drm/xe: Do not wake device during a GT reset

Waking the device during a GT reset can lead to unintended memory
allocation, which is not allowed since GT resets occur in the reclaim
path. Prevent this by holding a PM reference while a reset is in flight.

Fixes: dd08ebf ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: [email protected]
Signed-off-by: Matthew Brost <[email protected]>
Reviewed-by: Matthew Auld <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
(cherry picked from commit 480b358e7d8ef69fd8f1b0cad6e07c7d70a36ee4)
Signed-off-by: Lucas De Marchi <[email protected]>

Author: mbrost05

drivers/gpu/drm/xe/xe_gt.c

@@ -813,21 +813,23 @@ static int gt_reset(struct xe_gt *gt)
 	unsigned int fw_ref;
 	int err;
 
-	if (xe_device_wedged(gt_to_xe(gt)))
-		return -ECANCELED;
+	if (xe_device_wedged(gt_to_xe(gt))) {
+		err = -ECANCELED;
+		goto err_pm_put;
+	}
 
 	/* We only support GT resets with GuC submission */
-	if (!xe_device_uc_enabled(gt_to_xe(gt)))
-		return -ENODEV;
+	if (!xe_device_uc_enabled(gt_to_xe(gt))) {
+		err = -ENODEV;
+		goto err_pm_put;
+	}
 
 	xe_gt_info(gt, "reset started\n");
 
 	err = gt_wait_reset_unblock(gt);
 	if (!err)
 		xe_gt_warn(gt, "reset block failed to get lifted");
 
-	xe_pm_runtime_get(gt_to_xe(gt));
-
 	if (xe_fault_inject_gt_reset()) {
 		err = -ECANCELED;
 		goto err_fail;
@@ -874,6 +876,7 @@ static int gt_reset(struct xe_gt *gt)
 	xe_gt_err(gt, "reset failed (%pe)\n", ERR_PTR(err));
 
 	xe_device_declare_wedged(gt_to_xe(gt));
+err_pm_put:
 	xe_pm_runtime_put(gt_to_xe(gt));
 
 	return err;
@@ -895,7 +898,9 @@ void xe_gt_reset_async(struct xe_gt *gt)
 		return;
 
 	xe_gt_info(gt, "reset queued\n");
-	queue_work(gt->ordered_wq, &gt->reset.worker);
+	xe_pm_runtime_get_noresume(gt_to_xe(gt));
+	if (!queue_work(gt->ordered_wq, &gt->reset.worker))
+		xe_pm_runtime_put(gt_to_xe(gt));
 }
 
 void xe_gt_suspend_prepare(struct xe_gt *gt)