Bug 25154 - [i965 bisected] OGLC/pxtrans-tex.c fail on visuals without alpha
Summary: [i965 bisected] OGLC/pxtrans-tex.c fail on visuals without alpha
Status: CLOSED WONTFIX
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: unspecified
Hardware: All Linux (All)
: medium normal
Assignee: Ian Romanick
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
: 25192 (view as bug list)
Depends on:
Blocks:
 
Reported: 2009-11-17 19:17 UTC by zhao jian
Modified: 2015-06-16 02:02 UTC (History)
3 users (show)

See Also:
i915 platform:
i915 features:


Attachments
xorg.0.log (19.62 KB, text/plain)
2009-11-17 19:17 UTC, zhao jian
Details

Description zhao jian 2009-11-17 19:17:12 UTC
Created attachment 31287 [details]
xorg.0.log

System Environment:
--------------------------
Libdrm:		(master)c20706ffe56cc658a4814f89e391c14f6065ec57
Mesa:		(master)e015a4c29bf61077a50780cc99381510671b20ec
Xserver:		(master)fb95090730360b6b7f5429c40937e8fc9dfe5c14
Xf86_video_intel:	(master)c180baf43b8a0e407448018f3a7e42491cf974ae
Kernel:       (drm-intel-next)5b8f0be0dce012d190a53d55240fe3fde6306476

Bug detailed description:
-------------------------
This issue happens on G45.
There are 4 oglc cases(pxconv-gettex.c, pxstore-gettex.c, pxtrans-gettex.c, pxtrans-tex.c) regressed due to the following commit:
commit eb437fabe017611e1f855fffa45f59cd38709be8
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Mon Nov 16 14:49:40 2009 -0800

    AL1616: Add formats for GL_LUMINANCE16_ALPHA16 textures

The error output:
    File - util-pxconv.c, line - 855.
        PixelStore packing settings:  SWAP_BYTES is TRUE.  LSB_FIRST is
        FALSE. Skip 0 pixels, 0 rows, and 0 images.  Alignment is 1 bytes.
        Row length is 0; image height is 0. SWAP_BYTES GetTexImage
        GL_TEXTURE_1D (GL_LUMINANCE12_ALPHA4) test failed at (0,0) with
        format GL_RGBA and type GL_HALF_FLOAT_ARB.  Test generated an RGBA
        color of (0.210, 0.744, 1.000, 0.321). Read back {0.491, 0.000,
        0.000, 0.000}. Expected {0.210, 0.000, 0.000, 0.321}.
    State Report. (Current) <-> (Default)
        GL_ATTRIB_STACK_DEPTH (State Information)
            1.00 <-> 0.00
        GL_DITHER (State Information)
            GL_FALSE <-> GL_TRUE
        GL_LINE_STIPPLE_PATTERN (State Information)
            0.00 <-> 65535.00
        GL_PACK_ALIGNMENT (State Information)
            1.00 <-> 4.00
        GL_PACK_SWAP_BYTES (State Information)
            GL_TRUE <-> GL_FALSE
        GL_PROJECTION_MATRIX (State Information)
            +0.020 +0.000 +0.000 +0.000 <-> +1.000 +0.000 +0.000 +0.000
            +0.000 +0.020 +0.000 +0.000 <-> +0.000 +1.000 +0.000 +0.000
            +0.000 +0.000 -1.000 +0.000 <-> +0.000 +0.000 +1.000 +0.000
            -1.000 -1.000 +0.000 +1.000 <-> +0.000 +0.000 +0.000 +1.000
        GL_POLYGON_STIPPLE (State Information)
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
             0  0  0  0  0  0  0  0 <-> FF FF FF FF FF FF FF FF
        GL_CLIENT_ATTRIB_STACK_DEPTH (State Information)
            1.00 <-> 0.00

Reproduce steps:
-------------------------
1. xinit&
2. run oglc cases
Comment 1 zhao jian 2009-11-17 22:19:20 UTC
The glean case pixelFormats on G45 also failed due to this commit. 
Comment 2 Ian Romanick 2009-11-17 22:35:40 UTC
(In reply to comment #0)

> commit eb437fabe017611e1f855fffa45f59cd38709be8
> Author: Ian Romanick <ian.d.romanick@intel.com>
> Date:   Mon Nov 16 14:49:40 2009 -0800
> 
>     AL1616: Add formats for GL_LUMINANCE16_ALPHA16 textures

