[llvm-testresults] evchen-ppc32 nightly tester results

Daniel Dunbar daniel_dunbar at apple.com
Sat Feb 13 19:24:12 PST 2010


llvm-gcc-x86_64-darwin10 is also sporadically failing.

 - Daniel

On Feb 13, 2010, at 7:16 PM, Daniel Dunbar wrote:

> 
> On Feb 13, 2010, at 6:59 PM, Dale Johannesen wrote:
> 
>> 
>>>> On Feb 12, 2010, at 5:14 PMPST, daniel_dunbar at apple.com wrote:
>>>> 
>>>>> http://smooshlab.apple.com/perf/db_nt_internal/nightlytest/5438/
>>>>> Name:  evchen.apple.com
>>>>> Nickname: evchen-ppc32:1
>>>>> 
>>>>> Run: 5438, Start Time: 2010-02-12 12:38:20, End Time: 2010-02-12 17:14:17
>>>> 
>>>> Assuming this time is local and accurate, this is before my R2 patch went in.  Whew.
>>> 
>>>>> Comparing To: 5427, Start Time: 2010-02-12 00:39:48, End Time: 2010-02-12 05:15:04
>>>>> 
>>>>> --- Changes Summary ---
>>>>> New Test Passes: 0
>>>>> New Test Failures: 7
>> 
>> I have not seen another report from this bot(?), but all these new failures have been fixed except 300.twolf llc build.  That is a crash in LSR that looks like memory corruption.  
>> 7647273.
>> 
> 
> Should we disable LSR until this is resolved? It is also causing problems during clang bootstrap (although that could also be clangs fault). However, the current powerpc bootstrap failure looks a similar issue to what clang hits:
> --
> source='../../llvm-gcc.src/libdecnumber/decNumber.c' object='decNumber.o' libtool=no /Volumes/SandBox/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj/./prev-gcc/xgcc -B/Volumes/SandBox/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.obj/./prev-gcc/ -B/Volumes/SandBox/zorg/buildbot/smooshlab/slave/build.llvm-gcc-powerpc-darwin9/llvm-gcc.install/powerpc-apple-darwin9/bin/  -I../../llvm-gcc.src/libdecnumber -I.  -g -O2 -mdynamic-no-pic -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -Wmissing-format-attribute -Wcast-qual -pedantic -Wno-long-long  -I../../llvm-gcc.src/libdecnumber -I.  -c ../../llvm-gcc.src/libdecnumber/decNumber.c
> Instruction does not dominate all uses!
>  %scevgep135136 = bitcast i32* %scevgep135 to i8* ; <i8*> [#uses=1]
>  %uglygep137 = getelementptr i8* %scevgep135136, i32 2 ; <i8*> [#uses=1]
> Broken module found, compilation aborted!
> ../../llvm-gcc.src/libdecnumber/decNumber.c:5591: internal compiler error: Abort trap
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://llvm.org/bugs/> for instructions.
> --
> 
> - Daniel
> 
>>>>> Externals/SPEC/CINT2000/176_gcc/176_gcc [LLC, LLC compile]
>>>>> Externals/SPEC/CINT2000/197_parser/197_parser [LLC, LLC compile]
>>>>> Externals/SPEC/CINT2000/300_twolf/300_twolf [LLC, JIT, LLC compile]
>> 
> 





More information about the llvm-testresults mailing list