@@ -1882,6 +1882,7 @@ struct drm_ioctl_desc i915_ioctls[] = {
DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_CREATE, i915_gem_context_create_ioctl, DRM_UNLOCKED),
DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_DESTROY, i915_gem_context_destroy_ioctl, DRM_UNLOCKED),
DRM_IOCTL_DEF_DRV(I915_REG_READ, i915_reg_read_ioctl, DRM_UNLOCKED),
+ DRM_IOCTL_DEF_DRV(I915_GET_RESET_STATS, i915_get_reset_stats_ioctl, DRM_UNLOCKED),
};
int i915_max_ioctl = DRM_ARRAY_SIZE(i915_ioctls);
@@ -823,6 +823,9 @@ int i915_reset(struct drm_device *dev)
i915_gem_reset(dev);
+ /* Count unsuccessful ones */
+ dev_priv->reset_count++;
+
ret = -ENODEV;
if (get_seconds() - dev_priv->gpu_error.last_reset < 5)
DRM_ERROR("GPU hanging too fast, declaring wedged!\n");
@@ -1228,3 +1231,42 @@ int i915_reg_read_ioctl(struct drm_device *dev,
return 0;
}
+
+int i915_get_reset_stats_ioctl(struct drm_device *dev,
+ void *data, struct drm_file *file)
+{
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct intel_ring_buffer *ring;
+ struct drm_i915_file_private *file_priv = file->driver_priv;
+ struct drm_i915_reset_stats *args = data;
+ struct i915_reset_stats *rs = NULL;
+ int ret;
+
+ ret = mutex_lock_interruptible(&dev->struct_mutex);
+ if (ret)
+ return ret;
+
+ if (args->ctx_id == 0) {
+ rs = &file_priv->reset_stats;
+ ret = 0;
+ goto out;
+ }
+
+ ring = &dev_priv->ring[RCS];
+
+ ret = i915_gem_context_get_reset_stats(ring,
+ file,
+ args->ctx_id,
+ &rs);
+out:
+ if (rs && ret == 0) {
+ args->global_lost = dev_priv->reset_count;
+ args->total_lost = rs->total;
+ args->innocent_lost = rs->innocent;
+ args->guilty_lost = rs->guilty;
+ }
+
+ mutex_unlock(&dev->struct_mutex);
+
+ return ret ? -EINVAL : 0;
+}
@@ -1040,6 +1040,10 @@ typedef struct drm_i915_private {
/* Old dri1 support infrastructure, beware the dragons ya fools entering
* here! */
struct i915_dri1_state dri1;
+
+ /* get_reset_stats ioctl */
+ u32 reset_count;
+
} drm_i915_private_t;
/* Iterate over initialised rings */
@@ -1832,6 +1836,8 @@ extern int intel_enable_rc6(const struct drm_device *dev);
extern bool i915_semaphore_is_enabled(struct drm_device *dev);
int i915_reg_read_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
+int i915_get_reset_stats_ioctl(struct drm_device *dev, void *data,
+ struct drm_file *file);
/* overlay */
#ifdef CONFIG_DEBUG_FS
@@ -198,6 +198,7 @@ typedef struct _drm_i915_sarea {
#define DRM_I915_GEM_SET_CACHING 0x2f
#define DRM_I915_GEM_GET_CACHING 0x30
#define DRM_I915_REG_READ 0x31
+#define DRM_I915_GET_RESET_STATS 0x32
#define DRM_IOCTL_I915_INIT DRM_IOW( DRM_COMMAND_BASE + DRM_I915_INIT, drm_i915_init_t)
#define DRM_IOCTL_I915_FLUSH DRM_IO ( DRM_COMMAND_BASE + DRM_I915_FLUSH)
@@ -247,6 +248,7 @@ typedef struct _drm_i915_sarea {
#define DRM_IOCTL_I915_GEM_CONTEXT_CREATE DRM_IOWR (DRM_COMMAND_BASE + DRM_I915_GEM_CONTEXT_CREATE, struct drm_i915_gem_context_create)
#define DRM_IOCTL_I915_GEM_CONTEXT_DESTROY DRM_IOW (DRM_COMMAND_BASE + DRM_I915_GEM_CONTEXT_DESTROY, struct drm_i915_gem_context_destroy)
#define DRM_IOCTL_I915_REG_READ DRM_IOWR (DRM_COMMAND_BASE + DRM_I915_REG_READ, struct drm_i915_reg_read)
+#define DRM_IOCTL_I915_GET_RESET_STATS DRM_IOWR (DRM_COMMAND_BASE + DRM_I915_GET_RESET_STATS, struct drm_i915_reset_stats)
/* Allow drivers to submit batchbuffers directly to hardware, relying
* on the security mechanisms provided by hardware.
@@ -980,4 +982,21 @@ struct drm_i915_reg_read {
__u64 offset;
__u64 val; /* Return value */
};
+
+struct drm_i915_reset_stats {
+ __u32 ctx_id;
+ __u32 flags;
+
+ /* For all contexts */
+ __u32 global_lost;
+
+ /* For this context */
+ __u32 total_lost;
+ __u32 innocent_lost;
+ __u32 guilty_lost;
+
+ /* unknown_lost ==
+ * total - (innocent + guilty) */
+};
+
#endif /* _UAPI_I915_DRM_H_ */
This ioctl returns reset status for specified context. Returned status is context loss state with regards to previous call to this ioctl. It can one off: no_error : no context lost detected guilty : context was lost and this context was one submitting a batch which hung the gpu innocent : context was lost but this context fell victim to some other context which submitted batch which hung the gpu v2: get rid of state tracking completely and deliver only counts. Idea from Chris Wilson. Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> --- drivers/gpu/drm/i915/i915_dma.c | 1 + drivers/gpu/drm/i915/i915_drv.c | 42 +++++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 6 ++++++ include/uapi/drm/i915_drm.h | 19 ++++++++++++++++++ 4 files changed, 68 insertions(+)