diff --git a/.pick_status.json b/.pick_status.json
index a122307bcce..25f60f3e9a6 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -4524,7 +4524,7 @@
         "description": "anv: Make sure all previous vm binds are done before execute perf query pool",
         "nominated": true,
         "nomination_type": 0,
-        "resolution": 0,
+        "resolution": 1,
         "main_sha": null,
         "because_sha": null,
         "notes": null
diff --git a/src/intel/vulkan/xe/anv_batch_chain.c b/src/intel/vulkan/xe/anv_batch_chain.c
index e2e1df92108..540dcb1fb0e 100644
--- a/src/intel/vulkan/xe/anv_batch_chain.c
+++ b/src/intel/vulkan/xe/anv_batch_chain.c
@@ -315,11 +315,14 @@ xe_queue_exec_locked(struct anv_queue *queue,
                        perf_query_pass, &exec);
 
    if (perf_query_pool && perf_query_pass >= 0 && cmd_buffer_count) {
+      struct drm_xe_sync xe_syncs[1] = {};
       struct drm_xe_exec perf_query_exec = {
             .exec_queue_id = queue->exec_queue_id,
             .num_batch_buffer = 1,
             .address = perf_query_pool->bo->offset +
                        khr_perf_query_preamble_offset(perf_query_pool, perf_query_pass),
+            .num_syncs = 1,
+            .syncs = (uintptr_t)xe_syncs,
       };
       assert(perf_query_pass < perf_query_pool->n_passes);
       struct intel_perf_query_info *query_info = perf_query_pool->pass_query[perf_query_pass];
@@ -340,6 +343,11 @@ xe_queue_exec_locked(struct anv_queue *queue,
          }
       }
 
+      xe_syncs[0].type = DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ;
+      xe_syncs[0].flags = 0;/* wait */
+      xe_syncs[0].handle = intel_bind_timeline_get_syncobj(&device->bind_timeline);
+      xe_syncs[0].timeline_value = intel_bind_timeline_get_last_point(&device->bind_timeline);
+
       if (!device->info->no_hw && result == VK_SUCCESS) {
          if (intel_ioctl(device->fd, DRM_IOCTL_XE_EXEC, &perf_query_exec))
             result = vk_device_set_lost(&device->vk, "perf_query_exec failed: %m");