[llvm-testresults] buildbot failure in smooshlab on llvm-gcc-powerpc-darwin9

Fri Feb 26 09:41:11 PST 2010

On Thu, Feb 25, 2010 at 11:26 PM, Evan Cheng <evan.cheng at apple.com> wrote:
> I just did a successful self-host with my patch re-applied.

Just a reminder, its more than a self-host, the failure was in the
second llvm-gcc build. You would need to do
  llvm -> llvm-gcc -> llvm -> llvm-gcc
to even possibly replicate the problem.

 - Daniel

> Evan
>
> On Feb 24, 2010, at 6:15 PM, Daniel Dunbar wrote:
>
>> On Wed, Feb 24, 2010 at 5:50 PM, Stuart Hastings <stuart at apple.com> wrote:
>>>
>>> On Feb 24, 2010, at 1:38 PM, Stuart Hastings wrote:
>>>
>>> On Feb 24, 2010, at 12:15 PM, Daniel Dunbar wrote:
>>>
>>> Just a reminder, this isn't a simple bootstrap, it is failing in the stage2
>>> llm-gcc compile. My guess is the first LLVM-gcc is miscompiling the second
>>> stage LLVM. That won't show up in a normal LLVM-gcc bootstrap.
>>>
>>> I have an llvmgcc bootstrap of r97011 running right now on smoosh-14.
>>>
>>> A straightforward FSF-style bootstrap of llvm-gcc at revision 97011 worked
>>> O.K.  It looks like I'll need to duplicate the build-install-build procedure
>>> used by the buildbot scripts.
>>> Daniel, is it feasible to run the buildbot script by hand in a directory of
>>> my own choosing?  If not, I think I can duplicate it by scanning the logs...
>>> stuart
>>
>> No, but its pretty straightforward, just do another configure and make
>> of llvm using the built llvm-gcc, then another llvm-gcc with the
>> second llvm and using the first llvm-gcc.
>>
>> And yes, you can see what it is doing by checking the first two lines
>> of the logs, the first line is the command that is run, the second is
>> the directory it ran it in.
>>
>> - Daniel
>>
>>>
>>> stuart
>>>
>>>
>>> - Daniel
>>>
>>>
>>> On Feb 24, 2010, at 11:12 AM, Evan Cheng <evan.cheng at apple.com> wrote:
>>>
>>> Well, I can't reproduce on my machine. My ppc machine self-hosted fine last
>>> night. I think there is something else going on. Stuart, Dale, can one of
>>> you guys reproduce this look at this a bit? Is it llvmCore being miscompiled
>>> or llvmgcc42? Please get me a test case if that's possible. Thanks.
>>>
>>> Dan, can you audit my patch carefully? Thanks.
>>>
>>> Evan
>>>
>>> On Feb 24, 2010, at 10:55 AM, Dale Johannesen wrote:
>>>
>>>
>>> On Feb 24, 2010, at 10:49 AMPST, Bob Wilson wrote:
>>>
>>> The ppc buildbot just turned green (with 97011 reverted).
>>>
>>> Yep.  I'm satisfied that was it.  Thanks Daniel.
>>>
>>> On Feb 24, 2010, at 10:43 AM, Dale Johannesen wrote:
>>>
>>>
>>> On Feb 24, 2010, at 9:34 AMPST, Evan Cheng wrote:
>>>
>>> On Feb 24, 2010, at 9:02 AM, Daniel Dunbar <daniel_dunbar at apple.com> wrote:
>>>
>>> On Feb 24, 2010, at 1:47 AM, Evan Cheng wrote:
>>>
>>> I saw this on x86 last night. Then it disappeared. I am moving tomorrow so I
>>> can't deal with it. Dale, could you help? Thanks.
>>>
>>> Ok, I'm going to try another speculative revert of r97011 then.
>>>
>>> You are not understanding me. I saw this assertion on x86 the night before.
>>> It was well before I commit the patch.
>>>
>>> You may have, but the x86 buildbots are not failing.  The ppc buildbots
>>> (both) are failing consistently, and logs indicate the breakage was between
>>> 97006 and 97012.   I suppose it could be nondeterminism but the observed
>>> behavior is quite deterministic.  I'll go talk to Stuart...
>>>
>>> Someone should actually debug the failure.
