AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

tc: don't reuse first rp info on batch if there is work pending

Browse Source 
this otherwise causes a mismatch if drivers try to access previous info
during a set_framebuffer_state call (e.g., to flush clears)

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35777>
This commit is contained in:
Mike Blumenkrantz
2025-06-26 16:55:25 -04:00
committed by Marge Bot
parent f0c58518cc
commit 9482ad212d
1 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/gallium/auxiliary/util/u_threaded_context.c
+4 -2
View File
@@ -1452,8 +1452,10 @@ tc_set_framebuffer_state(struct pipe_context *_pipe,
{
struct threaded_context *tc = threaded_context(_pipe);
unsigned next = tc->next;
/* always skip incrementing if this is the first call on a batch */
bool set_skip_first_increment = !tc->batch_slots[tc->next].num_total_slots;
/* always skip incrementing if this is the first call on a batch AND there is no pending work from previous batch */
bool set_skip_first_increment = !tc->batch_slots[tc->next].num_total_slots &&
(!tc->renderpass_info_recording ||
!TC_RENDERPASS_INFO_HAS_WORK(tc->renderpass_info_recording->data32[0]));
struct tc_framebuffer *p =
tc_add_call_no_copy(tc, TC_CALL_set_framebuffer_state, tc_framebuffer);
unsigned nr_cbufs = fb->nr_cbufs;