I think the breakage caused by this commit is different.  This commit added some values to the gl_format enum.  This caused almost all of the existing enums to change, and this caused a bunch of array index values to break.  This can be verified by trying this commit:

commit 3325dc91be2534079ebf7997700b6e5f17a75283
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Mon Nov 16 14:53:12 2009 -0800

    AL1616: Add TexImage storage path

We've seen a lot of these texture format errors, and they all seem to involve GL_HALF_FLOAT.  I wonder if we should disable GL_ARB_half_float_pixel in the i965 driver.
Comment 3 Ian Romanick 2009-11-17 23:00:36 UTC
glean's pixelFormats test passes with this commit (to mesa_7_7_branch):

commit 3325dc91be2534079ebf7997700b6e5f17a75283
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Mon Nov 16 14:53:12 2009 -0800

    AL1616: Add TexImage storage path
Comment 4 Ian Romanick 2009-11-18 14:25:34 UTC
The irony of a cut-and-paste error in the bug close message for a cut-and-paste bug.  This bug is actually fixed by *THIS* commit to the mesa_7_7_branch:

commit 70dca0c273d681d004b014dd8d4434be664cb202
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Tue Nov 17 22:53:06 2009 -0800

    AL1616: Fix cut-and-paste bug
    
    One of the PACK_COLOR_88 cases was left over from copying
    _mesa_texstore_al88 to _mesa_texstore_al1616.
Comment 5 Ian Romanick 2009-11-19 19:50:10 UTC
*** Bug 25192 has been marked as a duplicate of this bug. ***
Comment 6 zhao jian 2009-11-20 00:20:30 UTC
(In reply to comment #3)
> glean's pixelFormats test passes with this commit (to mesa_7_7_branch):
> commit 3325dc91be2534079ebf7997700b6e5f17a75283
> Author: Ian Romanick <ian.d.romanick@intel.com>
> Date:   Mon Nov 16 14:53:12 2009 -0800
>     AL1616: Add TexImage storage path

Yes. This case first be fixed by this commit. But it was failed again with the new commit of fdfbae3381553fc93202560abe3b41b4d543f0bf. So I reopen it. 
commit fdfbae3381553fc93202560abe3b41b4d543f0bf
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Mon Nov 16 15:09:24 2009 -0800

    i965: Use MESA_FORMAT_AL1616 when appropriate
Comment 7 Ian Romanick 2009-11-21 17:12:34 UTC
(In reply to comment #6)
> (In reply to comment #3)
> > glean's pixelFormats test passes with this commit (to mesa_7_7_branch):
> > commit 3325dc91be2534079ebf7997700b6e5f17a75283
> > Author: Ian Romanick <ian.d.romanick@intel.com>
> > Date:   Mon Nov 16 14:53:12 2009 -0800
> >     AL1616: Add TexImage storage path
> 
> Yes. This case first be fixed by this commit. But it was failed again with the
> new commit of fdfbae3381553fc93202560abe3b41b4d543f0bf. So I reopen it. 
> commit fdfbae3381553fc93202560abe3b41b4d543f0bf
> Author: Ian Romanick <ian.d.romanick@intel.com>
> Date:   Mon Nov 16 15:09:24 2009 -0800
> 
>     i965: Use MESA_FORMAT_AL1616 when appropriate
> 

Re-read comment #4.  Do you have the correct fix in your tree?  With the commit mentioned in comment #4, I am not able to reproduce this bug, but I was able to reproduce it without that commit.
Comment 8 zhao jian 2009-11-22 18:01:09 UTC
(In reply to comment #7)
> (In reply to comment #6)
> > (In reply to comment #3)
> > > glean's pixelFormats test passes with this commit (to mesa_7_7_branch):
> > > commit 3325dc91be2534079ebf7997700b6e5f17a75283
> > > Author: Ian Romanick <ian.d.romanick@intel.com>
> > > Date:   Mon Nov 16 14:53:12 2009 -0800
> > >     AL1616: Add TexImage storage path
> > 
> > Yes. This case first be fixed by this commit. But it was failed again with the
> > new commit of fdfbae3381553fc93202560abe3b41b4d543f0bf. So I reopen it. 
> > commit fdfbae3381553fc93202560abe3b41b4d543f0bf
> > Author: Ian Romanick <ian.d.romanick@intel.com>
> > Date:   Mon Nov 16 15:09:24 2009 -0800
> > 
> >     i965: Use MESA_FORMAT_AL1616 when appropriate
> > 
> Re-read comment #4.  Do you have the correct fix in your tree?  With the commit
> mentioned in comment #4, I am not able to reproduce this bug, but I was able to
> reproduce it without that commit.

