mirror of
https://gitlab.freedesktop.org/mesa/mesa.git
synced 2026-05-05 13:58:04 +02:00
v3d: Block bin on render when doing vertex texturing.
The kernel by default serializes the BCL on previous BCLs submitted on this FD, but not RCLs. For now this fix is conservative and blocks on last RCL if any vertex texturing is done, which fails to get bin/render overlap if there was an intermediate job that doesn't draw to the BCL's buffer. I've dropped a perf_debug() in here to note that as a potential future improvement. Fixes intermittent failures in KHR-GLES3.copy_tex_image_conversions.required.*
This commit is contained in:
parent
34cefa7fe0
commit
0a3f653180
1 changed files with 14 additions and 0 deletions
|
|
@ -425,6 +425,20 @@ v3d_draw_vbo(struct pipe_context *pctx, const struct pipe_draw_info *info)
|
|||
|
||||
struct v3d_job *job = v3d_get_job_for_fbo(v3d);
|
||||
|
||||
/* If vertex texturing depends on the output of rendering, we need to
|
||||
* ensure that that rendering is complete before we run a coordinate
|
||||
* shader that depends on it.
|
||||
*
|
||||
* Given that doing that is unusual, for now we just block the binner
|
||||
* on the last submitted render, rather than tracking the last
|
||||
* rendering to each texture's BO.
|
||||
*/
|
||||
if (v3d->verttex.num_textures) {
|
||||
perf_debug("Blocking binner on last render "
|
||||
"due to vertex texturing.\n");
|
||||
job->submit.in_sync_bcl = v3d->out_sync;
|
||||
}
|
||||
|
||||
/* Get space to emit our draw call into the BCL, using a branch to
|
||||
* jump to a new BO if necessary.
|
||||
*/
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue