On Tuesday 22 January 2008 16:41:01 Ian Romanick wrote:
> Brian Paul wrote:
> | Ian Romanick wrote:
> |> Every program in progs/glsl either segfaults or hits an assertion.  Is
> |> there some magic set of configure options that I should use with LLVM to
> |> just make it work?  I only ask because one of the assertions was:
> |>
> |> LLVM JIT requested to do lazy compilation of function 'execute_shader4'
> |> when lazy compiles are disabled!
> |>
> |> This was with texdemo1.
> |
> | Does something simple like trivial/tri.c work?
>
> Everything that I tried in progs/trivial seemed to work.  I tried tri,
> tri-tri, tri-z, tri-fog, and quad-tex-2d.
>
> | I think Zack will be back on LLVM work shortly.  What's in the gallium
> | tree is a bit stale, I think.  He'll probably comment when he comes
> | on-line...
>
> Is there any path that is known to work?  I hit assertions without LLVM
> (SSE path, presumably) as well.

Yes, this is a bit of a mess right now and I'm just about to fix it. There's a 
few issues here:
a) LLVM/SSE fragment shaders don't handle x/y coordinates (this shouldn't 
cause segfaults, just no rendering)
b) the lazy compilation is disabled for debugging purposes (this way I can 
dump the generated assembly code in gdb right away), unfortunately it means 
that every shader but the the initial one will cause an assert (this is the 
ee->DisableLazyCompilation() call in gallivm.cpp)
c) haven't committed the change of layout yet and everything depends on that

I'm back on llvm code this week, so please give me a few days.

z

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Mesa3d-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev

Reply via email to