radv: emit PFP_SYNC_ME right after STRMOUT_BUFFER_UPDATE is emitted

This is likely less frequent than the draws, and it's only needed
when the VA is used.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40341>
This commit is contained in:
Samuel Pitoiset 2026-03-20 10:57:37 +01:00 committed by Marge Bot
parent a0471ddad8
commit 57e2b272d5

View file

@ -15496,6 +15496,7 @@ radv_CmdEndTransformFeedbackEXT(VkCommandBuffer commandBuffer, uint32_t firstCou
const struct radv_physical_device *pdev = radv_device_physical(device);
struct radv_streamout_state *so = &cmd_buffer->state.streamout;
struct radv_cmd_stream *cs = cmd_buffer->cs;
bool needs_pfp_sync_me = false;
assert(firstCounterBuffer + counterBufferCount <= MAX_SO_BUFFERS);
@ -15532,6 +15533,8 @@ radv_CmdEndTransformFeedbackEXT(VkCommandBuffer commandBuffer, uint32_t firstCou
va += vk_buffer_address(&buffer->vk, counter_buffer_offset);
radv_cs_add_buffer(device->ws, cs->b, buffer->bo);
needs_pfp_sync_me = true;
}
if (pdev->info.gfx_level >= GFX12) {
@ -15573,6 +15576,15 @@ radv_CmdEndTransformFeedbackEXT(VkCommandBuffer commandBuffer, uint32_t firstCou
assert(cs->b->cdw <= cdw_max);
if (needs_pfp_sync_me && pdev->info.has_load_ctx_reg_pkt) {
/* Make sure that PFP waits for ME to avoid a race condition because the data is written by
* STRMOUT_BUFFER_UPDATE/COPY_DATA in ME, but LOAD_CONTEXT_REG_INDEX loads the value from
* memory between PFP and ME.
*/
radeon_check_space(device->ws, cs->b, 2);
ac_emit_cp_pfp_sync_me(cs->b, false);
}
radv_set_streamout_enable(cmd_buffer, false);
}
@ -15602,12 +15614,6 @@ radv_emit_strmout_buffer(struct radv_cmd_buffer *cmd_buffer, const struct radv_d
radeon_end();
if (pdev->info.has_load_ctx_reg_pkt) {
/* Make sure that PFP waits for ME to avoid a race condition because the data is written by
* STRMOUT_BUFFER_UPDATE in ME, but LOAD_CONTEXT_REG_INDEX loads the value from memory between
* PFP and ME.
*/
ac_emit_cp_pfp_sync_me(cs->b, false);
ac_emit_cp_load_context_reg_index(cs->b, R_028B2C_VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE, 1,
draw_info->strmout_va, false);
} else {