OK using llvm r81322 xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ ./llc -mattr=+neon /home/xerxe s/4821.bc xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ cat /home/xerxes/4821.s .file "/home/xerxes/4821.bc" .syntax unified .eabi_attribute 10, 2 .eabi_attribute 20, 1 .eabi_attribute 21, 1 .eabi_attribute 23, 3 .eabi_attribute 24, 1 .eabi_attribute 25, 1 .text .globl aaa .align 2 .code 16 .thumb_func aaa: vmov.i8 q0, #-1 vstmia r0, {d0,d1} .size aaa, .-aaa -- 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 9 06:04:30 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 06:04:30 -0500 Subject: [LLVMbugs] [Bug 4593] ARM JIT BrainF -jit hello.bf Asserts while deleting putchar In-Reply-To: Message-ID: <200909091104.n89B4UeC006669@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4593 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #2 from Xerxes R??nby 2009-09-09 06:04:29 --- *** This bug has been marked as a duplicate of bug 4406 *** -- 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 9 06:10:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 06:10:12 -0500 Subject: [LLVMbugs] [Bug 4926] Assertion in rewriteARMFrameIndex() In-Reply-To: Message-ID: <200909091110.n89BACxC007841@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4926 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |xerxes at zafena.se Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Xerxes R??nby 2009-09-09 06:10:12 --- Looks OK using rev 81322 from TOT xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ ./llc /home/xerxes/4926.bc xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ cat /home/xerxes/4926.s .file "/home/xerxes/4926.bc" .eabi_attribute 20, 1 .eabi_attribute 21, 1 .eabi_attribute 23, 3 .eabi_attribute 24, 1 .eabi_attribute 25, 1 .text .globl test_vget_lanep16 .align 2 test_vget_lanep16: sub sp, sp, #10 ldrh r0, [sp, #+4] strh r0, [sp] .LBB1_1: @ %return add sp, sp, #10 bx lr .size test_vget_lanep16, .-test_vget_lanep16 xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ -- 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 9 06:13:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 06:13:12 -0500 Subject: [LLVMbugs] [Bug 4566] Invalid asmprinting for Thumb RSB instruction [ generic asmprinter construction problem?] In-Reply-To: Message-ID: <200909091113.n89BDCYW008382@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4566 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |xerxes at zafena.se Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #7 from Xerxes R??nby 2009-09-09 06:13:11 --- testcase looks OK using rev 81322 from TOT xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ ./llc -march=pic16 /home/xerxe s/4566.bc xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ cat /home/xerxes/4566.s #include P16F1937.INC ;Function Declarations - BEGIN. global @main global @main.ret. global @main.args. ;Function Declarations - END. ;Exported Variables - BEGIN global @i global @j global @pc ;Exported Variables - END @idata.0.# IDATA @i dl 4294967286 ; 0xFFFFFFF6 @j dl 4294967276 ; 0xFFFFFFEC @pc dw 160 @main.frame_section.# UDATA_OVR @main.frame.: @main.ret.: @main.args. RES 0 @main.temp. RES 6 @main.autos_section.# UDATA_OVR @main.auto.k RES 4 @main.code_section.# CODE retlw low(@main.frame.) retlw high(@main.frame.) @main: banksel @i movf @i + 3, W banksel @main.temp. movwf @main.temp. + 0 banksel @i movf @i + 2, W banksel @main.temp. movwf @main.temp. + 1 banksel @i movf @i + 0, W banksel @main.temp. movwf @main.temp. + 2 banksel @i movf @i + 1, W banksel @main.temp. movwf @main.temp. + 3 movf @main.temp. + 2, W banksel @j addwf @j + 0, W banksel @main.temp. movwf @main.temp. + 2 movf @main.temp. + 3, W banksel @j addwfc @j + 1, W banksel @main.temp. movwf @main.temp. + 3 movf @main.temp. + 1, W banksel @j addwfc @j + 2, W banksel @main.temp. movwf @main.temp. + 1 movf @main.temp. + 0, W banksel @j addwfc @j + 3, W banksel @main.auto.k movwf @main.auto.k + 3 banksel @main.temp. movf @main.temp. + 1, W banksel @main.auto.k movwf @main.auto.k + 2 banksel @main.temp. movf @main.temp. + 3, W banksel @main.auto.k movwf @main.auto.k + 1 banksel @main.temp. movf @main.temp. + 2, W banksel @main.auto.k movwf @main.auto.k + 0 banksel @pc movf @pc + 0, W banksel @main.temp. movwf @main.temp. + 4 banksel @pc movf @pc + 1, W banksel @main.temp. movwf @main.temp. + 5 movf @main.temp. + 4, W movwf FSR0L movf @main.temp. + 5, W movwf FSR0H movf @main.temp. + 2, W addlw 32 movwi 0[FSR0] banksel @pc movf @pc + 0, W movwf FSR0L movf @pc + 1, W movwf FSR0H moviw 0[FSR0] addlw -1 movwi 0[FSR0] return END xerxes at overo:/usr/src/openembedded/verdexpro/tmp/work/armv5te-angstrom-linux-gnueabi/llvm2.6-2.5+svnr81322-r1/trunk/build/bin$ -- 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 9 06:18:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 06:18:06 -0500 Subject: [LLVMbugs] [Bug 4926] Assertion in rewriteARMFrameIndex() In-Reply-To: Message-ID: <200909091118.n89BI6xs009241@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4926 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #2 from Anton Korobeynikov 2009-09-09 06:18:00 --- (In reply to comment #1) > Looks OK using rev 81322 from TOT No it is not. You forgot to enable NEON via -mattr=+neon -- 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 9 09:21:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 09:21:45 -0500 Subject: [LLVMbugs] [Bug 4935] New: ARM HowToUseJIT Result: 10 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4935 Summary: ARM HowToUseJIT Result: 10 Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: xerxes at zafena.se CC: llvmbugs at cs.uiuc.edu Consider these two consecutive runs of HowToUseJIT on ARM: one returns 11 (correct) the other returns 10 (wrong). It fails on about 1 of 4 runs. tested using rev 81338 xerxes at sheeva:/wd/llvm/Debug/examples$ ./HowToUseJIT We just constructed this LLVM module: ; ModuleID = 'test' define i32 @add1(i32 %AnArg) { EntryBlock: %addresult = add i32 1, %AnArg ; [#uses=1] ret i32 %addresult } define i32 @foo() { EntryBlock: %add1 = tail call i32 @add1(i32 10) ; [#uses=1] ret i32 %add1 } declare void @abort() Running foo: Result: 11 xerxes at sheeva:/wd/llvm/Debug/examples$ ./HowToUseJIT We just constructed this LLVM module: ; ModuleID = 'test' define i32 @add1(i32 %AnArg) { EntryBlock: %addresult = add i32 1, %AnArg ; [#uses=1] ret i32 %addresult } define i32 @foo() { EntryBlock: %add1 = tail call i32 @add1(i32 10) ; [#uses=1] ret i32 %add1 } declare void @abort() Running foo: Result: 10 xerxes at sheeva:/wd/llvm/Debug/examples$ -- 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 9 09:30:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 09:30:51 -0500 Subject: [LLVMbugs] [Bug 4891] Backend producing incorrect code for aliased buffers In-Reply-To: Message-ID: <200909091430.n89EUpLd002362@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4891 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #15 from Dan Gohman 2009-09-09 09:30:50 --- This is fixed on TOT in r81343 here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086743.html GenWidenVectorLoads itself was fine, but the code that called it was accidentally using the wrong chain. The original testcase is now compiled to this, with both loads happening before either store: movl (%rdi), %eax movl (%rsi), %ecx movl %ecx, (%rdi) movl %eax, (%rsi) And, Dave's reset.bad testcase is now compiled correctly. -- 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 9 11:11:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 11:11:05 -0500 Subject: [LLVMbugs] [Bug 4936] New: Win64 visual studio build does not compile ASM correctly Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4936 Summary: Win64 visual studio build does not compile ASM correctly Product: new-bugs Version: trunk Platform: PC OS/Version: Windows NT Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: gareth_morgan77 at hotmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3487) --> (http://llvm.org/bugs/attachment.cgi?id=3487) Working cmake file The visual studio files generated by cmake will not compile X86CompilationCallback_Win64.asm correctly. To get this to work I had to add the following code to lib\Target\X86\CMakeLists.txt (full file attached): if( CMAKE_CL_64 ) enable_language(ASM_MASM) ADD_CUSTOM_COMMAND( OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/X86CompilationCallback_Win64.obj COMMAND "C:/Program Files (x86)/Microsoft Visual Studio 9.0/VC/bin/amd64/ml64" /Fo ${CMAKE_CURRENT_BINARY_DIR}/X86CompilationCallback_Win64.obj /c ${CMAKE_CURRENT_SOURCE_DIR}/X86CompilationCallback_Win64.asm DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/X86CompilationCallback_Win64.asm ) set(sources ${sources} ${CMAKE_CURRENT_BINARY_DIR}/X86CompilationCallback_Win64.obj) endif() I suspect the issue is with the Cmake system, not our the LLVM 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 9 12:45:53 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 12:45:53 -0500 Subject: [LLVMbugs] [Bug 4810] [2.6] ExecutionEngine test failures on mixed 64/32 system In-Reply-To: Message-ID: <200909091745.n89HjrtD032171@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4810 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #9 from Tanya Lattner 2009-09-09 12:45:51 --- Ok to close 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 Wed Sep 9 13:20:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 13:20:44 -0500 Subject: [LLVMbugs] [Bug 4934] -loop-unswitch: cLoopUnswitch.cpp:204: virtual bool:: LoopUnswitch::runOnLoop(llvm::Loop*, llvm::LPPassManager&): Assertion `currentLoop->isLCSSAForm()' failed. In-Reply-To: Message-ID: <200909091820.n89IKiOG005769@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4934 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Dan Gohman 2009-09-09 13:20:44 --- Fixed, here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086773.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 9 13:43:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 13:43:39 -0500 Subject: [LLVMbugs] [Bug 4937] New: bugpoint doesn't binary search with run-custom Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4937 Summary: bugpoint doesn't binary search with run-custom Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu, nlewycky at google.com Running bugpoint on the bugpoint-passinput.bc from http://www.fileden.com/files/2009/9/1/2561232/bugpoint-passinput.bc which is 3.5M results a bugpoint-tooptimize.bc that is 3.4M. I think that if bugpoint did a binary search on the functions it could narrow it down more. My custom.sh looks like this: #!/bin/sh llvm-dis $1 -o -|grep extractvalue|grep undef >/dev/null test $? -eq 0 && exit 1 echo "OK" exit 0 Basically -ipsccp transforms extractvalue %949, 1 into extractvalue undef, 1, and I want to reduce a testcase for that. Thus if there is an extractvalue..undef in the output, then its bugged. $ bugpoint -disable-loop-extraction -mlimit=0 -ipsccp --append-exit-code -run-custom -safe-run-custom -exec-command=./custom.sh bugpoint-passinput.bc ... override triple is Read input file : 'bugpoint-passinput.bc' *** All input ok Running selected passes on program to test for crash: override triple is Success! Initializing execution environment: Found command in: ./custom.sh Running the code generator to test for a crash: Generating reference output from raw program: Reference output is: bugpoint.reference.out-sxXqxN *** Checking the code generator... *** Output matches: Debugging miscompilation! Checking to see if '' compiles correctly: override triple is yup. Checking to see if '-ipsccp' compiles correctly: override triple is nope. *** Found miscompiling pass: -ipsccp Emitted bitcode to 'bugpoint-passinput.bc' *** You can reproduce the problem with: opt bugpoint-passinput.bc -ipsccp Checking to see if the program is misoptimized when these functions are run through the pass: _ZN4llvm17AsmMatcherEmitter3runERNS_11raw_ostreamE _GLOBAL__I__ZN4llvm17AsmMatcherEmitter3runERNS_11raw_ostreamE _ZNK12_GLOBAL__N_19ClassInfo10isSubsetOfERKS0_ _ZNSt8_Rb_treeISsSsSt9_IdentityISsESt4lessISsESaISsEE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordES2_St9_IdentityIS2_ESt4lessIS2_ESaIS2_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISt3setIPN4llvm6RecordESt4lessIS3_ESaIS3_EES7_St9_IdentityIS7_ES4_IS7_ESaIS7_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordESt4pairIKS2_St3setIS2_St4lessIS2_ESaIS2_EEESt10_Select1stISA_ES7_SaISA_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISt3setIPN4llvm6RecordESt4lessIS3_ESaIS3_EESt4pairIKS7_PN12_GLOBAL__N_19ClassInfoEESt10_Select1stISD_ES4_IS7_ESaISD_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordESt4pairIKS2_PN12_GLOBAL__N_19ClassInfoEESt10_Select1stIS8_ESt4lessIS2_ESaIS8_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISsSt4pairIKSsPN12_GLOBAL__N_19ClassInfoEESt10_Select1stIS5_ESt4lessISsESaIS5_EE7_S_leftEPSt18_Rb_tree_node_base... <1642 total> Optimizing functions being tested: override triple is done. Checking to see if the merged program executes correctly: yup. *** The following functions are being miscompiled: _ZN4llvm17AsmMatcherEmitter3runERNS_11raw_ostreamE _GLOBAL__I__ZN4llvm17AsmMatcherEmitter3runERNS_11raw_ostreamE _ZNK12_GLOBAL__N_19ClassInfo10isSubsetOfERKS0_ _ZNSt8_Rb_treeISsSsSt9_IdentityISsESt4lessISsESaISsEE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordES2_St9_IdentityIS2_ESt4lessIS2_ESaIS2_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISt3setIPN4llvm6RecordESt4lessIS3_ESaIS3_EES7_St9_IdentityIS7_ES4_IS7_ESaIS7_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordESt4pairIKS2_St3setIS2_St4lessIS2_ESaIS2_EEESt10_Select1stISA_ES7_SaISA_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISt3setIPN4llvm6RecordESt4lessIS3_ESaIS3_EESt4pairIKS7_PN12_GLOBAL__N_19ClassInfoEESt10_Select1stISD_ES4_IS7_ESaISD_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeIPN4llvm6RecordESt4pairIKS2_PN12_GLOBAL__N_19ClassInfoEESt10_Select1stIS8_ESt4lessIS2_ESaIS8_EE7_S_leftEPSt18_Rb_tree_node_base _ZNSt8_Rb_treeISsSt4pairIKSsPN12_GLOBAL__N_19ClassInfoEESt10_Select1stIS5_ESt4lessISsESaIS5_EE7_S_leftEPSt18_Rb_tree_node_base... <1642 total> Checking to see if the program is misoptimized when all blocks are extracted. override triple is Optimizing functions being tested: override triple is done. Checking to see if the merged program executes correctly: yup. Checking to see if the program is misoptimized when all but these 63372 blocks are extracted: entry bb2.i invcont8 bb12 bb.i.i363 bb.i.i.i364 bb1.i.i.i365 _ZN9__gnu_cxxL27__exchange_and_add_dispatchEPii.exit.i.i366 bb2.i.i367 bb17 ... override triple is Optimizing functions being tested: override triple is done. Checking to see if the merged program executes correctly: yup. Outputting reduced bitcode files which expose the problem: Non-optimized portion: Emitted bitcode to 'bugpoint-tonotoptimize.bc' Portion that is input to optimizer: Emitted bitcode to 'bugpoint-tooptimize.bc' *** You can reproduce the problem with: opt bugpoint-tooptimize.bc -ipsccp -- 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 9 18:01:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 18:01:04 -0500 Subject: [LLVMbugs] [Bug 4939] New: incorrect ARM code for SETOLE floating-point comparisons Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4939 Summary: incorrect ARM code for SETOLE floating-point comparisons Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: bob.wilson at apple.com CC: llvmbugs at cs.uiuc.edu The FPCCToARMCC function in ARMISelLowering.cpp handles SETOLE comparisons by translating to ARMCC::GT and indicating that "...the operands should be inverted". That comment is ambiguous and the code using this function appears to interpret it in two different ways. For SELECT_CC, the "true" and "false" operands are swapped when FPCCToARMCC returns true. For BR_CC, the values being compared are swapped. Unfortunately, neither interpretation is correct! If the values being compared are swapped, SETOLE should be translated to SETOGE or ARMCC::GE. If the SELECT_CC result "true" and "false" result values are swapped, SETOLE should be translated to SETUGT or ARMCC::HI. Unless I'm missing something here, there's no reason to swap anything, since SETOLE for floating-point values can be directly translated to ARMCC::LS. This problem was exposed by a failure of the SPEC 450.soplex test. -- 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 9 18:16:08 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 18:16:08 -0500 Subject: [LLVMbugs] [Bug 4939] incorrect ARM code for SETOLE floating-point comparisons In-Reply-To: Message-ID: <200909092316.n89NG86h017232@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4939 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Bob Wilson 2009-09-09 18:16:06 --- Fixed in svn revision 81397. -- 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 9 20:27:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 9 Sep 2009 20:27:14 -0500 Subject: [LLVMbugs] [Bug 4812] ARM large JIT regression between 2009-07-08 rev 74992 and 2009-07-10 rev 75243 In-Reply-To: Message-ID: <200909100127.n8A1RELj001912@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4812 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #16 from Evan Cheng 2009-09-09 20:27:14 --- r81403 and 81422 as well. -- 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 10 00:21:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 00:21:14 -0500 Subject: [LLVMbugs] [Bug 4942] New: Crash during bootstrap Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4942 Summary: Crash during bootstrap Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: dpatel at apple.com, llvmbugs at cs.uiuc.edu, deeppatel1987 at gmail.com This is a minimized test case that Sandeep me based on a crash he is seeing during bootstrap of llvm-gcc. It is pretty aggressively minimized, and is very sensitive to any further changes in the types. I have attached the version of llvm-gcc that this reproduces with (on Darwin). -- ddunbar at giles:delta$ ./cc1plus -g t.cpp bool EVT::isPow2VectorType() const void DAGCombiner::AddToWorkList(SDNode*) SDValue DAGCombiner::CombineTo(SDValue) SDValue f0(SDValue) Analyzing compilation unitPerforming interprocedural optimizations Assembling functions: SDValue DAGCombiner::CombineTo(SDValue) t.cpp: In member function 'SDValue DAGCombiner::CombineTo(SDValue)': t.cpp:24: internal compiler error: Bus error Please submit a full bug report, with preprocessed source if appropriate. See for instructions. -- Sandeep, can you see if this test case also crashes with your cc1plus? Devang, can you try looking at this test case in gdb to determine why it is crashing? Hopefully that will give enough information to explain what the crash is, and how to fix it. -- 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 10 09:29:07 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 09:29:07 -0500 Subject: [LLVMbugs] [Bug 4943] New: llvm-ld crashing when linking FreeBSD kernel Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4943 Summary: llvm-ld crashing when linking FreeBSD kernel Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Transformation Utilities AssignedTo: unassignedbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu I am getting this when I try to link FreeBSD kernel: Assertion failed: (*PrevPtr == this && "List invariant broken"), function RemoveFromUseList, file Value.cpp, line 458. I can provide tarball with all the .o files and exact command line but it's huge. -- 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 10 11:25:58 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 11:25:58 -0500 Subject: [LLVMbugs] [Bug 4943] llvm-ld crashing when linking FreeBSD kernel In-Reply-To: Message-ID: <200909101625.n8AGPwMT005499@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4943 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #2 from Chris Lattner 2009-09-10 11:25:57 --- Yeah, we definitely need more info than 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 10 11:31:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 11:31:45 -0500 Subject: [LLVMbugs] [Bug 4943] llvm-ld crashing when linking FreeBSD kernel In-Reply-To: Message-ID: <200909101631.n8AGVjCw006328@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4943 Roman Divacky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|INVALID | --- Comment #3 from Roman Divacky 2009-09-10 11:31:43 --- sorry for not providing any info this is the command line: Assertion failed: (*PrevPtr == this && "List invariant broken"), function RemoveFromUseList, file Value.cpp, line 458. Stack dump: 0. Program arguments: llvm-ld -native -strip-all -export-dynamic -o kernel.debug locore.o dsfield.o dsinit.o dsmethod.o dsmthdat.o dsobject.o dsopcode.o dsutils.o dswexec.o dswload.o dswscope.o dswstate.o evevent.o evgpe.o evgpeblk.o evmisc.o evregion.o evrgnini.o evsci.o evxface.o evxfevnt.o evxfregn.o exconfig.o exconvrt.o excreate.o exdump.o exfield.o exfldio.o exmisc.o exmutex.o exnames.o exoparg1.o exoparg2.o exoparg3.o exoparg6.o exprep.o exregion.o exresnte.o exresolv.o exresop.o exstore.o exstoren.o exstorob.o exsystem.o exutils.o hwacpi.o hwgpe.o hwregs.o hwsleep.o hwtimer.o hwvalid.o hwxface.o nsaccess.o nsalloc.o nsdump.o nseval.o nsinit.o nsload.o nsnames.o nsobject.o nsparse.o nspredef.o nssearch.o nsutils.o nswalk.o nsxfeval.o nsxfname.o nsxfobj.o psargs.o psloop.o psopcode.o psparse.o psscope.o pstree.o psutils.o pswalk.o psxface.o rsaddr.o rscalc.o rscreate.o rsdump.o rsinfo.o rsio.o rsirq.o rslist.o rsmemory.o rsmisc.o rsutils.o rsxface.o tbfadt.o tbfind.o tbinstal.o tbutils.o tbxface.o tbxfroot.o utalloc.o utcache.o utcopy.o utdebug.o utdelete.o uteval.o utglobal.o utinit.o utlock.o utmath.o utmisc.o utmutex.o utobject.o utresrc.o utstate.o utxface.o rijndael-alg-fst.o rijndael-api-fst.o sha2.o db_access.o db_break.o db_capture.o db_command.o db_examine.o db_expr.o db_input.o db_lex.o db_main.o db_output.o db_print.o db_ps.o db_run.o db_script.o db_sym.o db_thread.o db_textdump.o db_variables.o db_watch.o db_write_cmd.o OsdDebug.o OsdHardware.o OsdInterrupt.o OsdMemory.o OsdSchedule.o OsdStream.o OsdSynch.o OsdTable.o acpi.o acpi_acad.o acpi_battery.o acpi_button.o acpi_cmbat.o acpi_cpu.o acpi_ec.o acpi_hpet.o acpi_isab.o acpi_lid.o acpi_package.o acpi_pci.o acpi_pci_link.o acpi_pcib.o acpi_pcib_acpi.o acpi_pcib_pci.o acpi_perf.o acpi_powerres.o acpi_quirk.o acpi_resource.o acpi_smbat.o acpi_thermal.o acpi_throttle.o acpi_timer.o agp.o agp_if.o ata_if.o ata-all.o ata-dma.o ata-lowlevel.o ata-queue.o ata-sata.o ata-isa.o ata-pci.o ata-ahci.o ata-acard.o ata-acerlabs.o ata-adaptec.o ata-amd.o ata-ati.o ata-cenatek.o ata-cypress.o ata-cyrix.o ata-highpoint.o ata-intel.o ata-ite.o ata-jmicron.o ata-marvell.o ata-micron.o ata-national.o ata-netcell.o ata-nvidia.o ata-promise.o ata-serverworks.o ata-siliconimage.o ata-sis.o ata-via.o ata-disk.o atapi-tape.o ichss.o eisa_if.o led.o memdev.o mmcbr_if.o mmcbus_if.o null.o card_if.o power_if.o fixup_pci.o hostb_pci.o ignore_pci.o isa_pci.o pci.o pci_if.o pci_pci.o pci_user.o pcib_if.o vga_pci.o pty.o harvest.o hash.o probe.o randomdev.o randomdev_soft.o yarrow.o schistory.o scmouse.o scterm.o scvidctl.o syscons.o sysmouse.o uart_bus_acpi.o uart_bus_isa.o uart_bus_pci.o uart_core.o uart_dev_ns8250.o uart_if.o uart_subr.o uart_tty.o watchdog.o dead_vnops.o devfs_devs.o devfs_rule.o devfs_vfsops.o devfs_vnops.o fifo_vnops.o geom_ctl.o geom_dev.o geom_disk.o geom_dump.o geom_event.o geom_io.o geom_kern.o geom_slice.o geom_subr.o geom_vfs.o g_label.o g_label_ext2fs.o g_label_iso9660.o g_label_msdosfs.o g_label_ntfs.o g_label_reiserfs.o g_label_ufs.o g_label_gpt.o g_part.o g_part_if.o g_part_bsd.o g_part_ebr.o g_part_gpt.o g_part_mbr.o isa_if.o isa_common.o isahint.o orm.o bus_if.o clock_if.o cpufreq_if.o device_if.o imgact_elf.o imgact_shell.o init_main.o init_sysent.o ksched.o kern_acct.o kern_clock.o kern_condvar.o kern_conf.o kern_cons.o kern_cpu.o kern_cpuset.o kern_context.o kern_descrip.o kern_environment.o kern_event.o kern_exec.o kern_exit.o kern_fail.o kern_fork.o kern_idle.o kern_intr.o kern_jail.o kern_kthread.o kern_ktrace.o kern_linker.o kern_lock.o kern_lockf.o kern_malloc.o kern_mbuf.o kern_mib.o kern_module.o kern_mtxpool.o kern_mutex.o kern_ntptime.o kern_osd.o kern_physio.o kern_pmc.o kern_priv.o kern_proc.o kern_prot.o kern_resource.o kern_rmlock.o kern_rwlock.o kern_sema.o kern_shutdown.o kern_sig.o kern_subr.o kern_switch.o kern_sx.o kern_synch.o kern_syscalls.o kern_sysctl.o kern_tc.o kern_thr.o kern_thread.o kern_time.o kern_timeout.o kern_umtx.o kern_uuid.o kern_xxx.o link_elf.o linker_if.o md5c.o p1003_1b.o posix4_mib.o sched_ule.o serdev_if.o stack_protector.o subr_acl_posix1e.o subr_autoconf.o subr_blist.o subr_bus.o subr_bufring.o subr_clock.o subr_devstat.o subr_disk.o subr_eventhandler.o subr_fattime.o subr_hints.o subr_kdb.o subr_kobj.o subr_lock.o subr_log.o subr_module.o subr_msgbuf.o subr_param.o subr_pcpu.o subr_power.o subr_prf.o subr_prof.o subr_rman.o subr_rtc.o subr_sbuf.o subr_scanf.o subr_sglist.o subr_sleepqueue.o subr_smp.o subr_stack.o subr_taskqueue.o subr_trap.o subr_turnstile.o subr_unit.o sys_generic.o sys_pipe.o sys_process.o sys_socket.o sysv_ipc.o sysv_msg.o sysv_sem.o sysv_shm.o tty.o tty_info.o tty_inq.o tty_outq.o tty_pts.o tty_tty.o tty_ttydisc.o uipc_accf.o uipc_debug.o uipc_domain.o uipc_mbuf.o uipc_mbuf2.o uipc_shm.o uipc_sockbuf.o uipc_socket.o uipc_syscalls.o uipc_usrreq.o vfs_acl.o vfs_bio.o vfs_cache.o vfs_cluster.o vfs_default.o vfs_export.o vfs_extattr.o vfs_hash.o vfs_init.o vfs_lookup.o vfs_mount.o vfs_subr.o vfs_syscalls.o vfs_vnops.o arc4random.o bcd.o bsearch.o crc32.o fnmatch.o gets.o index.o inet_ntoa.o memcmp.o qsort.o qsort_r.o random.o rindex.o scanc.o skpc.o strcasecmp.o strcat.o strcmp.o strcpy.o strcspn.o strdup.o strlcat.o strlcpy.o strlen.o strncmp.o strncpy.o strsep.o strspn.o strstr.o strtol.o strtoq.o strtoul.o strtouq.o strvalid.o bpf.o bpf_buffer.o bpf_filter.o bpf_zerocopy.o if.o if_clone.o if_dead.o if_ethersubr.o if_loop.o if_llatbl.o if_media.o if_mib.o if_tun.o netisr.o pfil.o radix.o radix_mpath.o raw_cb.o raw_usrreq.o route.o rtsock.o if_ether.o igmp.o in.o ip_id.o in_mcast.o in_pcb.o in_proto.o in_rmx.o ip_ecn.o ip_encap.o ip_fastfwd.o ip_icmp.o ip_input.o ip_options.o ip_output.o raw_ip.o tcp_hostcache.o tcp_input.o tcp_lro.o tcp_output.o tcp_offload.o tcp_reass.o tcp_sack.o tcp_subr.o tcp_syncache.o tcp_timer.o tcp_timewait.o tcp_usrreq.o udp_usrreq.o dest6.o frag6.o icmp6.o in6.o in6_cksum.o in6_ifattach.o in6_mcast.o in6_pcb.o in6_proto.o in6_rmx.o in6_src.o ip6_forward.o ip6_id.o ip6_input.o ip6_output.o mld6.o nd6.o nd6_nbr.o nd6_rtr.o raw_ip6.o route6.o scope6.o udp6_usrreq.o audit.o audit_arg.o audit_bsm.o audit_bsm_domain.o audit_bsm_errno.o audit_bsm_fcntl.o audit_bsm_klib.o audit_bsm_socket_type.o audit_bsm_token.o audit_pipe.o audit_syscalls.o audit_trigger.o audit_worker.o mac_syscalls.o teken.o ffs_alloc.o ffs_balloc.o ffs_inode.o ffs_snapshot.o ffs_softdep.o ffs_subr.o ffs_tables.o ffs_vfsops.o ffs_vnops.o ufs_acl.o ufs_bmap.o ufs_dirhash.o ufs_extattr.o ufs_gjournal.o ufs_inode.o ufs_lookup.o ufs_quota.o ufs_vfsops.o ufs_vnops.o default_pager.o device_pager.o phys_pager.o sg_pager.o swap_pager.o uma_core.o uma_dbg.o vm_contig.o vm_fault.o vm_glue.o vm_init.o vm_kern.o vm_map.o vm_meter.o vm_mmap.o vm_object.o vm_page.o vm_pageout.o vm_pager.o vm_phys.o vm_reserv.o vm_unix.o vm_zeroidle.o vnode_pager.o OsdEnvironment.o acpi_machdep.o acpi_switch.o acpi_wakeup.o madt.o amd64_mem.o atomic.o autoconf.o bios.o busdma_machdep.o cpu_switch.o db_disasm.o db_interface.o db_trace.o dump_machdep.o elf_machdep.o exception.o fpu.o identcpu.o in_cksum.o initcpu.o intr_machdep.o io.o io_apic.o legacy.o local_apic.o machdep.o mca.o mem.o minidump_machdep.o mp_machdep.o mpboot.o msi.o nexus.o pmap.o sigtramp.o stack_machdep.o support.o sys_machdep.o trap.o tsc.o uio_machdep.o uma_machdep.o vm_machdep.o clock.o elcr.o isa.o isa_dma.o nmi.o pci_bus.o pci_cfgreg.o acpi_if.o acpi_wmi_if.o agp_amd64.o agp_i810.o agp_intel.o agp_via.o atkbd.o atkbd_atkbdc.o atkbdc.o atkbdc_isa.o atkbdc_subr.o psm.o fb.o vga.o iodev.o fdc.o fdc_acpi.o fdc_isa.o kbd.o memutil.o scterm-teken.o scvgarndr.o scvtb.o uart_cpu_amd64.o atrtc.o syscons_isa.o vga_isa.o link_elf_obj.o powernow.o est.o hwpstate.o p4tcc.o memmove.o memset.o config.o env.o hints.o vnode_if.o hack.So vers.o and here you can download all the files: http://www.vlakno.cz/~rdivacky/llvm.bug.tgz -- 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 10 15:21:20 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 15:21:20 -0500 Subject: [LLVMbugs] [Bug 4945] New: clang: -fomit-frame-pointer must be enabled at -O1 and higher on x86-64 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4945 Summary: clang: -fomit-frame-pointer must be enabled at -O1 and higher on x86-64 Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu llvm-gcc and gcc enable -fomit-frame-pointer at -O1 and higher (including -Os) on x86-64; clang does not. Testcase: $ echo 'int main(){return 0;}'|~/llvm-git/install/bin/clang -march=x86-64 -O1 -x c - -S -o -|grep rbp pushq %rbp movq %rsp, %rbp popq %rbp $ echo 'int main(){return 0;}'|~/llvm-git/install/bin/llvm-gcc -march=x86-64 -O1 -x c - -S -o -|grep rbp Note that this is a driver issue, if I use -fomit-frame-pointer clang works correctly too: $ echo 'int main(){return 0;}'|~/llvm-git/install/bin/clang -march=x86-64 -fomit-frame-pointer -O1 -x c - -S -o -|grep rbp -- 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 10 20:48:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 10 Sep 2009 20:48:37 -0500 Subject: [LLVMbugs] [Bug 4890] Assert when generating debug info for C++ destructor In-Reply-To: Message-ID: <200909110148.n8B1mbnS012291@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4890 Anders Carlsson changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |andersca at mac.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Anders Carlsson 2009-09-10 20:48:37 --- http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090907/021191.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 11 01:17:33 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 01:17:33 -0500 Subject: [LLVMbugs] [Bug 4946] New: tblgen generating insane function in cellspu dag isel Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4946 Summary: tblgen generating insane function in cellspu dag isel Product: tools Version: 2.2 Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: TableGen AssignedTo: unassignedbugs at nondot.org ReportedBy: clattner at apple.com CC: llvmbugs at cs.uiuc.edu In the Select_SPUISD_MUL64_MARKER_i64 function, tblgen is generating a call to "Emit_201" passing it many many many arguments. Emit_201 only has one call site, so it could be inlined into the callee. It isn't clear what is going on. Another issue is that we're passing around EVT values by value in dagisel matcher code, and EVT doesn't qualify to get the default copy ctors, so they are not passed by value. This is "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 Fri Sep 11 02:40:58 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 02:40:58 -0500 Subject: [LLVMbugs] [Bug 4947] New: LLVM-GCC generates wrong type for GV zeroinitializer ( LLVM 2.6 regression) Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4947 Summary: LLVM-GCC generates wrong type for GV zeroinitializer (LLVM 2.6 regression) Product: new-bugs Version: 2.6 Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jlerouge at apple.com CC: llvmbugs at cs.uiuc.edu Using LLVM-GCC that comes with LLVM 2.6, the following C code: typedef struct { char a; int b; } foo; foo myFoo[1] = {{0}}; generates the following IR using llvm-gcc -c -emit-llvm: %0 = type { i8, [7 x i8] } %struct.foo = type { i8, i32 } @myFoo = global [1 x %0] zeroinitializer, align 4 The expected result is (what LLVM-GCC 2.5 gives): %struct.foo = type { i8, i32 } @myFoo = global [1 x %struct.foo] zeroinitializer so that myFoo ends up having the correct type ([1 x %struct.foo]*). If myFoo is completely initialized, the behaviour is correct: typedef struct { char a; int b; } foo; foo myFoo[1] = {{0, 0}}; will give: %struct.foo = type { i8, i32 } @myFoo = global [1 x %struct.foo] zeroinitializer Thanks, Julien -- 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 11 09:05:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 09:05:13 -0500 Subject: [LLVMbugs] [Bug 4948] New: --time-report crashes llvm-gcc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4948 Summary: --time-report crashes llvm-gcc 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: baldrick at free.fr CC: llvmbugs at cs.uiuc.edu Compiling any file with --time-report and checks enabled: $ llvm-gcc --time-report -c empty.c Execution times (seconds) llvm backend per file : 0.01 (100%) usr 0.00 ( 0%) sys 0.00 ( 0%) wall 0 kB ( 0%) ggc TOTAL : 0.01 0.00 0.01 989 kB Extra diagnostic checks enabled; compiler may run slowly. Configure with --disable-checking to disable checks. cc1: llvm/include/llvm/Support/Timer.h:177: llvm::TimerGroup::~TimerGroup(): Assertion `NumTimers == 0 && "TimerGroup destroyed before all contained timers!"' 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 Fri Sep 11 09:41:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 09:41:27 -0500 Subject: [LLVMbugs] [Bug 4949] New: llc crashes with Intrinsic functions cannot be mangled by Mangler Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4949 Summary: llc crashes with Intrinsic functions cannot be mangled by Mangler Product: new-bugs Version: 2.6 Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: martin.nowack at inf.tu-dresden.de CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3492) --> (http://llvm.org/bugs/attachment.cgi?id=3492) bugpoint file created using bugpoint -run-llc only Trying to create assembler code from a bc file, llc aborts. Bugpoint reduced bc file is 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 Fri Sep 11 12:06:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 12:06:12 -0500 Subject: [LLVMbugs] [Bug 4949] verifier doesn't reject address of intrinsic In-Reply-To: Message-ID: <200909111706.n8BH6C9g019463@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4949 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Summary|llc crashes with Intrinsic |verifier doesn't reject |functions cannot be mangled |address of intrinsic |by Mangler | --- Comment #2 from Chris Lattner 2009-09-11 12:06:12 --- Actually, that .bc file is invalid. You can't take the address of an intrinsic. Did you create this IR directly, or did some pass do it? If so, the pass is broken. I made the verifier abort on this so that you can hopefully track it down: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086904.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 11 12:25:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 12:25:56 -0500 Subject: [LLVMbugs] [Bug 4948] --time-report crashes llvm-gcc In-Reply-To: Message-ID: <200909111725.n8BHPurT022374@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4948 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Duncan Sands 2009-09-11 12:25:55 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086909.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 11 13:42:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 13:42:37 -0500 Subject: [LLVMbugs] [Bug 4820] [2.6] make clean removes llvm-config before it's done using it In-Reply-To: Message-ID: <200909111842.n8BIgbIj002501@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4820 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Bob Wilson 2009-09-11 13:42:36 --- Fixed in svn r81547. -- 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 11 16:54:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 16:54:14 -0500 Subject: [LLVMbugs] [Bug 4953] New: Integrate calling convention into function types Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4953 Summary: Integrate calling convention into function types Product: clang Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu Function types should include the calling convention as part of the type. This is necessary for proper handling of things like arrays of functions pointers using a different calling convention. -- 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 11 17:14:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 17:14:43 -0500 Subject: [LLVMbugs] [Bug 4954] New: llvm-ld: Assertion `isa(Val) && "cast() argument of incompatible type!"' Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4954 Summary: llvm-ld: Assertion `isa(Val) && "cast() argument of incompatible type!"' Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: nicolas.geoffray at lip6.fr CC: llvmbugs at cs.uiuc.edu The two following .bc files (produced by llvm-gcc on linux/x86_32) give an assertion error when trying to llvm-ld them together in that particular order: > llvm-ld first.bc second.bc Doing llvm-ld second.bc first.bc works fine. The full backtrace is: llvm-ld: /home/varth/project/llvm-svn/llvm/include/llvm/Support/Casting.h:199: typename llvm::cast_retty::ret_type llvm::cast(const Y&) [with X = llvm::CompositeType, Y = const llvm::Type*]: Assertion `isa(Val) && "cast() argument of incompatible type!"' failed. 0 llvm-ld 0x083eaa6a 1 llvm-ld 0x083eb6e3 2 0xb7f96420 __kernel_sigreturn + 0 3 libc.so.6 0xb7d1da01 abort + 257 4 libc.so.6 0xb7d1510e __assert_fail + 238 5 llvm-ld 0x08391526 llvm::ConstantFoldGetElementPtr(llvm::LLVMContext&, llvm::Constant const*, bool, llvm::Constant* const*, unsigned int) + 470 6 llvm-ld 0x082e029f llvm::ConstantExpr::getGetElementPtrTy(llvm::Type const*, llvm::Constant*, llvm::Value* const*, unsigned int) + 175 7 llvm-ld 0x083438b8 llvm::ConvertConstantType::convert(llvm::ConstantExpr*, llvm::Type const*) + 408 8 llvm-ld 0x083484be llvm::ValueMap::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 126 9 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 10 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 11 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 12 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 13 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 14 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 15 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 16 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 17 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 18 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 19 llvm-ld 0x0836fdcc llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2652 20 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 21 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 22 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 23 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 24 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 25 llvm-ld 0x083713ac llvm::TypeMap::RefineAbstractType(llvm::FunctionType*, llvm::DerivedType const*, llvm::Type const*) + 1644 26 llvm-ld 0x0836b4a9 llvm::FunctionType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 27 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 28 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 29 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 30 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 31 llvm-ld 0x0836fd77 llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2567 32 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 33 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 34 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 35 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 36 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 37 llvm-ld 0x0836fdcc llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2652 38 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 39 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 40 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 41 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 42 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 43 llvm-ld 0x0836fd77 llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2567 44 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 45 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 46 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 47 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 48 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 49 llvm-ld 0x0836fdcc llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2652 50 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 51 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 52 llvm-ld 0x0836fd77 llvm::TypeMap::RefineAbstractType(llvm::StructType*, llvm::DerivedType const*, llvm::Type const*) + 2567 53 llvm-ld 0x0836b3e9 llvm::StructType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 54 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 55 llvm-ld 0x08371b7e llvm::TypeMap::RefineAbstractType(llvm::PointerType*, llvm::DerivedType const*, llvm::Type const*) + 1358 56 llvm-ld 0x0836b609 llvm::PointerType::refineAbstractType(llvm::DerivedType const*, llvm::Type const*) + 57 57 llvm-ld 0x0836b173 llvm::DerivedType::unlockedRefineAbstractTypeTo(llvm::Type const*) + 291 58 llvm-ld 0x0836b4fe llvm::DerivedType::refineAbstractTypeTo(llvm::Type const*) + 78 59 llvm-ld 0x080ef158 60 llvm-ld 0x080f71d6 llvm::Linker::LinkModules(llvm::Module*, llvm::Module*, std::string*) + 694 61 llvm-ld 0x080edbd3 llvm::Linker::LinkInFile(llvm::sys::Path const&, bool&) + 1299 62 llvm-ld 0x080eed5a llvm::Linker::LinkInItems(std::vector, std::allocator > > const&, std::vector, std::allocator > >&) + 282 63 llvm-ld 0x080cce5b main + 4027 64 libc.so.6 0xb7d07450 __libc_start_main + 224 65 llvm-ld 0x080c4b71 __gxx_personality_v0 + 293 -- 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 11 18:00:58 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 18:00:58 -0500 Subject: [LLVMbugs] [Bug 4953] Integrate calling convention into function types In-Reply-To: Message-ID: <200909112300.n8BN0wYY006124@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4953 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #4 from Chris Lattner 2009-09-11 18:00:57 --- This is not the solution you're looking for. function pointers (arrays or otherwise) work fine because the call should have the calling conv specified on it. -- 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 11 18:18:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 18:18:03 -0500 Subject: [LLVMbugs] [Bug 4957] New: find-cycles.pl incompatible with MinGW headers Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4957 Summary: find-cycles.pl incompatible with MinGW headers Product: new-bugs Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Keywords: build-problem, portability Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu When I build a stage2 compiler hosted on MinGW, find-cycles.pl fails. (Stage 1 is darwin->mingw32.) The culprits are the ctype.h functions provided in MinGW. They're "inline" in C++ and an out-of-line copy of isalnum() is generated inside libLLVMCppBackend.a (among others). MinGW uses "extern __inline__" and "extern inline __attribute__((__gnu_inline__)) in C depending on the version of GCC, but just "inline" in C++. Some crazy thoughts: 1. Build with -ffunction-sections and have a libbing step remove all out-of-line ctype functions during MinGW builds. 2. Implement the gnu_inline attribute in llvm-gcc and clang and then change _mingw.h accordingly. The original implementation of gnu_inline is after GCC went GPLv3, so this would have to be done from scratch. 3. Find some other way to mark these functions with availableExternally instead of linkonce_odr. This blocks non-debug builds on MinGW. -- 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 11 18:42:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 11 Sep 2009 18:42:13 -0500 Subject: [LLVMbugs] [Bug 4953] Integrate calling convention into function types In-Reply-To: Message-ID: <200909112342.n8BNgDQL011415@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4953 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|INVALID | --- Comment #5 from Daniel Dunbar 2009-09-11 18:42:13 --- This was and still is a clang bug, not an LLVM 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 Sat Sep 12 04:12:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 04:12:48 -0500 Subject: [LLVMbugs] [Bug 4958] New: "opt -scalarrepl" produces wrong code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4958 Summary: "opt -scalarrepl" produces wrong code Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: d.g.gorbachev at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3498) --> (http://llvm.org/bugs/attachment.cgi?id=3498) testcase LLVM-GCC and LLVM r81611. -- 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 12 13:17:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 13:17:56 -0500 Subject: [LLVMbugs] [Bug 4959] New: update llvm-gcc OpenBSD target configurations Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4959 Summary: update llvm-gcc OpenBSD target configurations Product: new-bugs Version: trunk Platform: PC OS/Version: OpenBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jsg at openbsd.org CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3499) --> (http://llvm.org/bugs/attachment.cgi?id=3499) update openbsd llvm-gcc config The attached diff updates OpenBSD target configurations for llvm-gcc based on the gcc 4.2 port, and most importantly enables pic when building an llvm enabled configuration. I see no llvm-gcc specific list or maintainer so I'll submit it here. -- 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 12 13:25:54 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 13:25:54 -0500 Subject: [LLVMbugs] [Bug 4960] New: [2.6] ARM Linux no implementation for llvm::sys::Memory:: InvalidateInstructionCache Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4960 Summary: [2.6] ARM Linux no implementation for llvm::sys::Memory::InvalidateInstructionCache Product: libraries Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: blocker Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: xerxes at zafena.se CC: llvmbugs at cs.uiuc.edu Blocks: 4886 llvm::sys::Memory::InvalidateInstructionCache in llvm/lib/System/Memory.cpp have no implementation to clear the instruction cache on ARM Linux. This BUG causes sporadic ExecutionEngine regressions failures on ARM Linux since the instruction cache can be out of sync and make the ARM JIT crash with a sigsegv on the first instruction when returning from a callback jited basic block. llvm::sys::Memory::InvalidateInstructionCache might be implemented on ARM Linux using gcc's builtin __clear_cache() -- 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 12 13:33:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 13:33:47 -0500 Subject: [LLVMbugs] [Bug 4772] ARM unittest JIT. GlobalInFunction fail during ARM Machine Code Emitter In-Reply-To: Message-ID: <200909121833.n8CIXlHr007113@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4772 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |DUPLICATE --- Comment #9 from Xerxes R??nby 2009-09-12 13:33:46 --- After pussling on this one, the jited code actually looks fine. The SIGSEGV are most likely caused by missing code to invalidate the instruction cache on ARM Linux. *** This bug has been marked as a duplicate of bug 4960 *** -- 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 12 14:42:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 14:42:03 -0500 Subject: [LLVMbugs] [Bug 4961] New: [2.6] ARM GCC __sync_synchronize does not generate code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4961 Summary: [2.6] ARM GCC __sync_synchronize does not generate code Product: libraries Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: xerxes at zafena.se CC: llvmbugs at cs.uiuc.edu Blocks: 4886 http://gcc.gnu.org/ml/gcc-patches/2009-08/msg00600.html GCC __sync_synchronize does not generate code for ARM. This basically means LLVM_MULTITHREADED==1 are broken for ARM until we implement some OS specific code to implement sys::MemoryFence() in lib/System/Atomic.cpp for for ARM Darwin and ARM Linux when using GCC. For ARM Linux this can be implemented by a system call to __kernel_dmb a linux kernel helper located at fixed address 0xffff0fa0 that can perform the memmorybarrier for userspace applications. For ARM Darwin i have no clue, please assist. -- 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 12 16:19:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 16:19:00 -0500 Subject: [LLVMbugs] [Bug 4962] New: [2.6] tests need backporting Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4962 Summary: [2.6] tests need backporting Product: Test Suite Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu The attached patch fixes several test cases on the 2.6 branch, which need back porting to work on 2.6 (for llc input changes). -- 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 12 16:25:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 16:25:27 -0500 Subject: [LLVMbugs] [Bug 4963] New: constant folding insertvalue loses packedness of struct Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4963 Summary: constant folding insertvalue loses packedness of struct Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Keywords: miscompilation 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 Consider the following: D code using C binding (much like C): auto i = LLVMInt32Type(); auto t = LLVMStructType(&i, 1, 1); auto tmp = LLVMGetUndef(t); LLVMDumpType(LLVMTypeOf(tmp)); uint idx = 0; tmp = LLVMConstInsertValue(tmp, LLVMConstInt(i, 0, 0), &idx, 1); LLVMDumpType(LLVMTypeOf(tmp)); Dumps: <{ i32 }> { i32 } This patch is against revision 79806, but it is also present in 2.6, it's a pretty serious bug, and I think it should be fixed in 2.6 as well. =================================================================== --- ../lib/VMCore/ConstantFold.cpp (revision 79806) +++ ../lib/VMCore/ConstantFold.cpp (working copy) @@ -520,8 +520,8 @@ UndefValue::get(MemberTy); Ops[i] = const_cast(Op); } - if (isa(AggTy)) - return ConstantStruct::get(Context, Ops); + if (const StructType* ST = dyn_cast(AggTy)) + return ConstantStruct::get(Context, Ops, ST->isPacked()); else return ConstantArray::get(cast(AggTy), Ops); } @@ -549,8 +549,8 @@ Constant::getNullValue(MemberTy); Ops[i] = const_cast(Op); } - if (isa(AggTy)) - return ConstantStruct::get(Context, Ops); + if (const StructType* ST = dyn_cast(AggTy)) + return ConstantStruct::get(Context, Ops, ST->isPacked()); else return ConstantArray::get(cast(AggTy), Ops); } @@ -566,8 +566,8 @@ Ops[i] = const_cast(Op); } Constant *C; - if (isa(Agg->getType())) - C = ConstantStruct::get(Context, Ops); + if (const StructType* ST = dyn_cast(Agg->getType())) + C = ConstantStruct::get(Context, Ops, ST->isPacked()); else C = ConstantArray::get(cast(Agg->getType()), Ops); return 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 Sat Sep 12 17:24:19 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 17:24:19 -0500 Subject: [LLVMbugs] [Bug 4965] New: Invalid register assignment due to regclass copy coalesced away Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4965 Summary: Invalid register assignment due to regclass copy coalesced away Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3502) --> (http://llvm.org/bugs/attachment.cgi?id=3502) .ll file Consider the bytecode attached. Currently it fails due to invalid subreg index (there is on 32-bit subreg of Q8 reG) despite of explicit copy_to_regclass node. It seems the cross-regclass copy is removed during 2addr instruction pass: %reg1063 = INSERT_SUBREG %reg1061, %reg1062, 4 prepend: %reg1063 = VMOVQ %reg1061 rewrite to: %reg1063 = INSERT_SUBREG %reg1063, %reg1062, 4 -- 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 12 17:54:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 17:54:36 -0500 Subject: [LLVMbugs] [Bug 4962] [2.6] tests need backporting In-Reply-To: Message-ID: <200909122254.n8CMsaDL008749@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4962 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |tonic at nondot.org Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Tanya Lattner 2009-09-12 17:54:36 --- This is in 2.6 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 Sat Sep 12 18:32:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 18:32:15 -0500 Subject: [LLVMbugs] [Bug 4966] New: assembly printer fails to print reference to global var Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4966 Summary: assembly printer fails to print reference to global var Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: nunoplopes at sapo.pt CC: llvmbugs at cs.uiuc.edu There was a recent regression in the assembly code printer. I believe it was in r81508 (2009-09-11 06:58:44 +0100 (Fri, 11 Sep 2009)), but I didn't confirm. $ llc foo.ll $ cat foo.ll @"__PRETTY_FUNCTION__.int main(int, char **)" = private constant [23 x i8] c"int main(int, char **)\00" ; <[23 x i8]*> [#uses=1] declare void @__assert_fail(i8*) define void @main() nounwind { entry: call void @__assert_fail(i8* getelementptr inbounds ([23 x i8]* @"__PRETTY_FUNCTION__.int main(int, char **)", i32 0, i32 0)) ret void } $ cat foo.ll.s (...) main: # @main .LBB1_0: # %entry subl $4, %esp movl $.L__PRETTY_FUNCTION__.int main(int, char **), (%esp) (...) .section .rodata.str1.16,"aMS", at progbits,1 .align 16 .L__PRETTY_FUNCTION__.int_20_main_28_int_2C__20_char_20__2A__2A__29_: # @"__PRETTY_FUNCTION__.int main(int, char **)" .asciz "int main(int, char **)" .size .L__PRETTY_FUNCTION__.int_20_main_28_int_2C__20_char_20__2A__2A__29_, 23 note that the GV symbol name in the movl is not mangled, and in the GV decl is. -- 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 12 19:22:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 19:22:11 -0500 Subject: [LLVMbugs] [Bug 4967] New: Can't compiler floating point related code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4967 Summary: Can't compiler floating point related code Product: libraries Version: 2.5 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: yonggangluo at hotmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3504) --> (http://llvm.org/bugs/attachment.cgi?id=3504) the file to be compiled llvm-gcc internal compiling error! cc1plus: /home/dreamkxd/src/llvm-2.5/lib/Target/X86/X86FloatingPoint.cpp:1072: void::FPS::handleSpecialFP(llvm::ilist_iterator&): Assertion `Op.isUse() && (Op.isKill() || getFPReg(Op) == FirstFPRegOp || MI->killsRegister(Op.getReg())) && "Ret only defs operands, and values aren't live beyond it"' failed. /home/dreamkxd/src/Amaya/wxWidgets/src/common/matrix.cpp:603: internal compiler error: Aborted Please submit a full bug report, with preprocessed source if appropriate. See for instructions. because of the generated position is not correct. So I attached the compiled 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 Sat Sep 12 20:01:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 20:01:32 -0500 Subject: [LLVMbugs] [Bug 4965] Invalid register assignment due to regclass copy coalesced away In-Reply-To: Message-ID: <200909130101.n8D11WGE024749@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4965 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Anton Korobeynikov 2009-09-12 20:01:31 --- Fixed in: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086978.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086985.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 12 20:24:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 20:24:44 -0500 Subject: [LLVMbugs] [Bug 4873] Infinite loop in APFloat::convertFromDecimalString() around APFloat.cpp:2378 In-Reply-To: Message-ID: <200909130124.n8D1Oi3d027641@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4873 Dmitry Gorbachev changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #10 from Dmitry Gorbachev 2009-09-12 20:24:44 --- Dup. of bug 4958, I believe. *** This bug has been marked as a duplicate of bug 4958 *** -- 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 12 21:18:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 21:18:44 -0500 Subject: [LLVMbugs] [Bug 4968] New: gtest death tests give annoying warning Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4968 Summary: gtest death tests give annoying warning Product: Test Suite Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu This warning shows up a lot on some platforms (e.g. the ARM buildbot): [WARNING] gtest-death-test.cc:458: Death tests use fork(), which is unsafe particularly in a threaded context. For this test, Google Test couldn't detect the number of threads. It should be 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 Sat Sep 12 21:19:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 12 Sep 2009 21:19:45 -0500 Subject: [LLVMbugs] [Bug 4960] [2.6] ARM Linux no implementation for llvm::sys::Memory:: InvalidateInstructionCache In-Reply-To: Message-ID: <200909130219.n8D2JjI7002931@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4960 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Daniel Dunbar 2009-09-12 21:19:45 --- Fixed in r81655. -- 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 13 02:47:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 02:47:09 -0500 Subject: [LLVMbugs] [Bug 4536] clang __gnu_inline__ semantics do not match gcc In-Reply-To: Message-ID: <200909130747.n8D7l9mC018351@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4536 Douglas Gregor changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #9 from Douglas Gregor 2009-09-13 02:47:08 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090907/021268.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 Sun Sep 13 03:28:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 03:28:37 -0500 Subject: [LLVMbugs] [Bug 4969] New: Once compiling a file comes from Amaya, using -O2 fail but without it succeed. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4969 Summary: Once compiling a file comes from Amaya, using -O2 fail but without it succeed. Product: libraries Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Interprocedural Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: yonggangluo at hotmail.com CC: llvmbugs at cs.uiuc.edu dreamkxd at dreamkxd-desktop:~/src/Amaya/Amaya/objs/thotlib$ /usr/lib/llvm2.6/gcc-4.2/bin/llvm-g++ -emit-llvm -S -v -O2 -Wall -x c++ -D__cplusplus -D_UNIX -D_GL -D_WX -DHAVE_CONFIG_H -I.. -I../../amaya/xpm -I../../thotlib/include -I../../thotlib/internals/var -I../../thotlib/internals/h -I../../thotlib/internals/f -I/home/dreamkxd/src/Amaya/wxWidgets/src/png -I/home/dreamkxd/src/Amaya/wxWidgets/src/jpeg -I/home/dreamkxd/src/Amaya/wxWidgets/src/tiff -I/home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 -I/home/dreamkxd/src/Amaya/wxWidgets/include -I/home/dreamkxd/src/Amaya/wxWidgets/contrib/include -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES -D__WXGTK__ -pthread -I/home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include -I/usr/include/freetype2 -c ../../thotlib/presentation/presrules.c -o presentation/presrules.o Using built-in specs. Target: i686-linux-gnu Configured with: /mnt/src/llvm-gcc/configure --host=i686-linux-gnu --build=i686-linux-gnu --prefix=/usr/lib/llvm2.6/gcc-4.2 --enable-languages=c,c++ --program-prefix=llvm- --enable-llvm=/usr/lib/llvm2.6 --enable-threads --disable-nls --disable-shared --disable-multilib --disable-bootstrap Thread model: posix gcc version 4.2.1 (Based on Apple Inc. build 5649) (LLVM build) /usr/lib/llvm2.6/gcc-4.2/libexec/gcc/i686-linux-gnu/4.2.1/cc1plus -quiet -v -I.. -I../../amaya/xpm -I../../thotlib/include -I../../thotlib/internals/var -I../../thotlib/internals/h -I../../thotlib/internals/f -I/home/dreamkxd/src/Amaya/wxWidgets/src/png -I/home/dreamkxd/src/Amaya/wxWidgets/src/jpeg -I/home/dreamkxd/src/Amaya/wxWidgets/src/tiff -I/home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 -I/home/dreamkxd/src/Amaya/wxWidgets/include -I/home/dreamkxd/src/Amaya/wxWidgets/contrib/include -I/home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include -I/usr/include/freetype2 -D_GNU_SOURCE -D_REENTRANT -D__cplusplus -D_UNIX -D_GL -D_WX -DHAVE_CONFIG_H -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES -D__WXGTK__ ../../thotlib/presentation/presrules.c -emit-llvm -quiet -dumpbase presrules.c -mtune=generic -auxbase-strip presentation/presrules.o -O2 -Wall -version -o presentation/presrules.o ignoring nonexistent directory "/usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../i686-linux-gnu/include" ignoring nonexistent directory "/home/dreamkxd/src/Amaya/wxWidgets/contrib/include" #include "..." search starts here: #include <...> search starts here: .. ../../amaya/xpm ../../thotlib/include ../../thotlib/internals/var ../../thotlib/internals/h ../../thotlib/internals/f /home/dreamkxd/src/Amaya/wxWidgets/src/png /home/dreamkxd/src/Amaya/wxWidgets/src/jpeg /home/dreamkxd/src/Amaya/wxWidgets/src/tiff /home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 /home/dreamkxd/src/Amaya/wxWidgets/include /home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include /usr/include/freetype2 /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1 /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1/i686-linux-gnu /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1/backward /usr/local/include /usr/lib/llvm2.6/gcc-4.2/include /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/include /usr/include End of search list. GNU C++ version 4.2.1 (Based on Apple Inc. build 5649) (LLVM build) (i686-linux-gnu) compiled by GNU C version 4.2.1 (Based on Apple Inc. build 5636) (LLVM build). GGC heuristics: --param ggc-min-expand=64 --param ggc-min-heapsize=63683 Compiler executable checksum: a994527e259f35a25d2a7ed449fb1e8b ../../thotlib/presentation/presrules.c:5016: internal compiler error: Segmentation fault Please submit a full bug report, with preprocessed source if appropriate. See for instructions. Because of dependence is too much, so I didn't attach other files. dreamkxd at dreamkxd-desktop:~/src/Amaya/Amaya/objs/thotlib$ /usr/lib/llvm2.6/gcc-4.2/bin/llvm-g++ -emit-llvm -S -v -Wall -x c++ -D__cplusplus -D_UNIX -D_GL -D_WX -DHAVE_CONFIG_H -I.. -I../../amaya/xpm -I../../thotlib/include -I../../thotlib/internals/var -I../../thotlib/internals/h -I../../thotlib/internals/f -I/home/dreamkxd/src/Amaya/wxWidgets/src/png -I/home/dreamkxd/src/Amaya/wxWidgets/src/jpeg -I/home/dreamkxd/src/Amaya/wxWidgets/src/tiff -I/home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 -I/home/dreamkxd/src/Amaya/wxWidgets/include -I/home/dreamkxd/src/Amaya/wxWidgets/contrib/include -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES -D__WXGTK__ -pthread -I/home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include -I/usr/include/freetype2 -c ../../thotlib/presentation/presrules.c -o presentation/presrules.o Using built-in specs. Target: i686-linux-gnu Configured with: /mnt/src/llvm-gcc/configure --host=i686-linux-gnu --build=i686-linux-gnu --prefix=/usr/lib/llvm2.6/gcc-4.2 --enable-languages=c,c++ --program-prefix=llvm- --enable-llvm=/usr/lib/llvm2.6 --enable-threads --disable-nls --disable-shared --disable-multilib --disable-bootstrap Thread model: posix gcc version 4.2.1 (Based on Apple Inc. build 5649) (LLVM build) /usr/lib/llvm2.6/gcc-4.2/libexec/gcc/i686-linux-gnu/4.2.1/cc1plus -quiet -v -I.. -I../../amaya/xpm -I../../thotlib/include -I../../thotlib/internals/var -I../../thotlib/internals/h -I../../thotlib/internals/f -I/home/dreamkxd/src/Amaya/wxWidgets/src/png -I/home/dreamkxd/src/Amaya/wxWidgets/src/jpeg -I/home/dreamkxd/src/Amaya/wxWidgets/src/tiff -I/home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 -I/home/dreamkxd/src/Amaya/wxWidgets/include -I/home/dreamkxd/src/Amaya/wxWidgets/contrib/include -I/home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include -I/usr/include/freetype2 -D_GNU_SOURCE -D_REENTRANT -D__cplusplus -D_UNIX -D_GL -D_WX -DHAVE_CONFIG_H -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES -D__WXGTK__ ../../thotlib/presentation/presrules.c -emit-llvm -quiet -dumpbase presrules.c -mtune=generic -auxbase-strip presentation/presrules.o -Wall -version -o presentation/presrules.o ignoring nonexistent directory "/usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../i686-linux-gnu/include" ignoring nonexistent directory "/home/dreamkxd/src/Amaya/wxWidgets/contrib/include" #include "..." search starts here: #include <...