Bug 30122

Summary: r600 scrambled render when using 2d texture sample for 1d texture co-ord in fragment shader
Product: Mesa Reporter: Thomas Jones <thomas.jones>
Component: Drivers/DRI/R600Assignee: Default DRI bug account <dri-devel>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: medium    
Version: git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments: small program that triggers the bug
Screenshot showing the brokeness
The expected rendering
r600c incorrect rendering

Description Thomas Jones 2010-09-10 11:40:26 UTC
Created attachment 38612 [details]
small program that triggers the bug

Using a shader like this results in a strange scrambled rendering:


uniform sampler2D src;
uniform sampler1D pal;
void main() {
	gl_FragColor = texture1D(pal, texture2D(src, gl_TexCoord[0].xy).x);
}

it looks like in addition to just being messed up it also gets the texture co-ordinates wrong (I'm rending with GL_QUADS, the two triangles that make up the quad seem to be interpolating differently across the middle)
Comment 1 Thomas Jones 2010-09-10 11:43:09 UTC
Created attachment 38613 [details]
Screenshot showing the brokeness
Comment 2 Thomas Jones 2010-09-10 11:47:20 UTC
Created attachment 38614 [details]
The expected rendering
Comment 3 Thomas Jones 2010-09-10 11:51:09 UTC
Also noticed, much the same happens with r600 classic too.
Comment 4 Pierre-Eric Pelloux-Prayer 2011-06-15 02:04:40 UTC
Created attachment 47987 [details]
r600c incorrect rendering

Tested with HD4850 + mesa from git :
  * r600g rendering is correct
  * r600c rendering is still wrong (see attached image)

Could you confirm it is fixed for you as well with r600g ?

If it's the case, maybe this bug should be assigned to r600c ?
Comment 5 Thomas Jones 2011-06-15 08:36:29 UTC
(In reply to comment #4)
> Created an attachment (id=47987) [details]
> r600c incorrect rendering
> 
> Tested with HD4850 + mesa from git :
>   * r600g rendering is correct
>   * r600c rendering is still wrong (see attached image)
> 
> Could you confirm it is fixed for you as well with r600g ?
> 
> If it's the case, maybe this bug should be assigned to r600c ?

Indeed you are correct, it does seem to be fixed on r600c in git, any idea what commit did it? It would probably be useful to point it out so whoever tries to fix r600c can see it.
Comment 6 Andreas Boll 2012-09-12 15:31:45 UTC
(In reply to comment #4)
> Created attachment 47987 [details]
> r600c incorrect rendering
> 
> Tested with HD4850 + mesa from git :
>   * r600g rendering is correct
>   * r600c rendering is still wrong (see attached image)
> 
> Could you confirm it is fixed for you as well with r600g ?
> 
> If it's the case, maybe this bug should be assigned to r600c ?

Note: classic r600 driver has been abandoned.

Closing.

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.