thr3ads.net - llvm dev - [LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi [Nov 2011]

If this information is useful, please help other people find it:
Share via:

spop at codeaurora.org

2011-Nov-04 22:11 UTC

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

Hi Daniel,
> Sebastian, this looks like it is most likely some kind of fallout from
> your changes.
Thanks for letting me know about these failing testcases.

In the logs of the buildbot:
http://lab.llvm.org:8011/builders/llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi/builds/273/steps/run.build.step.configure_llvm_1/logs/stdio
I see that the bot is configuring llvm with:
--build=x86_64-apple-darwin10 --host=x86_64-apple-darwin10
--target=i686-pc-mingw32

Before my patches, specifying --target had no effect as it was the
--host value that was taken.  So by default we used to compile code
for the host that is "x86_64-apple-darwin10".

After my patches, the value set with --target is used.  So now, by
default this build bot will generate code for "i686-pc-mingw32".  I
think that these fails are due to the fact that the testcases are not
working when the target is specified to be different than host: in the
following list of failing testcases I see the JIT cases failing:

Failing Tests (26):
    LLVM :: ExecutionEngine/2002-12-16-ArgTest.ll
    LLVM :: ExecutionEngine/2003-01-04-ArgumentBug.ll
    LLVM :: ExecutionEngine/2003-01-04-LoopTest.ll
    LLVM :: ExecutionEngine/2003-01-15-AlignmentTest.ll
    LLVM :: ExecutionEngine/2003-05-06-LivenessClobber.ll
    LLVM :: ExecutionEngine/2003-05-07-ArgumentTest.ll
    LLVM :: ExecutionEngine/2003-06-04-bzip2-bug.ll
    LLVM :: ExecutionEngine/2003-08-15-AllocaAssertion.ll
    LLVM :: ExecutionEngine/2003-08-21-EnvironmentTest.ll
    LLVM :: ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll
    LLVM ::
ExecutionEngine/2003-10-18-PHINode-ConstantExpr-CondCode-Failure.ll
    LLVM :: ExecutionEngine/2005-12-02-TailCallBug.ll
    LLVM :: ExecutionEngine/hello.ll
    LLVM :: ExecutionEngine/hello2.ll
    LLVM :: ExecutionEngine/stubs.ll
    LLVM :: ExecutionEngine/test-call.ll
    LLVM :: ExecutionEngine/test-fp.ll
    LLVM :: ExecutionEngine/test-loadstore.ll
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/JIT.GlobalInFunction
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.AvailableExternallyGlobalIsntEmitted
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.EscapedLazyStubStillCallable
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FarCallToKnownFunction
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FunctionPointersOutliveTheirCreator
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/LazyLoadedJITTest.EagerCompiledRecursionThroughGhost
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.EagerMode
    LLVM-Unit ::
ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.LazyMode

I think that for JIT, the compiler should figure out what the host
is with a *runtime* check (i.e., the JIT should not use the info from
the configure flag --host.)

Thanks,
Sebastian
>
>  - Daniel
>
> On Nov 1, 2011, at 5:26 PM, llvm.buildmaster at lab.llvm.org wrote:
>
>> The Buildbot has detected a new failure on builder
>> llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi while building LLVM.
>> Full details are available at:
>>
http://lab.llvm.org:8011/builders/llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi/builds/273
>>
>> Buildbot URL: http://lab.llvm.org:8011/
>>
>> Buildslave for this Build: kistanova1
>>
>> Build Reason: scheduler
>> Build Source Stamp: 143501
>> Blamelist: ddunbar,efriedma,spop
>>
>> BUILD FAILED: failed run.build.step.test_llvm_1
>>
>> sincerely,
>> -The Buildbot
>>
>>
>>
>
>

Eric Christopher

2011-Nov-04 22:15 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

On Nov 4, 2011, at 3:11 PM, spop at codeaurora.org wrote:
> I think that for JIT, the compiler should figure out what the host
> is with a *runtime* check (i.e., the JIT should not use the info from
> the configure flag --host.)
That seems reasonable. Would you update this please?

Thanks!

-eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111104/2d3d82b5/attachment.html>

Eli Friedman

2011-Nov-04 22:16 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

On Fri, Nov 4, 2011 at 3:11 PM,  <spop at codeaurora.org>
wrote:> Hi Daniel,
>
>> Sebastian, this looks like it is most likely some kind of fallout from
>> your changes.
>
> Thanks for letting me know about these failing testcases.
>
> In the logs of the buildbot:
>
http://lab.llvm.org:8011/builders/llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi/builds/273/steps/run.build.step.configure_llvm_1/logs/stdio
> I see that the bot is configuring llvm with:
> --build=x86_64-apple-darwin10 --host=x86_64-apple-darwin10
> --target=i686-pc-mingw32
>
> Before my patches, specifying --target had no effect as it was the
> --host value that was taken.  So by default we used to compile code
> for the host that is "x86_64-apple-darwin10".
>
> After my patches, the value set with --target is used.  So now, by
> default this build bot will generate code for "i686-pc-mingw32".
 I
> think that these fails are due to the fact that the testcases are not
> working when the target is specified to be different than host: in the
> following list of failing testcases I see the JIT cases failing:
>
> Failing Tests (26):
>    LLVM :: ExecutionEngine/2002-12-16-ArgTest.ll
>    LLVM :: ExecutionEngine/2003-01-04-ArgumentBug.ll
>    LLVM :: ExecutionEngine/2003-01-04-LoopTest.ll
>    LLVM :: ExecutionEngine/2003-01-15-AlignmentTest.ll
>    LLVM :: ExecutionEngine/2003-05-06-LivenessClobber.ll
>    LLVM :: ExecutionEngine/2003-05-07-ArgumentTest.ll
>    LLVM :: ExecutionEngine/2003-06-04-bzip2-bug.ll
>    LLVM :: ExecutionEngine/2003-08-15-AllocaAssertion.ll
>    LLVM :: ExecutionEngine/2003-08-21-EnvironmentTest.ll
>    LLVM :: ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll
>    LLVM ::
> ExecutionEngine/2003-10-18-PHINode-ConstantExpr-CondCode-Failure.ll
>    LLVM :: ExecutionEngine/2005-12-02-TailCallBug.ll
>    LLVM :: ExecutionEngine/hello.ll
>    LLVM :: ExecutionEngine/hello2.ll
>    LLVM :: ExecutionEngine/stubs.ll
>    LLVM :: ExecutionEngine/test-call.ll
>    LLVM :: ExecutionEngine/test-fp.ll
>    LLVM :: ExecutionEngine/test-loadstore.ll
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/JIT.GlobalInFunction
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.AvailableExternallyGlobalIsntEmitted
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.EscapedLazyStubStillCallable
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FarCallToKnownFunction
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FunctionPointersOutliveTheirCreator
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/LazyLoadedJITTest.EagerCompiledRecursionThroughGhost
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.EagerMode
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.LazyMode
>
> I think that for JIT, the compiler should figure out what the host
> is with a *runtime* check (i.e., the JIT should not use the info from
> the configure flag --host.)
Err, what?  We can and should compute the host with a compile-time
check; how would you possibly run lli on a platform that isn't the
same as --host?

-Eli

spop at codeaurora.org

2011-Nov-04 22:22 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

>
> On Nov 4, 2011, at 3:11 PM, spop at codeaurora.org wrote:
>
>> I think that for JIT, the compiler should figure out what the host
>> is with a *runtime* check (i.e., the JIT should not use the info from
>> the configure flag --host.)
>
> That seems reasonable. Would you update this please?
>
Yes, I would be happy to find the best fix for this.

Thinking a bit more, I don't see a good reason to not use the value
set by configure --host.  At configure time, we know the kind machine
on which the compiler will run, so that's the machine we should target
when JITting.  Opinions?

Thanks,
Sebastian

Eric Christopher

2011-Nov-04 22:22 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

On Nov 4, 2011, at 3:16 PM, Eli Friedman wrote:
> Err, what?  We can and should compute the host with a compile-time
> check; how would you possibly run lli on a platform that isn't the
> same as --host?
lli in interp mode sure, but generating native code?

-eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111104/fff7bf81/attachment.html>

Sebastian Pop

2011-Nov-04 22:28 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

On Fri, Nov 4, 2011 at 5:16 PM, Eli Friedman <eli.friedman at gmail.com>
wrote:>> I think that for JIT, the compiler should figure out what the host
>> is with a *runtime* check (i.e., the JIT should not use the info from
>> the configure flag --host.)
>
> Err, what?  We can and should compute the host with a compile-time
> check; how would you possibly run lli on a platform that isn't the
> same as --host?
You are right, when we use the compiler as a JIT, runtime = compile time
so we would need to figure out the host at that time.

