tu/kgsl: Spin unti KGSL reports queue timestamp during profiling

KGSL writes the profiling values asynchronously while we read them
immediately after the IOCTL returns which can result in the struct
not being filled in by the time we read it, this results in AGI not
correctly processing any timestamps from larger submits which take
longer to queue. To fix this, we now busy-wait on until the value
has been written out by KGSL.

Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30147>

src/freedreno/vulkan/tu_knl_kgsl.cc

@@ -1220,6 +1220,7 @@ kgsl_queue_submit(struct tu_queue *queue, struct vk_queue_submit *vk_submit)
       };
       profiling_buffer =
          (struct kgsl_cmdbatch_profiling_buffer *) tu_suballoc_bo_map(bo);
+      memset(profiling_buffer, 0, sizeof(*profiling_buffer));
    }
 
    if (tu_autotune_submit_requires_fence(cmd_buffers, cmdbuf_count)) {
@@ -1300,7 +1301,13 @@ kgsl_queue_submit(struct tu_queue *queue, struct vk_queue_submit *vk_submit)
 
    uint64_t gpu_offset = 0;
 #if HAVE_PERFETTO
-   if (profiling_buffer && profiling_buffer->gpu_ticks_queued) {
+   if (profiling_buffer) {
+      /* We need to wait for KGSL to queue the GPU command before we can read
+       * the timestamp. Since this is just for profiling and doesn't take too
+       * long, we can just busy-wait for it.
+       */
+      while (p_atomic_read(&profiling_buffer->gpu_ticks_queued) == 0);
+
       struct kgsl_perfcounter_read_group perf = {
          .groupid = KGSL_PERFCOUNTER_GROUP_ALWAYSON,
          .countable = 0,