[llvm-commits] Nightly Test Regressions

Dan Gohman gohman at apple.com
Mon Sep 21 13:37:02 CDT 2009


When I run 483.xalancbmk, it gets an opt verifier error:
"Invoke result does not dominate all uses!" Anyone have any
guesses as to what triggered that?

Dan

On Sep 21, 2009, at 11:17 AM, Bill Wendling wrote:

> All of my EH changes were reverted, so it's something else.
>
> -bw
>
> On Sep 21, 2009, at 9:30 AM, Evan Cheng wrote:
>
>> 483_xalancbmk is still failing as far as I can tell. The rest are  
>> fixed.
>>
>> Evan
>>
>> On Sep 19, 2009, at 10:12 AM, Daniel Dunbar wrote:
>>
>>> Hi All,
>>>
>>> There were a large number of regressions on the nightly tests last
>>> night. I'd appreciate some help in tracking down the failures.
>>>
>>> This is a summary of the important new failures:
>>>
>>> This is failing in many places:
>>> SingleSource/Regression/C++/EH/exception_spec_test [LLC, LLC-BETA,
>>> LLC compile, LLC_BETA compile]
>>> and these are failing in some places:
>>> SingleSource/Benchmarks/Shootout-C++/except [LLC, LLC-BETA, LLC
>>> compile, LLC_BETA compile]
>>> SingleSource/Regression/C++/EH/ctor_dtor_count-2 [LLC, LLC-BETA, LLC
>>> compile, LLC_BETA compile]
>>> SingleSource/Regression/C++/EH/throw_rethrow_test [LLC, LLC-BETA,
>>> LLC compile, LLC_BETA compile]
>>> and this is failing when built with clang:
>>> SingleSource/UnitTests/ObjC/exceptions [LLC, LLC-BETA, LLC compile,
>>> LLC_BETA compile]
>>>
>>> Bill, Chris, Eric, Duncan: Can you guys sort out someone to look  
>>> at it?
>>>
>>> This is failing in a number of places:
>>> Externals/SPEC/CINT2006/483_xalancbmk/483_xalancbmk [JIT, JIT
>>> codegen, LLC, LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA  
>>> compile]
>>> Externals/SPEC/CINT2006/483_xalancbmk/483_xalancbmk [JIT, CBE, JIT
>>> codegen, LLC, LLC-BETA, LLC compile, Bitcode, GCCAS, LLC_BETA  
>>> compile]
>>> and these are failing in at least one place:
>>> Externals/SPEC/CINT2006/471_omnetpp/471_omnetpp [LLC, LLC-BETA, LLC
>>> compile, LLC_BETA compile]
>>> Externals/SPEC/CINT2000/252_eon/252_eon [JIT codegen]
>>>
>>> Evan or Dan, can you guys point a finger or take a look?
>>>
>>> Thanks,
>>> - Daniel
>>
>




More information about the llvm-commits mailing list