[PATCH 1/3] drm/amdgpu: Add debug messages for MES queue reset

[Amber Lin <amber.lin@amd.com>]

This patch adds debug messages to collect some MES queue reset data.

Signed-off-by: Amber Lin <amber.lin@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 12 ++++++++++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c |  1 +
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 0ae6e990c310..0c1c8ae822d4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -2495,15 +2495,21 @@ int amdgpu_gfx_reset_mes_compute(struct amdgpu_device *adev,
 	 */
 	r = amdgpu_mes_suspend(adev, 0);
 	/* if suspend all success, it should no hang queue */
-	if (!r)
+	if (!r) {
+		dev_dbg(adev->dev, "Suspend all succeeds. xcc_id %d\n", xcc_id);
 		/* always reset the KCQ/userq since we need to signal the fence
 		 * and we could be stuck in a loop which is preemptable.
 		 */
 		goto fence_reset;
+	}
 	r = amdgpu_mes_detect_and_reset_hung_queues(adev, AMDGPU_RING_TYPE_COMPUTE,
 						    true, &num_hung, adev->gfx.mec.mes_hung_db_array, 0);
-	if (r)
+	if (r) {
+		dev_dbg(adev->dev, "MES fails on detecting hung queues\n");
 		goto out;
+	}
+	dev_dbg(adev->dev, "Detected %d hung queues\n", num_hung);
+
 	if (hung_queue_count)
 		*hung_queue_count = num_hung;
 
@@ -2530,6 +2536,8 @@ int amdgpu_gfx_reset_mes_compute(struct amdgpu_device *adev,
 		pipe = hqd_info[i].pipe_index;
 		queue = hqd_info[i].queue_index;
 		queue_type = hqd_info[i].queue_type;
+		dev_dbg(adev->dev, "Hung queue: queue_id %d at pipe %d type %d\n",
+					queue, pipe, queue_type);
 
 		/* reset any KCQs */
 		r = amdgpu_gfx_reset_mes_kcq(adev, ring,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
index 2ff1816adfc4..58766b27e150 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -557,6 +557,7 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
 	for (i = 0; i < adev->mes.hung_queue_hqd_info_offset; i++) {
 		/* Finding hung queues where db_array[i] is a valid doorbell */
 		if (db_array[i] != AMDGPU_MES_INVALID_DB_OFFSET) {
+			dev_dbg(adev->dev, "MES: hung queue doorbell 0x%x\n", db_array[i]);
 			hung_db_array[i] = db_array[i];
 			*hung_db_num += 1;
 		}
-- 
2.43.0