Bug 91039 - [d3d9] D3D9 state tracker crashes due to stack misalignment if Mesa is compiled with -mfpmath=sse
Summary: [d3d9] D3D9 state tracker crashes due to stack misalignment if Mesa is compil...
Status: RESOLVED MOVED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Gallium/StateTracker/galliumnine (show other bugs)
Version: 10.6
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: mesa-dev
QA Contact: mesa-dev
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-06-20 19:06 UTC by Itzamna
Modified: 2019-09-18 17:52 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:

Attachments

Description Itzamna 2015-06-20 19:06:20 UTC 
Description: 
The "Nine" D3D9 state tracker crashes when Wine loads any D3D program if Mesa is compiled with -mfpmath=sse. This is because the Linux ABI assumes a stack alignment to 16 bytes, whereas the Win32 ABI assumes a stack alignment to 4 bytes. This causes a stack misalignment and thus a general protection fault.


Additional information: 
Also see the following external bug reports:

https://bugs.winehq.org/show_bug.cgi?id=22475
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40838


Specifications:
OS: Gentoo GNU/Linux x86_64
Kernel version: Linux 4.0.4-hardened-r3
Mesa version: 10.6.0


Solution:
The crashes no longer occur when Mesa's stack is realigned to 4 bytes. Therefore, always pass -mstackrealign to CFLAGS if -mfpmath=sse is enabled; this should probably be handled in the configure script.
Comment 1 Itzamna 2015-06-20 19:27:19 UTC 
(In reply to Itzamna from comment #0)

> Solution:
> The crashes no longer occur when Mesa's stack is realigned to 4 bytes.
> Therefore, always pass -mstackrealign to CFLAGS if -mfpmath=sse is enabled;
> this should probably be handled in the configure script.

Erratum: -mstackrealign realigns the incoming (4 byte-aligned) stack to 16 bytes, see https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html .
Comment 2 Axel Davy 2018-05-09 06:10:53 UTC 
To my knowledge we've been handling function stack realignment in nine for a few years now, and thus I assume this issue is solved. Do you confirm ?
Comment 3 GitLab Migration User 2019-09-18 17:52:49 UTC 
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/122.

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.