iris: Expose PIPE_CAP_DEVICE_RESET_STATUS_QUERY

This provides a way for the application to query whether any resets have
happened, which lets us expose "robust" contexts.  This also enables the
KHR_robust_buffer_access_behavior tests.

4 files changed, 72 insertions, 0 deletions

diff --git a/src/gallium/drivers/iris/iris_batch.c b/src/gallium/drivers/iris/iris_batch.c
index 881531b9782..bbcedf4bfc0 100644
--- a/src/gallium/drivers/iris/iris_batch.c
+++ b/src/gallium/drivers/iris/iris_batch.c
@@ -475,6 +475,40 @@ replace_hw_ctx(struct iris_batch *batch)
    return true;
 }
 
+enum pipe_reset_status
+iris_batch_check_for_reset(struct iris_batch *batch)
+{
+   struct iris_screen *screen = batch->screen;
+   enum pipe_reset_status status = PIPE_NO_RESET;
+   struct drm_i915_reset_stats stats = { .ctx_id = batch->hw_ctx_id };
+
+   if (drmIoctl(screen->fd, DRM_IOCTL_I915_GET_RESET_STATS, &stats))
+      DBG("DRM_IOCTL_I915_GET_RESET_STATS failed: %s\n", strerror(errno));
+
+   if (stats.batch_active != 0) {
+      /* A reset was observed while a batch from this hardware context was
+       * executing.  Assume that this context was at fault.
+       */
+      status = PIPE_GUILTY_CONTEXT_RESET;
+   } else if (stats.batch_pending != 0) {
+      /* A reset was observed while a batch from this context was in progress,
+       * but the batch was not executing.  In this case, assume that the
+       * context was not at fault.
+       */
+      status = PIPE_INNOCENT_CONTEXT_RESET;
+   }
+
+   if (status != PIPE_NO_RESET) {
+      /* Our context is likely banned, or at least in an unknown state.
+       * Throw it away and start with a fresh context.  Ideally this may
+       * catch the problem before our next execbuf fails with -EIO.
+       */
+      replace_hw_ctx(batch);
+   }
+
+   return status;
+}
+
 /**
  * Submit the batch to the GPU via execbuffer2.
  */
diff --git a/src/gallium/drivers/iris/iris_batch.h b/src/gallium/drivers/iris/iris_batch.h
index 747ad72d166..44e27e79623 100644
--- a/src/gallium/drivers/iris/iris_batch.h
+++ b/src/gallium/drivers/iris/iris_batch.h
@@ -150,6 +150,8 @@ bool iris_batch_references(struct iris_batch *batch, struct iris_bo *bo);
 void iris_use_pinned_bo(struct iris_batch *batch, struct iris_bo *bo,
                         bool writable);
 
+enum pipe_reset_status iris_batch_check_for_reset(struct iris_batch *batch);
+
 static inline unsigned
 iris_batch_bytes_used(struct iris_batch *batch)
 {
diff --git a/src/gallium/drivers/iris/iris_context.c b/src/gallium/drivers/iris/iris_context.c
index 5d4d27edf4e..b56e746ea76 100644
--- a/src/gallium/drivers/iris/iris_context.c
+++ b/src/gallium/drivers/iris/iris_context.c
@@ -101,6 +101,40 @@ iris_lost_context_state(struct iris_batch *batch)
    memset(ice->state.last_grid, 0, sizeof(ice->state.last_grid));
 }
 
+static enum pipe_reset_status
+iris_get_device_reset_status(struct pipe_context *ctx)
+{
+   struct iris_context *ice = (struct iris_context *)ctx;
+
+   enum pipe_reset_status worst_reset = PIPE_NO_RESET;
+
+   /* Check the reset status of each batch's hardware context, and take the
+    * worst status (if one was guilty, proclaim guilt).
+    */
+   for (int i = 0; i < IRIS_BATCH_COUNT; i++) {
+      /* This will also recreate the hardware contexts as necessary, so any
+       * future queries will show no resets.  We only want to report once.
+       */
+      enum pipe_reset_status batch_reset =
+         iris_batch_check_for_reset(&ice->batches[i]);
+
+      if (batch_reset == PIPE_NO_RESET)
+         continue;
+
+      if (worst_reset == PIPE_NO_RESET) {
+         worst_reset = batch_reset;
+      } else {
+         /* GUILTY < INNOCENT < UNKNOWN */
+         worst_reset = MIN2(worst_reset, batch_reset);
+      }
+   }
+
+   if (worst_reset != PIPE_NO_RESET && ice->reset.reset)
+      ice->reset.reset(ice->reset.data, worst_reset);
+
+   return worst_reset;
+}
+
 static void
 iris_set_device_reset_callback(struct pipe_context *ctx,
                                const struct pipe_device_reset_callback *cb)
@@ -223,6 +257,7 @@ iris_create_context(struct pipe_screen *pscreen, void *priv, unsigned flags)
    ctx->destroy = iris_destroy_context;
    ctx->set_debug_callback = iris_set_debug_callback;
    ctx->set_device_reset_callback = iris_set_device_reset_callback;
+   ctx->get_device_reset_status = iris_get_device_reset_status;
    ctx->get_sample_position = iris_get_sample_position;
 
    ice->shaders.urb_size = devinfo->urb.size;
diff --git a/src/gallium/drivers/iris/iris_screen.c b/src/gallium/drivers/iris/iris_screen.c
index b42c359d833..df0c2a81404 100644
--- a/src/gallium/drivers/iris/iris_screen.c
+++ b/src/gallium/drivers/iris/iris_screen.c
@@ -160,6 +160,7 @@ iris_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_INT64_DIVMOD:
    case PIPE_CAP_SAMPLER_VIEW_TARGET:
    case PIPE_CAP_ROBUST_BUFFER_ACCESS_BEHAVIOR:
+   case PIPE_CAP_DEVICE_RESET_STATUS_QUERY:
    case PIPE_CAP_COPY_BETWEEN_COMPRESSED_AND_PLAIN_FORMATS:
    case PIPE_CAP_FRAMEBUFFER_NO_ATTACHMENT:
    case PIPE_CAP_CULL_DISTANCE: