[Mesa-dev] [PATCH 3/6] mesa/st: Factor out array and buffer setup from st_atom_array.c.

[Chris Wilson]

Quoting Mathias Fröhlich (2018-11-23 17:14:45)
> Hi Chris,
> 
> On Friday, 23 November 2018 16:12:38 CET Chris Wilson wrote:
> > 
> > Something to note here is that valgrind reports
> > (piglit/bin/drawoverhead):
> > 
> > ==492== Use of uninitialised value of size 8
> > ==492==    at 0x6EB8FA9: cso_hash_find_node (cso_hash.c:198)
> > ==492==    by 0x6EB9026: cso_hash_insert (cso_hash.c:213)
> > ==492==    by 0x6EB6730: cso_set_vertex_elements (cso_context.c:1092)
> > ==492==    by 0x714C76F: set_vertex_attribs (st_atom_array.c:384)
> > ==492==    by 0x714C76F: st_update_array (st_atom_array.c:512)
> > ==492==    by 0x71073F3: st_validate_state (st_atom.c:261)
> > ==492==    by 0x70615D1: prepare_draw (st_draw.c:123)
> > ==492==    by 0x70615D1: st_draw_vbo (st_draw.c:149)
> > ==492==    by 0x70F5BF4: _mesa_validated_drawrangeelements (draw.c:850)
> > ==492==    by 0x70F5BF4: _mesa_validated_drawrangeelements (draw.c:782)
> > ==492==    by 0x70F5F32: _mesa_DrawElements (draw.c:1004)
> > ==492==    by 0x48F6C74: stub_glDrawElements (piglit-dispatch-gen.c:12618)
> > ==492==    by 0x10B3F4: draw (drawoverhead.c:275)
> > ==492==    by 0x10D070: perf_measure_rate (common.c:56)
> > ==492==    by 0x10C69E: perf_run (drawoverhead.c:645)
> > ==492==  Uninitialised value was created by a stack allocation
> > ==492==    at 0x714C25D: st_update_array (st_atom_array.c:389)
> > 
> > from velements being used sparsely.
> > 
> > Does your patch prevent this?
> 
> I tried to reproduce this, but valgrind does not show any failures with drawoverhead
> on radeonsi.
> What driver backend do you use?

iris, but we don't hit backends before the error on this path. It may
just be a bug fixed not yet merged into iris. Hmm, or a local patch,
perhaps:

commit 3492e5dfd8e8a7b0f37a3be2a18010628d46f695
Author: Kenneth Graunke <kenneth at whitecape.org>
Date:   Fri Aug 24 15:06:35 2018 -0700

    gallium, st/mesa: Support edge flags as a vertex element property

Thanks for checking.
-Chris