Bug 50422 - Furmark.exe crash with wine (32 bits under 64 bits) with llvm-3.1 and r600-llvm-compiler enabled
Summary: Furmark.exe crash with wine (32 bits under 64 bits) with llvm-3.1 and r600-ll...
Status: RESOLVED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/Gallium/r600 (show other bugs)
Version: git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-05-28 03:19 UTC by Laurent carlier
Modified: 2012-10-31 16:35 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
wine output and backtrace (18.21 KB, text/plain)
2012-05-28 03:19 UTC, Laurent carlier
Details
output with R600_DUMP_SHADER=1 (45.68 KB, text/plain)
2012-05-28 03:21 UTC, Laurent carlier
Details
furmark output with llvm-3.1 and r600-llvm-compiler enabled (1.47 MB, image/png)
2012-10-31 12:53 UTC, Laurent carlier
Details

Description Laurent carlier 2012-05-28 03:19:24 UTC
Created attachment 62162 [details]
wine output and backtrace

* glxinfo32:
OpenGL renderer string: Gallium 0.4 on AMD RV770
OpenGL version string: 2.1 Mesa 8.1-devel (git-6404095)
OpenGL shading language version string: 1.30

* uname -a
Linux archMain 3.4.0-1-ARCH #1 SMP PREEMPT Mon May 21 10:10:13 CEST 2012 x86_64 GNU/Linux
Comment 1 Laurent carlier 2012-05-28 03:21:21 UTC
Created attachment 62163 [details]
output with R600_DUMP_SHADER=1
Comment 2 Tom Stellard 2012-06-19 19:11:00 UTC
I've been seeing a few strange crashes in the LLVM backend when using wine.  What version of wine are you using?  Does it work with an older version of wine?
Comment 3 Henri Verbeet 2012-06-20 08:41:51 UTC
This isn't just another instance of "Clang is broken" / bug 44466, right?
Comment 4 Laurent carlier 2012-06-24 06:20:02 UTC
(In reply to comment #2)
> I've been seeing a few strange crashes in the LLVM backend when using wine. 
> What version of wine are you using?  Does it work with an older version of
> wine?

I've tried with several version of wine without much success
Comment 5 Laurent carlier 2012-06-24 06:33:47 UTC
(In reply to comment #3)
> This isn't just another instance of "Clang is broken" / bug 44466, right?

llvm and mesa are all built with a multilib gcc, perhaps it's the culprit ? Another bug similer to https://bugs.freedesktop.org/show_bug.cgi?id=44466 ?

I've had to force host and target to i386-pc-linux-gnu to get a proper lib32 llvm.....


related ? http://llvm.org/bugs/show_bug.cgi?id=11145
Comment 6 Johannes Obermayr 2012-10-28 22:13:27 UTC
Does it crash with DRAW_USE_LLVM=0?

If not, please try whether this patch can provide more information:
https://bugs.freedesktop.org/show_bug.cgi?id=48351#c10
Comment 7 Laurent carlier 2012-10-31 12:53:04 UTC
Created attachment 69350 [details]
furmark output with llvm-3.1 and r600-llvm-compiler enabled

Wine doesn't fail anymore, now rendering only garbage, so there is progress!
Comment 8 Laurent carlier 2012-10-31 13:01:01 UTC
I don't know if it's related, but etqw gives me also similar garbaged output
Comment 9 Laurent carlier 2012-10-31 16:30:50 UTC
The culprit for broken output is (after bisecting):
5ab82e0ccf84855e9311ebfc58d1b57b437ed991 is the first bad commit
commit 5ab82e0ccf84855e9311ebfc58d1b57b437ed991
Author: Vincent Lejeune <vljn@ovi.com>
Date:   Fri Oct 19 15:49:06 2012 +0200

    r600g: tgsi-to-llvm emits right input intrinsics
    
    Reviewed-by: Tom Stellard <thomas.stellard at amd.com>

:040000 040000 e678b643c10733ff7ef7163f3e21aee0a16ebb6e 4d10bf29f966beb56e12c385529284c53db1f950 M      src
Comment 10 Laurent carlier 2012-10-31 16:35:01 UTC
No more crash, so closing, a new bug is opened for broken output.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.