drm/amdgpu: return -ENODEV to user space when vram is lost v2

below ioctl will return -ENODEV:
amdgpu_cs_ioctl
amdgpu_cs_wait_ioctl
amdgpu_cs_wait_fences_ioctl
amdgpu_gem_va_ioctl
amdgpu_info_ioctl

v2: only for map and replace cases in amdgpu_gem_va_ioctl

Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -824,6 +824,7 @@ struct amdgpu_fpriv {
 	struct mutex		bo_list_lock;
 	struct idr		bo_list_handles;
 	struct amdgpu_ctx_mgr	ctx_mgr;
+	u32			vram_lost_counter;
 };
 
 /*
@@ -1528,6 +1529,7 @@ struct amdgpu_device {
 	atomic64_t			num_bytes_moved;
 	atomic64_t			num_evictions;
 	atomic_t			gpu_reset_counter;
+	atomic_t			vram_lost_counter;
 
 	/* data for buffer migration throttling */
 	struct {
@@ -1914,6 +1916,8 @@ static inline bool amdgpu_has_atpx(void) { return false; }
 extern const struct drm_ioctl_desc amdgpu_ioctls_kms[];
 extern const int amdgpu_max_kms_ioctl;
 
+bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
+			  struct amdgpu_fpriv *fpriv);
 int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags);
 void amdgpu_driver_unload_kms(struct drm_device *dev);
 void amdgpu_driver_lastclose_kms(struct drm_device *dev);

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -1097,6 +1097,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	union drm_amdgpu_cs *cs = data;
 	struct amdgpu_cs_parser parser = {};
 	bool reserved_buffers = false;
@@ -1104,6 +1105,8 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 
 	if (!adev->accel_working)
 		return -EBUSY;
+	if (amdgpu_kms_vram_lost(adev, fpriv))
+		return -ENODEV;
 
 	parser.adev = adev;
 	parser.filp = filp;
@@ -1165,12 +1168,15 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
 {
 	union drm_amdgpu_wait_cs *wait = data;
 	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
 	struct amdgpu_ring *ring = NULL;
 	struct amdgpu_ctx *ctx;
 	struct dma_fence *fence;
 	long r;
 
+	if (amdgpu_kms_vram_lost(adev, fpriv))
+		return -ENODEV;
 	r = amdgpu_cs_get_ring(adev, wait->in.ip_type, wait->in.ip_instance,
 			       wait->in.ring, &ring);
 	if (r)
@@ -1344,12 +1350,15 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *filp)
 {
 	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	union drm_amdgpu_wait_fences *wait = data;
 	uint32_t fence_count = wait->in.fence_count;
 	struct drm_amdgpu_fence *fences_user;
 	struct drm_amdgpu_fence *fences;
 	int r;
 
+	if (amdgpu_kms_vram_lost(adev, fpriv))
+		return -ENODEV;
 	/* Get the fences from userspace */
 	fences = kmalloc_array(fence_count, sizeof(struct drm_amdgpu_fence),
 			GFP_KERNEL);

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -2839,8 +2839,10 @@ retry:
 			if (r)
 				goto out;
 			vram_lost = amdgpu_check_vram_lost(adev);
-			if (vram_lost)
+			if (vram_lost) {
 				DRM_ERROR("VRAM is lost!\n");
+				atomic_inc(&adev->vram_lost_counter);
+			}
 			r = amdgpu_ttm_recover_gart(adev);
 			if (r)
 				goto out;

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c

@@ -597,6 +597,11 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 			args->operation);
 		return -EINVAL;
 	}
+	if ((args->operation == AMDGPU_VA_OP_MAP) ||
+	    (args->operation == AMDGPU_VA_OP_REPLACE)) {
+		if (amdgpu_kms_vram_lost(adev, fpriv))
+			return -ENODEV;
+	}
 
 	INIT_LIST_HEAD(&list);
 	if ((args->operation != AMDGPU_VA_OP_CLEAR) &&

drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c

@@ -235,6 +235,7 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct drm_amdgpu_info *info = data;
 	struct amdgpu_mode_info *minfo = &adev->mode_info;
 	void __user *out = (void __user *)(uintptr_t)info->return_pointer;
@@ -247,6 +248,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 
 	if (!info->return_size || !info->return_pointer)
 		return -EINVAL;
+	if (amdgpu_kms_vram_lost(adev, fpriv))
+		return -ENODEV;
 
 	switch (info->query) {
 	case AMDGPU_INFO_ACCEL_WORKING:
@@ -747,6 +750,12 @@ void amdgpu_driver_lastclose_kms(struct drm_device *dev)
 	vga_switcheroo_process_delayed_switch();
 }
 
+bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
+			  struct amdgpu_fpriv *fpriv)
+{
+	return fpriv->vram_lost_counter != atomic_read(&adev->vram_lost_counter);
+}
+
 /**
  * amdgpu_driver_open_kms - drm callback for open
  *
@@ -799,6 +808,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 
 	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr);
 
+	fpriv->vram_lost_counter = atomic_read(&adev->vram_lost_counter);
 	file_priv->driver_priv = fpriv;
 
 out_suspend: