[Mesa-dev] Merge of glsl2 branch to master

Ian Romanick idr at freedesktop.org
Fri Aug 6 15:24:50 PDT 2010

Hash: SHA1

As of today the glsl2 branch has three piglit regressions WRT master,
and there is only one crashing piglit tests.  As before, we have tested
swrast and i965.  For master I used commit 27041d7cb, and for glsl2 I
used commit c234d0b2.  The results are available at
http://people.freedesktop.org/~idr/results/.  The regressed tests cases are:

 * draw_buffers-05.vert - As mentioned previously, the spec is ambiguous
and Nvidia and AMD produce different results for this test.  Until the
spec ambiguity is resolved, I don't care about this failure.

* glsl1-texcoord varying - This test accesses gl_TexCoord[] using a
non-constant index without dimensioning the array. The new compiler
correctly rejects this shader. Nvidia accepts it, but I believe they're
going outside the spec (which they do a lot). I haven't tested this on
AMD or Apple yet. Is this shader from an application?  We've changed the
test in piglit commit c6146f121, but I think it still isn't quite right.

 * fbo-drawbuffers-maxtargets - Fails on i965 because we don' have loop
unrolling in the compiler or good array handling in the driver.  We'll
get loop unolling in the compiler first.  Proper array handling will
come to the driver once we move away from using Mesa IR.

I propose that we merge master to glsl2 on *Friday, August 13th* (one
week from today).  Barring unforeseen issues, I propose that we merge
glsl2 to master on *Monday, August 16th*.

I believe that all of the concerns people raised before have either been
resolved or are in the process of being resolved:

 * Jakob is testing the glsl2 branch on softpipe and (presumably)
llvmpipe (as per Michel Dänzer)

 * We merged master to our branch about a week ago, and we will do so
again (with a retest) before we merge the branch to master.  We would
like to perform these merges in just over a week (as per Dave Airlie).

 * We have not added any tests from mesa-demos to piglit.  They have
been run, and we can run them again before merging.  The ones that
people really care about should be added to piglit (as per Brian Paul).

 * Using MESA_GLX_FORCE_ALPHA=y seems to partially resolve the alpha
read-back issue with swrast but not completely (as per Brian Paul).

 * There have been some regressions reported in r300, but these look to
be manageable.  I don't have the bug numbers handy (as per Tom Stellard).

 * Geometry shaders and GLSL 1.30 support are at the top of our post-Q3
todo list.  Since all of the geometry shader extensions have explicit
dependencies on either GLSL 1.30 or GL_EXT_gpu_shader4, it's not clear
to me how Mesa current supports them.  Since none of the hardware
drivers currently expose this extension, it seems that it be missed if
we lose this functionality for a quarter or two (as per Zack Rusin).

 * talloc has been ported to Windows (as per José Fonseca).

Have I missed anyone or any issue that was raised?

Between now and the merge, we have the following items on our list:

 * Fix the couple of build bugs (#29207 and #29209) that have been reported.

 * Fix the regression in fbo-drawbuffers-maxtargets.

Between now and shipping Mesa 7.9, we have the following major items on
our list:

 * Implement GLSL ES support in the front-end.

 * Implement loop analysis so that we can implement some loop
optimizations and loop unroll for architectures (e.g., i915) that lack

 * More tests.  We're working with a couple applications to get some of
their shaders contributed to piglit for testing.  We'll see what happens

As far as pushing for a Q3 release of Mesa 7.9 with the new compiler, I
don't think it's going to be a problem.  There are still some rough
spots, and there's still work to be done.  Overall, though, the new
compiler is much better than the existing compiler.  I think this meets
the criteria for a development release. :)

Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/


More information about the mesa-dev mailing list