[llvm-commits] [llvm] r155589 - /llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll

[alex_rosenberg at playstation.sony.com]

We've also noticed this. If -mcpu isn't set explicitly, it's seeded with info from the local machine, which never makes sense if an explicit triple is specified. This pretty much breaks all X86 cross-compilation (e.g. like a bootstrapping cross to Windows host).

Seems like a change should happen in X86Subtarget where sys::getHostCPUName() is called. Some test will need to set their Features more accurately at the same time.

Alex

> Date: Wed, 25 Apr 2012 21:22:33 -0000
> To: llvm-commits at cs.uiuc.edu
> From: Jakob Stoklund Olesen <stoklund at 2pi.dk>
> 
> Author: stoklund
> Date: Wed Apr 25 16:22:33 2012
> New Revision: 155589
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=155589&view=rev
> Log:
> Try to fix llvm-arm-linux builder with -mcpu.
> 
> Modified:
>     llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll
> 
> Modified: llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll?rev=155589&r1=155588&r2=155589&view=diff
> ==============================================================================
> --- llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll (original)
> +++ llvm/trunk/test/CodeGen/X86/2012-02-20-MachineCPBug.ll Wed Apr 25 16:22:33 2012
> @@ -1,4 +1,4 @@
> -; RUN: llc < %s -mtriple=i386-apple-macosx -mattr=+sse | FileCheck %s
> +; RUN: llc < %s -mtriple=i386-apple-macosx -mcpu=core2 -mattr=+sse | FileCheck %s
>  ; PR11940: Do not optimize away movb %al, %ch
>  
>  %struct.APInt = type { i64* }
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits