oizma / angleproject

Automatically exported from code.google.com/p/angleproject
Other
0 stars 0 forks source link

gl_FragCoord emulation problem in ANGLE #73

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
There appear to be additional problems with the gl_FragCoord emulation in 
ANGLE. Please see http://crbug.com/63269 for the test case. I've tried 
reverting http://code.google.com/p/angleproject/source/detail?r=473 in my local 
workspace and the problem still occurs, so I don't believe I introduced the 
problem with that fix.

Would appreciate it if someone from TransGaming could look into this.

Original issue reported on code.google.com by kbr@chromium.org on 16 Nov 2010 at 1:54

GoogleCodeExporter commented 9 years ago
I found out that it has something to do with the dx_Depth vector. If I replace 
its components with 0.5 (the default) in the HLSL, the correct image is 
rendered.

Oddly when I check the values that are being set for dx_Depth, they're also 
0.5. The shader assembly looks sound too. I'll check whether something is going 
wrong with the uniform to constant register mapping...

Original comment by nicolas....@gmail.com on 16 Nov 2010 at 1:20

GoogleCodeExporter commented 9 years ago
Found the issue. Program::getActiveUniform isn't correclty skipping internally 
used uniforms (like dx_Depth). So even though the application retrieves the 
correct uniform index, Chromium's internal list of uniforms didn't have the 
right names.

This mismatch caused the vertex shader's transform matrix to never be set, and 
so a null matrix is used and the triangles weren't visible.

This is a pretty serious bug and I'll fix it as soon as possible.

Original comment by nicolas....@gmail.com on 16 Nov 2010 at 4:55

GoogleCodeExporter commented 9 years ago

Original comment by nicolas....@gmail.com on 17 Nov 2010 at 1:11

GoogleCodeExporter commented 9 years ago
in r483

Original comment by dan...@transgaming.com on 17 Nov 2010 at 1:13