anv: Fix context id or exec queue used to open perf stream

It was always using device->context_id what is not valid in i915 when
has_vm_control is true or when running with Xe KMD.

But anv_AcquireProfilingLockKHR() don't have the queue information so
at least for now we will only support queries in a single queue.

And for consistency doing the same in
anv_QueueSetPerformanceConfigurationINTEL() although here we have the
queue parameter but queries are only supported in render engine
so it would only expose other queues if user set some parameters.

Cc: 24.2 <mesa-stable>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30652>
(cherry picked from commit c5d79d533a)

.pick_status.json

@@ -4544,7 +4544,7 @@
         "description": "anv: Fix context id or exec queue used to open perf stream",
         "nominated": true,
         "nomination_type": 0,
-        "resolution": 0,
+        "resolution": 1,
         "main_sha": null,
         "because_sha": null,
         "notes": null

src/intel/vulkan/anv_device.c

@@ -2293,6 +2293,7 @@ anv_physical_device_init_queue_families(struct anv_physical_device *pdevice)
                           protected_flag,
             .queueCount = gc_count,
             .engine_class = INTEL_ENGINE_CLASS_RENDER,
+            .supports_perf = true,
          };
       }
       if (g_count > 0) {

src/intel/vulkan/anv_perf.c

@@ -95,13 +95,30 @@ anv_device_perf_init(struct anv_device *device)
 }
 
 static int
-anv_device_perf_open(struct anv_device *device, uint64_t metric_id)
+anv_device_perf_open(struct anv_device *device, struct anv_queue *queue, uint64_t metric_id)
 {
+   uint32_t context_or_exec_queue_id;
    uint64_t period_exponent = 31; /* slowest sampling period */
+   int ret;
 
-   return intel_perf_stream_open(device->physical->perf, device->fd,
-                                 device->context_id, metric_id,
-                                 period_exponent, true, true);
+   switch (device->physical->info.kmd_type) {
+   case INTEL_KMD_TYPE_I915:
+      context_or_exec_queue_id = device->physical->has_vm_control ?
+                                 queue->context_id : device->context_id;
+      break;
+   case INTEL_KMD_TYPE_XE:
+      context_or_exec_queue_id = queue->exec_queue_id;
+      break;
+   default:
+      unreachable("missing");
+      context_or_exec_queue_id = 0;
+   }
+
+   ret = intel_perf_stream_open(device->physical->perf, device->fd,
+                                context_or_exec_queue_id, metric_id,
+                                period_exponent, true, true);
+
+   return ret;
 }
 
 /* VK_INTEL_performance_query */
@@ -215,6 +232,20 @@ VkResult anv_ReleasePerformanceConfigurationINTEL(
    return VK_SUCCESS;
 }
 
+static struct anv_queue *
+anv_device_get_perf_queue(struct anv_device *device)
+{
+   for (uint32_t i = 0; i < device->queue_count; i++) {
+      struct anv_queue *queue = &device->queues[i];
+      const struct anv_queue_family *family = queue->family;
+
+      if (family->supports_perf)
+         return queue;
+   }
+
+   return NULL;
+}
+
 VkResult anv_QueueSetPerformanceConfigurationINTEL(
     VkQueue                                     _queue,
     VkPerformanceConfigurationINTEL             _configuration)
@@ -223,9 +254,12 @@ VkResult anv_QueueSetPerformanceConfigurationINTEL(
    ANV_FROM_HANDLE(anv_performance_configuration_intel, config, _configuration);
    struct anv_device *device = queue->device;
 
+   if (queue != anv_device_get_perf_queue(device))
+      return VK_ERROR_UNKNOWN;
+
    if (!INTEL_DEBUG(DEBUG_NO_OACONFIG)) {
       if (device->perf_fd < 0) {
-         device->perf_fd = anv_device_perf_open(device, config->config_id);
+         device->perf_fd = anv_device_perf_open(device, queue, config->config_id);
          if (device->perf_fd < 0)
             return VK_ERROR_INITIALIZATION_FAILED;
       } else {
@@ -367,7 +401,11 @@ VkResult anv_AcquireProfilingLockKHR(
    assert(device->perf_fd == -1);
 
    if (!INTEL_DEBUG(DEBUG_NO_OACONFIG)) {
-      fd = anv_device_perf_open(device, first_metric_set->oa_metrics_set_id);
+      struct anv_queue *queue = anv_device_get_perf_queue(device);
+
+      if (queue == NULL)
+         return VK_ERROR_UNKNOWN;
+      fd = anv_device_perf_open(device, queue, first_metric_set->oa_metrics_set_id);
       if (fd < 0)
          return VK_TIMEOUT;
    }

src/intel/vulkan/anv_private.h

@@ -947,6 +947,7 @@ struct anv_queue_family {
    uint32_t       queueCount;
 
    enum intel_engine_class engine_class;
+   bool supports_perf;
 };
 
 #define ANV_MAX_QUEUE_FAMILIES 5