Bug 45422 - [bisected i965] oglc lodbias(basic.allCases) regressed
Summary: [bisected i965] oglc lodbias(basic.allCases) regressed
Status: VERIFIED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: git
Hardware: All Linux (All)
: high major
Assignee: Ian Romanick
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 42993
  Show dependency tree / graph
 
Reported: 2012-01-30 22:37 UTC by fangxun
Modified: 2014-07-17 09:20 UTC (History)
3 users (show)

See Also:
i915 platform:
i915 features:

Attachments

Description fangxun 2012-01-30 22:37:10 UTC 
System Environment:
--------------------------
Arch:           x86_64
Platform:       Sugarbay
Libdrm:         (master)2.4.30-18-gb643b0713aefdc0611e47654e88263b53b0de6f5
Mesa:           (8.0)caebd7929dca802ece8ef36b0f85094d66133b57
Xserver:      (server-1.11-branch)xorg-server-1.11.3
Xf86_video_intel: (master)2.17.0
Kernel:   (drm-intel-fixes)a4ea430853b71753103ec693acfc8624bd3e748e

Bug detailed description:
------------------------- 
It fails on Sandybridge and Ironlake. 
Bisect shows e10c47adffb0371cba3d6a2c80313489e6fa8767 is the first bad commit.
commit e10c47adffb0371cba3d6a2c80313489e6fa8767
Author:     Eric Anholt <eric@anholt.net>
AuthorDate: Fri Jan 20 13:43:11 2012 -0800
Commit:     Eric Anholt <eric@anholt.net>
CommitDate: Fri Jan 27 11:29:27 2012 -0800

    mesa: Fix CopyTex{Sub,}Image error checks for integer vs non-integer.

    Fixes Intel oglconform negative.typeFormatMismatch.copyteximage.

    NOTE: This is a candidate for the 8.0 branch.

    Reviewed-by: Brian Paul <brianp@vmware.com>
    (cherry picked from commit 422b18794eacc8f0be5b4e9611e2726f4e0d86b6)


Reproduce steps:
----------------
1. start X
2. ./oglconform -z -s -suite all -v 2 -test lodbias basic.allCases
Comment 1 Ian Romanick 2012-02-02 11:29:40 UTC 
I cannot reproduce this failure on my mobile SNB (8086:0126 (rev 09)).  If you are still able to reproduce this either at current tip of 8.0 (0e08205) or at the bisected commit, can you provide the '-v 4' output from one failing visual?
Comment 2 Gordon Jin 2012-02-05 18:30:21 UTC 
Xun is out today. 

Ouping, can you check it today?
Comment 3 Gordon Jin 2012-02-06 01:04:49 UTC 
(In reply to comment #2)
> Xun is out today. 
> Ouping, can you check it today?

Oops, just realized Ouping is also out today. Let's wait for Xun tomorrow.
Comment 4 fangxun 2012-02-07 01:59:48 UTC 
This failure is unstable on mobile SNB. It is easy to reproduce this failure on sugarbay at current tip of 8.0(65b50085). 
The bisect commit seems not correct because autobisect find another bad commit.
commit 289a7e74b32267b6b65a02f387f6701d70aa2eaf
Author: Brian Paul 
Date: Mon Jan 16 10:54:01 2012 -0700

swrast: remove unused _swrast_put_row()
(cherry picked from commit 64be85540f30e904784c6a72f2ba2784d5ccb36d)


Output with '-v 4':
./oglconform -z -s -suite all -v 4 -D 123 -test lodbias basic.allCases
Intel OpenGL Conformance Test
Version ENG (Feb  6 2012 17:15:38)

CLI options echo:
oglconform -z -s -suite all -v 4 -D 123 -test lodbias basic.allCases

Window will be recreated 1 times.
  Window 0 will run 1 testcases on config with id 123.
Total of 1 testcases will be executed.

WARNING: OpenCL is not supported.
Setup Report.
    Verbose level = 4.
    Random number seed = 1.
    Path inactive.
    Test Repetition Number = 1.

Visual Report for ID 123 (24 bits).
ID      |ACCELERA|DB      |REND_T  |SURF_T  |C_BUF_T |BUF_S   |RED_S   |
     123|       1|       1|      gl|  wipbpx|    rgba|      24|       8|

GREEN_S |BLUE_S  |ALPHA_S |DEPTH_S |STENC_S |ACCUM_S |SPL_BUF |SAMPLES |
       8|       8|       0|       0|       0|       0|       0|       0|

SRGB    |TEX_RGB |TEX_RGBA|CAVEAT  |SWAP    |M_PBUF_W|M_PBUF_H|M_PBUF_P
      -1|       0|       0|    none|   undef|       0|       0|       0

Epsilon Report.
    zero error epsilon = 0.000122.
    RGBA error epsilon = 0.00404, 0.00404, 0.00404, 0.000122.
    Depth buffer error epsilon = 0.000122.
    Accumulation error epsilon = 0.000122, 0.000122, 0.000122, 0.000122.

OpenGL Report.
    Vendor - 'Tungsten Graphics, Inc'
    Renderer - 'Mesa DRI Intel(R) Sandybridge Desktop '
    Version - '2.1 Mesa 8.0-rc2 (git-65b5008)'
    GLSL Version - '1.30'

>> TextureLodBias (lodbias)  test:

PATHS ENABLED:
PATHS DEFAULT: PATH_ALIAS PATH_ALPHA PATH_BLEND PATH_DEPTH PATH_DITHER PATH_FOG PATH_LOGICOP PATH_SHADE PATH_STENCIL PATH_STIPPLE PATH_CLAMP_COLOR PATH_MSAA
--> 1.1 - basic.allCases subcase:
    File - /GFX/build/testsuite/Oglconform_31/oglconform_31/src/OGLconform/lodbias.c, line - 305.
        Texture target is GL_TEXTURE_1D. Map size is 32 by 32. Magnification
        filter is GL_LINEAR. Minification filter is GL_NEAREST_MIPMAP_LINEAR.
        Lod bias is -14. Color is (0, 1, 0), should be (1, 0, 0).
Test Run() failed
--< 1.1 - basic.allCases subcase failed.
<< TextureLodBias (lodbias)  test failed. (1 of 1 subcases)

Intel Conformance failed.
Total Passed : 0
Total Failed : 1
Total Not run: 0
Comment 5 Ian Romanick 2012-02-07 17:29:25 UTC 
(In reply to comment #4)
> OpenGL Report.
>     Vendor - 'Tungsten Graphics, Inc'
>     Renderer - 'Mesa DRI Intel(R) Sandybridge Desktop '
>     Version - '2.1 Mesa 8.0-rc2 (git-65b5008)'
                 ^^^
This should say 3.0.  Are you not building with --enable-texture-float?

>     GLSL Version - '1.30'
Comment 6 fangxun 2012-02-07 18:28:30 UTC 
We are not building with '--enable-texture-float' on Mesa 8.0 branch.
I build mesa 8.0 with '--enable-texture-float' and retest the case.

Intel OpenGL Conformance Test
Version ENG (Feb  7 2012 18:51:01)

CLI options echo:
oglconform -z -s -suite all -v 4 -D 123 -test lodbias basic.allCases

Window will be recreated 1 times.
  Window 0 will run 1 testcases on config with id 123.
Total of 1 testcases will be executed.

WARNING: OpenCL is not supported.
Setup Report.
    Verbose level = 4.
    Random number seed = 1.
    Path inactive.
    Test Repetition Number = 1.

