[llvm-testresults] llvm-arm-linux buildbot regression

[Jim Grosbach]

On Dec 2, 2010, at 12:47 PM, Eric Christopher wrote:

>> 
>> I agree. Not being able to reproduce the failures locally is indeed very inconvenient.
>> 
>> I marked the ARM ExecutionEngine tests as XFAIL for now.
>> 
>> There is one other failure which looks unrelated and is likely a bug of a different sort.
>> 
>> ******************** TEST 'LLVM :: CodeGen/X86/apm.ll' FAILED ********************Script:
>> --
>> llc < /wd/buildbot/llvm-arm-linux/llvm/test/CodeGen/X86/apm.ll -o - -march=x86-64 | FileCheck /wd/buildbot/llvm-arm-linux/llvm/test/CodeGen/X86/apm.ll
>> --
>> Exit Code: 1
>> Command Output (stderr):
>> --
>> llc: X86ISelLowering.cpp:9441: llvm::MachineBasicBlock* llvm::X86TargetLowering::EmitMonitor(llvm::MachineInstr*, llvm::MachineBasicBlock*) const: Assertion `Subtarget->hasSSE3() && "Target must have SSE3 features enabled"' failed.
>> 0  llc       0x0112e01c
>> 1  llc       0x0112e9a4
>> 2  libc.so.6 0x4021cc10 __default_sa_restorer_v2 + 0
>> Stack dump:
>> 0.	Program arguments: llc -o - -march=x86-64 
>> 1.	Running pass 'Function Pass Manager' on module '<stdin>'.
>> 2.	Running pass 'Expand ISel Pseudo-instructions' on function '@foo'
>> /wd/buildbot/llvm-arm-linux/llvm/test/CodeGen/X86/apm.ll:4:10: error: expected string not found in input
>> ; CHECK: foo:
>>         ^
>> <stdin>:1:1: note: scanning from here
> 
> This one is definitely me, but I think Duncan fixed them a couple days ago? If not I'll look.

Very possible it's just lagging behind. I just clicked on the most recent failing build and looked at the failures.  That arm tester runs veeeeerrrrryyyyyy slooooowwwwwwwllllly. It takes a while to cycle, too.