fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-07 11:28:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

zink: when sparse unbinding, always wait on main timeline semaphore

Browse source 
this ensures synchronization between gfx/compute work and the sparse queue

fixes KHR-GL46.sparse_buffer_tests.BufferStorageTest_cases_*. on turnip

cc: mesa-stable

Closes: #13747
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36954>
(cherry picked from commit 39ffc3f629)
This commit is contained in:
Mike Blumenkrantz 2025-08-23 07:32:56 -04:00 • committed by Eric Engestrom
parent 6bf190f73f
commit e12c95c4bf
2 changed files with 41 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.pick_status.json
View file

@ -4674,7 +4674,7 @@
"description": "zink: when sparse unbinding, always wait on main timeline semaphore",
"nominated": true,
"nomination_type": 1,
"resolution": 0,
"resolution": 1,
"main_sha": null,
"because_sha": null,
"notes": null

40
src/gallium/drivers/zink/zink_bo.c
View file

@ -761,13 +761,49 @@ track_freed_sparse_bo(struct zink_context *ctx, struct zink_sparse_backing *back
util_dynarray_append(&ctx->bs->freed_sparse_backing_bos, struct zink_bo*, backing->bo);
}
static VkTimelineSemaphoreSubmitInfo
init_timeline_wait(struct zink_context *ctx, struct zink_resource *res, bool commit, VkSemaphore *wait)
{
struct zink_screen *screen = zink_screen(ctx->base.screen);
VkTimelineSemaphoreSubmitInfo timeline = {
VK_STRUCTURE_TYPE_TIMELINE_SEMAPHORE_SUBMIT_INFO,
NULL,
1,
&screen->curr_batch
};
/* nothing to do here */
if (*wait)
return timeline;
/* commit || no usage = no wait */
if (commit || !zink_resource_has_usage(res) || zink_resource_usage_check_completion_fast(screen, res, ZINK_RESOURCE_ACCESS_RW))
return timeline;
if (zink_resource_usage_is_unflushed(res) && !zink_resource_usage_matches(res, ctx->bs)) {
/* assuming this is a batch that is doing an async submit: unlock wait for that to finish */
simple_mtx_unlock(&screen->queue_lock);
/* note that this can deadlock if multi-context */
zink_resource_usage_unflushed_wait(ctx, res, ZINK_RESOURCE_ACCESS_RW);
/* make sure to lock again and take queue ownership */
simple_mtx_lock(&screen->queue_lock);
}
*wait = screen->sem;
timeline.pWaitSemaphoreValues = &screen->curr_batch;
return timeline;
}
static VkSemaphore
buffer_commit_single(struct zink_context *ctx, struct zink_resource *res, struct zink_bo *bo, uint32_t bo_offset, uint32_t offset, uint32_t size, bool commit, VkSemaphore wait)
{
struct zink_screen *screen = zink_screen(ctx->base.screen);
VkSemaphore sem = zink_create_semaphore(screen);
VkTimelineSemaphoreSubmitInfo timeline = init_timeline_wait(ctx, res, commit, &wait);
VkBindSparseInfo sparse = {0};
sparse.sType = VK_STRUCTURE_TYPE_BIND_SPARSE_INFO;
sparse.pNext = &timeline;
sparse.bufferBindCount = res->obj->storage_buffer ? 2 : 1;
sparse.waitSemaphoreCount = !!wait;
sparse.pWaitSemaphores = &wait;
@ -922,8 +958,10 @@ texture_commit_single(struct zink_context *ctx, struct zink_resource *res, VkSpa
{
struct zink_screen *screen = zink_screen(ctx->base.screen);
VkSemaphore sem = zink_create_semaphore(screen);
VkTimelineSemaphoreSubmitInfo timeline = init_timeline_wait(ctx, res, commit, &wait);
VkBindSparseInfo sparse = {0};
sparse.sType = VK_STRUCTURE_TYPE_BIND_SPARSE_INFO;
sparse.pNext = &timeline;
sparse.imageBindCount = 1;
sparse.waitSemaphoreCount = !!wait;
sparse.pWaitSemaphores = &wait;
@ -948,8 +986,10 @@ texture_commit_miptail(struct zink_context *ctx, struct zink_resource *res, stru
{
struct zink_screen *screen = zink_screen(ctx->base.screen);
VkSemaphore sem = zink_create_semaphore(screen);
VkTimelineSemaphoreSubmitInfo timeline = init_timeline_wait(ctx, res, commit, &wait);
VkBindSparseInfo sparse = {0};
sparse.sType = VK_STRUCTURE_TYPE_BIND_SPARSE_INFO;
sparse.pNext = &timeline;
sparse.imageOpaqueBindCount = 1;
sparse.waitSemaphoreCount = !!wait;
sparse.pWaitSemaphores = &wait;