Bug 45887 - [PNV bisected] oglc fbo-multisample(basic.drawpixels.msaatosingle.depth) regressed
Summary: [PNV bisected] oglc fbo-multisample(basic.drawpixels.msaatosingle.depth) regr...
Status: CLOSED WONTFIX
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i915 (show other bugs)
Version: git
Hardware: All Linux (All)
: low normal
Assignee: Chad Versace
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 42993
  Show dependency treegraph
 
Reported: 2012-02-10 04:15 UTC by fangxun
Modified: 2014-01-21 00:23 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments

Description fangxun 2012-02-10 04:15:20 UTC
System Environment:
--------------------------
Arch:           i386
Platform:       Pineview
Libdrm:         (master)2.4.31-1-g2cfac57d364d0166ed9472b086c16aea376f495a
Mesa:           (8.0)e86d90eb208292916c8a04c5d75499492003dd16
Xserver:      (server-1.11-branch)xorg-server-1.11.3
Xf86_video_intel:  (master)2.17.0-644-g4d8369f8e60fd4f5a0ef49f3e9866ea5ecb21927
Kernel:         (drm-intel-fixes) 617cf884810b44384fe8e9431e9babeb80a2ff37

Bug detailed description:
------------------------- 
It fails at mesa 8.0 and master branch on pineview. Bisect fails due to some mesa commit build failure. The last known bad commit is
Mesa(8.0)c85402aba91755808729fadf57a9f92285f4e61a, the last known good commit is Mesa(7.11)7722a54e60c9da080b557d6596e4c9d363ead515.


Reproduce steps:
----------------
1. start X
2. ./oglconform -z -s -suite all -v 2 -test fbo-multisample basic.drawpixels.msaatosingle.depth
Comment 1 Ian Romanick 2012-02-10 16:27:33 UTC
I just tried the "last known good commit" (7722a54).  Did this test ever actually work on i915 platforms?

oglconform64: main/texfetch_tmp.h:2317: fetch_texel_2d_f_s8_z24: Assertion `texImage->TexFormat == MESA_FORMAT_S8_Z24' failed.
Comment 2 Ian Romanick 2012-02-10 16:28:38 UTC
(In reply to comment #1)
> I just tried the "last known good commit" (7722a54).  Did this test ever
> actually work on i915 platforms?

The missing sentence there is "and got the following assertion failure."

> oglconform64: main/texfetch_tmp.h:2317: fetch_texel_2d_f_s8_z24: Assertion
> `texImage->TexFormat == MESA_FORMAT_S8_Z24' failed.
Comment 3 Ian Romanick 2012-02-10 17:10:41 UTC
Never mind.  I was able to bisect this bug to the following commit:

commit af9548d3358884b408fb4c3cc69028befc5abe1d
Author: Eric Anholt <eric@anholt.net>
Date:   Tue Jul 12 10:20:53 2011 -0700

    i915: Add support for HW rendering with no color draw buffer.
    
    This is useful for shadow map generation.  Tested with glsl-bug-22603,
    which rendered the depth textures with fallbacks before.
    
    Acked-by: Chad Versace <chad@chad-versace.us>
Comment 4 Chad Versace 2012-08-06 16:25:38 UTC
Assigning to self (chad). This bug hasn't been updated in 6 months, and it appears to be low-hanging fruit.
Comment 5 Chad Versace 2012-08-15 00:15:05 UTC
Changing importance=medium/minor.

This test no longer aborts due to assertion failure. It fails due to unexpected depth values. 

The test should skip because MSAA is not supported on Pineview. Yet, oglconform runs the test anyway because 

But I don't know whose fault this is: Mesa's or OGLconform's.
Comment 6 Chad Versace 2012-08-15 00:22:36 UTC
Undo last edit. I accidentally hit the Enter key. Oops.
Comment 7 Paul Berry 2012-08-15 13:38:23 UTC
Note: this comment may help explain why oglconform is running msaa tests on Pineview (which doesn't support MSAA):

https://bugs.freedesktop.org/show_bug.cgi?id=50165#c7

(Briefly, Mesa exposes EXT_framebuffer_multisample even on platforms that don't support multisampling, but it sets GL_MAX_SAMPLES_EXT to 1 to indicate that multisampling isn't actually supported.  This is necessary because EXT_framebuffer_multisample is a required subpart of ARB_framebuffer_object, which those platforms do support).

Unfortunately, that comment doesn't explain why the oglconform test has regressed.
Comment 8 Chad Versace 2013-01-29 22:41:16 UTC
Dropping important to low/normal because this is an oglconform regression on Pineview.
Comment 9 Chad Versace 2013-08-22 17:58:02 UTC
Status ASSIGNED -> NEW, because I'm actively working on it.
Comment 10 Chad Versace 2014-01-20 20:08:05 UTC
NEW -> RESOLVED/WONTFIX

This is an oglconform bug on Pineview. No one's working on it, and I doubt anyone ever will.


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.