> search starts here: .. ../../amaya/xpm ../../thotlib/include ../../thotlib/internals/var ../../thotlib/internals/h ../../thotlib/internals/f /home/dreamkxd/src/Amaya/wxWidgets/src/png /home/dreamkxd/src/Amaya/wxWidgets/src/jpeg /home/dreamkxd/src/Amaya/wxWidgets/src/tiff /home/dreamkxd/src/Amaya/Amaya/objs/wxWidgets_RELEASE/lib/wx/include/gtk2-unicode-release-static-2.8 /home/dreamkxd/src/Amaya/wxWidgets/include /home/dreamkxd/src/Amaya/Amaya/objs/Mesa/include /usr/include/freetype2 /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1 /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1/i686-linux-gnu /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/../../../../include/c++/4.2.1/backward /usr/local/include /usr/lib/llvm2.6/gcc-4.2/include /usr/lib/llvm2.6/gcc-4.2/lib/gcc/i686-linux-gnu/4.2.1/include /usr/include End of search list. GNU C++ version 4.2.1 (Based on Apple Inc. build 5649) (LLVM build) (i686-linux-gnu) compiled by GNU C version 4.2.1 (Based on Apple Inc. build 5636) (LLVM build). GGC heuristics: --param ggc-min-expand=64 --param ggc-min-heapsize=127366 Compiler executable checksum: a994527e259f35a25d2a7ed449fb1e8b As you can see, once I cancel the -O2 option, it's can be compiled correctly:) So it's must be the opt problem... I generate the ll file. By the way, the source code of presrules.c can be download from http://www.w3.org/Amaya/User/SourceDist.html The exact source code I downloaded is http://wam.inrialpes.fr/software/amaya/amaya-fullsrc-11.3-pre.tgz -- 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 13 11:53:31 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 11:53:31 -0500 Subject: [LLVMbugs] [Bug 4970] New: Verifier segfaults Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4970 Summary: Verifier segfaults Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu This happened after updating the snv checkout. It was ok with a 20 days-old checkout. If the problem is not obvious I can try to bisect. Stack dump: #0 0x00007f22d1099142 in llvm::Type::getForwardedType (this=0x31) at /home/oscar/dev/llvm/head/include/llvm/Type.h:302 No locals. #1 0x00007f22d19dcfbd in llvm::Type::getForwardedTypeInternal (this=0x3b0f4f0) at /home/oscar/dev/llvm/head/lib/VMCore/Type.cpp:238 RealForwardedType = (const class llvm::Type *) 0x7fffda7b1680 __PRETTY_FUNCTION__ = "const llvm::Type* llvm::Type::getForwardedTypeInternal() const" #2 0x00007f22d109915e in llvm::Type::getForwardedType (this=0x3b0f4f0) at /home/oscar/dev/llvm/head/include/llvm/Type.h:303 No locals. #3 0x00007f22d109936a in llvm::PATypeHolder::get (this=0x7fffda7b16f0) at /home/oscar/dev/llvm/head/include/llvm/Type.h:435 NewTy = (const llvm::Type *) 0x0 #4 0x00007f22d10993af in llvm::PATypeHolder::operator llvm::Type* ( this=0x7fffda7b16f0) at /home/oscar/dev/llvm/head/include/llvm/AbstractTypeUser.h:151 No locals. #5 0x00007f22d1a07fdc in std::less::operator() ( this=0xd168840, __x=@0x7fffda7b16f0, __y=@0xd16b5e0) at /usr/include/c++/4.3/bits/stl_function.h:230 No locals. #6 0x00007f22d1a085b0 in std::_Rb_tree, std::less, std::allocator >::_M_insert_unique (this=0xd168840, __v=@0x7fffda7b16f0) at /usr/include/c++/4.3/bits/stl_tree.h:1141 __x = (class std::_Rb_tree_node *) 0xd16b5c0 __y = (class std::_Rb_tree_node *) 0xd16b5c0 __comp = true __j = {_M_node = 0x7fffda7b14e0} #7 0x00007f22d1a0875d in std::set, std::allocator >::insert (this=0xd168840, __x=@0x7fffda7b16f0) at /usr/include/c++/4.3/bits/stl_set.h:381 __p = {first = {_M_node = 0x7fffda7b1530}, second = 168} #8 0x00007f22d1a087e4 in llvm::SmallSet::insert ( this=0xd1687a0, V=@0x7fffda7b16f0) at /home/oscar/dev/llvm/head/include/llvm/ADT/SmallSet.h:60 I = (const llvm::PATypeHolder *) 0x7fffda7b16f0 #9 0x00007f22d19fecc1 in VerifyType (this=0xd168690, Ty=0x3b0f4f0) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:1462 No locals. #10 0x00007f22d19ff01b in VerifyType (this=0xd168690, Ty=0x3b0f2c0) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:1499 PTy = (const llvm::PointerType *) 0x3b0f2c0 #11 0x00007f22d1a003ac in visitInstruction (this=0xd168690, I=@0x3b0f4b8) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:1456 BB = (class llvm::BasicBlock *) 0x3ac28b0 #12 0x00007f22d1a00a83 in visitBitCastInst (this=0xd168690, I=@0x3b0f4b8) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:1003 SrcTy = (const class llvm::Type *) 0x2af8ab0 DestTy = (const class llvm::Type *) 0x3b0f2c0 SrcBitSize = 0 DestBitSize = 0 #13 0x00007f22d1a00aab in visitBitCast (this=0xd168690, I=@0x3b0f4b8) at /home/oscar/dev/llvm/head/include/llvm/Instruction.def:154 No locals. #14 0x00007f22d1a06262 in visit (this=0xd168690, I=@0x3b0f4b8) at /home/oscar/dev/llvm/head/include/llvm/Instruction.def:154 No locals. #15 0x00007f22d1a063c2 in visit (this=0xd168690, I=@0x3b0f4b8) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:360 No locals. #16 0x00007f22d1a06407 in visit > ( this=0xd168690, Start= {> = {}, NodePtr = 0x3b0f588}, End= {> = {}, NodePtr = 0x3ac28b0}) at /home/oscar/dev/llvm/head/include/llvm/Support/InstVisitor.h:89 No locals. #17 0x00007f22d1a0645c in visit (this=0xd168690, BB=@0x3ac28b0) at /home/oscar/dev/llvm/head/include/llvm/Support/InstVisitor.h:104 No locals. #18 0x00007f22d1a064a1 in visit > ( this=0xd168690, Start= {> = {}, NodePtr = 0x3ac2780}, End= {> = {}, NodePtr = 0x3b21508}) at /home/oscar/dev/llvm/head/include/llvm/Support/InstVisitor.h:89 No locals. #19 0x00007f22d1a064f6 in visit (this=0xd168690, F=@0x3b214d0) at /home/oscar/dev/llvm/head/include/llvm/Support/InstVisitor.h:100 No locals. #20 0x00007f22d1a0654f in runOnFunction (this=0xd168690, F=@0x3b214d0) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:168 No locals. #21 0x00007f22d19ce22a in llvm::FPPassManager::runOnFunction (this=0xd168f80, F=@0x3b214d0) at /home/oscar/dev/llvm/head/lib/VMCore/PassManager.cpp:1371 X = { = { _vptr.PrettyStackTraceEntry = 0x7f22d212fe90, NextEntry = 0x7fffda7b20b0}, P = 0xd168690, V = 0x3b214d0, M = 0x0} FP = (class llvm::FunctionPass *) 0xd168690 Index = 2 Changed = false #22 0x00007f22d19ce448 in llvm::FPPassManager::runOnModule (this=0xd168f80, M=@0x2c6daa0) at /home/oscar/dev/llvm/head/lib/VMCore/PassManager.cpp:1394 I = {> = {}, NodePtr = 0x3b214d0} E = {> = {}, NodePtr = 0x2c6da60} Changed = false #23 0x00007f22d19cc93c in llvm::MPPassManager::runOnModule (this=0xd1684b0, M=@0x2c6daa0) at /home/oscar/dev/llvm/head/lib/VMCore/PassManager.cpp:1447 X = { = { _vptr.PrettyStackTraceEntry = 0x7f22d212fe90, NextEntry = 0x0}, P = 0xd168f80, V = 0x0, M = 0x2c6daa0} MP = (llvm::ModulePass *) 0xd168f80 Index = 0 Changed = false #24 0x00007f22d19d0221 in llvm::PassManagerImpl::run (this=0xd165ca0, M=@0x2c6daa0) at /home/oscar/dev/llvm/head/lib/VMCore/PassManager.cpp:1528 Index = 0 Changed = false #25 0x00007f22d19d0283 in llvm::PassManager::run (this=0x7fffda7b21f0, M=@0x2c6daa0) at /home/oscar/dev/llvm/head/lib/VMCore/PassManager.cpp:1557 No locals. #26 0x00007f22d19f7a04 in llvm::verifyModule (M=@0x2c6daa0, action=llvm::AbortProcessAction, ErrorInfo=0x0) at /home/oscar/dev/llvm/head/lib/VMCore/Verifier.cpp:1853 PM = { = { _vptr.PassManagerBase = 0x7f22d212fc10}, PM = 0xd165ca0} V = ((anonymous namespace)::Verifier *) 0xd168690 #27 0x00007f22d1089ecb in lp0::LlvmCompile (resultTipo= {static tipoBaseCount = 1397, static tipoBaseMask = 32767, static kindMask = 983040, static isRefMask = 2147483648, static isConstMask = 1073741824, static hasInitializerMask = 536870912, data = 3665503712}, commands=@0x72963c0, source=0x8c0be0) at /home/oscar/dev/idb-llvm-head/lp0/llvm.cpp:1243 M = (class llvm::Module *) 0x2c6daa0 coreBe = { stack = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = 0xb5fac00, _M_finish = 0xb5fac00, _M_end_of_storage = 0xb5fb000}}, }, module = 0x2c6daa0, currentFunction = 0xd157390, scheduledFunctions = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = 0xc12e590, _M_finish = 0xc1893e8, _M_end_of_storage = 0xc1be590}}, }, functionMap = { _M_t = { _M_impl = { > >> = {<__gnu_cxx::new_allocator > >> = {}, }, _M_key_compare = {> = {}, }, _M_header = { _M_color = std::_S_red, _M_parent = 0x2a6bed0, _M_left = 0x9365430, _M_right = 0x90b4e90}, _M_node_count = 5171}}}, printFunction = false} EE = (class llvm::ExecutionEngine *) 0x3624ed0 mainF = (class llvm::Function *) 0x3b214d0 t1 = 24440000 be = { pointerMap = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = 0x0, _M_finish = 0x0, _M_end_of_storage = 0x0}}, }, corebe = @0x7fffda7b2370, commands = @0x72963c0, jumped = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = { = {> = {}, _M_p = 0x2a3f9f0, _M_offset = 0}, }, _M_finish = { = {> = {}, _M_p = 0x2a40008, _M_offset = 29}, }, _M_end_of_storage = 0x2a40010}}, }, cmdInfos = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = 0x9083200, _M_finish = 0x90b3fd0, _M_end_of_storage = 0x90b3fd0}}, }, bb = 0x0, nextIsReturn = false, bbTerminated = false} MP = (llvm::ExistingModuleProvider *) 0x698f450 error = {static npos = 18446744073709551615, _M_dataplus = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_p = 0x8b54b8 ""}} mf = (class llvm::Function *) 0x7f22d27b0000 __PRETTY_FUNCTION__ = "lp0::LLVMFunction* lp0::LlvmCompile(lp0::Tipo, lp0::Commands&, lp0::Source*)" Context = {pImpl = 0x3d72790} instructionGenMapReady = true #28 0x00007f22d10ab7af in LLVMCode (this=0x2a1e390, ee=@0x8c0c20, code=@0x28463c0, s=0x8c0be0) at /home/oscar/dev/idb-llvm-head/lp0/api.cpp:691 No locals. -- 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 13 13:05:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 13:05:06 -0500 Subject: [LLVMbugs] [Bug 4966] assembly printer fails to print reference to global var In-Reply-To: Message-ID: <200909131805.n8DI56eT001654@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4966 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-13 13:05:06 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/087002.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 Sun Sep 13 16:08:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 16:08:35 -0500 Subject: [LLVMbugs] [Bug 4970] Verifier segfaults In-Reply-To: Message-ID: <200909132108.n8DL8Zcq025704@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4970 Nick Lewycky changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #4 from Nick Lewycky 2009-09-13 16:08:35 --- Fixed in r81716. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/087028.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 Sun Sep 13 16:48:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 16:48:12 -0500 Subject: [LLVMbugs] [Bug 4971] New: Emitting LLVM assembler code segfaults. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4971 Summary: Emitting LLVM assembler code segfaults. Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu While dumping a LLVM module as LLVM assembler a segfault occurs: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7f28cb72a700 (LWP 31697)] 0x00007f28ca02da76 in llvm::PATypeHandle::operator llvm::Type* ( this=0x737365726464) at /home/oscar/dev/llvm/head/include/llvm/AbstractTypeUser.h:106 106 inline operator Type *() const { return const_cast(Ty); } This is the code that dumps the module M: llvm::raw_fd_ostream f("code.llvm", errorInfo); f << *M; Backtrace: #0 0x00007f28ca02da76 in llvm::PATypeHandle::operator llvm::Type* ( this=0x737365726464) at /home/oscar/dev/llvm/head/include/llvm/AbstractTypeUser.h:106 No locals. #1 0x00007f28ca8eacc4 in IncorporateType (this=0x7fffd3754af0, Ty=0x33955b0) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:372 I = (llvm::PATypeHandle *) 0x737365726464 E = (llvm::PATypeHandle *) 0x737958a98ab4 #2 0x00007f28ca8eacd0 in IncorporateType (this=0x7fffd3754af0, Ty=0x3395380) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:372 I = (llvm::PATypeHandle *) 0x33953c8 E = (llvm::PATypeHandle *) 0x33953d8 #3 0x00007f28ca8eaffa in Run (this=0x7fffd3754af0, M=@0x2761030) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:348 I = (const class llvm::Instruction &) @0x3395578: { = { = { _vptr.Value = 0x7f28cb0cf350, SubclassID = 61 '=', HasValueHandle = 0 '\0', SubclassOptionalData = 0 '\0', SubclassData = 0, VTy = {Ty = 0x3395380}, UseList = 0x3395630, Name = 0x33952a0}, OperandList = 0x3395560, NumOperands = 1}, > = {> = {Prev = 0x3395488}, Next = 0x3395648}, Parent = 0x3574590} II = {> = {}, NodePtr = 0x3395578} E = {> = {}, NodePtr = 0x3574590} BB = {> = {}, NodePtr = 0x3574590} E = {> = {}, NodePtr = 0x35cab28} FI = {> = {}, NodePtr = 0x35caaf0} E = {> = {}, NodePtr = 0x2760ff0} ST = (const class llvm::TypeSymbolTable &) @0x37f87c0: { = { _vptr.AbstractTypeUser = 0x7f28cb0d1a70}, tmap = {_M_t = { _M_impl = {, std::allocator >, const llvm::Type*> > >> = {<__gnu_cxx::new_allocator, std::allocator >, const llvm::Type*> > >> = {}, }, _M_key_compare = {, std::allocator >, const std::basic_string, std::allocator >, bool>> = {}, }, _M_header = {_M_color = std::_S_red, _M_parent = 0x0, _M_left = 0x37f87d0, _M_right = 0x37f87d0}, _M_node_count = 0}}}, LastUnique = 0} #4 0x00007f28ca8eb6db in AddModuleTypesToPrinter (TP=@0x7fffd3754d48, NumberedTypes=@0x7fffd3754d58, M=0x2761030) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:440 ST = (const class llvm::TypeSymbolTable &) @0x37f87c0: { = { _vptr.AbstractTypeUser = 0x7f28cb0d1a70}, tmap = {_M_t = { _M_impl = {, std::allocator >, const llvm::Type*> > >> = {<__gnu_cxx::new_allocator, std::allocator >, const llvm::Type*> > >> = {}, }, _M_key_compare = {, std::allocator >, const std::basic_string, std::allocator >, bool>> = {}, }, _M_header = {_M_color = std::_S_red, _M_parent = 0x0, _M_left = 0x37f87d0, _M_right = 0x37f87d0}, _M_node_count = 0}}}, LastUnique = 0} #5 0x00007f28ca8eb7cd in AssemblyWriter (this=0x7fffd3754d30, o=@0x7fffd3754db0, Mac=@0x7fffd3754ca0, M=0x2761030, AAW=0x0) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:1263 No locals. #6 0x00007f28ca8f1551 in llvm::Module::print (this=0x2761030, ROS=@0x7fffd3754ff0, AAW=0x0) at /home/oscar/dev/llvm/head/lib/VMCore/AsmWriter.cpp:1993 SlotTable = {TheModule = 0x2761030, TheFunction = 0x0, FunctionProcessed = false, TheMDNode = 0x0, TheNamedMDNode = 0x0, mMap = { NumBuckets = 64, Buckets = 0xd767e40, NumEntries = 0, NumTombstones = 0}, mNext = 0, fMap = {NumBuckets = 64, Buckets = 0xd774900, NumEntries = 0, NumTombstones = 0}, fNext = 0, mdnMap = {NumBuckets = 64, Buckets = 0xd729e30, NumEntries = 0, NumTombstones = 0}, mdnNext = 0} OS = { = {_vptr.raw_ostream = 0x7f28cb0d2bf0, OutBufStart = 0xd421390 "??????=\r", OutBufEnd = 0xd422390 "\020\020", OutBufCur = 0xd421390 "??????=\r", BufferMode = llvm::raw_ostream::InternalBuffer, Error = false}, static DELETE_STREAM = , static PRESERVE_STREAM = , TheStream = 0x7fffd3754ff0, DeleteStream = false, ColumnScanned = 0, Scanned = 0x0} TheModule = 0x2761030, TypePrinter = {TypeNames = 0x9353270}, AnnotationWriter = 0x0, NumberedTypes = { >> = { _M_impl = {> = {<__gnu_cxx::new_allocator> = {}, }, _M_start = 0xd59c360, _M_finish = 0xd59c3b0, _M_end_of_storage = 0xd59c3e0}}, }, MDNodes = {_M_t = { _M_impl = { > >> = {<__gnu_cxx::new_allocator > >> = {}, }, _M_key_compare = {> = {}, }, _M_header = {_M_color = std::_S_red, _M_parent = 0x0, _M_left = 0x7fffd3754d78, _M_right = 0x7fffd3754d78}, _M_node_count = 0}}}, MetadataIDNo = 0} #7 0x00007f28ca03ba37 in llvm::operator<< (O=@0x7fffd3754ff0, M=@0x2761030) at /home/oscar/dev/llvm/head/include/llvm/Module.h:486 No locals. -- 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 13 17:42:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 13 Sep 2009 17:42:15 -0500 Subject: [LLVMbugs] [Bug 4972] New: [2.6] llvm-gcc fails to compile Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4972 Summary: [2.6] llvm-gcc fails to compile Product: tools Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: foldr at codedgers.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3508) --> (http://llvm.org/bugs/attachment.cgi?id=3508) .bc file that is causing the error While trying to compile llvm-gcc (branches/release_26) against the LLVM 2.6 branch, I encountered the following error: [..]./gcc/xgcc -shared-libgcc [...] -fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -O2 -D_GNU_SOURCE -c ../../../../llvm-gcc/libstdc++-v3/libsupc++/eh_personality.cc -fPIC -DPIC -o eh_personality.o Personality function should be set! UNREACHABLE executed at /home/gman/code/codedgers/morpher/src/lib/CodeGen/MachineModuleInfo.cpp:298! ../../../../llvm-gcc/libstdc++-v3/libsupc++/eh_personality.cc:759: internal compiler error: Aborted This happens both with and without '--disable-bootstrap'. Trunk llc is able to handle the problematic .bc file, though. -- 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 14 00:04:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 00:04:16 -0500 Subject: [LLVMbugs] [Bug 4973] New: missed load through bitcasted global Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4973 Summary: missed load through bitcasted global Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu This load should get eliminated: -- ddunbar at giles:tmp$ cat t.ll ; ModuleID = 't.cpp' target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" target triple = "x86_64-apple-darwin10.0" %0 = type { i8, [3 x i8] } %struct.s0 = type <{ i8, [3 x i8] }> @g0 = private constant %0 { i8 13, [3 x i8] zeroinitializer }, align 4 define i8 @f0() nounwind { %t0 = load i8* getelementptr (%struct.s0* bitcast (%0* @g0 to %struct.s0*), i64 0, i32 0), align 4 ret i8 %t0 } ddunbar at giles:tmp$ opt < t.ll -S -O3 ; ModuleID = '' target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" target triple = "x86_64-apple-darwin10.0" %0 = type { i8, [3 x i8] } %struct.s0 = type <{ i8, [3 x i8] }> @g0 = private constant %0 { i8 13, [3 x i8] zeroinitializer }, align 4 ; <%0*> [#uses=1] define i8 @f0() nounwind readnone { %t0 = load i8* getelementptr (%struct.s0* bitcast (%0* @g0 to %struct.s0*), i64 0, i32 0), align 4 ; [#uses=1] ret i8 %t0 } -- -- 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 14 00:15:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 00:15:39 -0500 Subject: [LLVMbugs] [Bug 4974] New: Assertion failed: (InReg && "Value not in map!") Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4974 Summary: Assertion failed: (InReg && "Value not in map!") Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: evan.cheng at apple.com, llvmbugs at cs.uiuc.edu, bob.wilson at apple.com ddunbar at giles:tmp$ cat t.ll ; ModuleID = 'inputs/test.2.b.c' target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32" target triple = "thumbv7-apple-darwin9.3.0" %0 = type <{ [0 x i32] }> %struct.T0 = type { } @.str = private constant [25 x i8] c"fn2: testing arguments.\0A\00" ; <[25 x i8]*> [#uses=1] @T0_values = internal global [1 x %struct.T0] zeroinitializer, align 1 ; <[1 x %struct.T0]*> [#uses=2] define arm_apcscc void @test_fn2() nounwind { entry: %agg.tmp = alloca %struct.T0 ; <%struct.T0*> [#uses=2] %call = call arm_apcscc i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([25 x i8]* @.str, i32 0, i32 0)) ; [#uses=0] %tmp = bitcast %struct.T0* %agg.tmp to i8* ; [#uses=1] call void @llvm.memcpy.i32(i8* %tmp, i8* bitcast ([1 x %struct.T0]* @T0_values to i8*), i32 0, i32 1) %0 = bitcast %struct.T0* %agg.tmp to %0* ; <%0*> [#uses=1] %1 = load %0* %0, align 1 ; <%0> [#uses=1] call arm_apcscc void @fn2(%0 %1) ret void } declare arm_apcscc i32 @printf(i8*, ...) declare arm_apcscc void @fn2(%0) declare void @llvm.memcpy.i32(i8* nocapture, i8* nocapture, i32, i32) nounwind ddunbar at giles:tmp$ llc < t.ll .syntax unified Assertion failed: (InReg && "Value not in map!"), function getValue, file /Volumes/Data/Users/ddunbar/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp, line 933. 0 llc 0x0000000100b73187 PrintStackTrace(void*) + 38 1 llc 0x0000000100b73715 SignalHandler(int) + 336 2 libSystem.B.dylib 0x00007fff8726214a _sigtramp + 26 3 libSystem.B.dylib 0x00007fff00000000 _sigtramp + 2027544272 4 libSystem.B.dylib 0x00007fff872de104 __pthread_markcancel + 0 5 llc 0x00000001007a4d4e llvm::SelectionDAGLowering::getValue(llvm::Value const*) + 4080 6 llc 0x00000001007a9bb0 llvm::SelectionDAGLowering::LowerCallTo(llvm::CallSite, llvm::SDValue, bool, llvm::MachineBasicBlock*) + 304 7 llc 0x00000001007baba9 llvm::SelectionDAGLowering::visitCall(llvm::CallInst&) + 3167 8 llc 0x00000001007c676a llvm::SelectionDAGLowering::visit(unsigned int, llvm::User&) + 1242 9 llc 0x00000001007c682e llvm::SelectionDAGLowering::visit(llvm::Instruction&) + 44 10 llc 0x00000001007dffb8 llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::ilist_iterator, llvm::ilist_iterator) + 96 11 llc 0x00000001007e0b8b llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 2439 12 llc 0x00000001007e1a30 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 982 13 llc 0x00000001008a8a59 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 83 14 llc 0x0000000100af229b llvm::FPPassManager::runOnFunction(llvm::Function&) + 341 15 llc 0x0000000100af2889 llvm::FunctionPassManagerImpl::run(llvm::Function&) + 111 16 llc 0x0000000100af2a38 llvm::FunctionPassManager::run(llvm::Function&) + 112 17 llc 0x000000010002f458 main + 3233 18 llc 0x000000010002e0a8 start + 52 Stack dump: 0. Program arguments: llc 1. Running pass 'ARM Instruction Selection' on function '@test_fn2' Abort trap -- 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 14 00:18:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 00:18:35 -0500 Subject: [LLVMbugs] [Bug 4975] New: Assertion failed: (Begin + idx < End), function operator[] Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4975 Summary: Assertion failed: (Begin + idx < End), function operator[] Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: evan.cheng at apple.com, llvmbugs at cs.uiuc.edu, bob.wilson at apple.com ddunbar at giles:tmp$ cat t.ll ; ModuleID = 'inputs/test.4.a.c' target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32" target triple = "thumbv7-apple-darwin9.3.0" %0 = type <{ [0 x i32] }> %union.T0 = type { } @.str = private constant [23 x i8] c" fn4: arg0 (empty)\0A\00" ; <[23 x i8]*> [#uses=1] define arm_apcscc void @fn4(%0) nounwind { entry: %arg0 = alloca %union.T0 ; <%union.T0*> [#uses=1] %1 = bitcast %union.T0* %arg0 to %0* ; <%0*> [#uses=1] store %0 %0, %0* %1, align 1 %call = call arm_apcscc i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([23 x i8]* @.str, i32 0, i32 0)) ; [#uses=0] ret void } declare arm_apcscc i32 @printf(i8*, ...) ddunbar at giles:tmp$ llc < t.ll .syntax unified Assertion failed: (Begin + idx < End), function operator[], file /Volumes/Data/Users/ddunbar/llvm/include/llvm/ADT/SmallVector.h, line 124. 0 llc 0x0000000100b73187 PrintStackTrace(void*) + 38 1 llc 0x0000000100b73715 SignalHandler(int) + 336 2 libSystem.B.dylib 0x00007fff8726214a _sigtramp + 26 3 libSystem.B.dylib 0x0000000102411010 _sigtramp + 2065362656 4 libSystem.B.dylib 0x00007fff872de104 __pthread_markcancel + 0 5 llc 0x0000000100735125 llvm::SmallVectorImpl::operator[](unsigned int) + 85 6 llc 0x00000001007a62a0 llvm::SelectionDAGISel::LowerArguments(llvm::BasicBlock*) + 2958 7 llc 0x00000001007e0394 llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 400 8 llc 0x00000001007e1a30 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 982 9 llc 0x00000001008a8a59 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 83 10 llc 0x0000000100af229b llvm::FPPassManager::runOnFunction(llvm::Function&) + 341 11 llc 0x0000000100af2889 llvm::FunctionPassManagerImpl::run(llvm::Function&) + 111 12 llc 0x0000000100af2a38 llvm::FunctionPassManager::run(llvm::Function&) + 112 13 llc 0x000000010002f458 main + 3233 14 llc 0x000000010002e0a8 start + 52 Stack dump: 0. Program arguments: llc 1. Running pass 'ARM Instruction Selection' on function '@fn4' Abort trap -- 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 14 07:28:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 07:28:29 -0500 Subject: [LLVMbugs] [Bug 4976] New: [PPC] An asm causes an assert in the code generator Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4976 Summary: [PPC] An asm causes an assert in the code generator Product: new-bugs Version: trunk Platform: PC OS/Version: Linux 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 Created an attachment (id=3511) --> (http://llvm.org/bugs/attachment.cgi?id=3511) The file that causes the abort. The attached .ll file causes the PPC code generator to abort: [~/ellcc/test/source] main% ~/ellcc/ellcc/powerpc-elf-ecc _exit.i -c [~/ellcc/test/source] main% llc _exit.o llc: /home/rich/llvm-trunk-new/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodesEmit.cpp:122: void llvm::ScheduleDAGSDNodes::EmitCopyFromReg(llvm::SDNode*, unsigned int, bool, bool, unsigned int, llvm::DenseMap, llvm::DenseMapInfo >&): Assertion `Emitted && "Unable to issue a copy instruction!\n"' failed. 0 llc 0x0902b0c0 1 llc 0x0902b68c 2 0x00362400 __kernel_sigreturn + 0 3 libc.so.6 0x00bcce28 abort + 392 4 libc.so.6 0x00bc440e __assert_fail + 238 5 llc 0x08bca5c5 llvm::ScheduleDAGSDNodes::EmitCopyFromReg(llvm::SDNode*, unsigned int, bool, bool, unsigned int, llvm::DenseMap, llvm::DenseMapInfo >&) + 1655 6 llc 0x08bcae86 llvm::ScheduleDAGSDNodes::EmitNode(llvm::SDNode*, bool, bool, llvm::DenseMap, llvm::DenseMapInfo >&) + 2044 7 llc 0x08bcb428 llvm::ScheduleDAGSDNodes::EmitSchedule() + 438 8 llc 0x08c390cb llvm::SelectionDAGISel::CodeGenAndEmitDAG() + 4753 9 llc 0x08c3ab15 llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::ilist_iterator, llvm::ilist_iterator) + 487 10 llc 0x08c3b5e0 llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 2484 11 llc 0x08c3c462 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 886 12 llc 0x088fdf24 13 llc 0x08d2c9ea llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 86 14 llc 0x08fa1d45 llvm::FPPassManager::runOnFunction(llvm::Function&) + 299 15 llc 0x08fa2848 llvm::FunctionPassManagerImpl::run(llvm::Function&) + 124 16 llc 0x08fa2a15 llvm::FunctionPassManager::run(llvm::Function&) + 135 17 llc 0x0847bfa4 main + 3468 18 libc.so.6 0x00bb66e5 __libc_start_main + 229 19 llc 0x0847a1d1 Stack dump: 0. Program arguments: llc _exit.o 1. Running pass 'PowerPC DAG->DAG Pattern Instruction Selection' on function '@_exit' Abort [~/ellcc/test/source] main% -- 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 14 10:06:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 10:06:16 -0500 Subject: [LLVMbugs] [Bug 4977] New: CallGraphSCC expensive checks failure Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4977 Summary: CallGraphSCC expensive checks failure Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: clattner at apple.com, llvmbugs at cs.uiuc.edu Clang has one test failure when compiled with ENABLE_EXPENSIVE_CHECKS: -- ******************** TEST '/home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c' FAILED ******************** Script: -- /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/Debug+Checks/bin/clang-cc /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c -triple i386-unknown-unknown -O3 -emit-llvm -o /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp not grep '@g0' /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/Debug+Checks/bin/clang-cc /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c -triple x86_64-unknown-unknown -O3 -emit-llvm -o /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp not grep '@g0' /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/Debug+Checks/bin/clang-cc /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c -triple powerpc-unknown-unknown -O3 -emit-llvm -o /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp not grep '@g0' /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp true -- Exit Code: 134 Command Output (stdout): -- -- Command Output (stderr): -- /usr/include/c++/4.3/debug/safe_iterator.h:433:error: attempt to compare a singular iterator to a past-the-end iterator. Objects involved in the operation: iterator "lhs" @ 0x0x2954fe0 { type = N11__gnu_debug14_Safe_iteratorIN9__gnu_cxx17__normal_iteratorIPSt4pairIN4llvm6WeakVHEPNS4_13CallGraphNodeEENSt6__norm6vectorIS8_SaIS8_EEEEENSt7__debug6vectorIS8_SC_EEEE (mutable iterator); state = singular; references sequence with type `NSt7__debug6vectorISt4pairIN4llvm6WeakVHEPNS2_13CallGraphNodeEESaIS6_EEE' @ 0x0x2954fe0 } iterator "rhs" @ 0x0x7fff1c075150 { type = N11__gnu_debug14_Safe_iteratorIN9__gnu_cxx17__normal_iteratorIPSt4pairIN4llvm6WeakVHEPNS4_13CallGraphNodeEENSt6__norm6vectorIS8_SaIS8_EEEEENSt7__debug6vectorIS8_SC_EEEE (mutable iterator); state = past-the-end; references sequence with type `NSt7__debug6vectorISt4pairIN4llvm6WeakVHEPNS2_13CallGraphNodeEESaIS6_EEE' @ 0x0x7fff1c075150 } 0 clang-cc 0x0000000001723946 1 clang-cc 0x0000000001723fd5 2 libpthread.so.0 0x00002b508ec60080 3 libc.so.6 0x00002b508f84efb5 gsignal + 53 4 libc.so.6 0x00002b508f850bc3 abort + 387 5 libstdc++.so.6 0x00002b508f0d387c __gnu_debug::_Error_formatter::_M_error() const + 508 6 clang-cc 0x0000000000a1dafa 7 clang-cc 0x00000000015435d9 8 clang-cc 0x00000000015435f9 9 clang-cc 0x0000000001547af4 10 clang-cc 0x0000000001547b84 11 clang-cc 0x0000000001547eb9 12 clang-cc 0x000000000153e04d 13 clang-cc 0x0000000001674a18 14 clang-cc 0x0000000001676d69 15 clang-cc 0x0000000001676dcb 16 clang-cc 0x00000000004a7d51 17 clang-cc 0x00000000004a7ecb 18 clang-cc 0x00000000007667bf 19 clang-cc 0x000000000047d6d6 20 clang-cc 0x000000000047edc5 main + 3209 21 libc.so.6 0x00002b508f83a5a6 __libc_start_main + 230 22 clang-cc 0x0000000000475ff9 Stack dump: 0. Program arguments: /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/Debug+Checks/bin/clang-cc /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c -triple i386-unknown-unknown -O3 -emit-llvm -o /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp 1. parser at end of file 2. Per-module optimization passes 3. Running pass 'CallGraph Pass Manager' on module '/home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c'. /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.script: line 7: 10747 Aborted /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/Debug+Checks/bin/clang-cc /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/CodeGen/parameter-passing.c -triple i386-unknown-unknown -O3 -emit-llvm -o /home/buildbot/buildslave/clang-x86_64-linux-checks/llvm/tools/clang/test/Output/CodeGen/parameter-passing.c.tmp -- ******************** -- 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 14 11:49:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 11:49:48 -0500 Subject: [LLVMbugs] [Bug 4973] missed optzn on load through bitcasted global In-Reply-To: Message-ID: <200909141649.n8EGnmqh030012@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4973 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |LATER --- Comment #1 from Chris Lattner 2009-09-14 11:49:48 --- This is already in lib/Target/README.txt -- 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 14 19:34:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 19:34:22 -0500 Subject: [LLVMbugs] [Bug 4981] New: [Sema] invalid use of '__builtin_offsetof' crashes clang-cc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4981 Summary: [Sema] invalid use of '__builtin_offsetof' crashes clang-cc Product: clang Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: kremenek at apple.com CC: llvmbugs at cs.uiuc.edu The following 3 line file crashes clang-cc: $ cat t.c void foo() { int c = __builtin_offsetof(struct p, x); } $ clang-cc t.c Assertion failed: ((!isa(LookupCtx) || LookupCtx->isDependentContext() || cast(LookupCtx)->isDefinition() || Context.getTypeDeclType(cast(LookupCtx))->getAs() ->isBeingDefined()) && "Declaration context must already be complete!"), function LookupQualifiedName, file /Users/kremenek/llvm/tools/clang/lib/Sema/SemaLookup.cpp, line 1026. 0 clang-cc 0x00dd7be8 PrintStackTrace(void*) + 40 ... -- 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 14 20:22:42 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 14 Sep 2009 20:22:42 -0500 Subject: [LLVMbugs] [Bug 4880] Assertion in instruction selector In-Reply-To: Message-ID: <200909150122.n8F1Mgk1001050@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4880 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Dan Gohman 2009-09-14 20:22:42 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087096.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 15 00:40:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 00:40:59 -0500 Subject: [LLVMbugs] [Bug 4977] CallGraphSCC expensive checks failure In-Reply-To: Message-ID: <200909150540.n8F5exfg001438@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4977 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Chris Lattner 2009-09-15 00:40:58 --- Fixed: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087112.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 15 01:15:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 01:15:09 -0500 Subject: [LLVMbugs] [Bug 4982] New: -regalloc= simple creates invalid code crashing fp stackifier Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4982 Summary: -regalloc=simple creates invalid code crashing fp stackifier Product: libraries Version: 1.0 Platform: PC OS/Version: All Status: NEW Keywords: compile-fail Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: clattner at apple.com CC: llvmbugs at cs.uiuc.edu These work: $ cat bug.ll 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 float @f() nounwind { entry: ret float 0.000000e+00 } $ llc bug.ll -o - -regalloc=local $ llc bug.ll -o - This crashes: $ llc bug.ll -o - -regalloc=simple I suspect a kill/dead marker problem, but haven't looked into it. -- 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 15 01:15:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 01:15:40 -0500 Subject: [LLVMbugs] [Bug 4967] Can't compiler floating point related code In-Reply-To: Message-ID: <200909150615.n8F6FebU005988@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4967 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Chris Lattner 2009-09-15 01:15:39 --- Ok, the original bug here was fixed sometime since 2.5. I split the regalloc=simple bug out to its own PR to simplify things. Bug 4982 -- 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 15 01:28:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 01:28:40 -0500 Subject: [LLVMbugs] [Bug 4963] constant folding insertvalue loses packedness of struct In-Reply-To: Message-ID: <200909150628.n8F6SeCx007650@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4963 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Chris Lattner 2009-09-15 01:28:39 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087113.html Thanks Tomas! -- 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 15 08:54:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 08:54:00 -0500 Subject: [LLVMbugs] [Bug 4935] ARM HowToUseJIT Result: 10 In-Reply-To: Message-ID: <200909151354.n8FDs0TG013351@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4935 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Xerxes R??nby 2009-09-15 08:54:00 --- This bug have been fixed in TOT I cant reproduce this bug when using r81758 and r81851 on arm the bug are present in r81518 possibly fixed by Anton in r81635 and r81657 http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086978.