fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-05 22:38:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

anv: workaround flaky xfb query results on Gfx11

Browse source 
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29836>
This commit is contained in:
Lionel Landwerlin 2024-06-21 15:24:57 +03:00 • committed by Marge Bot
parent feaa5ce1ec
commit 884397b587
1 changed files with 20 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
src/intel/vulkan/genX_query.c
View file

@ -1317,6 +1317,26 @@ void genX(CmdEndQueryIndexedEXT)(
ANV_PIPE_CS_STALL_BIT |
ANV_PIPE_STALL_AT_SCOREBOARD_BIT);
emit_xfb_query(&b, index, anv_address_add(query_addr, 16));
#if GFX_VER == 11
/* Running the following CTS pattern on ICL will likely report a failure :
*
* dEQP-VK.transform_feedback.primitives_generated_query.get.queue_reset.32bit.geom.*
*
* If you dump the returned values in genX(GetQueryPoolResults)(), you
* will notice that the last 64bit value is 0 and rereading the value
* once more will return a non-zero value. This seems to indicate that
* the memory writes are not ordered somehow... Otherwise the
* availability write below would ensure the previous writes above have
* completed.
*
* So as a workaround, we stall CS to make sure the previous writes have
* landed before emitting the availability.
*/
genx_batch_emit_pipe_control(&cmd_buffer->batch,
cmd_buffer->device->info,
cmd_buffer->state.current_pipeline,
ANV_PIPE_CS_STALL_BIT);
#endif
emit_query_mi_availability(&b, query_addr, true);
break;