Sorry, I omitted your update in comment #4. This case is first regressed by commit eb437fabe017611e1f855fffa45f59cd38709be8 and fixed by commit 3325dc91be2534079ebf7997700b6e5f17a75283, then reintroduced by commit fdfbae3381553fc93202560abe3b41b4d543f0bf and finally was fixed by commit 70dca0c273d681d004b014dd8d4434be664cb202 now. It works well both on master and mesa_7_7 branch now. So verified. 
Comment 9 Shuang He 2009-12-09 18:33:29 UTC
For you reference, those 4 OGLC regression, which is the original issue reported in this bug, is just fixed a few days ago by following commit:
commit db352f58fab419c475b89418cd27b35f5f5d3822
Author: Eric Anholt <eric@anholt.net>
Date:   Wed Dec 2 12:42:36 2009 -0800

    mesa: Fix bad conversion in AL1616_REV texstore.
Comment 10 Shuang He 2009-12-09 18:41:58 UTC
(In reply to comment #9)
> For you reference, those 4 OGLC regression, which is the original issue
> reported in this bug, is just fixed a few days ago by following commit:
> commit db352f58fab419c475b89418cd27b35f5f5d3822
> Author: Eric Anholt <eric@anholt.net>
> Date:   Wed Dec 2 12:42:36 2009 -0800
> 
>     mesa: Fix bad conversion in AL1616_REV texstore.
> 

Sorry, only 3 of them fixed, pxtrans-tex.c still fails
Comment 11 Ian Romanick 2009-12-11 14:10:01 UTC
pxtrans-tex.c only fails on visuals without alpha, so it's failure seems unrelated to the other failures.
Comment 12 Kristian Høgsberg 2010-05-10 17:18:40 UTC
(In reply to comment #10)
> (In reply to comment #9)
> > For you reference, those 4 OGLC regression, which is the original issue
> > reported in this bug, is just fixed a few days ago by following commit:
> > commit db352f58fab419c475b89418cd27b35f5f5d3822
> > Author: Eric Anholt <eric@anholt.net>
> > Date:   Wed Dec 2 12:42:36 2009 -0800
> > 
> >     mesa: Fix bad conversion in AL1616_REV texstore.
> > 
> 
> Sorry, only 3 of them fixed, pxtrans-tex.c still fails

Is this still the case?  I can't reproduce any failure with the pxtrans-tex.c case.
Comment 13 Eric Anholt 2010-06-04 12:52:04 UTC
Same here, looks like this testcase is fine now, tested on G45.
Comment 14 zhao jian 2010-06-10 20:36:12 UTC
It failed on visuals without alpha channel on all our platforms(G45, Piketon, Pineview). With mesa both on master and 7.8 branch. 
Libdrm:		(master)2.4.20-22-ge65caeba9ed0e6c53830d944248aaae2228351ab
Mesa:		(master)mesa_7_6_1_rc1-8765-g9ef6d34f7e03f3d33c0ebad4191f3300a9062c4a
Xserver:		(master)xorg-server-1.8.0-363-g8e97e5f9425639ad0a084150d0b232cad417595d
Xf86_video_intel:		(master)2.11.0-166-g8c1a8d22978f143044cc3cd0ff2ade5736ee9d48
Kernel:       (for-linus)c496fa1fff0248ef8cd637efb52b70dea7afaa9d
Comment 15 Ouping Zhang 2011-04-28 02:49:57 UTC
System Environment:
--------------------------
Libdrm:    (master)2.4.25
Mesa:      (7.10)73f4273b8647abe0020ad3c2d593e009092d2be5
Xserver:   (server-1.10-branch)xorg-server-1.10.1
Xf86_video_intel:   (master)2.15.0-6-g67e5a74e997f199327f9115c7ba867df3c49da8d

case pxtrans-tex still fails on PineView, G45 and Piketon.
Comment 16 Gordon Jin 2015-06-16 02:01:54 UTC
closing as won't fix.


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.