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086985.html I will examine if this bug still exists in 2.6 release branch. -- 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 15 10:47:21 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 10:47:21 -0500 Subject: [LLVMbugs] [Bug 4857] Check/fix sys::getHostTriple on subtargets. In-Reply-To: Message-ID: <200909151547.n8FFlLEI028764@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4857 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Daniel Dunbar 2009-09-15 10:47:20 --- I have decided that LLVM ARM triples are an abomination, which I hope to eventually eliminate. That makes this bug done! :) -- 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 15 10:47:54 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 10:47:54 -0500 Subject: [LLVMbugs] [Bug 4877] globaldce : Assertion failed: (DbgNode && "Invalid DIArray"), function getNumElements, file /Volumes/Data/Users/ddunbar/llvm/lib/ Analysis/DebugInfo.cpp, line 258. In-Reply-To: Message-ID: <200909151547.n8FFlsqo028850@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4877 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #12 from Daniel Dunbar 2009-09-15 10:47:54 --- Devang fixed 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 Tue Sep 15 11:02:20 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 11:02:20 -0500 Subject: [LLVMbugs] [Bug 4954] llvm-ld: Assertion `isa(Val) && "cast() argument of incompatible type!"' In-Reply-To: Message-ID: <200909151602.n8FG2KPE030718@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4954 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #11 from Dan Gohman 2009-09-15 11:02:20 --- This is now fixed, here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087134.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 15 11:34:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 11:34:14 -0500 Subject: [LLVMbugs] [Bug 4942] Crash during bootstrap In-Reply-To: Message-ID: <200909151634.n8FGYEHf002820@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4942 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from devang.patel 2009-09-15 11:34:14 --- Fixed. r81586. -- 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 15 11:46:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 11:46:55 -0500 Subject: [LLVMbugs] [Bug 4984] New: llc -O0: incorrect emission of negative offsets Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4984 Summary: llc -O0: incorrect emission of negative offsets Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu Compiling attached testcase results in these assembler warnings on x86-64 $ llc -O0 >> <<<) .text .align 16 .globl foo .type foo, at function foo: # @foo .LBB1_0: # %entry pushq %rbx movq %rdi, %rbx movl $40, %esi movq %rbx, %rdi call strchr movabsq $2, %rcx addq %rbx, %rcx cmpq %rax, %rcx ja .LBB1_2 .LBB1_1: # %"" >>>>> movb 4294967294(%rax), %al <<<<<<<<<<<<<<<<<<<<<<<<< cmpb $33, %al sete %al movl $1, %ecx testb %al, %al movl $0, %eax cmovne %ecx, %eax popq %rbx ret .LBB1_2: # %"" xorl %eax, %eax popq %rbx ret .size foo, .-foo .section .note.GNU-stack,"", at progbits Without -O0, llc by default outputs this where as doesn't warn: .text .align 16 .globl foo .type foo, at function foo: # @foo .LBB1_0: # %entry pushq %rbx movq %rdi, %rbx movl $40, %esi movq %rbx, %rdi call strchr addq $2, %rbx cmpq %rax, %rbx ja .LBB1_2 .LBB1_1: # %"" cmpb $33, -2(%rax) sete %al movzbl %al, %eax popq %rbx ret .LBB1_2: # %"" xorl %eax, %eax popq %rbx ret .size foo, .-foo .section .note.GNU-stack,"", at progbits -- 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 15 12:23:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 12:23:34 -0500 Subject: [LLVMbugs] [Bug 3239] The x86-32 JIT does not work when compiled as PIC In-Reply-To: Message-ID: <200909151723.n8FHNYXP009743@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3239 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #3 from Anton Korobeynikov 2009-09-15 12:23:34 --- *** This bug has been marked as a duplicate of bug 3801 *** -- 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 15 12:39:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 12:39:40 -0500 Subject: [LLVMbugs] [Bug 4986] New: assert in LiveIntervalAnalysis Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4986 Summary: assert in LiveIntervalAnalysis Product: new-bugs Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3514) --> (http://llvm.org/bugs/attachment.cgi?id=3514) reduction Assertion failed: (it != mi2iMap_.end() && "Invalid instruction!"), function getInstructionIndex, file /Volumes/Boot/llvm-top/llvm/include/llvm/CodeGen/LiveIntervalAnalysis.h, line 226. -- 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 15 13:21:21 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 13:21:21 -0500 Subject: [LLVMbugs] [Bug 4987] New: RUN: line parsing bug Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4987 Summary: RUN: line parsing bug Product: Test Suite Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: clattner at apple.com CC: llvmbugs at cs.uiuc.edu At least with testrunner.sh in llvm, This run line was parsed wrong: ; RUN: llc < %s -march=x86-64 -O0| FileCheck %s --check-prefix=X64 It was passing --check-prefix= and other stuff to llc. Putting a space before the | fixed it. -- 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 15 13:27:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 13:27:51 -0500 Subject: [LLVMbugs] [Bug 4984] llc -O0: incorrect emission of negative offsets In-Reply-To: Message-ID: <200909151827.n8FIRpnv018495@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4984 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-15 13:27:51 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087152.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 15 13:34:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 13:34:34 -0500 Subject: [LLVMbugs] [Bug 4929] [2.6] MultiSource/Applications/lemon JIT regression In-Reply-To: Message-ID: <200909151834.n8FIYYQf019481@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4929 Dale Johannesen changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #10 from Dale Johannesen 2009-09-15 13:34:33 --- Using separate resolving function looks a fairly invasive change and I doubt I could get it right for all situations. I've settled for changing the marker byte, which is quite safe and doesn't make things any kludgier than they were before:) http://llvm.org/viewvc/llvm-project?rev=81888&view=rev -- 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 15 21:46:07 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 21:46:07 -0500 Subject: [LLVMbugs] [Bug 4989] New: 2-element vector initialization expression doesn't work. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4989 Summary: 2-element vector initialization expression doesn't work. Product: clang Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: major Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: syoyofujita at gmail.com CC: llvmbugs at cs.uiuc.edu float2 a = (float2){val, val} expression seems doesn't work. a.x should be 1.0 but it is initialized as nan. CPU: x86 OS : Linux 32bit CMD: clang main.c Rev: 81991 // main.c #include typedef float float2 __attribute__((ext_vector_type(2))); float2 test(float2 a, float2 b) { return a + b; } int main( int argc, char **argv) { int val = 1; float2 a = (float2){val, val}; // Expects (1.0, 1.0), but gets (nan, 1.0) printf("a = %f, %f\n", a.x, a.y); 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 15 23:54:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 15 Sep 2009 23:54:47 -0500 Subject: [LLVMbugs] [Bug 4987] RUN: line parsing bug In-Reply-To: Message-ID: <200909160454.n8G4slul002780@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4987 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Daniel Dunbar 2009-09-15 23:54:43 --- No, that is how Tcl parses. The run line just happens to be arguments which the exec function interprets, but in Tcl the | doesn't have special handling. -- ddunbar at ozzy:clang (master)$ wish % exec echo a| b a| b % echo echo a | b couldn't execute "b": no such file or directory % -- -- 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 16 01:17:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:17:55 -0500 Subject: [LLVMbugs] [Bug 4932] clang: doesn't honor -fno-asynchronous-unwind-tables In-Reply-To: Message-ID: <200909160617.n8G6HteU013315@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4932 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Daniel Dunbar 2009-09-16 01:17:54 --- This is bad, but I haven't fully deciphered the logic to match gcc with -funwind-tables -fasynchronous-unwind-tables -fnon-call-execptions. I think the situation is that on Darwin we set -fasynchronous-unwind-tables by default (on x86_64), which gets us unwind-tables, unless we have -mkernel or -fapple-kext going on. The state machine of what actually causes llvm-gcc to get unwind tables seems way more complicated than necessary, but I don't really know enough about these options to know they aren't being used. Fixed in r82021. I took the small incremental improvement line, so that we always honor -funwind-tables and -fno-unwind-tables, but use -fasyncronous-unwind-tables as the default if those aren't present. -- 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 16 01:21:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:21:11 -0500 Subject: [LLVMbugs] [Bug 3034] Document clang C status In-Reply-To: Message-ID: <200909160621.n8G6LBKD013792@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3034 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #6 from Daniel Dunbar 2009-09-16 01:21:10 --- I've not done this long enough that so much stuff works its unneeded. Most known missing stuff is documented in the users manual. -- 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 16 01:23:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:23:09 -0500 Subject: [LLVMbugs] [Bug 3306] gcc compat: empty declarations are errors not warnings In-Reply-To: Message-ID: <200909160623.n8G6N9dD014077@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3306 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WONTFIX --- Comment #1 from Daniel Dunbar 2009-09-16 01:23:09 --- Closing until someone cares. -- 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 16 01:27:18 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:27:18 -0500 Subject: [LLVMbugs] [Bug 3480] clang miscompiles MultiSource/Applications/oggenc? In-Reply-To: Message-ID: <200909160627.n8G6RI0g014705@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3480 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Daniel Dunbar 2009-09-16 01:27:17 --- As far as I know, this is fixed. If not I will reopen when I get back around to checking nightlytester results on Linux. -- 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 16 01:33:54 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:33:54 -0500 Subject: [LLVMbugs] [Bug 3603] Finish dependency file support In-Reply-To: Message-ID: <200909160633.n8G6XsWS015560@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3603 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #12 from Daniel Dunbar 2009-09-16 01:33:53 --- Treating this as fixed, we are only missing -MG and -MQ, both of which are slightly icky, and neither of which anyone has asked for. -- 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 16 01:41:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:41:27 -0500 Subject: [LLVMbugs] [Bug 4910] fatal error: error in backend: Ran out of registers during register allocation! In-Reply-To: Message-ID: <200909160641.n8G6fREM016576@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4910 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #2 from Daniel Dunbar 2009-09-16 01:41:27 --- This works for me on TOT. -- 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 16 01:43:12 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 01:43:12 -0500 Subject: [LLVMbugs] [Bug 4910] fatal error: error in backend: Ran out of registers during register allocation! In-Reply-To: Message-ID: <200909160643.n8G6hCes016835@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4910 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|WORKSFORME | --- Comment #3 from Daniel Dunbar 2009-09-16 01:43:11 --- Err, no it doesn't. -- 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 16 02:10:02 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 02:10:02 -0500 Subject: [LLVMbugs] [Bug 4881] LLVM does not compile libruntime, opt problem. In-Reply-To: Message-ID: <200909160710.n8G7A2Tg020174@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4881 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Daniel Dunbar 2009-09-16 02:10:02 --- I believe this was due to a metadata bug that Devang fixed, please reopen if 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 16 02:55:29 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 02:55:29 -0500 Subject: [LLVMbugs] [Bug 4901] llvm-ld: Segmentation fault running pass ' Dead Global Elimination' In-Reply-To: Message-ID: <200909160755.n8G7tTWB001550@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4901 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #2 from Daniel Dunbar 2009-09-16 02:55:28 --- This works for me on TOT. -- 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 16 03:00:28 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 03:00:28 -0500 Subject: [LLVMbugs] [Bug 4990] New: 'unavailable' attribute documented as error but is only warning Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4990 Summary: 'unavailable' attribute documented as error but is only warning Product: clang Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Documentation AssignedTo: unassignedclangbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu http://clang.llvm.org/docs/LanguageExtensions.html void honeypot(...) __attribute__((overloadable, unavailable)); // calling me is an error However clang diagnoses it only as a warning: void honeypot(...) __attribute__((overloadable, unavailable)); // calling me is an error void foo() { honeypot(); } $ clang foo.c -c p.c:2:14: warning: 'honeypot' is unavailable [-Wunavailable-declarations] void foo() { honeypot(); } ^ p.c:1:6: note: function has been explicitly marked unavailable here void honeypot(...) __attribute__((overloadable, unavailable)); // calling me is an error ^ 2 diagnostics generated. -- 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 16 03:10:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 03:10:23 -0500 Subject: [LLVMbugs] [Bug 4991] New: __TIMESTAMP__ has trailing newline Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4991 Summary: __TIMESTAMP__ has trailing newline Product: clang Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: preprocessor AssignedTo: unassignedclangbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu __TIMESTAMP__ has a trailing newline on Linux/x86-64 (that is what ctime/ctime_r returns), which causes invalid C code to be generated (clang accepts it though). Compiling this code with clang outputs a trailing newline: #include int main() { puts(__TIMESTAMP__); } Compiling it like this works: $ clang foo.c && ./a.out Wed Sep 16 11:08:01 2009 $ However if I preprocess first, clang doesn't accept its own (!) preprocessed output: clang -E foo.c|clang -x c - In file included from foo.c:1: In file included from foo.c:1: foo.c:4:10: error: missing terminating '"' character puts("Wed Sep 16 11:08:01 2009 ^ foo.c:4:10: error: expected expression foo.c:5:1: error: missing terminating '"' character "); ^ foo.c:6:2: error: expected '}' } ^ 4 diagnostics generated. -- 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 16 08:11:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 08:11:32 -0500 Subject: [LLVMbugs] [Bug 4991] __TIMESTAMP__ has trailing newline In-Reply-To: Message-ID: <200909161311.n8GDBWIv015177@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4991 Benjamin Kramer changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |benny.kra at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Benjamin Kramer 2009-09-16 08:11:30 --- fixed in r82043. http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090914/021405.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 16 09:10:05 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 09:10:05 -0500 Subject: [LLVMbugs] [Bug 4992] New: Tutorial examples are broken Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4992 Summary: Tutorial examples are broken Product: Documentation Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: minor Priority: P2 Component: General docs AssignedTo: unassignedbugs at nondot.org ReportedBy: graham.hemingway at gmail.com CC: llvmbugs at cs.uiuc.edu I am just working my way through the tutorial (Parts 1 and 2 at http://llvm.org/docs/tutorial) using TOT, and I am having to make a number of corrections for proper compilation. For example in Part 1 there is the line: Constant* c = mod->getOrInsertFunction("mul_add", /*ret type*/ IntegerType::get(32), /*args*/ IntegerType::get(32), IntegerType::get(32), IntegerType::get(32), /*varargs terminated with null*/ NULL); This gave me a number of errors. After changing IntegerType::get(32) to IntegerType::get(getGlobalContext(), 32) things worked better. Another example is in Part 2 (A more complicated function) there are the lines: BasicBlock* entry = BasicBlock::Create(getGlobalContext(), ("entry", gcd); BasicBlock* ret = BasicBlock::Create(getGlobalContext(), ("return", gcd); BasicBlock* cond_false = BasicBlock::Create(getGlobalContext(), ("cond_false", gcd); BasicBlock* cond_true = BasicBlock::Create(getGlobalContext(), ("cond_true", gcd); BasicBlock* cond_false_2 = BasicBlock::Create(getGlobalContext(), ("cond_false", gcd); The '(' and ')' on each line don't match. I removed the '(' after each getGlobalContext() and things seemed to work better. I admit that I am very new to llvm, but I think that the tutorial is out of sync with the current state of the API. If my assessment is correct, and you are OK with it, I am happy to make the necessary changes to the tutorials as I go along. Cheers, Graham -- 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 16 09:19:42 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 09:19:42 -0500 Subject: [LLVMbugs] [Bug 4992] Tutorial examples are broken In-Reply-To: Message-ID: <200909161419.n8GEJgd4023990@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4992 Xerxes R??nby changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |xerxes at zafena.se Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Xerxes R??nby 2009-09-16 09:19:42 --- > I admit that I am very new to llvm, but I think that the tutorial is out of > sync with the current state of the API. That are correct, the tutorial on the llvm.org homepage are only in sync with llvm 2.5. The examples on the homepage will be updated when llvm 2.6 are released and will then work with llvm 2.6. You need to use the tutorial text that came with your checkout from TOT for the tutorials to work with LLVM 2.7 svn trunk. Check the examples dir and the docs dir that came with your checkout and you will find the examples up to date. Good luck! -- 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 16 10:08:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 10:08:50 -0500 Subject: [LLVMbugs] [Bug 4927] assertion "no allocatable register" breaks Apple-style llvm-gcc build In-Reply-To: Message-ID: <200909161508.n8GF8oas030261@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4927 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Bob Wilson 2009-09-16 10:08:49 --- Yes, Dan fixed 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 Wed Sep 16 10:38:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 10:38:35 -0500 Subject: [LLVMbugs] [Bug 4993] New: Assertion `Reg && "this is not a register!"' failed during linux kernel building. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4993 Summary: Assertion `Reg && "this is not a register!"' failed during linux kernel building. Product: new-bugs Version: 2.5 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: joker at ispras.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3521) --> (http://llvm.org/bugs/attachment.cgi?id=3521) Preprocessed kernel file that causes fail. I use llvm-gcc to build linux kernel of one of the latest version (2.6.30.5). llvm-gcc can't process file drivers/acpi/sbshc.c and exit with assertion `Reg && "this is not a register!"'. Preprocessed file drivers/acpi/sbshc.c is attached. To process it command llvm-gcc -c drivers_acpi_sbshc.i is used. Note that there's no such problem if you just remove one case from switch: 916c916 < return ({ typeof(per_cpu__current_task) ret__; switch (sizeof(per_cpu__current_task)) { case 1: asm("mov" "b ""%%""gs"":%P" "1"",%0" : "=q" (ret__) : "m" (per_cpu__current_task)); break; case 2: asm("mov" "w ""%%""gs"":%P" "1"",%0" : "=r" (ret__) : "m" (per_cpu__current_task)); break; case 4: asm("mov" "l ""%%""gs"":%P" "1"",%0" : "=r" (ret__) : "m" (per_cpu__current_task)); break; case 8: asm("mov" "q ""%%""gs"":%P" "1"",%0" : "=r" (ret__) : "m" (per_cpu__current_task)); break; default: __bad_percpu_size(); } ret__; }); --- > return ({ typeof(per_cpu__current_task) ret__; switch (sizeof(per_cpu__current_task)) { case 1: asm("mov" "b ""%%""gs"":%P" "1"",%0" : "=q" (ret__) : "m" (per_cpu__current_task)); break; case 2: asm("mov" "w ""%%""gs"":%P" "1"",%0" : "=r" (ret__) : "m" (per_cpu__current_task)); break; case 4: asm("mov" "l ""%%""gs"":%P" "1"",%0" : "=r" (ret__) : "m" (per_cpu__current_task)); break; default: __bad_percpu_size(); } ret__; }); So it seems that this problem is connected with this case expression. -- 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 16 11:26:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 11:26:04 -0500 Subject: [LLVMbugs] [Bug 4994] New: Bug in convertion from vararg double to int in Sparc backend Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4994 Summary: Bug in convertion from vararg double to int in Sparc backend Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: quickslyver at free.fr CC: llvmbugs at cs.uiuc.edu Test case (C): #include int foo (char *fmt, ...) { va_list ap; va_start (ap, fmt); return va_arg (ap, double); } ----------------------------------------- Test case (LLVM): ; ModuleID = 'test_bug.c' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64" target triple = "sparc-unknown-unknown" define i32 @foo(i8* nocapture %fmt, ...) nounwind { entry: %ap = alloca i8*, align 4 ; [#uses=2] %ap1 = bitcast i8** %ap to i8* ; [#uses=1] call void @llvm.va_start(i8* %ap1) %0 = va_arg i8** %ap, double ; [#uses=1] %conv = fptosi double %0 to i32 ; [#uses=1] ret i32 %conv } declare void @llvm.va_start(i8*) nounwind -------------------------------------------- llc assertion fail: llc: LegalizeDAG.cpp:762: llvm::SDValue::SelectionDAGLegalize::LegalizeOp(llvm::SDValue): Assertion `(isTypeLegal(Node->getOperand(i).getValueType()) || Node->getOperand(i).getOpcode() == ISD::TargetConstant) && "Unexpected illegal type!"' failed. 0 llc 0x0907c98d 1 llc 0x0907cf59 2 0xb7f7a400 __kernel_sigreturn + 0 3 libc.so.6 0xb7cd2a85 abort + 389 4 libc.so.6 0xb7cc84ee __assert_fail + 238 5 llc 0x08c941bf 6 llc 0x08c9481c 7 llc 0x08c94f35 8 llc 0x08ca18ac 9 llc 0x08ca19ca llvm::SelectionDAG::Legalize(bool, llvm::CodeGenOpt::Level) + 48 10 llc 0x08c6a4ef llvm::SelectionDAGISel::CodeGenAndEmitDAG() + 3151 11 llc 0x08c6c57b llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::ilist_iterator, llvm::ilist_iterator) + 487 12 llc 0x08c6d036 llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 2468 13 llc 0x08c6deb2 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 886 14 llc 0x08d5f366 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 86 15 llc 0x08ff32dc llvm::FPPassManager::runOnFunction(llvm::Function&) + 306 16 llc 0x08ff3de0 llvm::FunctionPassManagerImpl::run(llvm::Function&) + 124 17 llc 0x08ff3fad llvm::FunctionPassManager::run(llvm::Function&) + 135 18 llc 0x085004ea main + 3466 19 libc.so.6 0xb7cba7a5 __libc_start_main + 229 20 llc 0x084fe721 Stack dump: 1. Running pass 'SPARC DAG->DAG Pattern Instruction Selection' on function '@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 Wed Sep 16 11:45:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 11:45:36 -0500 Subject: [LLVMbugs] [Bug 4993] Assertion `Reg && "this is not a register!"' failed during linux kernel building. In-Reply-To: Message-ID: <200909161645.n8GGjaVZ010719@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4993 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #3 from Anton Korobeynikov 2009-09-16 11:45:35 --- Works for me with LLVM TOT. -- 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 16 16:22:28 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 16:22:28 -0500 Subject: [LLVMbugs] [Bug 4972] [2.6] llvm-gcc fails to compile In-Reply-To: Message-ID: <200909162122.n8GLMSZw014460@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4972 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #20 from Chris Lattner 2009-09-16 16:22:27 --- great, fixed in 2.6 -- 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 16 19:57:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 19:57:40 -0500 Subject: [LLVMbugs] [Bug 4910] fatal error: error in backend: Ran out of registers during register allocation! In-Reply-To: Message-ID: <200909170057.n8H0veb2013176@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4910 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #10 from Evan Cheng 2009-09-16 19:57:39 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087299.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 16 19:59:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 19:59:57 -0500 Subject: [LLVMbugs] [Bug 4733] Ran out of registers during register allocation! In-Reply-To: Message-ID: <200909170059.n8H0xvCJ013554@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4733 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |evan.cheng at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Evan Cheng 2009-09-16 19:59:56 --- I believe this is fixed. Please verify. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087299.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 16 20:21:24 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 16 Sep 2009 20:21:24 -0500 Subject: [LLVMbugs] [Bug 4971] Dumping bitcode or LLVM assembler code segfaults. In-Reply-To: Message-ID: <200909170121.n8H1LOll016249@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4971 ??scar Fuentes changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #4 from ??scar Fuentes 2009-09-16 20:21:23 --- After working for a while on this, found that the problem happens on x86_64-linux but not on x86-mingw, that is related to opaque types used by function types, which ends being invalid or dangling pointers. When an opaque type is resolved and a function type that references it ends being a type that already exists, the problem arises. I tried to replicate the problem outside of my compiler with no luck. As this may be a bug on my side, I'll close the 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 Thu Sep 17 00:53:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 00:53:35 -0500 Subject: [LLVMbugs] [Bug 4982] -regalloc=simple creates invalid code crashing fp stackifier In-Reply-To: Message-ID: <200909170553.n8H5rZvW017736@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4982 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Evan Cheng 2009-09-17 00:53:34 --- Done. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087304.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 17 11:23:42 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 11:23:42 -0500 Subject: [LLVMbugs] [Bug 4999] New: llvm.eh.sjlj.longjmp is undocumented Message-ID: http://llvm.org/bugs/show_bug.cgi?id=4999 Summary: llvm.eh.sjlj.longjmp is undocumented Product: Documentation Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: General docs AssignedTo: unassignedbugs at nondot.org ReportedBy: idadesub at users.sourceforge.net CC: llvmbugs at cs.uiuc.edu In docs/ExceptionHandling.html, it looks like the intrinsic llvm.eh.sjlj.longjmp isn't documented. I assume it also follows the c longjmp implementation? -- 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 17 11:28:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 11:28:41 -0500 Subject: [LLVMbugs] [Bug 5000] New: [2.6] clang FileManager.cpp warnings Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5000 Summary: [2.6] clang FileManager.cpp warnings Product: new-bugs Version: 2.6 Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu Blocks: 4886 This is a release 64 bit build of clang. gcc 4.2 on a leopard machine. FileManager.cpp: In member function ???virtual int clang::MemorizeStatCalls::stat(const char*, stat*)???: FileManager.cpp:296: warning: ???empty$st_qspare$1??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_qspare$0??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_lspare??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_gen??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_flags??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_blksize??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_blocks??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_size??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_ctimespec$tv_nsec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_ctimespec$tv_sec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_mtimespec$tv_nsec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_mtimespec$tv_sec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_atimespec$tv_nsec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_atimespec$tv_sec??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_rdev??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_gid??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_uid??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_nlink??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_mode??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_ino??? is used uninitialized in this function FileManager.cpp:296: warning: ???empty$st_dev??? is used uninitialized in this function -- 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 17 11:32:19 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 11:32:19 -0500 Subject: [LLVMbugs] [Bug 5001] New: [2.6] test/CodeGen/X86/combiner-aa-1.ll fails Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5001 Summary: [2.6] test/CodeGen/X86/combiner-aa-1.ll fails Product: new-bugs Version: 2.6 Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu Blocks: 4886 debug or release: FAIL: /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/combiner-aa-1.ll for PR4880 Failed with exit(1) at line 1 while running: llc < /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/combiner-aa-1.ll --combiner-alias-analysis --combiner-global-alias-analysis llc: bitcode didn't read correctly. Reason: Bitcode stream should be a multiple of 4 bytes in length -- 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 17 11:34:10 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 11:34:10 -0500 Subject: [LLVMbugs] [Bug 5002] New: [2.6] test/CodeGen/X86/cmov-zext.ll fails Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5002 Summary: [2.6] test/CodeGen/X86/cmov-zext.ll fails Product: new-bugs Version: 2.6 Platform: Macintosh OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org CC: llvmbugs at cs.uiuc.edu Blocks: 4886 There are actually 2 tests failing that look similar, but if not, please separate into 2 bugs. FAIL: /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/cmov-zext.ll Failed with exit(1) at line 1 while running: llc < /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/cmov-zext.ll -march=x86-64 | FileCheck /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/cmov-zext.ll llc: bitcode didn't read correctly. Reason: Invalid bitcode signature /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/CodeGen/X86/cmov-zext.ll:7:15: error: expected string not found in input ; CHECK: cmovne %edi, %esi ^ :1:1: note: scanning from here FAIL: /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/Transforms/MemCpyOpt/align.ll Failed with exit(1) at line 1 while running: opt < /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/Transforms/MemCpyOpt/align.ll -S -memcpyopt | FileCheck /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/Transforms/MemCpyOpt/align.ll opt: Unknown command line argument '-S'. Try: 'opt --help' /Users/tonic/release/2.6/pre-release2/debug/llvm-2.6/test/Transforms/MemCpyOpt/align.ll:6:10: error: expected string not found in input ; CHECK: call void @llvm.memset.i64(i8* %a01, i8 0, i64 16, i32 4) ^ :1:1: note: scanning from here -- 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 17 12:58:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 12:58:34 -0500 Subject: [LLVMbugs] [Bug 4439] Cannot yet select error (llvm from subversion) In-Reply-To: Message-ID: <200909171758.n8HHwYXp027387@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4439 Ronald changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Ronald 2009-09-17 12:58:34 --- seems fixed after svn up 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 Thu Sep 17 13:15:10 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 13:15:10 -0500 Subject: [LLVMbugs] [Bug 4959] update llvm-gcc OpenBSD target configurations In-Reply-To: Message-ID: <200909171815.n8HIFACE029532@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4959 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Duncan Sands 2009-09-17 13:15:09 --- Thanks for the patch, applied here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087322.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 17 17:21:49 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 17:21:49 -0500 Subject: [LLVMbugs] [Bug 5003] New: JITResolver:: getGlobalValueIndirectSym may return incorrect value Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5003 Summary: JITResolver::getGlobalValueIndirectSym may return incorrect value Product: libraries Version: trunk Platform: PC OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Target-Independent JIT AssignedTo: unassignedbugs at nondot.org ReportedBy: pgurd at rapidmind.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3527) --> (http://llvm.org/bugs/attachment.cgi?id=3527) Patch to clear the Global Value Indirect Symbol map at the end of finishFunctiion In lib/ExecutionEngine/JIT/JIREmitter.cpp: The JITResolverState class has a GlobalToIndirectSymMap, which maps a GlobalValue pointer to its corresponding indirect symbol location. The GlobalToIndirectSymMap persistently records maps of GlobalValues to indirect symbol locations for all functions seen. That is. it seems that it is never cleared. The function JITResolver::getGlobalIndirectSym looks up its argument GlobalValue* in this map. If a GlobalValue* in a function was allocated with new() and later released when processing of that function is complete, then it is possible that the space so made available could be re-used later for a completely different GlobalValue* in a different function. When that happens (as it did for us!), the line of code void *&IndirectSym = state.getGlobalToIndirectSymMap(locked)[GV]; in getGlobalIndirectSym will find and return the long gone first use of that pointer value, rather than creating a new entry in the map and returning it, resulting in incorrect JIT code. This was noticed in x86_64 Mac OS X code, which needs to use the indirect symbol map to load addresses, persumably because there is no 64 bit immediate load instruction. We worked around this issue by adding code at the end of the finishFunction function to clear the map. I am guessing that this may not be the optimal way to fix this. In any case, I have attached the patch, in case it might be of any help. -- 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 17 19:48:21 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 19:48:21 -0500 Subject: [LLVMbugs] [Bug 5004] New: Bootstrap failure "Error: can not do 8 byte pc-relative relocation" Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5004 Summary: Bootstrap failure "Error: can not do 8 byte pc-relative relocation" Product: tools Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: pawel.worach at gmail.com CC: llvmbugs at cs.uiuc.edu Using llvm r82173 and gcc 4.2.1 20070719 as the host compiler the bootstrap of llvm-gcc r82176 fails to bootstrap on FreeBSD/amd64 8.0 /usr/src-local/llvm-gcc/obj/./gcc/xgcc -shared-libgcc -B/usr/src-local/llvm-gcc/obj/./gcc -nostdinc++ -L/usr/src-local/llvm-gcc/obj/x86_64-unknown-freebsd8.0/libstdc++-v3/src -L/usr/src-local/llvm-gcc/obj/x86_64-unknown-freebsd8.0/libstdc++-v3/src/.libs -B/opt/llvm-gcc/x86_64-unknown-freebsd8.0/bin/ -B/opt/llvm-gcc/x86_64-unknown-freebsd8.0/lib/ -isystem /opt/llvm-gcc/x86_64-unknown-freebsd8.0/include -isystem /opt/llvm-gcc/x86_64-unknown-freebsd8.0/sys-include -I/usr/src-local/llvm-gcc/libstdc++-v3/../gcc -I/usr/src-local/llvm-gcc/obj/x86_64-unknown-freebsd8.0/libstdc++-v3/include/x86_64-unknown-freebsd8.0 -I/usr/src-local/llvm-gcc/obj/x86_64-unknown-freebsd8.0/libstdc++-v3/include -I/usr/src-local/llvm-gcc/libstdc++-v3/libsupc++ -fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -O2 -c ../../../../libstdc++-v3/libsupc++/eh_alloc.cc -o eh_alloc.o /var/tmp//ccEBSLfL.s: Assembler messages: /var/tmp//ccEBSLfL.s:757: Error: can not do 8 byte pc-relative relocation /var/tmp//ccEBSLfL.s:772: Error: can not do 8 byte pc-relative relocation /var/tmp//ccEBSLfL.s:793: Error: can not do 8 byte pc-relative relocation gmake[4]: *** [eh_alloc.lo] Error 1 gmake[4]: Leaving directory `/usr/src-local/llvm-gcc/obj/x86_64-unknown-freebsd8.0/libstdc++-v3/libsupc++' -- 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 17 20:44:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 20:44:15 -0500 Subject: [LLVMbugs] [Bug 5005] New: Win64 stack corruption when spilling callee saved XMM registers. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5005 Summary: Win64 stack corruption when spilling callee saved XMM registers. Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: llvm at laurentm.net CC: llvmbugs at cs.uiuc.edu If callee saved XMM registers get spilled/restored in the prologue/epilogue, the instruction walker that skips the push/pop instructions will fail to recognize the movaps/movups instruction causing the stack update to be incorrectly placed. In X86RegisterInfo::emitPrologue: while (MBBI != MBB.end() && (MBBI->getOpcode() == X86::PUSH32r || MBBI->getOpcode() == X86::PUSH64r)) { PushedRegs = true; ++MBBI; This code should probably skip X86::MOVAPSmr instructions. Failing to recognize XMM saves/restores will produce the following incorrect code: pushq %rbp movq %rsp, %rbp subq $328, %rsp # <=== Incorrect position movups %xmm7, -80(%rbp) movups %xmm6, -64(%rbp) pushq %r15 pushq %r14 pushq %rsi pushq %rdi pushq %rbx The correct code should be: pushq %rbp movq %rsp, %rbp movups %xmm7, -80(%rbp) movups %xmm6, -64(%rbp) pushq %r15 pushq %r14 pushq %rsi pushq %rdi pushq %rbx subq $328, %rsp # <=== OK -- 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 17 23:26:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 23:26:45 -0500 Subject: [LLVMbugs] [Bug 5001] [2.6] test/CodeGen/X86/combiner-aa-1.ll fails In-Reply-To: Message-ID: <200909180426.n8I4QjOh010027@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5001 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Tanya Lattner 2009-09-17 23:26:45 --- Thanks Dan! -- 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 17 23:26:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 23:26:57 -0500 Subject: [LLVMbugs] [Bug 5002] [2.6] test/CodeGen/X86/cmov-zext.ll fails In-Reply-To: Message-ID: <200909180426.n8I4Qv8w010081@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5002 Tanya Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Tanya Lattner 2009-09-17 23:26:57 --- Thanks Dan! -- 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 17 23:51:28 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 17 Sep 2009 23:51:28 -0500 Subject: [LLVMbugs] [Bug 5000] [2.6] clang FileManager.cpp warnings In-Reply-To: Message-ID: <200909180451.n8I4pS9P013185@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5000 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Chris Lattner 2009-09-17 23:51:28 --- This should fix it: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090914/021485.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 18 11:00:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 18 Sep 2009 11:00:55 -0500 Subject: [LLVMbugs] [Bug 4217] Eliminate the Annotable and Annotation classes by making MachineFunction a pass In-Reply-To: Message-ID: <200909181600.n8IG0tPA014686@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4217 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Dan Gohman 2009-09-18 11:00:55 --- This as implemented 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 Fri Sep 18 12:07:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 18 Sep 2009 12:07:22 -0500 Subject: [LLVMbugs] [Bug 4776] msp430 backend: placing a variable at a specific address doesn' t seem to work In-Reply-To: Message-ID: <200909181707.n8IH7MVN023077@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4776 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #10 from Anton Korobeynikov 2009-09-18 12:07:21 --- Applied in http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087371.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 18 16:21:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 18 Sep 2009 16:21:39 -0500 Subject: [LLVMbugs] [Bug 5006] New: clang-cc -ast-print-xml get a Assertion failed error when operating on objective-C headers Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5006 Summary: clang-cc -ast-print-xml get a Assertion failed error when operating on objective-C headers Product: clang Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedclangbugs at nondot.org ReportedBy: kremenek at apple.com CC: llvmbugs at cs.uiuc.edu When using clang-cc "-ast-print-xml" argument, I experience an Assertion Failed error (see below). This is for llvm revision 81368 Here is the command I used on the Foundation.h header: ./clang-cc -ast-print-xml --mcpu=yonah -v -o /tmp/output.xml -x=objective-c /System/Library/Frameworks/Foundation.framework/Headers/Foundation.h This also is seen in other objective-C headers such as AppKit.h, SecurityFoundation.h, etc. Related: -- 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 19 03:14:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 03:14:43 -0500 Subject: [LLVMbugs] [Bug 5007] New: llvm-g++ -flto generates invalid object files Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5007 Summary: llvm-g++ -flto generates invalid object files Product: tools Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-g++ AssignedTo: unassignedbugs at nondot.org ReportedBy: bero at arklinux.org CC: llvmbugs at cs.uiuc.edu In current llvm and llvm-gcc 2.6 branches, llvm-g++ -flto generates invalid object files: $ cat test.cpp #include using namespace std; int main(int argc, char **argv) { cerr << "Microsoft sucks!" << endl; } $ llvm-g++ -flto test.cpp /tmp/ccboFcxx.o: file not recognized: File format not recognized $ llvm-g++ -flto -c test.cpp -o test.o $ file test.o test.o: data Without -flto, it works as expected. The size of test.o with -flto is almost twice that of test.o without -flto. -- 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 19 03:25:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 03:25:16 -0500 Subject: [LLVMbugs] [Bug 5007] llvm-g++ -flto generates invalid object files In-Reply-To: Message-ID: <200909190825.n8J8PGOF028222@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5007 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Duncan Sands 2009-09-19 03:25:15 --- With llvm-gcc, -flto produces LLVM bitcode, not object files. In order for this to work you need a linker that understands bitcode, see http://llvm.org/docs/GoldPlugin.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 19 04:53:31 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 04:53:31 -0500 Subject: [LLVMbugs] [Bug 4958] Buggy sdisel phi update code In-Reply-To: Message-ID: <200909190953.n8J9rVNt008055@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4958 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #15 from Evan Cheng 2009-09-19 04:53:27 --- It's nasty alright. Apparently codegen is filled with ugly and horrible code. Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087404.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087430.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 19 06:43:52 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 06:43:52 -0500 Subject: [LLVMbugs] [Bug 5008] New: linking with llvm-gcc results in "/usr/bin/ld: error: gcc: malformed archive header name at 1046" Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5008 Summary: linking with llvm-gcc results in "/usr/bin/ld: error: gcc: malformed archive header name at 1046" Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: bero at arklinux.org CC: llvmbugs at cs.uiuc.edu $ cat test.c #include int main(int argc, char **argv) { puts("test"); } $ llvm-gcc -o test.o test.c $ llvm-gcc -o test test.o /usr/bin/ld: error: gcc: malformed archive header name at 1046 /usr/bin/ld: error: gcc: malformed archive header name at 1046 collect2: ld returned 1 exit status $ llvm-gcc -use-gold-plugin -o test test.o /usr/bin/ld: error: gcc: malformed archive header name at 1046 /usr/bin/ld: error: gcc: malformed archive header name at 1046 collect2: ld returned 1 exit status $ gcc -o test test.o # THIS WORKS $ ld --version |head -n1 GNU gold (Linux/GNU Binutils 2.20.51.0.1.20090905) 1.9 $ llvm-gcc --version |head -n1 llvm-gcc (GCC) 4.2.1 (Based on Apple Inc. build 5649) (LLVM build 2.6) $ gcc --version |head -n1 gcc (GCC) 4.4.2 20090912 (prerelease) llvm and llvm-gcc are both 2.6 branch checkouts from today (llvm rev. 82192, llvm-gcc rev. 81414) -- 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 19 07:13:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 07:13:37 -0500 Subject: [LLVMbugs] [Bug 5008] linking with llvm-gcc results in "/usr/bin/ld: error: gcc: malformed archive header name at 1046" In-Reply-To: Message-ID: <200909191213.n8JCDbgF025716@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5008 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |INVALID --- Comment #4 from Duncan Sands 2009-09-19 07:13:36 --- $ llvm-gcc -o test.o test.c You forgot the -c in this line. -- 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 19 11:59:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 11:59:32 -0500 Subject: [LLVMbugs] [Bug 5008] linking with llvm-gcc results in "/usr/bin/ld: error: gcc: malformed archive header name at 1046" In-Reply-To: Message-ID: <200909191659.n8JGxWFs030061@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5008 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|INVALID | --- Comment #6 from Duncan Sands 2009-09-19 11:59:31 --- Reopening because the invalid manipulation in the report was a mistake in how the reporter reported the problem - there is a real problem here, apparently. -- 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 19 16:52:36 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 16:52:36 -0500 Subject: [LLVMbugs] [Bug 5009] New: inliner missing opportunity with CPS style code Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5009 Summary: inliner missing opportunity with CPS style code Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: idadesub at users.sourceforge.net CC: llvmbugs at cs.uiuc.edu I'm experimenting with the explicitly managed stack frames from http://nondot.org/sabre/LLVMNotes/ExplicitlyManagedStackFrames.txt, and I'm running into a simple case where the inliner appears to be missing an opportunity. Here's the CPS style code: /////////////////////////////////// #include struct cont_t { void (*f)(void*, int); void* sf; }; static void bar(struct cont_t* c, int y) { return c->f(c->sf, y+5); } struct foo_sf_t { struct cont_t* c; int x; }; static void foo2(struct foo_sf_t* sf, int y) { return sf->c->f(sf->c, sf->x * y); } static void foo(struct cont_t* c) { struct foo_sf_t sf = { c, 2 }; struct cont_t next = { (void (*)(void*, int))foo2, &sf }; return bar(&next, 14); } static void quit(struct cont_t* cont, int rcode) { return exit(rcode); } int main() { struct cont_t cont = { cont.f = (void (*)(void*, int))quit, NULL }; foo(&cont); return 0; } /////////////////////////////////// compiled with clang -O3, it generates the following code. You'll notice that it wasn't able to inline the function foo2: /////////////////////////////////// ; ModuleID = '' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" target triple = "i386-apple-darwin9.0" %struct.cont_t = type { void (i8*, i32)*, i8* } %struct.foo_sf_t = type { %struct.cont_t*, i32 } define i32 @main() nounwind { entry: %sf.i = alloca %struct.foo_sf_t, align 4 ; <%struct.foo_sf_t*> [#uses=3] %cont = alloca %struct.cont_t, align 4 ; <%struct.cont_t*> [#uses=3] %tmp = getelementptr inbounds %struct.cont_t* %cont, i32 0, i32 0 ; [#uses=1] store void (i8*, i32)* bitcast (void (%struct.cont_t*, i32)* @quit to void (i8*, i32)*), void (i8*, i32)** %tmp %tmp1 = getelementptr inbounds %struct.cont_t* %cont, i32 0, i32 1 ; [#uses=1] store i8* null, i8** %tmp1 %tmp.i = getelementptr inbounds %struct.foo_sf_t* %sf.i, i32 0, i32 0 ; <%struct.cont_t**> [#uses=1] store %struct.cont_t* %cont, %struct.cont_t** %tmp.i %tmp2.i = getelementptr inbounds %struct.foo_sf_t* %sf.i, i32 0, i32 1 ; [#uses=1] store i32 2, i32* %tmp2.i call void @foo2(%struct.foo_sf_t* %sf.i, i32 19) nounwind ret i32 0 } define internal void @quit(%struct.cont_t* nocapture %cont, i32 %rcode) nounwind { entry: tail call void @exit(i32 %rcode) nounwind ret void } define internal void @foo2(%struct.foo_sf_t* nocapture %sf, i32 %y) nounwind { entry: %tmp1 = getelementptr inbounds %struct.foo_sf_t* %sf, i32 0, i32 0 ; <%struct.cont_t**> [#uses=1] %tmp2 = load %struct.cont_t** %tmp1 ; <%struct.cont_t*> [#uses=2] %tmp3 = getelementptr inbounds %struct.cont_t* %tmp2, i32 0, i32 0 ; [#uses=1] %tmp4 = load void (i8*, i32)** %tmp3 ; [#uses=1] %conv = bitcast %struct.cont_t* %tmp2 to i8* ; [#uses=1] %tmp9 = getelementptr inbounds %struct.foo_sf_t* %sf, i32 0, i32 1 ; [#uses=1] %tmp10 = load i32* %tmp9 ; [#uses=1] %mul = mul i32 %tmp10, %y ; [#uses=1] tail call void %tmp4(i8* %conv, i32 %mul) nounwind ret void } declare void @exit(i32) /////////////////////////////////// However, if you change the function foo to manually inline bar as the following, the inliner then can fully evaluate the function: /////////////////////////////////// static void foo(struct cont_t* c) { struct foo_sf_t sf = { c, 2 }; struct cont_t next = { (void (*)(void*, int))foo2, &sf }; struct cont_t* nextp = &next; int y = 14; return nextp->f(nextp->sf, y+5); } /////////////////////////////////// /////////////////////////////////// ; ModuleID = '' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" target triple = "i386-apple-darwin9.0" %struct.cont_t = type { void (i8*, i32)*, i8* } define i32 @main() nounwind { entry: %cont = alloca %struct.cont_t, align 4 ; <%struct.cont_t*> [#uses=3] %tmp = getelementptr inbounds %struct.cont_t* %cont, i32 0, i32 0 ; [#uses=1] store void (i8*, i32)* bitcast (void (%struct.cont_t*, i32)* @quit to void (i8*, i32)*), void (i8*, i32)** %tmp %tmp1 = getelementptr inbounds %struct.cont_t* %cont, i32 0, i32 1 ; [#uses=1] store i8* null, i8** %tmp1 call void @quit(%struct.cont_t* %cont, i32 38) nounwind ret i32 0 } define internal void @quit(%struct.cont_t* nocapture %cont, i32 %rcode) nounwind { entry: tail call void @exit(i32 %rcode) nounwind ret void } declare void @exit(i32) /////////////////////////////////// -- 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 19 18:44:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 18:44:00 -0500 Subject: [LLVMbugs] [Bug 5010] New: llvm-gcc fails to build ffmpeg Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5010 Summary: llvm-gcc fails to build ffmpeg Product: new-bugs Version: 2.6 Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: bero at arklinux.org CC: llvmbugs at cs.uiuc.edu llvm-gcc -DHAVE_AV_CONFIG_H -I. -I"/usr/src/ark/BUILD/ffmpeg" -D_ISOC99_SOURCE -D_POSIX_C_SOURCE=200112 -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -m32 -O2 -march=i686 -fomit-frame-pointer -fweb -frename-registers -mmmx -g -fomit-frame-pointer -ffast-math --std=gnu99 -O2 -std=c99 -pthread -I/usr/include/dirac -g -Wdeclaration-after-statement -Wall -Wno-switch -Wdisabled-optimization -Wpointer-arith -Wredundant-decls -Wno-pointer-sign -Wcast-qual -Wwrite-strings -Wundef -O3 -fno-math-errno -fno-tree-vectorize -MMD -MF libavcodec/x86/dsputilenc_mmx.d -MT libavcodec/x86/dsputilenc_mmx.o -c -o libavcodec/x86/dsputilenc_mmx.o libavcodec/x86/dsputilenc_mmx.c cc1: error in backend: Ran out of registers during register allocation! Please check your inline asm statement for invalid constraints: INLINEASM , 10, %EAX, 9, %reg1348, 9, %reg1301, 9, %reg1345, 9, %reg1343, 9, %reg1346, 2147483657, %EAX, 14, %EFLAGS [dbg: /usr/src/ark/BUILD/ffmpeg/libavcodec/x86/mpegvideo_mmx_template.c,152,0] gcc 4.4.x compiles the code without problems. -- 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 19 19:49:18 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 19 Sep 2009 19:49:18 -0500 Subject: [LLVMbugs] [Bug 5011] New: Binaries generated by llvm-gcc break rpm debugedit Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5011 Summary: Binaries generated by llvm-gcc break rpm debugedit Product: new-bugs Version: 2.6 Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: bero at arklinux.org CC: llvmbugs at cs.uiuc.edu rpm debugedit is a tool that extracts debug info from binaries, allowing packages to include small, stripped binaries while still providing debug symbols etc. to those who need it, in a separate file. When using it on anything built with llvm-gcc (including stuff it works on when it was built with gcc, using identical makefiles), it complains that /usr/lib/rpm/bin/debugedit: canonicalization unexpectedly shrank by one character This is usually caused by an extra / in a path name; the typical way to trigger it would be mkdir src cat > src/foo.c < http://llvm.org/bugs/show_bug.cgi?id=5012 Summary: autoconf won't build 64-bit or universal properly on Snow Leopard Product: Build scripts Version: 2.6 Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: autoconf AssignedTo: unassignedbugs at nondot.org ReportedBy: ewmailing at gmail.com CC: llvmbugs at cs.uiuc.edu Using the autoconf scripts, I was trying to build llvm, clang, and llvm-gcc on Snow Leopard. By default, it seems that everything gets built as i386 instead of x86_64 on my 64-bit Mac. Since Snow Leopard's default is now 64-bit, this is not as useful for me. I tried setting the environment C*FLAGS and LDFLAGS to use -arch i386 -arch x86_64, but part way through the build, I got errors saying the system couldn't handle multiple architectures. I eventually just tried -arch x86_64. I think the llvm core built 64-bit, but I had problems with Clang and llvm-gcc. These latter components seemed to be trying to build 32-bit and failing. I am pretty sure llvm-gcc was ignoring all my C*FLAGS environment. I don't understand what Clang was doing. I eventually gave up and tried the CMake system. The CMake system worked perfectly for llvm and Clang, but llvm-gcc doesn't have a CMake description and I couldn't the llvm-gcc autoconf system to cooperate with the CMake build. Here is a little script of how I tried to build everything. #!/bin/sh CFLAGS="-isysroot /Developer/SDKs/MacOSX10.6.sdk -arch x86_64" CCFLAGS="-isysroot /Developer/SDKs/MacOSX10.6.sdk -arch x86_64" CXXFLAGS="-isysroot /Developer/SDKs/MacOSX10.6.sdk -arch x86_64" CPPFLAGS="-isysroot /Developer/SDKs/MacOSX10.6.sdk -arch x86_64" LDFLAGS="-arch x86_64" INSTALL_PATH=/opt/local LLVMDIR=llvm_release_26 LLVMGCCDIR=llvm-gcc-4.2_release_26 LLVMOBJDIR=`pwd`/build_llvm EXTRALANGS=,objc,obj-c++ TARGETOPTIONS='--with-arch=nocona --with-tune=generic' TRIPLE=i686-apple-darwin10 BUILDOPTIONS=LLVM_VERSION_INFO=2.6 mkdir -p build_llvm cd build_llvm ../$LLVMDIR/configure --prefix=$INSTALL_PATH --enable-optimized --enable-targets=all --with--llvmgccdir=/Users/ewing/Source/SVN/LLVM_RELEASE26/llvm-gcc-4.2_release_26/ make -j2 cd .. mkdir -p build_llvm-gcc cd build_llvm-gcc ../$LLVMGCCDIR/configure --enable-optimized --prefix=$INSTALL_PATH --program-prefix=llvm- --enable-llvm=$LLVMOBJDIR --enable-languages=c,c++$EXTRALANGS $TARGETOPTIONS --with-gxx-include-dir=/usr/include/c++/4.2.1 --build=$TRIPLE --host=$TRIPLE --target=$TRIPLE make ENABLE_OPTIMIZED=1 $BUILDOPTIONS -j2 make install cd .. ln -sf /usr/lib/libstdc++.6.dylib $INSTALL_PATH/lib ln -sf /usr/lib/libstdc++.6.dylib $INSTALL_PATH/lib/libstdc++.dylib -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Sun Sep 20 05:30:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 20 Sep 2009 05:30:56 -0500 Subject: [LLVMbugs] [Bug 5013] New: Can't build llvm-gcc from CMake build Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5013 Summary: Can't build llvm-gcc from CMake build Product: Build scripts Version: 2.6 Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: autoconf AssignedTo: unassignedbugs at nondot.org ReportedBy: ewmailing at gmail.com CC: llvmbugs at cs.uiuc.edu I built llvm on Snow Leopard using CMake. (The autoconf didn't work for me: bug 5012.) I am now trying to build llvm-gcc. As I understand it, I am required to pass the build directory of llvm using --enable-llvm=$LLVMOBJDIR I pointed it to my CMake build directory, but the build fails. gcc -c -DHAVE_CONFIG_H -g -O2 -I. -I../../../llvm-gcc-4.2_release_26/libiberty/../include -W -Wall -pedantic -Wwrite-strings -Wstrict-prototypes -Wc++-compat ../../../llvm-gcc-4.2_release_26/libiberty/pex-unix.c -o pex-unix.o rm -f ./libiberty.a pic/./libiberty.a ar rc ./libiberty.a \ ./regex.o ./cplus-dem.o ./cp-demangle.o ./md5.o ./alloca.o ./argv.o ./choose-temp.o ./concat.o ./cp-demint.o ./dyn-string.o ./fdmatch.o ./fibheap.o ./floatformat.o ./fnmatch.o ./fopen_unlocked.o ./getopt.o ./getopt1.o ./getpwd.o ./getruntime.o ./hashtab.o ./hex.o ./lbasename.o ./lrealpath.o ./make-relative-prefix.o ./make-temp-file.o ./objalloc.o ./obstack.o ./partition.o ./pexecute.o ./physmem.o ./pex-common.o ./pex-one.o ./pex-unix.o ./safe-ctype.o ./sort.o ./spaces.o ./splay-tree.o ./strerror.o ./strsignal.o ./ternary.o ./unlink-if-ordinary.o ./xatexit.o ./xexit.o ./xmalloc.o ./xmemdup.o ./xstrdup.o ./xstrerror.o ./xstrndup.o ./mempcpy.o ./stpncpy.o ./strndup.o ./strverscmp.o ranlib ./libiberty.a if [ x"" != x ]; then \ cd pic; \ ar rc ./libiberty.a \ ./regex.o ./cplus-dem.o ./cp-demangle.o ./md5.o ./alloca.o ./argv.o ./choose-temp.o ./concat.o ./cp-demint.o ./dyn-string.o ./fdmatch.o ./fibheap.o ./floatformat.o ./fnmatch.o ./fopen_unlocked.o ./getopt.o ./getopt1.o ./getpwd.o ./getruntime.o ./hashtab.o ./hex.o ./lbasename.o ./lrealpath.o ./make-relative-prefix.o ./make-temp-file.o ./objalloc.o ./obstack.o ./partition.o ./pexecute.o ./physmem.o ./pex-common.o ./pex-one.o ./pex-unix.o ./safe-ctype.o ./sort.o ./spaces.o ./splay-tree.o ./strerror.o ./strsignal.o ./ternary.o ./unlink-if-ordinary.o ./xatexit.o ./xexit.o ./xmalloc.o ./xmemdup.o ./xstrdup.o ./xstrerror.o ./xstrndup.o ./mempcpy.o ./stpncpy.o ./strndup.o ./strverscmp.o; \ ranlib ./libiberty.a; \ cd ..; \ else true; fi make[1]: *** [stage1-bubble] Error 2 make: *** [all] Error 2 /bin/sh ../llvm-gcc-4.2_release_26/mkinstalldirs /opt/local /opt/local /bin/sh: line 0: cd: ./fixincludes: No such file or directory make[1]: *** [install-fixincludes] Error 1 make: *** [install] Error 2 -- 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 20 16:36:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 20 Sep 2009 16:36:39 -0500 Subject: [LLVMbugs] [Bug 5012] autoconf won' t build 64-bit or universal properly on Snow Leopard In-Reply-To: Message-ID: <200909202136.n8KLadGM025921@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5012 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |daniel at zuster.org Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #1 from Daniel Dunbar 2009-09-20 16:36:38 --- Things should work fine if you use --build=x86_64-apple-darwin10 --host=x86_64-apple-darwin10 --target=x86_64-apple-darwin10 when configuring, this works fine for me, and my cadre of buildbots. Note that top-of-tree has also been switch to default to x86_64 on SnowLeopard. -- 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 20 19:37:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 20 Sep 2009 19:37:32 -0500 Subject: [LLVMbugs] [Bug 5016] New: Assertion failed: (castIsValid(getOpcode(), S, Ty) && " Illegal BitCast") Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5016 Summary: Assertion failed: (castIsValid(getOpcode(), S, Ty) && "Illegal BitCast") Product: new-bugs Version: trunk Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: pawel.worach at gmail.com CC: llvmbugs at cs.uiuc.edu Blocks: 3696 Created an attachment (id=3536) --> (http://llvm.org/bugs/attachment.cgi?id=3536) bugpoint-reduced-simplified.bc Assertion when building kern_sig.c from the FreeBSD kernel. LLVM r82419. # opt bugpoint-reduced-simplified.bc -gvn WARNING: You're attempting to print out a bitcode file. This is inadvisable as it may cause display problems. If you REALLY want to taste LLVM bitcode first-hand, you can force output with the `-f' option. Assertion failed: (castIsValid(getOpcode(), S, Ty) && "Illegal BitCast"), function BitCastInst, file Instructions.cpp, line 2605. Stack dump: 0. Running pass 'Function Pass Manager' on module 'bugpoint-reduced-simplified.bc'. 1. Running pass 'Global Value Numbering' on function '@kern_sigtimedwait' Abort (core dumped) -- 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 20 23:32:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 20 Sep 2009 23:32:50 -0500 Subject: [LLVMbugs] [Bug 4986] assert in LiveIntervalAnalysis In-Reply-To: Message-ID: <200909210432.n8L4WoDo014289@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4986 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Evan Cheng 2009-09-20 23:32:49 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090914/087569.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 21 00:59:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 00:59:37 -0500 Subject: [LLVMbugs] [Bug 4216] Consecutive small memory accesses not merged in a bigger access In-Reply-To: Message-ID: <200909210559.n8L5xbG7025267@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4216 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Chris Lattner 2009-09-21 00:59:36 --- The rest of the loads eliminated with this patch: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087571.html We now compile this down to one load on x86-32, which is much much better than GCC at least: _test_bitfield: movl 4(%esp), %eax movl %eax, %ecx andl $-65536, %ecx orl $32962, %eax andl $40186, %eax orl %ecx, %eax ret This should be only one and and one or, but hey, it's still progress :) -- Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug. From bugzilla-daemon at cs.uiuc.edu Mon Sep 21 08:07:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 08:07:22 -0500 Subject: [LLVMbugs] [Bug 5018] New: ffi/oprofile checked for even when explicitly disabled Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5018 Summary: ffi/oprofile checked for even when explicitly disabled Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu ./configure --enable-ffi=no --enable-oprofile=no still checks for ffi and oprofile and adds it to LIBS. The problem is here in configure.ac: dnl --enable-libffi : check whether the user wants to turn off libffi: AC_ARG_ENABLE(libffi,AS_HELP_STRING( --enable-libffi,[Check for the presence of libffi (default is YES)]),, enableval=yes) case "$enableval" in yes) llvm_cv_enable_libffi="yes" ;; no) llvm_cv_enable_libffi="no" ;; *) AC_MSG_ERROR([Invalid setting for --enable-libffi. Use "yes" or "no"]) ;; esac enableval should be tested inside AC_ARG_ENABLE, otherwise it contains something else: + eval 'enable_ffi=$ac_optarg' ++ enable_ffi=no ... + case "$withval" in + with_ocaml_libdir=auto + test '' = set + withval=default + case "$withval" in + WITH_BINUTILS_INCDIR=default + test xdefault '!=' xdefault + test '' = set + enableval=yes + case "$enableval" in + llvm_cv_enable_libffi=yes -- 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 21 12:19:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 12:19:41 -0500 Subject: [LLVMbugs] [Bug 5020] New: Nondeterministic behavior of opt -O3 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5020 Summary: Nondeterministic behavior of opt -O3 Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: d.g.gorbachev at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3538) --> (http://llvm.org/bugs/attachment.cgi?id=3538) ZIP archive with original BC file and optimized files Optimize with `opt -O3' or with `opt -instcombine -simplifycfg -inline -argpromotion -instcombine -jump-threading -simplifycfg' -- 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 21 12:24:46 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 12:24:46 -0500 Subject: [LLVMbugs] [Bug 5016] Assertion failed: (castIsValid(getOpcode(), S, Ty) && " Illegal BitCast") In-Reply-To: Message-ID: <200909211724.n8LHOkrn030317@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5016 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-21 12:24:45 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087594.html Thanks again for the reduced 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 Mon Sep 21 15:54:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 15:54:16 -0500 Subject: [LLVMbugs] [Bug 5022] New: setz+mozxb -> xor+setz optimization Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5022 Summary: setz+mozxb -> xor+setz optimization Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu This IR: define internal fastcc i32 @bc0f0(i8*) nounwind { %2 = getelementptr inbounds i8* %0, i32 80 ; [#uses=1] %3 = bitcast i8* %2 to i32** ; [#uses=1] %4 = load i32** %3 ; [#uses=1] %5 = getelementptr inbounds i32* %4, i32 3 ; [#uses=1] %6 = load i32* %5 ; [#uses=1] %7 = icmp eq i32 %6, 2 ; [#uses=1] %8 = zext i1 %7 to i32 ; [#uses=1] ret i32 %8 } Translates to this: mov 0x50(%rdi), %rax cmp $0x2, 0xc(%rax) setz %al movzxb %al, %eax ret This is fine, but could be tuned a bit further: Use some other register instead of rax in first mov (say rcx). Then at the same time you could xor rax, rax to clear it. Then a single setz %al would be enough, without the need for a movzxb. According to http://gmplib.org/~tege/x86-timing.pdf movz has timings: 1,2.5,1,2.5; 1,3,1,3 xor has timings: 1,2,1,2;1,3,1,3 So xor is both faster (on P4) and can be run in parallel with the mov/cmp, avoiding one data dependency (setz -> movvz). -- 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 21 16:10:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 16:10:40 -0500 Subject: [LLVMbugs] [Bug 5023] New: opt -scalarrepl triggers verifier error Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5023 Summary: opt -scalarrepl triggers verifier error Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: gohman at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3539) --> (http://llvm.org/bugs/attachment.cgi?id=3539) reduced testcase In the attached testcase, opt -scalarrepl results in the verifier complaining "Invoke result does not dominate all uses!" This is a reduced testcase from 483.xalancbmk. -- 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 21 17:39:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 17:39:57 -0500 Subject: [LLVMbugs] [Bug 5023] [483.xalancbmk] opt -scalarrepl triggers verifier error In-Reply-To: Message-ID: <200909212239.n8LMdv74008279@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5023 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #3 from Chris Lattner 2009-09-21 17:39:57 --- Fixed, thanks for the nice testcase: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087626.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 21 19:27:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 19:27:15 -0500 Subject: [LLVMbugs] [Bug 5024] New: Another regscavenger assertion Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5024 Summary: Another regscavenger assertion Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: ARM AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3540) --> (http://llvm.org/bugs/attachment.cgi?id=3540) Reduced test Consider the attached bytecode. Currently we're failing with: $ ./llc -mattr=+neon bugpoint-reduced-simplified.bc llc: /home/asl/proj/llvm/src_arm/lib/CodeGen/RegisterScavenging.cpp:219: void llvm::RegScavenger::forward(): Assertion `isUsed(Reg) && "Using an undefined register!"' 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 Mon Sep 21 21:14:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 21:14:37 -0500 Subject: [LLVMbugs] [Bug 5025] New: crash in objc messaging Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5025 Summary: crash in objc messaging Product: clang Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu This test case (originally from PR4894) is still crashing: -- ddunbar at ozzy:CodeGenObjC (master)$ cat t.m typedef const struct objc_selector { void *sel_id; const char *sel_types; } *SEL; @interface I2 +(id) dictionary; @end @implementation I3; +(void) initialize { I2 *a0 = [I2 dictionary]; } @end ddunbar at ozzy:CodeGenObjC (master)$ clang -c t.m t.m:11:17: warning: cannot find interface declaration for 'I3' @implementation I3; ^ Assertion failed: (Result->getType() == Ty && "Type specified is not correct!"), function Create, file /Volumes/Data/ddunbar/llvm/lib/VMCore/ConstantsContext.h, line 607. 0 clang-cc 0x00000001010e3c35 PrintStackTrace(void*) + 38 1 clang-cc 0x00000001010e41c3 SignalHandler(int) + 336 2 libSystem.B.dylib 0x00007fff8584414a _sigtramp + 26 3 clang-cc 0x0000000100b7d9d5 llvm::APInt::operator==(llvm::APInt const&) const + 81 4 libSystem.B.dylib 0x00007fff858c0104 __pthread_markcancel + 0 5 clang-cc 0x0000000100ff35fa llvm::ValueMap::Create(llvm::Type const*, llvm::ExprMapKeyType const&, std::_Rb_tree_iterator const, llvm::ConstantExpr*> >) + 172 6 clang-cc 0x0000000100ff37b2 llvm::ValueMap::getOrCreate(llvm::Type const*, llvm::ExprMapKeyType const&) + 248 7 clang-cc 0x0000000100fe7be0 getFoldedCast(llvm::Instruction::CastOps, llvm::Constant*, llvm::Type const*) + 382 8 clang-cc 0x0000000100fe7d0e llvm::ConstantExpr::getBitCast(llvm::Constant*, llvm::Type const*) + 230 9 clang-cc 0x0000000100fe8a75 llvm::ConstantExpr::getCast(unsigned int, llvm::Constant*, llvm::Type const*) + 587 10 clang-cc 0x0000000100fe9527 llvm::ConstantExpr::getPointerCast(llvm::Constant*, llvm::Type const*) + 235 11 clang-cc 0x0000000100fe206e llvm::ConstantFoldCastInstruction(llvm::LLVMContext&, unsigned int, llvm::Constant*, llvm::Type const*) + 499 12 clang-cc 0x0000000100fe7b11 getFoldedCast(llvm::Instruction::CastOps, llvm::Constant*, llvm::Type const*) + 175 13 clang-cc 0x0000000100fe7d0e llvm::ConstantExpr::getBitCast(llvm::Constant*, llvm::Type const*) + 230 14 clang-cc 0x000000010016c087 (anonymous namespace)::CGObjCNonFragileABIMac::EmitSelector(llvm::IRBuilder >&, clang::Selector) + 127 15 clang-cc 0x0000000100171c7b (anonymous namespace)::CGObjCNonFragileABIMac::GenerateMessageSend(clang::CodeGen::CodeGenFunction&, clang::QualType, clang::Selector, llvm::Value*, bool, llvm::SmallVector, 16u> const&, clang::ObjCMethodDecl const*) + 117 16 clang-cc 0x0000000100150e52 clang::CodeGen::CodeGenFunction::EmitObjCMessageExpr(clang::ObjCMessageExpr const*) + 958 17 clang-cc 0x00000001001444e8 (anonymous namespace)::ScalarExprEmitter::VisitObjCMessageExpr(clang::ObjCMessageExpr*) + 36 18 clang-cc 0x000000010014a13e clang::StmtVisitor<(anonymous namespace)::ScalarExprEmitter, llvm::Value*>::Visit(clang::Stmt*) + 4380 19 clang-cc 0x0000000100148cb3 (anonymous namespace)::ScalarExprEmitter::EmitCastExpr(clang::Expr const*, clang::QualType, clang::CastExpr::CastKind) + 1629 20 clang-cc 0x0000000100148ef8 (anonymous namespace)::ScalarExprEmitter::VisitCastExpr(clang::CastExpr const*) + 128 21 clang-cc 0x0000000100148f1f clang::StmtVisitor<(anonymous namespace)::ScalarExprEmitter, llvm::Value*>::VisitImplicitCastExpr(clang::ImplicitCastExpr*) + 29 22 clang-cc 0x0000000100149ca7 clang::StmtVisitor<(anonymous namespace)::ScalarExprEmitter, llvm::Value*>::Visit(clang::Stmt*) + 3205 23 clang-cc 0x000000010014a2b7 clang::CodeGen::CodeGenFunction::EmitScalarExpr(clang::Expr const*, bool) + 131 24 clang-cc 0x0000000100128320 clang::CodeGen::CodeGenFunction::EmitLocalBlockVarDecl(clang::VarDecl const&) + 2180 25 clang-cc 0x0000000100129375 clang::CodeGen::CodeGenFunction::EmitBlockVarDecl(clang::VarDecl const&) + 157 26 clang-cc 0x00000001001294de clang::CodeGen::CodeGenFunction::EmitDecl(clang::Decl const&) + 310 27 clang-cc 0x0000000100181af9 clang::CodeGen::CodeGenFunction::EmitDeclStmt(clang::DeclStmt const&) + 101 28 clang-cc 0x0000000100183174 clang::CodeGen::CodeGenFunction::EmitSimpleStmt(clang::Stmt const*) + 224 29 clang-cc 0x0000000100181ba6 clang::CodeGen::CodeGenFunction::EmitStmt(clang::Stmt const*) + 78 30 clang-cc 0x0000000100182f47 clang::CodeGen::CodeGenFunction::EmitCompoundStmt(clang::CompoundStmt const&, bool, llvm::Value*, bool) + 301 31 clang-cc 0x000000010018315a clang::CodeGen::CodeGenFunction::EmitSimpleStmt(clang::Stmt const*) + 198 32 clang-cc 0x0000000100181ba6 clang::CodeGen::CodeGenFunction::EmitStmt(clang::Stmt const*) + 78 33 clang-cc 0x0000000100150a7d clang::CodeGen::CodeGenFunction::GenerateObjCMethod(clang::ObjCMethodDecl const*) + 161 34 clang-cc 0x0000000100191d6f clang::CodeGen::CodeGenModule::EmitTopLevelDecl(clang::Decl*) + 955 35 clang-cc 0x00000001001a4d7c (anonymous namespace)::CodeGeneratorImpl::HandleTopLevelDecl(clang::DeclGroupRef) + 76 36 clang-cc 0x0000000100046f63 (anonymous namespace)::BackendConsumer::HandleTopLevelDecl(clang::DeclGroupRef) + 145 37 clang-cc 0x000000010027de82 clang::ParseAST(clang::Preprocessor&, clang::ASTConsumer*, clang::ASTContext&, bool, bool, clang::CodeCompleteConsumer* (*)(clang::Sema&, void*), void*) + 566 38 clang-cc 0x000000010003247c ProcessInputFile(clang::Preprocessor&, clang::PreprocessorFactory&, std::string const&, ProgActions, llvm::StringMap const&, llvm::LLVMContext&) + 3471 39 clang-cc 0x0000000100033b92 main + 3269 40 clang-cc 0x000000010002dc60 start + 52 41 clang-cc 0x0000000000000015 start + 4294779881 Stack dump: 0. Program arguments: /Volumes/Data/ddunbar/llvm.obj.64/Debug/bin/clang-cc -triple x86_64-apple-darwin10 -S -disable-free -main-file-name t.m --relocation-model pic -pic-level=1 --disable-fp-elim --unwind-tables=1 --mcpu=core2 --fmath-errno=0 -mmacosx-version-min=10.6.0 -fdiagnostics-show-option -o /var/folders/MP/MP5slYG8HRCD2v6HHtqYEE+++TI/-Tmp-/cc-PqdYj2.s -x objective-c t.m 1. t.m:15:1: current parser token '@' 2. t.m:12:1: LLVM IR generation of declaration 'I3::initialize' 3. t.m:12:20: LLVM IR generation of compound statement ('{}') clang: error: compiler command failed due to signal 6 (use -v to see invocation) -- Please enable it in the PR4894 test case when 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 21 21:19:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 21:19:15 -0500 Subject: [LLVMbugs] [Bug 4894] Crash while emitting debug information for complex Objective-C classes In-Reply-To: Message-ID: <200909220219.n8M2JFne004506@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4894 Daniel Dunbar changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #8 from Daniel Dunbar 2009-09-21 21:19:15 --- Fixed here: http://llvm.org/viewvc/llvm-project?view=rev&revision=82507 Test case (clang) here: http://llvm.org/viewvc/llvm-project?view=rev&revision=82510 David, please verify against your original 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 Mon Sep 21 21:48:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 21 Sep 2009 21:48:40 -0500 Subject: [LLVMbugs] [Bug 5026] New: Don't know how to fold this instruction! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5026 Summary: Don't know how to fold this instruction! Product: new-bugs Version: trunk 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 Jaunty on x86. regehr at john-home:~/volatile/tmp198$ clang -O1 small.c Don't know how to fold this instruction! UNREACHABLE executed at X86InstrInfo.cpp:2313! 0 clang-cc 0x09025c68 Stack dump: 0. Program arguments: /home/regehr/z/tmp/llvm-gcc-r82487-install/bin/../libexec/clang-cc -triple i386-pc-linux-gnu -S -disable-free -main-file-name small.c --relocation-model static --disable-fp-elim --unwind-tables=0 --mcpu=pentium4 --fmath-errno=1 -O1 -fdiagnostics-show-option -o /tmp/cc-ZMJk8h.s -x c small.c 1. parser at end of file 2. Code generation 3. Running pass 'Linear Scan Register Allocator' on function '@int165' clang: error: compiler command failed due to signal 6 (use -v to see invocation) regehr at john-home:~/volatile/tmp198$ clang -v clang version 1.1 (http://llvm.org/svn/llvm-project/cfe/trunk 82487) Target: i386-pc-linux-gnu Thread model: posix regehr at john-home:~/volatile/tmp198$ cat small.c long long int g_11; volatile const long long int g_77 = 1; int safe (int); unsigned char foo (unsigned char _ui1, unsigned char _ui2) { return _ui2 ? : 1; } void int165 (int uint32p57) { if (safe (g_77 <= foo (g_11 < safe (1), 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 Tue Sep 22 03:35:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 03:35:16 -0500 Subject: [LLVMbugs] [Bug 5024] Another regscavenger assertion In-Reply-To: Message-ID: <200909220835.n8M8ZGDw032506@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5024 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Evan Cheng 2009-09-22 03:35:15 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087661.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 22 04:07:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 04:07:48 -0500 Subject: [LLVMbugs] [Bug 5027] New: System library should support pipe in executeandwait Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5027 Summary: System library should support pipe in executeandwait Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: System Library AssignedTo: unassignedbugs at nondot.org ReportedBy: edwintorok at gmail.com CC: llvmbugs at cs.uiuc.edu ExecuteAndWait only supports redirection to files, not pipes. Pipes are available on both Unix and Windows, on Windows it could use CreatePipe. -- 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 22 08:38:54 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 08:38:54 -0500 Subject: [LLVMbugs] [Bug 5028] New: LLVM doesn't have -D_FORTIFY_SOURCE=x checks Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5028 Summary: LLVM doesn't have -D_FORTIFY_SOURCE=x checks Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Keywords: new-feature Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: iamleeg at gmail.com CC: llvmbugs at cs.uiuc.edu Summary: the compile-time and run-time safety tests enabled by use of the _FORTIFY_SOURCE macro are not included when compiling with LLVM-clang or LLVM-GCC. Steps to reproduce: take the following source code, and compile with the version of llvm-gcc or clang from /Developer/usr/bin in an Xcode 3.2 installation on Snow Leopard. #include #include int main (int argc, const char * argv[]) { char buffer[16]; strcpy(buffer, "hello"); strcpy(buffer, "This sentence is much too long for you."); strcpy(buffer, argv[1]); return 0; } Expected results: a warning is emitted by the compiler as it would be if compiled with GCC, due to the statically-testable buffer overrun at the second strcpy() invocation. If the compiled program is run, it should abort inside the libc overrun-check function. Actual results: no warning is emitted. The program crashes when the stack is smashed. -- 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 22 08:53:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 08:53:56 -0500 Subject: [LLVMbugs] [Bug 4216] Consecutive small memory accesses not merged in a bigger access In-Reply-To: Message-ID: <200909221353.n8MDruPA009423@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4216 Sylv??re Teissier changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #11 from Sylv??re Teissier 2009-09-22 08:53:56 --- that's stange with my target: with -01 I have this: clang-cc test2.c -emit-llvm -o - -O1 ; ModuleID = 'test2.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.control_reg = type { i8, i8, [2 x i8] } define void @test_bitfield(%struct.control_reg* noalias nocapture sret %agg.result, %struct.control_reg* nocapture byval %c) nounwind { entry: %tmp = bitcast %struct.control_reg* %c to i32* ; [#uses=1] %bf.prev.low = load i32* %tmp ; [#uses=1] %bf.prev.lo.cleared10 = or i32 %bf.prev.low, 32962 ; [#uses=1] %bf.new.lo11 = and i32 %bf.prev.lo.cleared10, -25350 ; [#uses=1] %0 = bitcast %struct.control_reg* %agg.result to i32* ; [#uses=1] store i32 %bf.new.lo11, i32* %0, align 4 ret void } that's ok, but with -O2 / -O3 I have this: $ clang-cc test2.c -emit-llvm -o - -O2 ; ModuleID = 'test2.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.control_reg = type { i8, i8, [2 x i8] } define void @test_bitfield(%struct.control_reg* noalias nocapture sret %agg.result, %struct.control_reg* nocapture byval %c) nounwind { entry: %tmp = bitcast %struct.control_reg* %c to i32* ; [#uses=1] %bf.prev.low = load i32* %tmp ; [#uses=3] %bf.prev.lo.cleared10 = or i32 %bf.prev.low, 32962 ; [#uses=1] %bf.new.lo11 = and i32 %bf.prev.lo.cleared10, -25350 ; [#uses=2] %trunc = trunc i32 %bf.new.lo11 to i8 ; [#uses=1] %tmp28 = lshr i32 %bf.new.lo11, 8 ; [#uses=1] %tmp29 = trunc i32 %tmp28 to i8 ; [#uses=1] %tmp30 = lshr i32 %bf.prev.low, 16 ; [#uses=1] %tmp31 = trunc i32 %tmp30 to i8 ; [#uses=1] %tmp32 = lshr i32 %bf.prev.low, 24 ; [#uses=1] %tmp33 = trunc i32 %tmp32 to i8 ; [#uses=1] %agg.result.0 = getelementptr %struct.control_reg* %agg.result, i32 0, i32 0 ; [#uses=1] store i8 %trunc, i8* %agg.result.0, align 4 %agg.result.1 = getelementptr %struct.control_reg* %agg.result, i32 0, i32 1 ; [#uses=1] store i8 %tmp29, i8* %agg.result.1, align 1 %agg.result.2.0 = getelementptr %struct.control_reg* %agg.result, i32 0, i32 2, i32 0 ; [#uses=1] store i8 %tmp31, i8* %agg.result.2.0, align 2 %agg.result.2.1 = getelementptr %struct.control_reg* %agg.result, i32 0, i32 2, i32 1 ; [#uses=1] store i8 %tmp33, i8* %agg.result.2.1, align 2 ret void } that's very bad :( do you have the same behavior ? -- 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 22 12:21:03 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 12:21:03 -0500 Subject: [LLVMbugs] [Bug 5029] New: tests should not use source code as input Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5029 Summary: tests should not use source code as input Product: Test Suite Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Nightly Tester AssignedTo: unassignedbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu Tests in the nightly test suite which need some input file for the test should *not* use their own source code. 1. If the test needs to be updated this introduces another variable into the performance. 2. If the test is being miscompiled its possible that editing the source (for debugging) will then mask the bug, but only because the source itself is the input. We could have a standard set of small/medium/large inputs for the programs which take some arbitrary input, but don't care what it is. -- 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 22 13:04:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 13:04:04 -0500 Subject: [LLVMbugs] [Bug 5024] Another regscavenger assertion In-Reply-To: Message-ID: <200909221804.n8MI44KU008836@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5024 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #5 from Anton Korobeynikov 2009-09-22 13:04:03 --- Thanks, Evan, this fixes the reduction, but doesn't fix the whole testcase. Another reduction 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 Tue Sep 22 16:10:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 16:10:47 -0500 Subject: [LLVMbugs] [Bug 5030] New: clang should get the target data from the LLVM back end Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5030 Summary: clang should get the target data from the LLVM back end Product: clang Version: unspecified Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: daniel at zuster.org CC: llvmbugs at cs.uiuc.edu clang should get the target data string from the LLVM back end, following llvm-gcc, instead of having it hardcoded. -- 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 22 18:43:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 18:43:00 -0500 Subject: [LLVMbugs] [Bug 5031] New: Invalid offset folding Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5031 Summary: Invalid offset folding Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3545) --> (http://llvm.org/bugs/attachment.cgi?id=3545) Testcase Consider the attached testcase. Let us compare the two outputs: 1. ./llc 1.bc -mcpu=cortex-a9 -soft-float -asm-verbose=0 -o good.s 2. ./llc 1.bc -mcpu=cortex-a9 -soft-float -asm-verbose=0 -relocation-model=static bad.s --- good.s 2009-09-23 03:35:50.000000000 +0400 +++ bad.s 2009-09-23 03:35:40.000000000 +0400 @@ -14,7 +14,6 @@ _ZN11FooD0Ev: stmfd sp!, {r4, lr} ldr r1, .LCPI1_0 - add r1, r1, #8 str r1, [r0] ldr r1, [r0, #+476] mov r4, r0 The difference is caused by the fact that we're not overriding default relocation model on ARM (this is separate bug, I will take care of it after this one will be fixed). This leads to different behaviour of TargetLowering::isOffsetFoldingLegal(): always true to static relocation model and false for default. It seems that folded offset is somehow not honored later which leads to invalid codegen. -- 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 22 19:24:10 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 19:24:10 -0500 Subject: [LLVMbugs] [Bug 5032] New: Bad DWARF encoding in EH frames. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5032 Summary: Bad DWARF encoding in EH frames. Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Keywords: miscompilation Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: wilsons at start.ca CC: llvmbugs at cs.uiuc.edu, wilsons at start.ca I am seeing the following exception table header generated on x86_64, GNU/Linux: GCC_except_table1: .byte 0x0 # Padding .byte 0x0 # Padding .Lexception1: .byte 0xFF # @LPStart format (DW_EH_PE_omit) .byte 0 # @TType format (DW_EH_PE_absptr) .uleb128 5 # @TType base offset Now, @TType format says to expect an absptr, but what follows is a uleb128. I am _not_ an expert on DWARF. But I think this is wrong. Shouldn't @TType format be DW_EH_PE_uleb128? Otherwise a dwarf reader would take @TType offset as an absoulte pointer address. Thanks, Steve -- 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 22 21:36:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 22 Sep 2009 21:36:17 -0500 Subject: [LLVMbugs] [Bug 5033] New: failure to warn about uninitialized induction var Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5033 Summary: failure to warn about uninitialized induction var Product: new-bugs Version: trunk 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'd expect llvm-gcc and clang to warn about use-before-init of i, but this does not happen, even at "-O -Wall", using r82487. void foo (void) { int i; for (; i<10; i++) { } } -- 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 23 01:29:09 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 01:29:09 -0500 Subject: [LLVMbugs] [Bug 5024] Another regscavenger assertion In-Reply-To: Message-ID: <200909230629.n8N6T9ir006372@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5024 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #8 from Evan Cheng 2009-09-23 01:29:08 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087714.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 23 08:40:13 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 08:40:13 -0500 Subject: [LLVMbugs] [Bug 5022] setz+mozxb -> xor+setz optimization In-Reply-To: Message-ID: <200909231340.n8NDeDpp009061@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5022 T??r??k Edwin changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #8 from T??r??k Edwin 2009-09-23 08:40:13 --- (In reply to comment #6) > In your numbers, the 95% confidence interval for the "rax -> rcx" case is > entirely contained within the 95% confidence interval for the "original" case, > so this isn't showing a significant difference. > Indeed, I was timing foo1 always. I fixed that, and also used rdtsc directly, results are: Timing original: 60300165 average, 189461 stddev, 95% confidence interval: 59921243 - 60679087; Timing rax -> rcx: 60303412 average, 214691 stddev, 95% confidence interval: 59874030 - 60732794; Timing rax -> rcx, setz+movz -> xor+setz: 70366981 average, 287094 stddev, 95% confidence interval: 69792793 - 70941169; So first two are the same, and the transform proposed here is actually slower. Closing this 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 eugeny.grishul at gmail.com Wed Sep 23 09:05:05 2009 From: eugeny.grishul at gmail.com (=?KOI8-R?B?5dfHxc7JyiDn0snb1czY?=) Date: Wed, 23 Sep 2009 18:05:05 +0400 Subject: [LLVMbugs] bug in sequence 'load --> bswap --> bitcast --> fptoui' Message-ID: Hi, Currently I develop own compiler using LLVM ( and MS Phoenix framework =) ) and found that 'fptoui' instruction usage in following situation crashes JIT: ;=========================================== @System.Test.ByteOrderTests.big2 = internal global i32 60994 @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" declare i32 @printf(i8* nocapture, ...) nounwind declare i32 @llvm.bswap.i32(i32) nounwind readnone define void @main() { %1 = load i32* @System.Test.ByteOrderTests.big2 ; [#uses=1] %2 = call i32 @llvm.bswap.i32(i32 %1) %3 = bitcast i32 %2 to float %4 = fptoui float %3 to i64 %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), i64 %4 ) ret void } ;Writes: ;0. Program arguments: lli test.bc ;00000000 (0x42EE0000 0x00000000 0x00D6FA60 0x01280E01) ;001D0023 (0x00C1F664 0x00C48F50 0x00C1F734 0x54AD8E88) ;0153245C (0x00C48F50 0x022941BC 0x00D61D18 0x54AD80F4), llvm::ExecutionEngine::runFunctionAsMain()+0940 bytes(s), e:\external\llvm\lib\executionengine\executionengine.cpp, line 387+0030 byte(s) ;01272C44 (0x00000002 0x00D680F8 0x00D61D18 0x54AD80A4), main()+2196 bytes(s), e:\external\llvm\tools\lli\lli.cpp, line 218+0064 byte(s) ;01C5FF78 (0x00C1F984 0x74DB3677 0x7EFDE000 0x00C1F9C4), __tmainCRTStartup()+0424 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 586+0025 byte(s) ;01C5FDBF (0x7EFDE000 0x00C1F9C4 0x77159D72 0x7EFDE000), mainCRTStartup()+0015 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 403 ;74DB3677 (0x7EFDE000 0x778EDA85 0x00000000 0x00000000), BaseThreadInitThunk()+0018 bytes(s) ;77159D72 (0x01C5FDB0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0099 bytes(s) ;77159D45 (0x01C5FDB0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0054 bytes(s) ;=========================================== But if I change 'fptoui' to 'fptosi' everything is OK: @System.Test.ByteOrderTests.big2 = internal global i32 60994 @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" declare i32 @printf(i8* nocapture, ...) nounwind declare i32 @llvm.bswap.i32(i32) nounwind readnone define void @main() { %1 = load i32* @System.Test.ByteOrderTests.big2 ; [#uses=1] %2 = call i32 @llvm.bswap.i32(i32 %1) %3 = bitcast i32 %2 to float %4 = fptosi float %3 to i64 %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), i64 %4 ) ret void } ;Writes: Big endian: 119 ;=========================================== I'm using llvm from trunk (23 sept 2009) on Windows 7 ( build 7600 ) To run test I used: llvm-as test.ll lli test.bc -- WBR, Eugeny Grishul -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvmbugs/attachments/20090923/2b4a8989/attachment.html From anton at korobeynikov.info Wed Sep 23 10:41:11 2009 From: anton at korobeynikov.info (Anton Korobeynikov) Date: Wed, 23 Sep 2009 19:41:11 +0400 Subject: [LLVMbugs] bug in sequence 'load --> bswap --> bitcast --> fptoui' In-Reply-To: References: Message-ID: Hello > @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" This looks completely wrong - MS CRT does not support this format specifier. It might be that you're just trashing the stack. -- With best regards, Anton Korobeynikov Faculty of Mathematics and Mechanics, Saint Petersburg State University From bugzilla-daemon at cs.uiuc.edu Wed Sep 23 11:59:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 11:59:15 -0500 Subject: [LLVMbugs] [Bug 5024] Another regscavenger assertion In-Reply-To: Message-ID: <200909231659.n8NGxFil003192@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5024 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED 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 Wed Sep 23 14:04:52 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 14:04:52 -0500 Subject: [LLVMbugs] [Bug 5031] Invalid offset folding In-Reply-To: Message-ID: <200909231904.n8NJ4qKB019256@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5031 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Anton Korobeynikov 2009-09-23 14:04:51 --- Fixed in http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087736.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 eugeny.grishul at gmail.com Wed Sep 23 14:22:03 2009 From: eugeny.grishul at gmail.com (=?KOI8-R?B?5dfHxc7JyiDn0snb1czY?=) Date: Wed, 23 Sep 2009 23:22:03 +0400 Subject: [LLVMbugs] bug in sequence 'load --> bswap --> bitcast --> fptoui' In-Reply-To: References: Message-ID: 1) MS CRT supports both %I64d (such not supports on osx/linux) and %lld 2) Extra arguments for __cdecl vararg functions can't hang program execution in such manner 3) even if change %lld to %I64d stack trace will be same 4) As I said problem is in 'fptoui' instruction. It's 100% reproduceable using provided test case. 2009/9/23 Anton Korobeynikov > Hello > > > @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" > This looks completely wrong - MS CRT does not support this format > specifier. It might be that you're just trashing the stack. > > -- > With best regards, Anton Korobeynikov > Faculty of Mathematics and Mechanics, Saint Petersburg State University > -- WBR, Eugeny Grishul -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.cs.uiuc.edu/pipermail/llvmbugs/attachments/20090923/d89e3fc5/attachment.html From bugzilla-daemon at cs.uiuc.edu Wed Sep 23 15:41:47 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 15:41:47 -0500 Subject: [LLVMbugs] [Bug 5036] New: llc crashes on Windows due to NULL X86AsmPrinter::MMI Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5036 Summary: llc crashes on Windows due to NULL X86AsmPrinter::MMI Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu Crashes on mingw and VC for every .bc file I tried, but not on Linux. This is the stack trace: llc.exe!llvm::MachineModuleInfo::getObjFileInfo() Line 167 + 0x3 bytes C++ llc.exe!llvm::X86AsmPrinter::runOnMachineFunction(llvm::MachineFunction & MF={...}) Line 145 + 0xb bytes C++ llc.exe!llvm::MachineFunctionPass::runOnFunction(llvm::Function & F={...}) Line 31 + 0x13 bytes C++ llc.exe!llvm::FPPassManager::runOnFunction(llvm::Function & F={...}) Line 1380 + 0x17 bytes C++ llc.exe!llvm::FunctionPassManagerImpl::run(llvm::Function & F={...}) Line 1332 + 0x1b bytes C++ llc.exe!llvm::FunctionPassManager::run(llvm::Function & F={...}) Line 1262 + 0xf bytes C++ llc.exe!main(int argc=3, char * * argv=0x001c4648) Line 393 C++ The problem is in X86AsmPrinter::runOnMachineFunction: if (Subtarget->isTargetCOFF()) { X86COFFMachineModuleInfo &COFFMMI = MMI->getObjFileInfo(); MMI is NULL. -- 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 23 16:35:38 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 16:35:38 -0500 Subject: [LLVMbugs] [Bug 5032] Bad DWARF encoding in EH frames. In-Reply-To: Message-ID: <200909232135.n8NLZc2Z006408@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5032 Bill Wendling changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |wendling at apple.com Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Bill Wendling 2009-09-23 16:35:37 --- (In reply to comment #0) > I am _not_ an expert on DWARF. But I think this is wrong. Shouldn't @TType > format be DW_EH_PE_uleb128? Otherwise a dwarf reader would take @TType offset > as an absoulte pointer address. Hi Steve, No. The "@TType format" applies to the encoding of functions within the exception table, not to the offsets. For instance on Darwin (with gcc, not llvm), we put the EH table in the __TEXT section. This requires the functions to be encoded in a PC-relative way. Something like this: .long L__ZId-. etc. Yes, it's confusing. But then so is this whole area of exception handling. :-) -- 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 23 17:23:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 17:23:44 -0500 Subject: [LLVMbugs] [Bug 5037] New: arm_neon.h float32_t is not compatible with float Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5037 Summary: arm_neon.h float32_t is not compatible with float Product: new-bugs Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: major Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3549) --> (http://llvm.org/bugs/attachment.cgi?id=3549) I think this repros the problem Assignments from float* to float32_t* cause errors with llvm-gcc, but work with GCC 4.4+VFP, FSF GCC trunk, and ARMCC. float32_t requires special mangling per the ABI. -- 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 23 21:27:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 23 Sep 2009 21:27:39 -0500 Subject: [LLVMbugs] [Bug 5024] Another regscavenger assertion In-Reply-To: Message-ID: <200909240227.n8O2Rd8u010824@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5024 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #10 from Evan Cheng 2009-09-23 21:27:38 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087771.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087772.