panvk: implement check_status on v10+

vk_common_*Wait* calls vk_device_check_status to detect device lost.
This allows device lost to be reported more timely.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32166>

src/panfrost/vulkan/csf/panvk_queue.h

@@ -69,4 +69,6 @@ VkResult panvk_per_arch(queue_init)(struct panvk_device *device,
                                     struct panvk_queue *queue, int idx,
                                     const VkDeviceQueueCreateInfo *create_info);
 
+VkResult panvk_per_arch(queue_check_status)(struct panvk_queue *queue);
+
 #endif

src/panfrost/vulkan/csf/panvk_vX_device.c

@@ -0,0 +1,29 @@
+/*
+ * Copyright © 2024 Collabora Ltd.
+ * SPDX-License-Identifier: MIT
+ */
+
+#include "panvk_device.h"
+#include "panvk_queue.h"
+
+VkResult
+panvk_per_arch(device_check_status)(struct vk_device *vk_dev)
+{
+   struct panvk_device *dev = to_panvk_device(vk_dev);
+   VkResult result = VK_SUCCESS;
+
+   for (uint32_t qfi = 0; qfi < PANVK_MAX_QUEUE_FAMILIES; qfi++) {
+      for (uint32_t q = 0; q < dev->queue_count[qfi]; q++) {
+         struct panvk_queue *queue = &dev->queues[qfi][q];
+         if (panvk_per_arch(queue_check_status)(queue) != VK_SUCCESS)
+            result = VK_ERROR_DEVICE_LOST;
+      }
+   }
+
+   if (pan_kmod_vm_query_state(dev->kmod.vm) != PAN_KMOD_VM_USABLE) {
+      vk_device_set_lost(&dev->vk, "vm state: not usable");
+      result = VK_ERROR_DEVICE_LOST;
+   }
+
+   return result;
+}

src/panfrost/vulkan/csf/panvk_vX_queue.c

@@ -779,3 +779,23 @@ panvk_per_arch(queue_finish)(struct panvk_queue *queue)
    drmSyncobjDestroy(dev->vk.drm_fd, queue->syncobj_handle);
    vk_queue_finish(&queue->vk);
 }
+
+VkResult
+panvk_per_arch(queue_check_status)(struct panvk_queue *queue)
+{
+   struct panvk_device *dev = to_panvk_device(queue->vk.base.device);
+   struct drm_panthor_group_get_state state = {
+      .group_handle = queue->group_handle,
+   };
+
+   int ret =
+      drmIoctl(dev->vk.drm_fd, DRM_IOCTL_PANTHOR_GROUP_GET_STATE, &state);
+   if (!ret && !state.state)
+      return VK_SUCCESS;
+
+   vk_queue_set_lost(&queue->vk,
+                     "group state: err=%d, state=0x%x, fatal_queues=0x%x", ret,
+                     state.state, state.fatal_queues);
+
+   return VK_ERROR_DEVICE_LOST;
+}

src/panfrost/vulkan/meson.build

@@ -65,6 +65,7 @@ csf_files = [
   'csf/panvk_vX_cmd_draw.c',
   'csf/panvk_vX_cmd_event.c',
   'csf/panvk_vX_cmd_query.c',
+  'csf/panvk_vX_device.c',
   'csf/panvk_vX_event.c',
   'csf/panvk_vX_queue.c',
 ]

src/panfrost/vulkan/panvk_device.h

@@ -90,6 +90,11 @@ panvk_per_arch(create_device)(struct panvk_physical_device *physical_device,
 
 void panvk_per_arch(destroy_device)(struct panvk_device *device,
                                     const VkAllocationCallbacks *pAllocator);
+
+#if PAN_ARCH >= 10
+VkResult panvk_per_arch(device_check_status)(struct vk_device *vk_dev);
+#endif
+
 #endif
 
 #endif

src/panfrost/vulkan/panvk_vX_device.c

@@ -250,6 +250,9 @@ panvk_per_arch(create_device)(struct panvk_physical_device *physical_device,
    device->vk.command_dispatch_table = &device->cmd_dispatch;
    device->vk.command_buffer_ops = &panvk_per_arch(cmd_buffer_ops);
    device->vk.shader_ops = &panvk_per_arch(device_shader_ops);
+#if PAN_ARCH >= 10
+   device->vk.check_status = panvk_per_arch(device_check_status);
+#endif
 
    device->kmod.allocator = (struct pan_kmod_allocator){
       .zalloc = panvk_kmod_zalloc,