Visual Report for ID 123 (24 bits).
ID      |ACCELERA|DB      |REND_T  |SURF_T  |C_BUF_T |BUF_S   |RED_S   |
     123|       1|       1|      gl|  wipbpx|    rgba|      24|       8|

GREEN_S |BLUE_S  |ALPHA_S |DEPTH_S |STENC_S |ACCUM_S |SPL_BUF |SAMPLES |
       8|       8|       0|      24|       8|      48|       0|       0|

SRGB    |TEX_RGB |TEX_RGBA|CAVEAT  |SWAP    |M_PBUF_W|M_PBUF_H|M_PBUF_P
      -1|       0|       0|    slow|   undef|       0|       0|       0

Epsilon Report.
    zero error epsilon = 0.000122.
    RGBA error epsilon = 0.00404, 0.00404, 0.00404, 0.000122.
    Depth buffer error epsilon = 0.000122.
    Accumulation error epsilon = 0.000137, 0.000137, 0.000137, 0.000122.

OpenGL Report.
    Vendor - 'Tungsten Graphics, Inc'
    Renderer - 'Mesa DRI Intel(R) Sandybridge Desktop '
    Version - '3.0 Mesa 8.0-devel (git-aed51e3)' (3.0)
    GLSL Version - '1.30'
    Context Flags - None

>> TextureLodBias (lodbias)  test:

PATHS ENABLED:
PATHS DEFAULT: PATH_ALIAS PATH_ALPHA PATH_BLEND PATH_DEPTH PATH_DITHER PATH_FOG PATH_LOGICOP PATH_SHADE PATH_STENCIL PATH_STIPPLE PATH_CLAMP_COLOR PATH_MSAA
--> 1.1 - basic.allCases subcase:
    File - /GFX/build/testsuite/Oglconform_31/oglconform_31/src/OGLconform/lodbias.c, line - 305.
        Texture target is GL_TEXTURE_1D. Map size is 32 by 32. Magnification
        filter is GL_LINEAR. Minification filter is GL_NEAREST_MIPMAP_LINEAR.
        Lod bias is -14. Color is (0, 1, 0), should be (1, 0, 0).
Test Run() failed
--< 1.1 - basic.allCases subcase failed.
<< TextureLodBias (lodbias)  test failed. (1 of 1 subcases)
Intel Conformance failed.
Total Passed : 0
Total Failed : 1
Total Not run: 0
Comment 7 Gordon Jin 2012-03-31 18:17:50 UTC 
(In reply to comment #4)
> This failure is unstable on mobile SNB. It is easy to reproduce this failure on
> sugarbay at current tip of 8.0(65b50085). 
> The bisect commit seems not correct because autobisect find another bad commit.
> commit 289a7e74b32267b6b65a02f387f6701d70aa2eaf
> Author: Brian Paul 
> Date: Mon Jan 16 10:54:01 2012 -0700
> swrast: remove unused _swrast_put_row()
> (cherry picked from commit 64be85540f30e904784c6a72f2ba2784d5ccb36d)

can we trust this bisect result?
Comment 8 Ian Romanick 2012-07-20 16:25:21 UTC 
I am unable to reproduce this failure on current Mesa master or Mesa 8.0.4 release on Sandybridge mobile.  Xun, can you confirm?
Comment 9 fangxun 2012-07-23 09:09:20 UTC 
It still fails on current Mesa master or Mesa 8.0.4 release on Sandybridge(include mobile), Ironlake and Ivybridge.
Comment 10 Eric Anholt 2012-07-31 17:32:21 UTC 
reproduces on my old oglconform snapshot on ivb, at least.
Comment 11 Anuj Phogat 2013-02-09 00:14:28 UTC 
Couldn't reproduce. Test passes with mesa master(1e85713) on mobile SNB.
Comment 12 fangxun 2013-02-21 08:40:33 UTC 
I am able to reproduce this failure on current Mesa master or Mesa 9.1 on  Ironlake, Sandybridge, Ivybridge and Haswell.
Comment 13 Tapani Pälli 2014-01-07 11:38:19 UTC 
tested on SNB desktop with oglconform 42.01.580 and mesa master [3.0 Mesa 10.1.0-devel (git-8d1400f], all 12 tests are passing.
Comment 14 Tapani Pälli 2014-01-07 11:44:34 UTC 
all tests passing also on IVB ultrabook, therefore resolving this
Comment 15 fangxun 2014-07-17 09:20:51 UTC 
Verified it on latest mesa master and 10.2 branch.

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.