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 24 00:45:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 00:45:22 -0500 Subject: [LLVMbugs] [Bug 5036] llc crashes on Windows due to NULL X86AsmPrinter::MMI In-Reply-To: Message-ID: <200909240545.n8O5jMmE002787@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5036 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Chris Lattner 2009-09-24 00:45:21 --- Fixed in r82684, I definitely made this change yesterday, I must have missed committing that dir :-( Sorry for the breakage! -- 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 24 04:17:30 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 04:17:30 -0500 Subject: [LLVMbugs] [Bug 5038] New: llvm-ld changes global varible with arguments of byval Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5038 Summary: llvm-ld changes global varible with arguments of byval Product: new-bugs Version: 2.5 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: hbrenkun at yahoo.cn CC: llvmbugs at cs.uiuc.edu I run my case: + rm optbug.llvm.x86.bc optbug.x86.bc optbug.x86.s optbug.x86.exe optbug.llvm.x86.ll optbug.x86.ll llvm-gcc4.2-2.5-x86-linux-RHEL4/bin/llvm-gcc -c -emit-llvm optbug.c -o optbug.llvm.x86.bc optbug.c: In function ???mycompare???: optbug.c:20: warning: incompatible implicit declaration of built-in function ???printf??? llvm-dis -f optbug.llvm.x86.bc -o optbug.llvm.x86.ll llvm-ld -ipsccp optbug.llvm.x86.bc -o optbug.x86.bc llvm-dis -f optbug.x86.bc.bc -o optbug.x86.ll llc -f optbug.x86.bc.bc -o optbug.x86.s llvm-gcc4.2-2.5-x86-linux-RHEL4/bin/llvm-gcc optbug.x86.s -o optbug.x86.exe + ./optbug.x86.exe in mycompare: line = 100, val1 = 99, val2 = 0 val1 != val2 /********my test case********/ //#include struct MYstr { unsigned char uc ; int ui; } mystr; void mytest(struct MYstr u) { u.ui = 99; u.uc = 'a'; } void mystructure() { mystr.ui = 0; mystr.uc = 'b'; mytest(mystr); mycompare(100, mystr.ui, 0); } int mycompare( int line, int val1, int val2) { printf("in mycompare: line = %d, val1 = %d, val2 = %d\n", line, val1, val2); if(val1 == val2) { printf("val1 == val2\n"); } else { printf("val1 != val2\n"); } return 1; } int main() { mystructure(); 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 Thu Sep 24 04:48:33 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 04:48:33 -0500 Subject: [LLVMbugs] [Bug 5038] llvm-ld changes global varible with arguments of byval In-Reply-To: Message-ID: <200909240948.n8O9mXXY014943@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5038 T??r??k Edwin changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #2 from T??r??k Edwin 2009-09-24 04:48:33 --- Fixed here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087797.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 24 05:51:56 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 05:51:56 -0500 Subject: [LLVMbugs] [Bug 5039] New: inefficient code generation of C bitfields Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5039 Summary: inefficient code generation of C bitfields Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu the attached test case produces on X86 this asm: setidt: .Leh_func_begin1: pushq %rbp .Llabel1: movq %rsp, %rbp .Llabel2: shlq $40, %rdx movabsq $34084860461056, %rax andq %rdx, %rax shlq $45, %rcx orq %rax, %rcx shlq $32, %r8 movabsq $30064771072, %rax andq %r8, %rax movl %esi, %edx orl $2097152, %edx andq $2162687, %rdx orq %rax, %rdx orq %rcx, %rdx movabsq $140737488355328, %rax orq %rdx, %rax movabsq $280405532016639, %rcx andq %rax, %rcx movq %rsi, %rax shlq $32, %rax movabsq $-281474976710656, %rdx andq %rax, %rdx orq %rcx, %rdx movq %rdx, idt(%rip) shrq $32, %rsi movabsq $-4294967296, %rax andq idt+8(%rip), %rax orq %rsi, %rax movq %rax, idt+8(%rip) popq %rbp ret while gcc generates much shorter: setidt: .LFB2: andl $3, %ecx andl $31, %edx movw %si, idt(%rip) sall $5, %ecx shrq $16, %rsi andl $7, %r8d orl %ecx, %edx movw %si, idt+6(%rip) shrq $16, %rsi orl $-128, %edx movw $32, idt+2(%rip) movb %r8b, idt+4(%rip) movb %dl, idt+5(%rip) movl %esi, idt+8(%rip) ret both is with -O2. -- 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 24 14:37:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 14:37:59 -0500 Subject: [LLVMbugs] [Bug 5042] New: 'Invoke' llvm instruction does not function per documentation Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5042 Summary: 'Invoke' llvm instruction does not function per documentation Product: libraries Version: trunk Platform: Other OS/Version: Linux Status: NEW Keywords: missing-feature Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: zrega at theophad.tmbx.com CC: llvmbugs at cs.uiuc.edu The invoke / unwind functionality does not work as per the documentation. It is now my understanding that exception handling in C, C++ is implemented using other libraries. Currently (in svn) when a unwind instruction is encountered, the program unwinds (Good), but the invoke does not handle it. The following code will throw a segmentation fault on my system. My system: triple=powerpc-unknown-linux-gnu llvm from svn 9/23/2009 Proposed resolution: * Change Documentation if this will not be supported without other libraries * Change llvm to adhear to the docs. Which could be hard to do. To Reproduce: Run the following code through llvm-as then lli or llc. //====---- Code Start -----========// @.str = internal constant [6 x i8] c"Hello\00" ; <[6 x i8]*> [#uses=1] @.strFunctionError = internal constant [36 x i8] c"Could not call function! Exiting...\00" ; <[36 x i8]*> [#uses=1] define i32 @main() nounwind { entry: %0 = tail call i32 @puts(i8* getelementptr ([6 x i8]* @.str, i32 0, i32 0)) nounwind ; [#uses=0] %a = invoke i32 @functionToFail() to label %normal unwind label %exception exception: tail call i32 @puts(i8* getelementptr ([36 x i8]* @.strFunctionError, i32 0, i32 0)) nounwind ret i32 -1 normal: %.fo = tail call i32 @puts(i8* getelementptr ([6 x i8]* @.str, i32 0, i32 0)) nounwind ret i32 0 } define i32 @functionToFail() { entry: unwind ret i32 0 } declare i32 @puts(i8*) //====---- Code End -----========// -- 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 24 15:32:23 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 15:32:23 -0500 Subject: [LLVMbugs] [Bug 5042] 'Invoke' llvm instruction does not function per documentation In-Reply-To: Message-ID: <200909242032.n8OKWN0v031709@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5042 Duncan Sands changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |baldrick at free.fr Status|NEW |RESOLVED Resolution| |LATER --- Comment #1 from Duncan Sands 2009-09-24 15:32:22 --- Currently the unwind instruction only allows you to rethrow an exception you just caught. It doesn't allow you to throw a new exception. Well, the interpreter supports this but the code generator does not. This is a known limitation that will hopefully be overcome one day. -- 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 24 16:14:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 16:14:55 -0500 Subject: [LLVMbugs] [Bug 5043] New: Error compiling ASTContext.cpp with VS2008 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5043 Summary: Error compiling ASTContext.cpp with VS2008 Product: clang Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: AST AssignedTo: unassignedclangbugs at nondot.org ReportedBy: Jeff.Krall at faircom.com CC: llvmbugs at cs.uiuc.edu Attempting to compile ASTContext.cpp at revision 82705 using the Visual Studio 2008 compiler fails with the following error: .\ASTContext.cpp(1107) : error C2371: 'F' : redefinition; different basic types .\ASTContext.cpp(1104) : see declaration of 'F' It seems that the scope of variable F declared on line 1104 extends to the block of code that begins on line 1107. I was able to work around the compile error by changing the variable F on lines 1104 and 1105 in my working copy to Fnop, as shown below: if (const FunctionNoProtoType *Fnop = T->getAs()) { ResultType = getFunctionNoProtoType(Fnop->getResultType(), true); } else { const FunctionProtoType *F = T->getAs(); -- 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 24 16:17:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 16:17:41 -0500 Subject: [LLVMbugs] [Bug 5042] 'Invoke' llvm instruction does not function per documentation In-Reply-To: Message-ID: <200909242117.n8OLHfDr005179@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5042 Daniel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|LATER | --- Comment #2 from Daniel 2009-09-24 16:17:40 --- Re-assign this bug to documentation. (In reply to comment #1) > Currently the unwind instruction only allows you to rethrow an > exception you just caught. I used unwind in this example. If I create a bug that creates an exceptional circumstance (ie a call to an unlinked function) invoke will still fail to catch this. Still, while this may be known to the developers, but if this has been the current operation from before v2.0, then it is well know, and thus should be documented as such. *To users this is an unknown issue.* > It doesn't allow you to throw a new > exception. Well, the interpreter supports this but the code > generator does not. This is a known limitation that will hopefully > be overcome one day. And regardless of how it is thrown, invoke doesn't work by itself, it needs external classes to function. Right? Re-assign this bug to documentation. Again, this may not be a code bug, it could be a documentation issue. From: http://llvm.org/docs/LangRef.html#i_invoke "http://llvm.org/docs/LangRef.html#i_invoke" 'exception label': the label reached when a callee returns with the unwind instruction. "The 'invoke' instruction causes control to transfer to a specified function, with the possibility of control flow transfer to either the 'normal' label or the 'exception' label. If the callee function returns with the "ret" instruction, control flow will return to the "normal" label. If the callee (or any indirect callees) returns with the "unwind" instruction, control is interrupted and continued at the dynamically nearest "exception" label." -- Rational: If I can't trust the documentation, how in the world am I to develop anything on this platform? The documentation for non pre-release software should be accurate as to the latest release. A work in progress should have final specs that are distinct from the documentation for non pre-release software. To fix: State current limitations of this instruction and when it can *actually* be used in "LangRef.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 24 16:40:58 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 16:40:58 -0500 Subject: [LLVMbugs] [Bug 5044] New: sqrt(-1.0) = 0 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5044 Summary: sqrt(-1.0) = 0 Product: new-bugs Version: trunk Platform: All OS/Version: All Status: NEW Severity: major Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: alexmac at adobe.com CC: llvmbugs at cs.uiuc.edu, alexmac at adobe.com Created an attachment (id=3555) --> (http://llvm.org/bugs/attachment.cgi?id=3555) svn diff output When called indirectly the sqrt function returns 0 for an input of -1.0 rather than NAN. compiling the following program with llvm-gcc -O3 demonstrates the problem ------------------------------------------- #include #include double mysqrt(double x) { return sqrt(x); } int main() { printf("sqrt(-1) == %G\n", mysqrt(-1.0)); } -------------------------------------------- A fix for the bug is 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 Thu Sep 24 16:50:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 16:50:00 -0500 Subject: [LLVMbugs] [Bug 5045] New: multiply by zero sometimes != zero Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5045 Summary: multiply by zero sometimes != zero Product: new-bugs Version: trunk Platform: All OS/Version: All Status: NEW Severity: major Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: alexmac at adobe.com CC: llvmbugs at cs.uiuc.edu, alexmac at adobe.com Created an attachment (id=3556) --> (http://llvm.org/bugs/attachment.cgi?id=3556) bugfix- svn diff output The constant folding code makes an invalid assumption that "X * 0 = 0", this is not true if X turns out to be a special number such as INF or NAN. Here is a test case that demonstrates the breakage: --------------------------------------------- #include #include double somefunc(int x) { return x ? INFINITY : NAN; } int main() { printf("inf * 0 = %G\n", somefunc(1) * 0); printf("NAN * 0 = %G\n", somefunc(0) * 0); } --------------------------------------------- fix is 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 Thu Sep 24 17:10:53 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 17:10:53 -0500 Subject: [LLVMbugs] [Bug 5044] sqrt(-1.0) = 0 In-Reply-To: Message-ID: <200909242210.n8OMArpB012198@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5044 Dale Johannesen changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Dale Johannesen 2009-09-24 17:10:52 --- Yes, in 82640. Example works for me. (The attached patch only affects vectors, and appears to be something else entirely.) -- 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 24 21:13:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 21:13:06 -0500 Subject: [LLVMbugs] [Bug 4865] [2.6] JIT dwarf exception handling seems broken In-Reply-To: Message-ID: <200909250213.n8P2D6xW010542@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4865 Laurent Sansonetti changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED 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 Thu Sep 24 23:00:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 24 Sep 2009 23:00:55 -0500 Subject: [LLVMbugs] [Bug 5046] New: Stack misadjustment when using sret. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5046 Summary: Stack misadjustment when using sret. Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3561) --> (http://llvm.org/bugs/attachment.cgi?id=3561) bitcode file that corresponds to the llvm assembler shown This code ; ModuleID = 'lp0' 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-win32" declare void @abort() define void @main([28 x i8]* sret) { call void @FEACA9C85_FFI_FN([28 x i8]* sret %0, i32 10) ret void } declare void @FEACA9C85_FFI_FN([28 x i8]* sret, i32) produces this: .text ALIGN 16 .globl _main _main: ; @main $BB1_0: subl $12, %esp ;; Why 12 bytes? $label1: movl 16(%esp), %eax movl %eax, (%esp) movl $10, 4(%esp) call _FEACA9C85_FFI_FN addl $8, %esp ;; 8 bytes freed, instead of 12 ret $4 ;; This is C callingconv. Why free 4 bytes of args? The net result is that the return jumps to some trash address. -- 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 25 00:21:00 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 00:21:00 -0500 Subject: [LLVMbugs] [Bug 4865] [2.6] JIT dwarf exception handling seems broken In-Reply-To: Message-ID: <200909250521.n8P5L0Tk001566@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4865 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #21 from Chris Lattner 2009-09-25 00:20:59 --- Ok, if this is fixed on mainline, then it is fixed. Unfortunately it is too late for 2.6. -- 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 25 03:06:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 03:06:55 -0500 Subject: [LLVMbugs] [Bug 5046] Stack misadjustment when using sret. In-Reply-To: Message-ID: <200909250806.n8P86t9h003103@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5046 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |asl at math.spbu.ru Status|NEW |RESOLVED Resolution| |INVALID --- Comment #2 from Anton Korobeynikov 2009-09-25 03:06:53 --- The code is fully correct according to x86 ABI. For struct return the caller allocates the place for hidden sret arg, but callee cleans the stack. You 'main' function has a non-standard signature, thus if there is something which relies on it, the things might be broken. -- 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 25 04:54:01 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 04:54:01 -0500 Subject: [LLVMbugs] [Bug 5047] New: [llvm 2.6 prerelease] Test failure on DebugInfo/ 2009-01-15-dbg_declare.ll Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5047 Summary: [llvm 2.6 prerelease] Test failure on DebugInfo/2009-01- 15-dbg_declare.ll Product: Test Suite Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: minor Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: voyageursp at gmail.com CC: llvmbugs at cs.uiuc.edu Hi, I am testing llvm packaging for Gentoo with current prereleases of llvm 2.6 (including prerelease 2). The test suite runs fine except for one small test: Running /var/tmp/portage/sys-devel/llvm-2.6_pre2/work/llvm-2.6/test/DebugInfo/dg.exp ... FAIL: /var/tmp/portage/sys-devel/llvm-2.6_pre2/work/llvm-2.6/test/DebugInfo/2009-01-15-dbg_declare.ll Failed with exit(1) at line 1 while running: llvm-as < /var/tmp/portage/sys-devel/llvm-2.6_pre2/work/llvm-2.6/test/DebugInfo/2009-01-15-dbg_declare.ll | llc -f -o /dev/null llc: error auto-selecting target for module 'No available targets are compatible with this triple'. Please use the -march option to explicitly pick a target. with configure run as: ./configure --prefix=/usr --build=i686-pc-linux-gnu --host=i686-pc-linux-gnu --mandir=/usr/share/man --infodir=/usr/share/info --datadir=/usr/share --sysconfdir=/etc --localstatedir=/var/lib --enable-optimized --disable-assertions --disable-expensive-checks --enable-targets=host-only This is fixed in llvm trunk (revs 80000 and 80210), but not merged in 2.6 branch This would be nice to have in final 2.6, so 'make check' runs ok without tweaking one of the tests -- 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 25 06:49:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 06:49:15 -0500 Subject: [LLVMbugs] [Bug 5043] Error compiling ASTContext.cpp with VS2008 In-Reply-To: Message-ID: <200909251149.n8PBnFWh031770@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5043 Benjamin Kramer changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |benny.kra at gmail.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Benjamin Kramer 2009-09-25 06:49:14 --- Fixed in r82762. http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090921/021738.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 25 09:51:26 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 09:51:26 -0500 Subject: [LLVMbugs] [Bug 4926] Assertion in rewriteARMFrameIndex() In-Reply-To: Message-ID: <200909251451.n8PEpQfE022910@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4926 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #11 from Bob Wilson 2009-09-25 09:51:25 --- Fixed in svn 82767. -- 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 25 10:13:57 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 10:13:57 -0500 Subject: [LLVMbugs] [Bug 605] [CBE] C Type layout doesn' t match LLVM type layout on Darwin in all cases In-Reply-To: Message-ID: <200909251513.n8PFDvfK025772@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=605 ????????? changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |yonggangluo at hotmail.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 25 11:26:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 11:26:59 -0500 Subject: [LLVMbugs] [Bug 605] [CBE] C Type layout doesn' t match LLVM type layout on Darwin in all cases In-Reply-To: Message-ID: <200909251626.n8PGQxNQ002580@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=605 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #4 from Chris Lattner 2009-09-25 11:26:59 --- This is not fixed, you're not using the C backend. -- 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 25 11:31:27 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 11:31:27 -0500 Subject: [LLVMbugs] [Bug 5045] multiply by zero sometimes != zero In-Reply-To: Message-ID: <200909251631.n8PGVRhS003189@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5045 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from Dan Gohman 2009-09-25 11:31:27 --- This is already fixed in 2.6; when fmul was split off from mul, a few problems of this nature were cleaned up. -- 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 25 12:08:11 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 12:08:11 -0500 Subject: [LLVMbugs] [Bug 5049] New: incorrect asserting conditions in TargetLowering.h Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5049 Summary: incorrect asserting conditions in TargetLowering.h Product: libraries Version: trunk Platform: PC OS/Version: Windows NT Status: NEW Severity: normal Priority: P2 Component: Target Description Classes AssignedTo: unassignedbugs at nondot.org ReportedBy: micah.villmow at amd.com CC: llvmbugs at cs.uiuc.edu setLoadExtAction(ISD::SEXTLOAD, MVT::v2f64, Expand) or setLoadExtAction(ISD::SEXTLOAD, MVT::v4f64, Expand) causes this assert to trigger: assert((unsigned)VT.SimpleTy < sizeof(LoadExtActions[0])*4 && ExtType < array_lengthof(LoadExtActions) && "Table isn't big enough!"); @ line 982 because the type used to be uint32_t. The problem comes that sizeof(LoadExtActions[0]) == 8 and 8 * 4 = 32, but both v2f64 and v4f64 have ID's 32 and 33 respectively, causing this assert to trigger. The correct method of doing this assert should be assert((unsigned)VT.SimpleTy < LAST_VALUETYPE && ExtType < array_lengthof(LoadExtActions) && "Table isn't big enough!"); Other asserts that also will be triggered because of this are in setTruncStoreAction, setConvertAction, setCondCodeAction, when the MemVT == (v2f64 || v4f64). The functions setIndexedLoadAction and setIndexedStoreAction actually get this correct, so someone has hit this assert before but neglected to updated all asserts in the 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 Fri Sep 25 12:52:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 12:52:35 -0500 Subject: [LLVMbugs] [Bug 5052] New: bug in sequence 'load --> bswap --> bitcast --> fptoui' Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5052 Summary: bug in sequence 'load --> bswap --> bitcast --> fptoui' Product: new-bugs Version: trunk Platform: PC OS/Version: other Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: eugeny.grishul at gmail.com CC: llvmbugs at cs.uiuc.edu Hi, Currently I develop own compiler using LLVM ( and MS Phoenix framework =) ) and found that 'fptoui' instruction usage in following situation crashes JIT: ;=========================================== @System.Test.ByteOrderTests.big2 = internal global i32 60994 @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" declare i32 @printf(i8* nocapture, ...) nounwind declare i32 @llvm.bswap.i32(i32) nounwind readnone define void @main() { %1 = load i32* @System.Test.ByteOrderTests.big2 ; [#uses=1] %2 = call i32 @llvm.bswap.i32(i32 %1) %3 = bitcast i32 %2 to float %4 = fptoui float %3 to i64 %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), i64 %4 ) ret void } ;Writes: ;0. Program arguments: lli test.bc ;00000000 (0x42EE0000 0x00000000 0x00D6FA60 0x01280E01) ;001D0023 (0x00C1F664 0x00C48F50 0x00C1F734 0x54AD8E88) ;0153245C (0x00C48F50 0x022941BC 0x00D61D18 0x54AD80F4), llvm::ExecutionEngine::runFunctionAsMain()+0940 bytes(s), e:\external\llvm\lib\executionengine\executionengine.cpp, line 387+0030 byte(s) ;01272C44 (0x00000002 0x00D680F8 0x00D61D18 0x54AD80A4), main()+2196 bytes(s), e:\external\llvm\tools\lli\lli.cpp, line 218+0064 byte(s) ;01C5FF78 (0x00C1F984 0x74DB3677 0x7EFDE000 0x00C1F9C4), __tmainCRTStartup()+0424 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 586+0025 byte(s) ;01C5FDBF (0x7EFDE000 0x00C1F9C4 0x77159D72 0x7EFDE000), mainCRTStartup()+0015 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 403 ;74DB3677 (0x7EFDE000 0x778EDA85 0x00000000 0x00000000), BaseThreadInitThunk()+0018 bytes(s) ;77159D72 (0x01C5FDB0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0099 bytes(s) ;77159D45 (0x01C5FDB0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0054 bytes(s) ;=========================================== But if I change 'fptoui' to 'fptosi' everything is OK: @System.Test.ByteOrderTests.big2 = internal global i32 60994 @str = internal constant [18 x i8] c"Big endian: %lld\0A\00" declare i32 @printf(i8* nocapture, ...) nounwind declare i32 @llvm.bswap.i32(i32) nounwind readnone define void @main() { %1 = load i32* @System.Test.ByteOrderTests.big2 ; [#uses=1] %2 = call i32 @llvm.bswap.i32(i32 %1) %3 = bitcast i32 %2 to float %4 = fptosi float %3 to i64 %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), i64 %4 ) ret void } ;Writes: Big endian: 119 ;=========================================== I'm using llvm from trunk (23 sept 2009) on Windows 7 ( build 7600 ) To run test I used: llvm-as test.ll lli test.bc WBR, Eugeny Grishul -- 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 25 13:33:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 13:33:06 -0500 Subject: [LLVMbugs] [Bug 5037] arm_neon.h float32_t is not compatible with float In-Reply-To: Message-ID: <200909251833.n8PIX6Zb019335@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5037 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Bob Wilson 2009-09-25 13:33:05 --- I was wrong about this issue applying to the integer vector element types. The proposed fix looks good to me, and I've applied it in svn 82787. -- 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 25 14:10:30 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 14:10:30 -0500 Subject: [LLVMbugs] [Bug 5053] New: incorrect 'frem' behaviour with floats Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5053 Summary: incorrect 'frem' behaviour with floats Product: new-bugs Version: trunk Platform: PC OS/Version: Windows NT Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: eugeny.grishul at gmail.com CC: llvmbugs at cs.uiuc.edu Use following test case: target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i686-pc-win32" @flt = internal global float 12.0e+0 @str = internal constant [18 x i8] c"Double value:%f \0A\00" declare i32 @printf(i8* nocapture, ...) nounwind define void @main() { %flt = load float* @flt %float2 = frem float %flt, 5.0 %double1 = fpext float %float2 to double %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), double %double1 ) ret void } Running with JIT: Stack dump: 0. Program arguments: lli test.bc 00000000 (0x41400000 0x40A00000 0x00FEFA40 0x0008D3B1) 00FF003B (0x01A1F6B4 0x01BFB4B0 0x01A1F784 0x5C1E1B36) 0033EA9C (0x01BFB4B0 0x00D81B5C 0x00FE1D20 0x5C1E15EA), llvm::ExecutionEngine::runFunctionAsMain()+0940 bytes(s), e:\external\llvm\lib\executionengine\executionengine.cpp, line 387+0030 byte(s) 00081894 (0x00000002 0x00FE7FD8 0x00FE1D20 0x5C1E155A), main()+2196 bytes(s), e:\external\llvm\tools\lli\lli.cpp, line 218+0064 byte(s) 00A6BFB8 (0x01A1F9D4 0x776F3677 0x7EFDE000 0x01A1FA14), __tmainCRTStartup()+0424 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 586+0025 byte(s) 00A6BDEF (0x7EFDE000 0x01A1FA14 0x77C99D72 0x7EFDE000), mainCRTStartup()+0015 bytes(s), f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c, line 403 776F3677 (0x7EFDE000 0x7623FA65 0x00000000 0x00000000), BaseThreadInitThunk()+0018 bytes(s) 77C99D72 (0x00A6BDE0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0099 bytes(s) 77C99D45 (0x00A6BDE0 0x7EFDE000 0x00000000 0x00000000), RtlInitializeExceptionChain()+0054 bytes(s) Running with interpreter: Double value:2.000000 But if I change floats to doubles - all is OK: target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple = "i686-pc-win32" @flt = internal global double 12.0e+0 @str = internal constant [18 x i8] c"Double value:%f \0A\00" declare i32 @printf(i8* nocapture, ...) nounwind define void @main() { %flt = load double* @flt %float2 = frem double %flt, 5.0 %yo = call i32(i8* nocapture, ...)* @printf( i8* getelementptr ([18 x i8]* @str, i32 0, i64 0), double %float2 ) ret void } -- 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 25 15:47:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 15:47:04 -0500 Subject: [LLVMbugs] [Bug 5037] arm_neon.h float32_t is not compatible with float In-Reply-To: Message-ID: <200909252047.n8PKl45D004481@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5037 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #3 from Bob Wilson 2009-09-25 15:47:03 --- The patch I applied breaks the ARM-specified mangling for NEON vector types: --- mangle-neon.s.good 2009-09-25 13:41:01.000000000 -0700 +++ mangle-neon.s 2009-09-25 13:41:57.000000000 -0700 @@ -80,11 +80,11 @@ bx lr - .globl __Z2f618__simd64_float32_t + .globl __Z2f6U8__vectorf .align 2 .code 16 - .thumb_func __Z2f618__simd64_float32_t -__Z2f618__simd64_float32_t: + .thumb_func __Z2f6U8__vectorf +__Z2f6U8__vectorf: sub sp, #2 * 4 fmdrr d0, r0, r1 fstd d0, [sp] @@ -227,11 +227,11 @@ bx lr - .globl __Z3f1519__simd128_float32_t + .globl __Z3f15U8__vectorf .align 2 .code 16 - .thumb_func __Z3f1519__simd128_float32_t -__Z3f1519__simd128_float32_t: + .thumb_func __Z3f15U8__vectorf +__Z3f15U8__vectorf: sub sp, #4 * 4 fmdrr d0, r2, r3 fmdrr d1, r0, r1 -- 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 25 18:40:02 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 18:40:02 -0500 Subject: [LLVMbugs] [Bug 5055] New: Another regscavenger assertion Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5055 Summary: Another regscavenger assertion Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: Alpha AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3564) --> (http://llvm.org/bugs/attachment.cgi?id=3564) Reduced bytecode Some recent commits broke llvm-gcc build for ARM. Original and reduced bytecode 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 Fri Sep 25 22:21:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 22:21:40 -0500 Subject: [LLVMbugs] [Bug 5058] New: Support for sret the msvc++ way. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5058 Summary: Support for sret the msvc++ way. Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: major Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu Contrary to gcc, on msvc++ is the caller the responsible of freeing the stack space used by the sret pointer parameter. Right now if LLVM code calls a function compiled with msvc++ that returns a struct, the application will crash due to the unbalanced stack (if the signature of the external function contains sret. If no sret is used and the address of the return struct is passed as a regular parameter, the call works). There are other differences on sret handling among gcc and msvc++. Those are listed on page 20 of http://www.agner.org/optimize/calling_conventions.pdf -- 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 25 23:24:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 25 Sep 2009 23:24:15 -0500 Subject: [LLVMbugs] [Bug 5059] New: Invalid child # of SDNodes Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5059 Summary: Invalid child # of SDNodes Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: PowerPC AssignedTo: unassignedbugs at nondot.org ReportedBy: wendling at apple.com CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3567) --> (http://llvm.org/bugs/attachment.cgi?id=3567) Bugpoint reduced program The attached program borks when compiled. This is from bootstrapping on a PowerBook G4. $ llc -o /dev/null bugpoint-reduced-simplified.bc /Users/wendling/llvm/llvm.src/lib/CodeGen/MachineInstr.cpp:76: failed assertion `isOnRegUseList() && "Reg operand is not on a use list"' Stack dump: 0. Program arguments: ./llvm.obj/Debug/bin/llc -o /dev/null bugpoint-reduced-simplified.bc 1. Running pass 'Live Interval Analysis' on function '@_nl_load_domain' Abort trap -- 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 26 00:47:06 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 00:47:06 -0500 Subject: [LLVMbugs] [Bug 5047] [llvm 2.6 prerelease] Test failure on DebugInfo/ 2009-01-15-dbg_declare.ll In-Reply-To: Message-ID: <200909260547.n8Q5l6fr007584@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5047 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WONTFIX --- Comment #1 from Chris Lattner 2009-09-26 00:47:05 --- Unfortunately, this is too late for 2.6, I'm glad it is fixed in trunk though. -- 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 26 11:47:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 11:47:44 -0500 Subject: [LLVMbugs] [Bug 5060] New: clang ignores attribute aligned on some variables Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5060 Summary: clang ignores attribute aligned on some variables Product: clang Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: gohman at apple.com CC: llvmbugs at cs.uiuc.edu For this C source: char *foo(void) { static char p __attribute__((aligned(32))); return &p; } void frob(char *); void bar(long n) { char p[n] __attribute__((aligned(32))); frob(p); } clang emits LLVM IR containing these two constructs, missing the alignment information. @foo.p = internal global i8 0 ; [#uses=1] %vla = alloca i8, i32 %n3, align 1 ; [#uses=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 Sat Sep 26 13:17:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 13:17:43 -0500 Subject: [LLVMbugs] [Bug 5060] clang ignores attribute aligned on some variables In-Reply-To: Message-ID: <200909261817.n8QIHhfM018315@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5060 Anders Carlsson changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |andersca at mac.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Anders Carlsson 2009-09-26 13:17:42 --- http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090921/021798.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 26 15:50:17 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 15:50:17 -0500 Subject: [LLVMbugs] [Bug 5062] New: builtin_trap() does not set noreturn Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5062 Summary: builtin_trap() does not set noreturn Product: clang Version: unspecified Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: Basic AssignedTo: unassignedclangbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu lev /tmp$ cat rw.c void foo(void) __attribute__((__noreturn__)); void foo(void) { __builtin_trap(); } lev /tmp$ gcc -c rw.c && echo yes yes lev /tmp$ clang -c rw.c && echo yes rw.c:6:1: warning: function declared 'noreturn' should not return [-Winvalid-noreturn] } ^ 1 diagnostic generated. the attached patch fixes 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 26 16:16:32 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 16:16:32 -0500 Subject: [LLVMbugs] [Bug 5062] builtin_trap() does not set noreturn In-Reply-To: Message-ID: <200909262116.n8QLGWYK009487@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5062 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-26 16:16:32 --- Nice, applied w/ testcase: http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090921/021804.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 26 22:15:35 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 26 Sep 2009 22:15:35 -0500 Subject: [LLVMbugs] [Bug 5064] New: Support byval the msvc++ way. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5064 Summary: Support byval the msvc++ way. Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: ofv at wanadoo.es CC: llvmbugs at cs.uiuc.edu When a C++ struct function argument has a destructor or copy constructor and is passed by value, GCC creates a temporary copy somewhere and passes a pointer to the temporary copy to the function. LLVM does not support this directly and is the front-end the responsible of creating the temporary and use the pointer as the function's parameter. MSVC++ does not use this trick: the C++ class is put on the stack along with the rest of function parameters. As LLVM is the responsible of arranging the parameters on the stack for the function call, this requires that LLVM know about the copy constructor and the destructor. Right now there is no workaround for this, so LLVM code is unable to call functions compiled with MSVC++ that take C++ classes by value (when those classes have a copy constructor and/or destructor). LLVM's byval should be extended to allow the front-end to specify a copy constructor and a destructor, and insert calls to them when the temporary is created and after the call is performed. -- 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 27 03:08:01 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 03:08:01 -0500 Subject: [LLVMbugs] [Bug 3852] symbol relocation; llvm-gcc V. gcc In-Reply-To: Message-ID: <200909270808.n8R881fO005938@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3852 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #2 from Chris Lattner 2009-09-27 03:08:00 --- We need more information. -- 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 27 03:09:04 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 03:09:04 -0500 Subject: [LLVMbugs] [Bug 3357] Kaleidoscope Tutorial Ch3 doesn't compile In-Reply-To: Message-ID: <200909270809.n8R89464006085@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=3357 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #4 from Chris Lattner 2009-09-27 03:09:04 --- I think this got fixed on mainline -- 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 27 13:29:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 13:29:16 -0500 Subject: [LLVMbugs] [Bug 5070] New: can't return vector of three ints Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5070 Summary: can't return vector of three ints Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca CC: llvmbugs at cs.uiuc.edu I tried a couple trivial non power of two vectors just to check whether they really work: define <3 x i8> @foo(<3 x i8> %x) { ret <3 x i8> %x } $ llvm-commit/Debug/bin/llc x.ll -o - .file "x.ll" Return operand #2 has unhandled type i8UNREACHABLE executed at CallingConvLower.cpp:93! 0 llc 0x08bd0635 1 llc 0x08bd0c01 2 0xffffe400 __kernel_sigreturn + 0 Stack dump: 0. Program arguments: llvm-commit/Debug/bin/llc x.ll -o - 1. Running pass 'X86 DAG->DAG Instruction Selection' on function '@foo' Aborted define <3 x i32> @foo(<3 x i32> %x) { ret <3 x i32> %x } $ llvm-commit/Debug/bin/llc x.ll -o - .file "x.ll" Return operand #2 has unhandled type i32UNREACHABLE executed at CallingConvLower.cpp:93! 0 llc 0x08bd0635 1 llc 0x08bd0c01 2 0xffffe400 __kernel_sigreturn + 0 Stack dump: 0. Program arguments: llvm-commit/Debug/bin/llc x.ll -o - 1. Running pass 'X86 DAG->DAG Instruction Selection' on function '@foo' Aborted This is on x86, not 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 Sun Sep 27 13:37:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 13:37:55 -0500 Subject: [LLVMbugs] [Bug 5070] can't return vector of three ints In-Reply-To: Message-ID: <200909271837.n8RIbthc013383@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5070 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Chris Lattner 2009-09-27 13:37:55 --- *** 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 Sun Sep 27 14:58:41 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 14:58:41 -0500 Subject: [LLVMbugs] [Bug 4895] Missed phi/phi/select folding opportunity In-Reply-To: Message-ID: <200909271958.n8RJwfQe023710@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4895 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #2 from Chris Lattner 2009-09-27 14:58:40 --- Committed with tweaks here, thanks! http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/087999.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 Sun Sep 27 15:01:50 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 15:01:50 -0500 Subject: [LLVMbugs] [Bug 4895] Missed phi/phi/select folding opportunity In-Reply-To: Message-ID: <200909272001.n8RK1ovs024163@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4895 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #3 from Chris Lattner 2009-09-27 15:01:50 --- The original testcase isn't handled here. Improving fix. -- 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 27 15:20:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 27 Sep 2009 15:20:34 -0500 Subject: [LLVMbugs] [Bug 4895] Missed phi/phi/select folding opportunity In-Reply-To: Message-ID: <200909272020.n8RKKYhP026640@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4895 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #4 from Chris Lattner 2009-09-27 15:20:34 --- Improved fix here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090921/088002.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 28 00:29:14 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 00:29:14 -0500 Subject: [LLVMbugs] [Bug 5055] Another regscavenger assertion In-Reply-To: Message-ID: <200909280529.n8S5TEic030275@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5055 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #7 from Evan Cheng 2009-09-28 00:29:13 --- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090928/088054.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 28 01:27:42 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 01:27:42 -0500 Subject: [LLVMbugs] [Bug 5055] Another regscavenger assertion In-Reply-To: Message-ID: <200909280627.n8S6Rg6k004933@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5055 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #8 from Anton Korobeynikov 2009-09-28 01:27:42 --- The second reduction is still failing... -- 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 28 01:43:45 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 01:43:45 -0500 Subject: [LLVMbugs] [Bug 5073] New: Assert failure: Operands of AddRec must be loop-invariant! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5073 Summary: Assert failure: Operands of AddRec must be loop- invariant! 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: baldrick at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3578) --> (http://llvm.org/bugs/attachment.cgi?id=3578) testcase .ll $ opt bugpoint-reduced-simplified.bc -indvars opt: llvm/include/llvm/Analysis/ScalarEvolutionExpressions.h:381: llvm::SCEVAddRecExpr::SCEVAddRecExpr(const llvm::FoldingSetNodeID&, const llvm::SmallVectorImpl&, const llvm::Loop*): Assertion `Operands[i]->isLoopInvariant(l) && "Operands of AddRec must be loop-invariant!"' 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 Mon Sep 28 02:29:22 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 02:29:22 -0500 Subject: [LLVMbugs] [Bug 5055] Another regscavenger assertion In-Reply-To: Message-ID: <200909280729.n8S7TMqE012869@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5055 Anton Korobeynikov changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #14 from Anton Korobeynikov 2009-09-28 02:29:22 --- (In reply to comment #13) > Ok. It's not safe to mark MOVTi16 rematerializable because of the register > input. I am testing a fix. Ok. I've disabled a remat of it 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 Mon Sep 28 13:13:53 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 13:13:53 -0500 Subject: [LLVMbugs] [Bug 5087] New: ASM instruction shouldn't crash JIT. Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5087 Summary: ASM instruction shouldn't crash JIT. Product: tools Version: trunk Platform: PC OS/Version: All Status: NEW Keywords: crash-on-invalid Severity: minor Priority: P2 Component: lli AssignedTo: kuba at gcc.gnu.org ReportedBy: kuba at gcc.gnu.org CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3579) --> (http://llvm.org/bugs/attachment.cgi?id=3579) test-case I noticed that ASM instruction crashes JIT on x86. When i was trying to use llvm_report_error() function instead of assert() i noticed that it caused other crash. -- 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 28 13:38:37 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 13:38:37 -0500 Subject: [LLVMbugs] [Bug 5088] New: verify that the jit doesn' t compile available_externally methods Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5088 Summary: verify that the jit doesn't compile available_externally methods Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: nicholas at mxc.ca ReportedBy: nlewycky at google.com CC: llvmbugs at cs.uiuc.edu Write a unit test which provides an externally visible function f() which returns 1 and a main() that runs the JIT. Inside the module, f() has an available_externally definition of f() which return 2, and function g() which calls 'return f()'. Verify that running JIT->getPointerToFunction(g) returns 1, not 2. -- 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 28 14:11:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 14:11:34 -0500 Subject: [LLVMbugs] [Bug 5089] New: extern inline problem Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5089 Summary: extern inline problem Product: clang Version: unspecified Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: Semantic Analyzer AssignedTo: unassignedclangbugs at nondot.org ReportedBy: rdivacky at freebsd.org CC: llvmbugs at cs.uiuc.edu witten /tmp# cat toplev.c extern inline int floor_log2 (unsigned int x) { } int floor_log2 (unsigned int x) { } witten /tmp# gcc -c toplev.c && echo yes yes witten /tmp# clang -c toplev.c && echo yes toplev.c:4:1: warning: control reaches end of non-void function [-Wreturn-type] } ^ toplev.c:7:1: error: redefinition of 'floor_log2' floor_log2 (unsigned int x) ^ toplev.c:2:1: note: previous definition is here floor_log2 (unsigned int x) ^ 3 diagnostics generated. -- 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 28 16:14:48 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 16:14:48 -0500 Subject: [LLVMbugs] [Bug 5089] extern inline problem In-Reply-To: Message-ID: <200909282114.n8SLEmE3015105@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5089 Dale Johannesen changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |dalej at apple.com Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #3 from Dale Johannesen 2009-09-28 16:14:45 --- *** This bug has been marked as a duplicate of bug 3341 *** -- 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 28 18:40:58 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 18:40:58 -0500 Subject: [LLVMbugs] [Bug 5092] New: Assertion failure in TranslateX86CC with SetCCOpcode == ISD ::SETOEQ Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5092 Summary: Assertion failure in TranslateX86CC with SetCCOpcode == ISD::SETOEQ Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: llvm at laurentm.net CC: llvmbugs at cs.uiuc.edu ----- bug.ll ----- define <4 x float> @bug(float %a) nounwind { entry: %cmp = fcmp oeq float %a, 0.000000e+00 ; [#uses=1] %temp = select i1 %cmp, <4 x float> , <4 x float> zeroinitializer ret <4 x float> %temp } ---------- $ cat bug.ll | llvm-as | llc -march=x86 .file "" Condcode should be pre-legalized away UNREACHABLE executed at X86ISelLowering.cpp:2267! 0 llc 0x00000000014608d8 1 llc 0x0000000001460e37 2 libpthread.so.0 0x00007faaf252b080 3 libc.so.6 0x00007faaf182efb5 gsignal + 53 4 libc.so.6 0x00007faaf1830bc3 abort + 387 5 llc 0x00000000014416bd llvm::llvm_report_error(llvm::Twine const&) + 0 6 llc 0x0000000000f7ffd7 7 llc 0x0000000000f816d3 llvm::X86TargetLowering::LowerSETCC(llvm::SDValue, llvm::SelectionDAG&) + 2149 8 llc 0x0000000000f82416 llvm::X86TargetLowering::LowerSELECT(llvm::SDValue, llvm::SelectionDAG&) + 186 9 llc 0x0000000000f835c6 llvm::X86TargetLowering::LowerOperation(llvm::SDValue, llvm::SelectionDAG&) + 1138 10 llc 0x00000000010c5720 11 llc 0x00000000010c5a25 12 llc 0x00000000010c5b23 llvm::SelectionDAG::LegalizeVectors() + 35 13 llc 0x000000000107fdac llvm::SelectionDAGISel::CodeGenAndEmitDAG() + 1936 14 llc 0x00000000010821a7 llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::ilist_iterator, llvm::ilist_iterator) + 673 15 llc 0x0000000001082cbc llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&, llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*, llvm::TargetInstrInfo const&) + 2598 16 llc 0x0000000001083b3e llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) + 934 17 llc 0x000000000117af4c llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 80 18 llc 0x00000000013e51a0 llvm::FPPassManager::runOnFunction(llvm::Function&) + 326 19 llc 0x00000000013e6eef llvm::FunctionPassManagerImpl::run(llvm::Function&) + 111 20 llc 0x00000000013e7096 llvm::FunctionPassManager::run(llvm::Function&) + 112 21 llc 0x0000000000915624 main + 3122 22 libc.so.6 0x00007faaf181a5a6 __libc_start_main + 230 23 llc 0x0000000000913b19 Stack dump: 0. Program arguments: /tmp/llvm/Debug/bin/llc -march=x86 1. Running pass 'X86 DAG->DAG Instruction Selection' on function '@bug' Aborted (core dumped) -- 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 28 19:39:55 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 19:39:55 -0500 Subject: [LLVMbugs] [Bug 5088] verify that the jit doesn't compile available_externally methods In-Reply-To: Message-ID: <200909290039.n8T0dtP4030724@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5088 Chris Lattner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Chris Lattner 2009-09-28 19:39:54 --- Please don't use bugzilla for your random personal todo list. -- 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 28 23:02:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 28 Sep 2009 23:02:43 -0500 Subject: [LLVMbugs] [Bug 5095] New: opt -loop-deletion crashes Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5095 Summary: opt -loop-deletion crashes Product: libraries Version: trunk Platform: PC OS/Version: MacOS X Status: NEW Severity: critical Priority: P2 Component: Loop Optimizer AssignedTo: unassignedbugs at nondot.org ReportedBy: sreeram at tachyontech.net CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3581) --> (http://llvm.org/bugs/attachment.cgi?id=3581) reduced file produced by bugpoint Running opt -loop-deletion on the attached .bc file crashes. I'm on Mac OS X 10.4. LLVM revision r82993. -- 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 29 01:28:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 01:28:59 -0500 Subject: [LLVMbugs] [Bug 5081] New: LLVM ERROR: Cannot yet select: 0x1ba3410: i64 = X86ISD:: Wrapper 0x1ba3b10 Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5081 Summary: LLVM ERROR: Cannot yet select: 0x1ba3410: i64 = X86ISD::Wrapper 0x1ba3b10 Product: new-bugs Version: 2.6 Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: martin.nowack at inf.tu-dresden.de CC: llvmbugs at cs.uiuc.edu llc crashes using the attached reduced bugpoint file. I do not have a glue what the reason is. The current llvm version I use is 2.6. Thanks for help. --- Comment #1 from Martin Nowack 2009-09-28 08:25:14 --- *** Bug 5083 has been marked as a duplicate of this bug. *** --- Comment #2 from Martin Nowack 2009-09-28 09:18:58 --- Currently, I cannot attach the file. Please email me and I will send it to you. Thanks. --- Comment #3 from Martin Nowack 2009-09-29 01:28:59 --- Created an attachment (id=3582) --> (http://llvm.org/bugs/attachment.cgi?id=3582) bugpoint 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 Tue Sep 29 02:08:34 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 02:08:34 -0500 Subject: [LLVMbugs] [Bug 4687] arm10tdmi (armv5) does not support strd In-Reply-To: Message-ID: <200909290708.n8T78YVB000432@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4687 Evan Cheng changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #6 from Evan Cheng 2009-09-29 02:08:33 --- Thanks. I've applied the patch. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090928/088107.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 29 06:44:39 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 06:44:39 -0500 Subject: [LLVMbugs] [Bug 5097] New: Buffer overflow in LoopSimplify Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5097 Summary: Buffer overflow in LoopSimplify 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: baldrick at free.fr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3583) --> (http://llvm.org/bugs/attachment.cgi?id=3583) testcase .ll Reproduce by building with expensive checking and doing: opt bugpoint-reduced-simplified.ll -loop-unswitch -loopsimplify -iv-users The problem seems to be that in lines 275...282 of LoopSimplify.cpp const std::vector *> &Children = Node->getChildren(); for (unsigned k = 0, g = Children.size(); k != g; ++k) { DT->changeImmediateDominator(Children[k], Node->getIDom()); if (DF) DF->changeImmediateDominator(Children[k]->getBlock(), Node->getIDom()->getBlock(), DT); } the code inside the loop mutates the Node->getChildren() vector. Since Children is a reference to this vector, it can change as the loop is executed. In the testcase Children has length 1, so the loop is entered. It looks like the first line resizes Children to be empty. Then the second line accesses off the end of Children. -- 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 29 11:20:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 11:20:16 -0500 Subject: [LLVMbugs] [Bug 5095] opt -loop-deletion crashes In-Reply-To: Message-ID: <200909291620.n8TGKGHv019158@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5095 devang.patel changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME --- Comment #1 from devang.patel 2009-09-29 11:20:16 --- Was this a transient issue? I do not see crash with r83066. -- 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 29 12:31:24 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 12:31:24 -0500 Subject: [LLVMbugs] [Bug 5073] Assert failure: Operands of AddRec must be loop-invariant! In-Reply-To: Message-ID: <200909291731.n8THVOPq028559@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5073 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |gohman at apple.com Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Dan Gohman 2009-09-29 12:31:23 --- This is fixed in r83011. It would be interesting to look at the generated code from the full testcase, since ScalarEvolution's handling of the polynomial recurrence has room for improvement. -- 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 29 15:17:15 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 15:17:15 -0500 Subject: [LLVMbugs] [Bug 5099] New: RewriteRope strictly undefined behavior, null 'this' Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5099 Summary: RewriteRope strictly undefined behavior, null 'this' Product: clang Version: unspecified Platform: All URL: http://clang.llvm.org/doxygen/RewriteRope_8h-source.html OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Basic AssignedTo: unassignedclangbugs at nondot.org ReportedBy: agrant at arm.com CC: llvmbugs at cs.uiuc.edu RewriteRope and RopeRefCountString rely on being able to call a non-virtual member function on a null pointer and then test 'this' in the member function implementation; see RopeRefCountString::addRef and dropRef. This is undefined behavior in standard C++ and implementations exist that will produce unexpected behavior (e.g. optimize away the test on 'this' because it's flagged as a non-null argument). The fix is to guard each call to addRef and dropRef with a test on the object pointer, so that it is no longer necessary to test 'this' - there is no net change in the number of runtime tests. -- 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 29 16:20:38 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 16:20:38 -0500 Subject: [LLVMbugs] [Bug 5100] New: several ARMv6T2 ops still missing Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5100 Summary: several ARMv6T2 ops still missing Product: new-bugs Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu BFI, SBFX, and UBFX patterns still need to be implemented. -- 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 29 16:21:51 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 29 Sep 2009 16:21:51 -0500 Subject: [LLVMbugs] [Bug 5101] New: ARMv6T2 and later should materialize GlobalAddresses using MOVW/MOVT Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5101 Summary: ARMv6T2 and later should materialize GlobalAddresses using MOVW/MOVT Product: new-bugs Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: deeppatel1987 at gmail.com CC: asl at math.spbu.ru, llvmbugs at cs.uiuc.edu -- 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 30 03:56:08 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 03:56:08 -0500 Subject: [LLVMbugs] [Bug 5104] New: printf returns 1 even when it fails if -simplify-libcalls is used Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5104 Summary: printf returns 1 even when it fails if -simplify- libcalls is used Product: new-bugs Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: timo.lindfors at iki.fi CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3589) --> (http://llvm.org/bugs/attachment.cgi?id=3589) testcase to reproduce the bug Steps to reproduce: 1) (testcase also as attachment) cat > printf-retval.c < int main(int argc, char *argv[]) { int ret; fclose(stdout); ret = printf("%c", 'A'); fprintf(stderr, "printf returned %d\n", ret); return 0; } EOF 2) clang -O0 printf-retval.c && ./a.out 3) clang -O3 printf-retval.c && ./a.out Expected results: 2 & 3) optimization does not change the return value of printf. It always returns a negative value on failure. stdout is closed so it will fail. Actual results: 2) printf returned -1 3) printf returned 1 More info: 1) It seems lib/Transforms/Scalar/SimplifyLibCalls.cpp converts printf to putchar in this case and ignores the return value of putchar completely. 2) The attached patch llvm-simplifylibcalls-putchar-retval1.patch works for me but probably needs to be polished by somebody more familiar with llvm internals. -- 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 30 06:54:24 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 06:54:24 -0500 Subject: [LLVMbugs] [Bug 5105] New: -O2 causes llvm/ clang to truncate the value passed to malloc Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5105 Summary: -O2 causes llvm/clang to truncate the value passed to malloc Product: clang Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: critical Priority: P2 Component: LLVM Codegen AssignedTo: unassignedclangbugs at nondot.org ReportedBy: js-llvm-bugzilla at webkeks.org CC: llvmbugs at cs.uiuc.edu This is on an AMD64 system. llvm revision: 83146 asgard:/tmp$ cat test.c #include #include #include int main() { printf("%p\n", malloc(SIZE_MAX)); return 0; } asgard:/tmp$ clang -S test.c -o test.S asgard:/tmp$ clang -S -O2 test.c -o test.O2.S asgard:/tmp$ cat test.S [???] movabsq $-1, %rax movq %rax, %rdi call malloc [???] asgard:/tmp$ cat test.O2.S [???] movl $4294967295, %edi call malloc [???] For some reason, llvm/clang truncates the value to 32 bits if -O2 is specified. -- 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 30 07:41:59 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 07:41:59 -0500 Subject: [LLVMbugs] [Bug 5105] -O2 causes llvm/clang to truncate the value passed to malloc In-Reply-To: Message-ID: <200909301241.n8UCfx7W016189@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5105 Benjamin Kramer changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |benny.kra at gmail.com Status|NEW |RESOLVED Resolution| |DUPLICATE --- Comment #1 from Benjamin Kramer 2009-09-30 07:41:58 --- *** This bug has been marked as a duplicate of bug 4664 *** -- 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 30 08:13:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 08:13:44 -0500 Subject: [LLVMbugs] [Bug 5107] New: Division by zero in SimpleRegisterCoalescing:: isWinToJoinCrossClass() Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5107 Summary: Division by zero in SimpleRegisterCoalescing::isWinToJoinCrossClass() Product: libraries Version: 2.6 Platform: PC OS/Version: Windows XP Status: NEW Severity: major Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: DevOmem at web.de CC: llvmbugs at cs.uiuc.edu Crash due to vivision by zero in SimpleRegisterCoalescing::isWinToJoinCrossClass() SmallSize can be some times zero and this will crash LLVM. Solution for now is to add this if(LargeSize<=0) LargeSize=1; if(SmallSize<=0) SmallSize=1; before if (SmallSize > Threshold || LargeSize > Threshold) if ((float)std::distance(mri_->use_begin(SmallReg),mri_->use_end()) / SmallSize < (float)std::distance(mri_->use_begin(LargeReg),mri_->use_end()) / LargeSize) -- 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 30 11:31:44 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 11:31:44 -0500 Subject: [LLVMbugs] [Bug 5109] New: long double type on ARM Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5109 Summary: long double type on ARM Product: tools Version: 2.5 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: jaykang10 at imrc.kist.re.kr CC: llvmbugs at cs.uiuc.edu Created an attachment (id=3591) --> (http://llvm.org/bugs/attachment.cgi?id=3591) bugpoint-reduced-simplified.bc Dear LLVM members. I am compiling coreutils-7.4 package for ARM linux using LLVM 2.5 version. When i compiled 'od' program in coreutils package using LLVM 2.5, i could see the error message on llc processing. > llvm-gcc -emit-llvm ./od.c -c -o ./od.bc -other-options... > llc -march=arm ./od.bc -f -o ./od.s llc: /home/jaykang10/HardDisk2/Projects/LLVM/src/llvm-2.5/lib/Target/TargetData.cpp:305: unsigned int llvm::TargetData::getAlignmentInfo(llvm::AlignTypeEnum, uint32_t, bool, const llvm::Type*) const: Assertion `AlignType == VECTOR_ALIGN && "Unknown alignment type!"' failed. The problem was that when i converted 'od' source code to llvm bitcode, long double type of 'od' source code was changed to x86_fp80 type in llvm bitcode and then llc on ARM target was not support x86_fp80 type. I think that long double type on ARM is 8 byte. (same as double type) When we use llvm on ARM target,(or other target not support x86_fp80) i think it might be better to convert long double type in C to double type in llvm bitcode on llvm-gcc or treat x86_fp80 type in llvm bitcode as double type on llc. (I wonder if we select the target at llvm-gcc build time) I would like to know how to solve the above problem. Thank you Best regards, Jin-Gu Kang -- 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 30 11:34:20 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 11:34:20 -0500 Subject: [LLVMbugs] [Bug 5110] New: LiveIntervals::getApproximateInstructionCount() some times return 0 (zero)! Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5110 Summary: LiveIntervals::getApproximateInstructionCount() some times return 0 (zero)! Product: libraries Version: 2.6 Platform: PC OS/Version: Windows NT Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: DevOmem at web.de CC: llvmbugs at cs.uiuc.edu LiveIntervals::getApproximateInstructionCount() some times return 0 (zero) and causes division by zero. Solution so far looks like this unsigned getApproximateInstructionCount(LiveInterval& I) { double IntervalPercentage = getScaledIntervalSize(I) / 1000.0; unsigned result = (unsigned)(IntervalPercentage * FunctionSize); if(result<=0) result = 1; return result; //return (unsigned)(IntervalPercentage * FunctionSize); } -- 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 30 11:38:38 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 11:38:38 -0500 Subject: [LLVMbugs] [Bug 5109] long double type on ARM In-Reply-To: Message-ID: <200909301638.n8UGcc01013952@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5109 Bob Wilson changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |bob.wilson at apple.com Status|NEW |RESOLVED Resolution| |INVALID --- Comment #1 from Bob Wilson 2009-09-30 11:38:38 --- It looks like you're using a copy of llvm-gcc that targets x86 to generate this bitcode file. You need to rebuild llvm-gcc to target arm. -- 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 30 13:53:40 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 13:53:40 -0500 Subject: [LLVMbugs] [Bug 5112] New: llc from LLVM 2.6 release branch segfaults on Linux Message-ID: http://llvm.org/bugs/show_bug.cgi?id=5112 Summary: llc from LLVM 2.6 release branch segfaults on Linux Product: new-bugs Version: 2.6 Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: collinwinter at google.com CC: llvmbugs at cs.uiuc.edu In the 2.6 release branch, llc segfaults when running the test suite on Linux (32/64 bit hybrid Ubuntu Dapper), failing 1400+ tests. Extracted from one of the failing tests: $ Release-Asserts/bin/llvm-as < /usr/local/google/collinwinter/llvm-26/test/Transforms/IndVarSimplify/2007-01-08-X86-64-Pointer.ll > /tmp/f.bc $ Release-Asserts/bin/llc -march=x86-64 < /tmp/f.bc Segmentation fault $ gdb Release-Asserts/bin/llc (gdb) run -march=x86-64 < /tmp/f.bc (gdb) run -march=x86-64 < /tmp/f.bc Starting program: /usr/local/google/collinwinter/llvm-26/Release-Asserts/bin/llc -march=x86-64 < /tmp/f.bc warning: Lowest section in system-supplied DSO at 0xffffe000 is .hash at ffffe0b4 [Thread debugging using libthread_db enabled] [New Thread 0xf7fe16c0 (LWP 17727)] Warning: guessed host libthread_db library "/usr/grte/v1/lib64/libthread_db.so.1". Using host libthread_db library "/usr/grte/v1/lib64/libthread_db.so.1". Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xf7fe16c0 (LWP 17727)] 0x46b0875c in memcpy () from /lib/tls/i686/cmov/libc.so.6 (gdb) bt #0 0x46b0875c in memcpy () from /lib/tls/i686/cmov/libc.so.6 #1 0x08d10d50 in llvm::Twine::printOneChild () #2 0x0818051e in ?? () #3 0x08e22b4c in ?? () #4 0xffffd7d0 in ?? () #5 0xffffd5c8 in ?? () #6 0x08d10da6 in llvm::Twine::print () #7 0x00000005 in ?? () #8 0x08e22b4c in ?? () #9 0xffffd5c8 in ?? () #10 0xffffd5c8 in ?? () #11 0x08d10cdd in llvm::Twine::printOneChild () #12 0x00000000 in ?? () (gdb) Configured with ./configure --enable-optimized --disable-assertions --enable-debug-symbols (the configuration we use for Unladen Swallow). This is using the Ubuntu system gcc 4.0.3, target i486-linux-gnu. Note that the problem goes away if I pass --enable-assertions to configure. This is http://llvm.org/svn/llvm-project/llvm/branches/release_26, r83156. -- 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 30 15:54:43 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 15:54:43 -0500 Subject: [LLVMbugs] [Bug 5097] Buffer overflow in LoopSimplify In-Reply-To: Message-ID: <200909302054.n8UKshnB013909@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=5097 Dan Gohman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Dan Gohman 2009-09-30 15:54:43 --- Fixed in r83166. -- 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 30 20:38:16 2009 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 30 Sep 2009 20:38:16 -0500 Subject: [LLVMbugs] [Bug 4401] Assertion failure in ScheduleDAGRList: instruction released too many times In-Reply-To: Message-ID: <200910010138.n911cG49017667@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=4401 reid.kleckner at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #5 from reid.kleckner at gmail.com 2009-09-30 20:38:16 --- This turns out to be a short integer overflow in ScheduleDAG.h, class SUnit. I committed a fix in r83159. -- 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.