https://bugs.freedesktop.org/show_bug.cgi?id=104143
Bug ID: 104143 Summary: r600/sb: clobbers gl_Position -> gl_FragCoord Product: Mesa Version: git Hardware: x86-64 (AMD64) OS: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 Assignee: dri-devel@lists.freedesktop.org Reporter: gw.fossdev@gmail.com QA Contact: dri-devel@lists.freedesktop.org
Created attachment 135998 --> https://bugs.freedesktop.org/attachment.cgi?id=135998&action=edit piglit drawing output with sb enabled
With a variation of the piglit
spec/glsl-1.10/execution/variable-indexing/vs-output-array-vec2-index-wr
r600/sb clobbers the gl_Position/gl_FragCoord.
The variation consists in replacing two arrays of vec2 by one array of vec4 and swizzling the elements to achieve the same result (by effectively interleaving the two arrays).
A few observations:
* As can be seen from the colour coding screen shots, the array content is correctly passed from the vertex shader to the fragment shader.
* The error only occurs for the uniform index value 0.
* When passing an additional parameter that contains a copy of gl_Position, then this parameter seems to have the correct value (i.e. the vertex shader correctly evaluates gl_Position).
* The byte code doesn't doesn't give any obvious indication why things go wrong with the optimized shader.
My mesa is at fa8c1b92b7.
best, Gert
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #1 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 135999 --> https://bugs.freedesktop.org/attachment.cgi?id=135999&action=edit Version of the original piglit that passes
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #2 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136000 --> https://bugs.freedesktop.org/attachment.cgi?id=136000&action=edit piglit screen output of simplified piglit
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #3 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136001 --> https://bugs.freedesktop.org/attachment.cgi?id=136001&action=edit Version of the piglit that uses interleaved array and fails with sb
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #4 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136002 --> https://bugs.freedesktop.org/attachment.cgi?id=136002&action=edit Version of the piglit that passes copy of gl_Position and tests it
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #5 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136003 --> https://bugs.freedesktop.org/attachment.cgi?id=136003&action=edit Piglit screen output with R600_DEBUG=nosb of shaders with extra pos parameter
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #6 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136004 --> https://bugs.freedesktop.org/attachment.cgi?id=136004&action=edit Piglit output with extra pos parameter and sb enabled
In this image on can see that the (corrected) position passed as extra parameter differs from the gl_Position for index=0 (colour coded difference), but only for index 0.
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #7 from Gert Wollny gw.fossdev@gmail.com --- Created attachment 136005 --> https://bugs.freedesktop.org/attachment.cgi?id=136005&action=edit Shader dump with pos test
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #8 from Gert Wollny gw.fossdev@gmail.com --- I found the problem:
if KC0[0].x == index (=0):
1 x: ADD_INT T0.x, KC0[0].x, [0xfffffffe -nan].x 2 x: MOVA_INT __.x, T0.x
Address register is now -2 and hence, in the next step R1 is unconditionally written, and this is actually the gl_Vertex value ...
3 z: MOV R[3+AR].z, 0 w: MOV R[3+AR].w, [0x3dcccccd 0.1].x
that is here used to evaluate the gl_Posuition.
5 x: MUL_IEEE T0.x, KC0[1].w, R1.x y: MUL_IEEE T0.y, KC0[1].z, R1.x 6 t: MULADD_IEEE T0.y, KC0[2].z, R1.y, T0.y SCL_212 ...
In the un-optimized shader R[3+AR].w is only written to if (KC0[0].x >= 2), and hence AR >= 0;
I.e. the sb optimizer is to aggressive in optimizing away the conditional blocks.
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #9 from Gert Wollny gw.fossdev@gmail.com --- Patch: https://patchwork.freedesktop.org/patch/192036/
https://bugs.freedesktop.org/show_bug.cgi?id=104143
--- Comment #10 from Emil Velikov emil.l.velikov@gmail.com --- Gert, should we close this considering the patch (fix?) has landed?
https://bugs.freedesktop.org/show_bug.cgi?id=104143
Gert Wollny gw.fossdev@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |FIXED Status|NEW |RESOLVED
--- Comment #11 from Gert Wollny gw.fossdev@gmail.com --- Fixed with commit 6c268ea79af80a65a89a23854bdbe8bc1e99ab23
dri-devel@lists.freedesktop.org
-
bugzilla-daemon＠freedesktop.org