Bug 23862 - [bisected G45] Blender segfaults
Summary: [bisected G45] Blender segfaults
Status: CLOSED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: unspecified
Hardware: Other All
: high normal
Assignee: Eric Anholt
QA Contact:
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2009-09-11 05:40 UTC by Sven Arvidsson
Modified: 2009-09-15 02:56 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
Viewport shading button (8.52 KB, image/png)
2009-09-11 05:40 UTC, Sven Arvidsson
Details
gdb log of blender crash on button push (5.33 KB, text/plain)
2009-09-11 13:54 UTC, Wolfgang Kufner
Details

Description Sven Arvidsson 2009-09-11 05:40:40 UTC
Created attachment 29412 [details]
Viewport shading button

Pressing the "Viewport Shading" (see screenshot if it's not apparent which one) button in Blender causes instant segfault. Works fine in software mode.

I think this is a regression, so I might try a bisect when I have time.

#0  0x00000000 in ?? ()
No symbol table info available.
#1  0xb5d3638a in _swrast_ReadPixels (ctx=0x85, x=337, y=263, width=133, 
    height=136, format=6408, type=5121, packing=0x8ef01fc, pixels=0xa054400)
    at swrast/s_readpix.c:605
        clippedPacking = {Alignment = 4, RowLength = 133, SkipPixels = 0, 
          SkipRows = 0, ImageHeight = 0, SkipImages = 0, SwapBytes = 0 '\0', 
          LsbFirst = 0 '\0', ClientStorage = 0 '\0', Invert = 0 '\0', 
          BufferObj = 0x8eafd18}
#2  0xb5c116cf in intelReadPixels (ctx=0x8ee32d8, x=337, y=263, width=133, 
    height=136, format=6408, type=5121, pack=0x8ef01fc, pixels=0xa054400)
    at intel_pixel_read.c:314
        __func__ = "intelReadPixels"
#3  0xb5caff36 in _mesa_ReadPixels (x=337, y=263, width=133, height=136, 
    format=6408, type=5121, pixels=0xa054400) at main/readpix.c:208
        ctx = 0x8ee32d8

System environment:
-- chipset: G45 / ICH10R
-- system architecture: 32-bit
-- Linux distribution: Debian unstable
-- Machine or mobo model: Asus P5Q-EM
-- Display connector: DVI
-- KMS: enabled
-- xf86-video-intel: efbcf29dd1a1ca058b7a2a93f0685102c06c9369
-- xserver: 1.6.3
-- mesa: 7c0152fbaeb21ab423a9de339b85c54d1713432b
-- drm: 67e4172394a88d4922fb8d9c7c3d96ce7e02c5a6
-- kernel: 2.6.31-rc9
Comment 1 Wolfgang Kufner 2009-09-11 13:53:48 UTC
This could be the same bug that I have observed and described in https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/427712 although here it does really not matter much what button in blender I click or even only mouse over. 

Anyway, this regression appeared between commits 4d85a6b3 and 18882f4d.
file: libgl1-mesa-dri

I attach my gdb log of the crash.

please let me know if you need further information
Comment 2 Wolfgang Kufner 2009-09-11 13:54:58 UTC
Created attachment 29426 [details]
gdb log of blender crash on button push
Comment 3 Sven Arvidsson 2009-09-11 14:02:15 UTC
(In reply to comment #1)
> This could be the same bug that I have observed and described in
> https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/427712 although here it
> does really not matter much what button in blender I click or even only mouse
> over. 

Doh! Indeed, almost any button will cause a segfault.

Comment 4 Sven Arvidsson 2009-09-11 14:40:34 UTC
Bisecting leads to this:

36dd53a3cded9d003ec418732b7fc93c1476aa9b is first bad commit
commit 36dd53a3cded9d003ec418732b7fc93c1476aa9b
Author: Eric Anholt <eric@anholt.net>
Date:   Thu Sep 10 09:26:38 2009 -0700

    intel: Don't forget to map the depth read buffer in spans.
    
    This broke BlitFramebufferEXT(GL_DEPTH_BUFFER_BIT).

:040000 040000 549d8f39d7717322b09dc656a5129f46162471f3 53fb80a11730f3ecfdf03231418e8dee0f7e984b M	src
Comment 5 Sven Arvidsson 2009-09-13 13:35:59 UTC
Might be the same as bug 23871
 
Comment 6 Gordon Jin 2009-09-14 20:18:49 UTC
Please retest with mesa master:

Branch: master
Commit: 8ec456c68cc69349c2506ce64047d06c7113780c
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=8ec456c68cc69349c2506ce64047d06c7113780c

Author: Brian Paul <brianp@vmware.com>
Date:   Mon Sep 14 12:40:53 2009 -0600

intel: fix renderbuffer map/unmap regression

Commit 36dd53a3cded9d003ec418732b7fc93c1476aa9b caused a few regressions
because the glReadBuffer() buffer wasn't getting mapped when GL_READ_BUFFER
!= GL_DRAW_BUFFER.
Comment 7 Wolfgang Kufner 2009-09-14 22:46:01 UTC
Yes, this bug is indeed fixed. :-)

I checked with fresh packages from ubuntu xorg-edgers ppa:
mesa (7.7.0~git20090914.d3f5a2e5-0ubuntu0tormod)
  * Checkout from git 20090914 (master branch) up to commit
    d3f5a2e5ab413ec0dab1efa8135705e5312d361d


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.