>>>
>>> Evan
>>>
>>>
>>> - Daniel
>>>
>>> Evan
>>>
>>> On Feb 24, 2010, at 12:44 AM, Daniel Dunbar <daniel_dunbar at apple.com> wrote:
>>>
>>> My revert of r97010 didn't work. Evan, 97011 seems the next obvious guess,
>>> can you take a look?
>>>
>>> - Daniel
>>>
>>> On Feb 23, 2010, at 10:45 PM, Daniel Dunbar wrote:
>>>
>>>
>>> On Feb 23, 2010, at 10:03 PM, Daniel Dunbar wrote:
>>>
>>> Evan, Jim, Bob, anyone have a guess what is going on here? I'm
>>>
>>> guessing Bob's patch is most likely since it touches an IR pass...
>>>
>>> I'll probably pull that in a bit if no one replies.
>>>
>>> I stashed off a .bc file, but I can't reproduce with it. I'm assuming this
>>> is most likely a miscompile of llvm stage1? Here is the info:
>>>
>>> --
>>>
>>> smoosh-14:libcpp buildslave$
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/./prev-gcc/xgcc
>>> -B/Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/./prev-gcc/
>>> -B/Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.install.2/powerpc-apple-darwin9/bin/
>>>  -I../../llvm-gcc.src/libcpp -I. -I../../llvm-gcc.src/libcpp/../include
>>> -I./../intl -I../../llvm-gcc.src/libcpp/include -DENABLE_LLVM -g -O2
>>> -mdynamic-no-pic -W -Wall -Wwrite-strings -Wstrict-prototypes
>>> -Wmissing-prototypes -Wold-style-definition -Wmissing-format-attribute
>>> -pedantic -Wno-long-long  -I../../llvm-gcc.src/libcpp -I.
>>> -I../../llvm-gcc.src/libcpp/../include -I./../intl
>>> -I../../llvm-gcc.src/libcpp/include -DENABLE_LLVM -c -o init.o -MT init.o
>>> -MMD -MP -MF .deps/init.Po ../../llvm-gcc.src/libcpp/init.c
>>>
>>> Assertion failed: (i < NumOperands && "getOperand() out of range!"),
>>> function getOperand, file
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm.src/include/llvm/User.h,
>>> line 104.
>>>
>>> ../../llvm-gcc.src/libcpp/init.c:666: internal compiler error: Abort trap
>>>
>>> Please submit a full bug report,
>>>
>>> with preprocessed source if appropriate.
>>>
>>> See <URL:http://llvm.org/bugs/> for instructions.
>>>
>>> smoosh-14:libcpp buildslave$
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/./prev-gcc/xgcc
>>> -B/Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/./prev-gcc/
>>> -B/Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.install.2/powerpc-apple-darwin9/bin/
>>>  -I../../llvm-gcc.src/libcpp -I. -I../../llvm-gcc.src/libcpp/../include
>>> -I./../intl -I../../llvm-gcc.src/libcpp/include -DENABLE_LLVM -g -O2
>>> -mdynamic-no-pic -W -Wall -Wwrite-strings -Wstrict-prototypes
>>> -Wmissing-prototypes -Wold-style-definition -Wmissing-format-attribute
>>> -pedantic -Wno-long-long  -I../../llvm-gcc.src/libcpp -I.
>>> -I../../llvm-gcc.src/libcpp/../include -I./../intl
>>> -I../../llvm-gcc.src/libcpp/include -DENABLE_LLVM -c -o init.o -MT init.o
>>> -MMD -MP -MF .deps/init.Po ../../llvm-gcc.src/libcpp/init.c -emit-llvm -o
>>> /tmp/t.bc
>>>
>>> (gdb) r
>>>
>>> Starting program:
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/prev-gcc/cc1
>>> -quiet -I../../llvm-gcc.src/libcpp -I.
>>> -I../../llvm-gcc.src/libcpp/../include -I./../intl
>>> -I../../llvm-gcc.src/libcpp/include -I../../llvm-gcc.src/libcpp -I.
>>> -I../../llvm-gcc.src/libcpp/../include -I./../intl
>>> -I../../llvm-gcc.src/libcpp/include -iprefix
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/prev-gcc/../lib/gcc/powerpc-apple-darwin9/4.2.1/
>>> -isystem
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj.2/./prev-gcc/include
>>> -MMD init.d -MF .deps/init.Po -MP -MT init.o -MQ init.o -D__DYNAMIC__
>>> -DENABLE_LLVM -DENABLE_LLVM ../../llvm-gcc.src/libcpp/init.c
>>> -feliminate-unused-debug-symbols -quiet -dumpbase init.c
>>> -mmacosx-version-min=10.5.8 -mdynamic-no-pic -auxbase-strip init.o -g -O2 -W
>>> -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
>>> -Wold-style-definition -Wmissing-format-attribute -pedantic -Wno-long-long
>>> -o /var/tmp//cc9W2r3q.s
>>>
>>> Reading symbols for shared libraries ++++.. done
>>>
>>> Assertion failed: (i < NumOperands && "getOperand() out of range!"),
>>> function getOperand, file
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm.src/include/llvm/User.h,
>>> line 104.
>>>
>>> Program received signal SIGABRT, Aborted.
>>>
>>> 0x90b76b50 in __kill ()
>>>
>>> (gdb) bt
>>>
>>> #0  0x90b76b50 in __kill ()
>>>
>>> #1  0x90c11c00 in abort ()
>>>
>>> #2  0x90c04c10 in __assert_rtn ()
>>>
>>> #3  0x00ea949c in llvm::InstCombiner::FoldGEPICmp ()
>>>
>>> #4  0x00eb1260 in llvm::InstCombiner::visitICmpInst ()
>>>
>>> #5  0x00ee5158 in llvm::InstVisitor<llvm::InstCombiner,
>>> llvm::Instruction*>::visit ()
>>>
>>> #6  0x00edf0c8 in llvm::InstCombiner::DoOneIteration ()
>>>
>>> #7  0x00ed9bc8 in llvm::InstCombiner::runOnFunction ()
>>>
>>> #8  0x010c5400 in llvm::FPPassManager::runOnFunction ()
>>>
>>> #9  0x00f2a190 in (anonymous namespace)::CGPassManager::runOnModule ()
>>>
>>> #10 0x010c4f58 in llvm::MPPassManager::runOnModule ()
>>>
>>> #11 0x010c67c0 in llvm::PassManagerImpl::run ()
>>>
>>> #12 0x010c6894 in llvm::PassManager::run ()
>>>
>>> #13 0x0087d6f4 in llvm_asm_file_end ()
>>>
>>> #14 0x0079eeec in compile_file ()
>>>
>>> #15 0x007a1f18 in do_compile ()
>>>
>>> #16 0x007a1fe0 in toplev_main ()
>>>
>>> #17 0x00122ffc in main ()
>>>
>>> --
>>>
>>> <t.bc>
>>>
>>> - Daniel
>>>
>>>
>>> - Daniel
>>>
>>> On Tue, Feb 23, 2010 at 6:48 PM,  <daniel_dunbar at apple.com> wrote:
>>>
>>> The Buildbot has detected a new failure of llvm-gcc-powerpc-darwin9 on
>>> smooshlab.
>>>
>>> Full details are available at:
>>>
>>> http://smooshlab.apple.com:8010/builders/llvm-gcc-powerpc-darwin9/builds/979
>>>
>>> Buildbot URL: http://smooshlab.apple.com:8010/
>>>
>>> Buildslave for this Build: smoosh-14.apple.com
>>>
>>> Build Reason:
>>>
>>> Build Source Stamp: 97012
>>>
>>> Blamelist: bwilson,evancheng,fjahanian,grosbach,johannes
>>>
>>> BUILD FAILED: failed test.llvm.stage2 compile.llvm-gcc.stage2
>>>
>>> sincerely,
>>>
>>> -The Buildbot
>>>
>>>
>>> ================================================================================
>>>
>>> CHANGES:
>>>
>>> Files:
>>>
>>> SingleSource/Benchmarks/Misc/Makefile
>>>
>>> SingleSource/UnitTests/Makefile
>>>
>>> At: Tue 23 Feb 2010 17:25:31
>>>
>>> Changed By: johannes
>>>
>>> Comments: Disable some tests on PPC Darwin that are never going to work.
>>>
>>> Properties:
>>>
>>>
>>>
>>>
>>> Files:
>>>
>>> lib/Frontend/RewriteObjC.cpp
>>>
>>> test/Rewriter/rewrite-block-pointer.mm
>>>
>>> At: Tue 23 Feb 2010 17:30:31
>>>
>>> Changed By: fjahanian
>>>
>>> Comments: Fix rewriting of a method when return type is
>>>
>>> a block pointer type. Fixes radar 7682149.
>>>
>>> Properties:
>>>
>>>
>>>
>>>
>>> File: test/Rewriter/rewrite-block-pointer.mm
>>>
>>> At: Tue 23 Feb 2010 17:40:33
>>>
>>> Changed By: fjahanian
>>>
>>> Comments: Correct radar no.
>>>
>>> Properties:
>>>
>>>
>>>
>>>
>>> Files:
>>>
>>> include/llvm/Analysis/PHITransAddr.h
>>>
>>> lib/Analysis/MemoryDependenceAnalysis.cpp
>>>
>>> lib/Analysis/PHITransAddr.cpp
>>>
>>> lib/Transforms/Scalar/GVN.cpp
>>>
>>> At: Tue 23 Feb 2010 17:40:33
>>>
>>> Changed By: bwilson
>>>
>>> Comments: Add an argument to PHITranslateValue to specify the DominatorTree.
>>>  If this
>>>
>>> argument is non-null, pass it along to PHITranslateSubExpr so that it can
>>>
>>> prefer using existing values that dominate the PredBB, instead of just
>>>
>>> blindly picking the first equivalent value that it finds on a uselist.
>>>
>>> Also when the DominatorTree is specified, have PHITranslateValue filter
>>>
>>> out any result that does not dominate the PredBB.  This is basically just
>>>
>>> refactoring the check that used to be in GetAvailablePHITranslatedSubExpr
>>>
>>> and also in GVN.
>>>
>>> Despite my initial expectations, this change does not affect the results
>>>
>>> of GVN for any testcases that I could find, but it should help compile time.
>>>
>>> Before this change, if PHITranslateSubExpr picked a value that does not
>>>
>>> dominate, PHITranslateWithInsertion would then insert a new value, which GVN
>>>
>>> would later determine to be redundant and would replace.  By picking a good
>>>
>>> value to begin with, we save GVN the extra work of inserting and then
>>>
>>> replacing a new value.
>>>
>>> Properties:
>>>
>>>
>>>
>>>
>>> Files:
>>>
>>> lib/CodeGen/SelectionDAG/DAGCombiner.cpp
>>>
>>> lib/CodeGen/SelectionDAG/TargetLowering.cpp
>>>
>>> lib/Target/X86/X86ISelLowering.cpp
>>>
>>> test/CodeGen/X86/2010-02-23-DAGCombineBug.ll
>>>
>>> test/CodeGen/X86/critical-edge-split.ll
>>>
>>> test/CodeGen/X86/ins_subreg_coalesce-3.ll
>>>
>>> test/CodeGen/X86/trunc-to-bool.ll
>>>
>>> test/CodeGen/X86/xor-icmp.ll
>>>
>>> At: Tue 23 Feb 2010 17:45:32
>>>
>>> Changed By: evancheng
>>>
>>> Comments: Re-apply 96540 and 96556 with fixes.Properties:
>>>
>>>
>>>
>>>
>>> Files:
>>>
>>> lib/Target/ARM/ARMISelLowering.cpp
>>>
>>> test/CodeGen/Thumb2/ldr-str-imm12.ll
>>>
>>> At: Tue 23 Feb 2010 17:45:32
>>>
>>> Changed By: grosbach
>>>
>>> Comments: LowerCall() should always do getCopyFromReg() to reference the
>>> stack pointer.
