[llvm] r253528 - [X86] Enable shrink-wrapping by default.

[Quentin Colombet via llvm-commits]

Hi Hans,

Thanks for the additional information.

Here is a small patch for you to try.

Thanks,
Q
-------------- next part --------------
A non-text attachment was scrubbed...
Name: stack_fix_up_clobbering_flags.patch
Type: application/octet-stream
Size: 1783 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20151130/4bbf09ec/attachment.obj>
-------------- next part --------------

> On Nov 30, 2015, at 1:29 PM, Hans Wennborg <hans at chromium.org> wrote:
> 
> On Mon, Nov 30, 2015 at 1:16 PM, Quentin Colombet <qcolombet at apple.com> wrote:
>>> On Nov 24, 2015, at 4:09 PM, Hans Wennborg <hans at chromium.org> wrote:
>>> On Wed, Nov 18, 2015 at 4:38 PM, Quentin Colombet via llvm-commits
>>> <llvm-commits at lists.llvm.org> wrote:
>>>> Author: qcolombet
>>>> Date: Wed Nov 18 18:38:00 2015
>>>> New Revision: 253528
>>>> 
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=253528&view=rev
>>>> Log:
>>>> [X86] Enable shrink-wrapping by default.
>>>> 
>>>> Differential Revision: http://reviews.llvm.org/D14156
>>> 
>>> I've reverted this in r254044 since it seems to be causing
>>> miscompiles. David bisected PR25607 to this patch, and I bisected a
>>> Chrome crash to it.
>>> 
>>> Sorry for not catching this sooner.
>> 
>> Does the chrome crash implies x86 32-bit or is it something different?
>> If the latter, could you provide a test case?
> 
> Yes, it was 32-bit x86.
> 
> I couldn't figure out the cause of the crash when I looked at it in a
> debugger, so I don't have a reproduction, sorry :-/
> 
> Maybe it's just the same problem as in PR25607? I'm happy to test any
> fixes on Chrome to see if they fix the crash I saw.
> 
> - Hans