diff --git a/src/broadcom/vulkan/v3dv_bo.c b/src/broadcom/vulkan/v3dv_bo.c
index 5a13479d004..0dc4b244d38 100644
--- a/src/broadcom/vulkan/v3dv_bo.c
+++ b/src/broadcom/vulkan/v3dv_bo.c
@@ -27,6 +27,7 @@
 #include <sys/mman.h>
 
 #include "drm-uapi/v3d_drm.h"
+#include "util/perf/cpu_trace.h"
 #include "util/u_memory.h"
 
 /* Default max size of the bo cache, in MB.
@@ -313,6 +314,7 @@ v3dv_bo_wait(struct v3dv_device *device,
              struct v3dv_bo *bo,
              uint64_t timeout_ns)
 {
+   MESA_TRACE_FUNC();
    struct drm_v3d_wait_bo wait = {
       .handle = bo->handle,
       .timeout_ns = timeout_ns,
diff --git a/src/broadcom/vulkan/v3dv_cmd_buffer.c b/src/broadcom/vulkan/v3dv_cmd_buffer.c
index 19f7a22d2b1..33bed9065ad 100644
--- a/src/broadcom/vulkan/v3dv_cmd_buffer.c
+++ b/src/broadcom/vulkan/v3dv_cmd_buffer.c
@@ -23,6 +23,7 @@
 
 #include "broadcom/common/v3d_csd.h"
 #include "v3dv_private.h"
+#include "util/perf/cpu_trace.h"
 #include "util/u_pack_color.h"
 #include "vk_common_entrypoints.h"
 #include "vk_util.h"
@@ -785,6 +786,7 @@ v3dv_job_init(struct v3dv_job *job,
               struct v3dv_cmd_buffer *cmd_buffer,
               int32_t subpass_idx)
 {
+   MESA_TRACE_FUNC();
    assert(job);
 
    /* Make sure we haven't made this new job current before calling here */
@@ -1109,6 +1111,7 @@ VKAPI_ATTR VkResult VKAPI_CALL
 v3dv_BeginCommandBuffer(VkCommandBuffer commandBuffer,
                         const VkCommandBufferBeginInfo *pBeginInfo)
 {
+   MESA_TRACE_FUNC();
    V3DV_FROM_HANDLE(v3dv_cmd_buffer, cmd_buffer, commandBuffer);
 
    /* If this is the first vkBeginCommandBuffer, we must initialize the
@@ -1885,6 +1888,7 @@ v3dv_CmdEndRenderPass2(VkCommandBuffer commandBuffer,
 VKAPI_ATTR VkResult VKAPI_CALL
 v3dv_EndCommandBuffer(VkCommandBuffer commandBuffer)
 {
+   MESA_TRACE_FUNC();
    V3DV_FROM_HANDLE(v3dv_cmd_buffer, cmd_buffer, commandBuffer);
 
    if (cmd_buffer->state.oom)
diff --git a/src/broadcom/vulkan/v3dv_pipeline.c b/src/broadcom/vulkan/v3dv_pipeline.c
index 02d5aa53ca2..0761bc22f3a 100644
--- a/src/broadcom/vulkan/v3dv_pipeline.c
+++ b/src/broadcom/vulkan/v3dv_pipeline.c
@@ -33,6 +33,7 @@
 
 #include "util/u_atomic.h"
 #include "util/os_time.h"
+#include "util/perf/cpu_trace.h"
 
 #include "vk_format.h"
 #include "vk_nir_convert_ycbcr.h"
@@ -3027,6 +3028,7 @@ v3dv_CreateGraphicsPipelines(VkDevice _device,
                              const VkAllocationCallbacks *pAllocator,
                              VkPipeline *pPipelines)
 {
+   MESA_TRACE_FUNC();
    V3DV_FROM_HANDLE(v3dv_device, device, _device);
    VkResult result = VK_SUCCESS;
 
@@ -3291,6 +3293,7 @@ v3dv_CreateComputePipelines(VkDevice _device,
                             const VkAllocationCallbacks *pAllocator,
                             VkPipeline *pPipelines)
 {
+   MESA_TRACE_FUNC();
    V3DV_FROM_HANDLE(v3dv_device, device, _device);
    VkResult result = VK_SUCCESS;
 
diff --git a/src/broadcom/vulkan/v3dv_queue.c b/src/broadcom/vulkan/v3dv_queue.c
index a4e635402f2..796c2041405 100644
--- a/src/broadcom/vulkan/v3dv_queue.c
+++ b/src/broadcom/vulkan/v3dv_queue.c
@@ -27,6 +27,7 @@
 #include "broadcom/clif/clif_dump.h"
 #include "util/libsync.h"
 #include "util/os_time.h"
+#include "util/perf/cpu_trace.h"
 #include "vk_drm_syncobj.h"
 
 #include <errno.h>
@@ -298,6 +299,7 @@ handle_reset_query_cpu_job(struct v3dv_queue *queue,
                            struct v3dv_submit_sync_info *sync_info,
                            bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
    struct v3dv_reset_query_cpu_job_info *info = &job->cpu.query_reset;
    assert(info->pool);
@@ -493,6 +495,7 @@ export_perfmon_last_job_sync(struct v3dv_queue *queue, struct v3dv_job *job, int
 static VkResult
 handle_end_query_cpu_job(struct v3dv_job *job, uint32_t counter_pass_idx)
 {
+   MESA_TRACE_FUNC();
    VkResult result = VK_SUCCESS;
 
    mtx_lock(&job->device->query_mutex);
@@ -549,6 +552,7 @@ handle_copy_query_results_cpu_job(struct v3dv_queue *queue,
                                   struct v3dv_submit_sync_info *sync_info,
                                   bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
    struct v3dv_copy_query_results_cpu_job_info *info =
       &job->cpu.query_copy_results;
@@ -713,6 +717,7 @@ handle_timestamp_query_cpu_job(struct v3dv_queue *queue,
                                struct v3dv_submit_sync_info *sync_info,
                                bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
 
    assert(job->type == V3DV_JOB_TYPE_CPU_TIMESTAMP_QUERY);
@@ -817,6 +822,7 @@ handle_csd_indirect_cpu_job(struct v3dv_queue *queue,
                             struct v3dv_submit_sync_info *sync_info,
                             bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
 
    assert(job->type == V3DV_JOB_TYPE_CPU_CSD_INDIRECT);
@@ -919,6 +925,7 @@ handle_cl_job(struct v3dv_queue *queue,
               struct v3dv_submit_sync_info *sync_info,
               bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
 
    struct drm_v3d_submit_cl submit = { 0 };
@@ -1056,6 +1063,7 @@ handle_tfu_job(struct v3dv_queue *queue,
                struct v3dv_submit_sync_info *sync_info,
                bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    assert(!V3D_DBG(DISABLE_TFU));
 
    struct v3dv_device *device = queue->device;
@@ -1095,6 +1103,7 @@ handle_csd_job(struct v3dv_queue *queue,
                struct v3dv_submit_sync_info *sync_info,
                bool signal_syncs)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_device *device = queue->device;
 
    struct drm_v3d_submit_csd *submit = &job->csd.submit;
@@ -1234,6 +1243,7 @@ VkResult
 v3dv_queue_driver_submit(struct vk_queue *vk_queue,
                          struct vk_queue_submit *submit)
 {
+   MESA_TRACE_FUNC();
    struct v3dv_queue *queue = container_of(vk_queue, struct v3dv_queue, vk);
    VkResult result;