>>>
>>> Machine instruction selection is much happier when operands are in virtual
>>>
>>> registers.Properties:
>>>
>>>
>>>
>>>
>>> LOGS:
>>>
>>> Last 10 lines of 'stdio':
>>>
>>> ********************
>>>
>>> Failing Tests (1):
>>>
>>>  LLVM :: FrontendC/2003-08-23-LocalUnionTest.c
>>>
>>> Expected Passes    : 4991
>>>
>>> Expected Failures  : 38
>>>
>>> Unsupported Tests  : 64
>>>
>>> Unexpected Failures: 1
>>>
>>> make[1]: *** [check-local-lit] Error 1
>>>
>>> make: *** [check-lit] Error 2
>>>
>>> Last 10 lines of 'fail':
>>>
>>> LLVM :: FrontendC/2003-08-23-LocalUnionTest.c
>>>
>>> Last 10 lines of 'xfail':
>>>
>>> LLVM :: LLVMC/MultiplePluginPriorities.td
>>>
>>> LLVM :: LLVMC/OutputSuffixHook.td
>>>
>>> LLVM :: MC/AsmParser/directive_lsym.s
>>>
>>> LLVM :: MC/AsmParser/exprs-invalid.s
>>>
>>> LLVM :: Transforms/GVN/rle-no-phi-translate.ll
>>>
>>> LLVM :: Transforms/GlobalOpt/2009-03-03-dbg.ll
>>>
>>> LLVM :: Transforms/GlobalOpt/2009-03-05-dbg.ll
>>>
>>> LLVM :: Transforms/IndVarSimplify/loop_evaluate_6.ll
>>>
>>> LLVM :: Transforms/TailCallElim/nocapture.ll
>>>
>>> LLVM :: Transforms/TailDup/2008-06-11-AvoidDupLoopHeader.ll
>>>
>>> Last 10 lines of 'unsupported':
>>>
>>> LLVM :: FrontendObjC/2009-04-27-bitfield-vs-ivar.m
>>>
>>> LLVM :: FrontendObjC/2009-04-28-bitfield-vs-vbc.m
>>>
>>> LLVM :: FrontendObjC/2009-08-05-utf16.m
>>>
>>> LLVM :: FrontendObjC/2009-08-17-DebugInfo.m
>>>
>>> LLVM :: FrontendObjC/2009-11-30-Objc-ID.m
>>>
>>> LLVM :: FrontendObjC/2010-02-01-utf16-with-null.m
>>>
>>> LLVM :: FrontendObjC/2010-02-11-fwritable-stringsBug.m
>>>
>>> LLVM :: FrontendObjC/2010-02-23-DbgInheritance.m
>>>
>>> LLVM :: LLVMC/ObjC++/hello.mm
>>>
>>> LLVM :: LLVMC/ObjC/hello.m
>>>
>>> Last 10 lines of '2003-08-23-localuniontest.c':
>>>
>>> --
>>>
>>> Command Output (stderr):
>>>
>>> --
>>>
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm.src/test/FrontendC/2003-08-23-LocalUnionTest.c:
>>> In function 'test':
>>>
>>> /Users/buildslave/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm.src/test/FrontendC/2003-08-23-LocalUnionTest.c:9:
>>> error: redefinition of 'union foo'
>>>
>>> --
>>>
>>> ********************
>>>
>>>
>>> Last 10 lines of 'stdio':
>>>
>>> configure: creating ./config.status
>>>
>>> config.status: creating Makefile
>>>
>>> config.status: creating gccbug
>>>
>>> config.status: creating mklibgcc
>>>
>>> config.status: creating libada-mk
>>>
>>> config.status: creating ada/Makefile
>>>
>>> config.status: creating auto-host.h
>>>
>>> config.status: executing default commands
>>>
>>> make[1]: *** [stage2-bubble] Error 2
>>>
>>> make: *** [all] Error 2
>>>
>>> Last 10 lines of 'warnings':
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:1781: warning: operand 3
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:1794: warning: operand 3
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:1807: warning: operand 3
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:1882: warning: operand 1
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:1889: warning: operand 1
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:2195: warning: destination
>>> missing a mode?
>>>
>>> ../../llvm-gcc.src/gcc/config/rs6000/altivec.md:2195: warning: operand 0
>>> missing mode?
>>>
>>> ../../llvm-gcc.src/gcc/sched-vis.c:628: warning: no previous prototype for
>>> 'print_insn'
>>>
>>> ../../llvm-gcc.src/gcc/libgcc2.c:412: warning: control reaches end of
>>> non-void function
>>>
>>> ../../llvm-gcc.src/gcc/libgcc2.c:2045: warning: 'noreturn' function does
>>> return
>>>
>>> _______________________________________________
>>>
>>> llvm-testresults mailing list
>>>
>>> llvm-testresults at cs.uiuc.edu
>>>
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> llvm-testresults mailing list
>>> llvm-testresults at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults
>>>
>>>
>>> _______________________________________________
>>> llvm-testresults mailing list
>>> llvm-testresults at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults
>>>
>>>
>
>