Sebastian
--
Qualcomm Innovation Center, Inc is a member of Code Aurora Forum

Daniel Dunbar

2011-Nov-04 22:28 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

I actually tend to agree with spop, it's cleaner to compute things at
runtime than at compile time. One particular reason is wanting to pick
the best target CPU for the current arch (which may not be what was
compiled for).

The current JIT target selection logic is really gross, I do believe
that it tried to do this, but it probably needs some spring cleaning.
Sebastian, can you try and take a look? We don't want to regress on
these tests, at least.

 - Daniel

On Fri, Nov 4, 2011 at 3:16 PM, Eli Friedman <eli.friedman at gmail.com>
wrote:> On Fri, Nov 4, 2011 at 3:11 PM,  <spop at codeaurora.org> wrote:
>> Hi Daniel,
>>
>>> Sebastian, this looks like it is most likely some kind of fallout
from
>>> your changes.
>>
>> Thanks for letting me know about these failing testcases.
>>
>> In the logs of the buildbot:
>>
http://lab.llvm.org:8011/builders/llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi/builds/273/steps/run.build.step.configure_llvm_1/logs/stdio
>> I see that the bot is configuring llvm with:
>> --build=x86_64-apple-darwin10 --host=x86_64-apple-darwin10
>> --target=i686-pc-mingw32
>>
>> Before my patches, specifying --target had no effect as it was the
>> --host value that was taken.  So by default we used to compile code
>> for the host that is "x86_64-apple-darwin10".
>>
>> After my patches, the value set with --target is used.  So now, by
>> default this build bot will generate code for
"i686-pc-mingw32".  I
>> think that these fails are due to the fact that the testcases are not
>> working when the target is specified to be different than host: in the
>> following list of failing testcases I see the JIT cases failing:
>>
>> Failing Tests (26):
>>    LLVM :: ExecutionEngine/2002-12-16-ArgTest.ll
>>    LLVM :: ExecutionEngine/2003-01-04-ArgumentBug.ll
>>    LLVM :: ExecutionEngine/2003-01-04-LoopTest.ll
>>    LLVM :: ExecutionEngine/2003-01-15-AlignmentTest.ll
>>    LLVM :: ExecutionEngine/2003-05-06-LivenessClobber.ll
>>    LLVM :: ExecutionEngine/2003-05-07-ArgumentTest.ll
>>    LLVM :: ExecutionEngine/2003-06-04-bzip2-bug.ll
>>    LLVM :: ExecutionEngine/2003-08-15-AllocaAssertion.ll
>>    LLVM :: ExecutionEngine/2003-08-21-EnvironmentTest.ll
>>    LLVM :: ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll
>>    LLVM ::
>> ExecutionEngine/2003-10-18-PHINode-ConstantExpr-CondCode-Failure.ll
>>    LLVM :: ExecutionEngine/2005-12-02-TailCallBug.ll
>>    LLVM :: ExecutionEngine/hello.ll
>>    LLVM :: ExecutionEngine/hello2.ll
>>    LLVM :: ExecutionEngine/stubs.ll
>>    LLVM :: ExecutionEngine/test-call.ll
>>    LLVM :: ExecutionEngine/test-fp.ll
>>    LLVM :: ExecutionEngine/test-loadstore.ll
>>    LLVM-Unit ::
>> ExecutionEngine/JIT/Debug+Asserts/JITTests/JIT.GlobalInFunction
>>    LLVM-Unit ::
>>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.AvailableExternallyGlobalIsntEmitted
>>    LLVM-Unit ::
>>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.EscapedLazyStubStillCallable
>>    LLVM-Unit ::
>>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FarCallToKnownFunction
>>    LLVM-Unit ::
>>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FunctionPointersOutliveTheirCreator
>>    LLVM-Unit ::
>>
ExecutionEngine/JIT/Debug+Asserts/JITTests/LazyLoadedJITTest.EagerCompiledRecursionThroughGhost
>>    LLVM-Unit ::
>> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.EagerMode
>>    LLVM-Unit ::
>> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.LazyMode
>>
>> I think that for JIT, the compiler should figure out what the host
>> is with a *runtime* check (i.e., the JIT should not use the info from
>> the configure flag --host.)
>
> Err, what?  We can and should compute the host with a compile-time
> check; how would you possibly run lli on a platform that isn't the
> same as --host?
>
> -Eli
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

Sebastian Pop

2011-Dec-27 19:36 UTC

head link

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

Hi,

On Fri, Nov 4, 2011 at 5:11 PM,  <spop at codeaurora.org>
wrote:> Hi Daniel,
>
>> Sebastian, this looks like it is most likely some kind of fallout from
>> your changes.
>
> Thanks for letting me know about these failing testcases.
>
> In the logs of the buildbot:
>
http://lab.llvm.org:8011/builders/llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi/builds/273/steps/run.build.step.configure_llvm_1/logs/stdio
> I see that the bot is configuring llvm with:
> --build=x86_64-apple-darwin10 --host=x86_64-apple-darwin10
> --target=i686-pc-mingw32
>
> Before my patches, specifying --target had no effect as it was the
> --host value that was taken.  So by default we used to compile code
> for the host that is "x86_64-apple-darwin10".
>
> After my patches, the value set with --target is used.  So now, by
> default this build bot will generate code for "i686-pc-mingw32".
 I
> think that these fails are due to the fact that the testcases are not
> working when the target is specified to be different than host: in the
> following list of failing testcases I see the JIT cases failing:
>
> Failing Tests (26):
>    LLVM :: ExecutionEngine/2002-12-16-ArgTest.ll
>    LLVM :: ExecutionEngine/2003-01-04-ArgumentBug.ll
>    LLVM :: ExecutionEngine/2003-01-04-LoopTest.ll
>    LLVM :: ExecutionEngine/2003-01-15-AlignmentTest.ll
>    LLVM :: ExecutionEngine/2003-05-06-LivenessClobber.ll
>    LLVM :: ExecutionEngine/2003-05-07-ArgumentTest.ll
>    LLVM :: ExecutionEngine/2003-06-04-bzip2-bug.ll
>    LLVM :: ExecutionEngine/2003-08-15-AllocaAssertion.ll
>    LLVM :: ExecutionEngine/2003-08-21-EnvironmentTest.ll
>    LLVM :: ExecutionEngine/2003-08-23-RegisterAllocatePhysReg.ll
>    LLVM ::
> ExecutionEngine/2003-10-18-PHINode-ConstantExpr-CondCode-Failure.ll
>    LLVM :: ExecutionEngine/2005-12-02-TailCallBug.ll
>    LLVM :: ExecutionEngine/hello.ll
>    LLVM :: ExecutionEngine/hello2.ll
>    LLVM :: ExecutionEngine/stubs.ll
>    LLVM :: ExecutionEngine/test-call.ll
>    LLVM :: ExecutionEngine/test-fp.ll
>    LLVM :: ExecutionEngine/test-loadstore.ll
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/JIT.GlobalInFunction
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.AvailableExternallyGlobalIsntEmitted
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.EscapedLazyStubStillCallable
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FarCallToKnownFunction
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/JITTest.FunctionPointersOutliveTheirCreator
>    LLVM-Unit ::
>
ExecutionEngine/JIT/Debug+Asserts/JITTests/LazyLoadedJITTest.EagerCompiledRecursionThroughGhost
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.EagerMode
>    LLVM-Unit ::
> ExecutionEngine/JIT/Debug+Asserts/JITTests/MultiJitTest.LazyMode
>
> I think that for JIT, the compiler should figure out what the host
> is with a *runtime* check (i.e., the JIT should not use the info from
> the configure flag --host.)
>
Until we fix this with the runtime check, here is the fix that introduces back
the function getHostTriple that returns the configure --host information
and uses it from the JIT.

Ok to commit this patch as an interim fix?

Thanks,
Sebastian
--
Qualcomm Innovation Center, Inc is a member of Code Aurora Forum
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-use-getHostTriple-instead-of-getDefaultTargetTriple-.patch
Type: text/x-diff
Size: 9047 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111227/5ae6f46f/attachment.patch>

Reasonably Related Threads

Search for more possibly parallel threads

llvm dev - Nov 2011 - [LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

[LLVMdev] JIT should query host info at runtime - Re: buildbot failure in LLVM on llvm-gcc-build-x86_64-darwin10-x-mingw32-x-armeabi

Reasonably Related Threads