Bug 19468

Summary: [965 OGLC] fptex.c test case fails
Product: Mesa Reporter: Ian Romanick <idr>
Component: Drivers/DRI/i965Assignee: Ian Romanick <idr>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: medium CC: brian.paul, haien.liu
Version: git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments: Check the texture coordinate and texture image unit separately.

Description Ian Romanick 2009-01-08 10:34:38 UTC 
Test case fptex.c fails.  This test case exercises available texture image units from ARB_fragment_program shaders.  I bisected the failure to the commit below.  I looked at the change, and I don't see anything obviously wrong.  Ugh. :(

commit 4a6ad999ea312f0af85de621c8b6a15a3d3b7ffd
Author: Brian Paul <brianp@vmware.com>
Date:   Thu Jan 1 14:04:57 2009 -0700

    i965: increase number of texture samplers to 16
    
    This lets GLSL shaders use up to 16 samplers.
    Fixed function is still limited to 8 textures.
    Tested with progs/glsl/samplers.c
Comment 1 Brian Paul 2009-01-08 13:02:43 UTC 
fptex.c?  I don't see a file by that name in the Mesa tree.
Comment 2 Eric Anholt 2009-01-13 17:25:37 UTC 
arbprogparse appears to be clueless about texture image units versus coord units.
Comment 3 Eric Anholt 2009-01-13 17:26:15 UTC 
Invalid texture unit index    File - fptex.c, line - 1075.
        !!ARBfp1.0
        OUTPUT out = result.color;
        TEMP textemp1, textemp2;
        TEX  textemp1, fragment.texcoord[7],texture[7], 1D;
        TEX  textemp2, fragment.texcoord[6],texture[8], 1D;
        ADD out, textemp1, textemp2;
        END
Comment 4 Ian Romanick 2009-01-13 18:11:05 UTC 
Created attachment 21961 [details] [review]
Check the texture coordinate and texture image unit separately.

Proposed fix.  Check the texture coordinate and texture image unit separately.  I haven't checked to see if nvprogparse.c needs the same treatment.  I haven't tested this yet either.
Comment 5 Brian Paul 2009-01-14 07:31:52 UTC 
Ian, I was about to post a patch as well.  Your's and mine look nearly identical.  If it indeed fixes the bug go ahead and commit.
Comment 6 Ian Romanick 2009-01-14 10:10:36 UTC 
I verified the proposed fix a couple minutes ago, and everything looks good.  I committed as 2549c26a8b1eec21bdd8f45d3b3dd06e17ac82ae.
Comment 7 Adam Jackson 2009-08-24 12:31:38 UTC 
Mass version move, cvs -> git

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.