Hi,
As one of the people responsible for getting those with the lit option
requiring which jit to use already complicating matters. I personally don't
think there's much of a problem use the interpreter in this case.
Cheers,
Dave
-----Original Message-----
From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On
Behalf Of Tim Northover
Sent: 14 February 2013 16:55
To: LLVM Developers Mailing List
Subject: [LLVMdev] How to XFAIL JIT tests for AArch64
Hi,
Currently, no tests that use lli without "-force-interpreter" are
expected to pass when executing on an AArch64 model. However, they
will pass if built and run on (say) X86, just setting the default
target triple.
So XFAIL gets unexpected passes on a compiler merely targetting
AArch64 and leaving the tests as they are gives unexpected failures
when they're run on a model.
Does anyone know of a good solution to this? I believe only four tests
are affected (test/ExecutionEngine/* has a more global solution and
isn't a problem):
LLVM :: Analysis/Profiling/load-branch-weights-ifs.ll
LLVM :: Analysis/Profiling/load-branch-weights-loops.ll
LLVM :: Analysis/Profiling/load-branch-weights-switches.ll
LLVM :: Transforms/LICM/2003-12-11-SinkingToPHI.ll
Would changing the tests to force the interpreter on lli invocations
break what the profile tests are trying to do? Would it be an
acceptable change if not?
Cheers.
Tim.
_______________________________________________
LLVM Developers mailing list
LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev