[llvm-commits] [llvm] r76558 - in /llvm/trunk: lib/CodeGen/VirtRegRewriter.cpp test/CodeGen/X86/2009-07-21-RewriterBug.ll

[Evan Cheng]

On Jul 21, 2009, at 10:52 AM, Chris Lattner wrote:

>
> On Jul 21, 2009, at 10:06 AM, Evan Cheng wrote:
>
>>>>
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=76558&view=rev
>>>> Log:
>>>> Another rewriter bug exposed by recent coalescer changes.
>>>> ReuseInfo::GetRegForReload() should make sure the "switched"
>>>> register is in the desired register class. I'm surprised this  
>>>> hasn't
>>>> caused more failures in the past.
>>>
>>> Thanks Evan, can you reduce this testcase though?  It is huge, and
>>> doesn't appear to have been run through bugpoint.
>>
>> That's a bugpoint reduced test case.
> Ok
>
>> I can't reduce it any further.
>
> I don't believe you.  It has tons of global variable inits etc that
> don't impact codegen.  Knowing the problem, can't you hand synthesize

Fine. I can remove some of the global variables.

> a testcase to expose the problem?

Trust me. It's not possible. I welcome anyone who can hand construct  
register allocation miscompilation problems.

Evan

>
> -Chris
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits