[PATCH v2] drm/amdgpu doorbell range should be set when gpu recovery

Lin.Cao lincao12 at amd.com
Mon Oct 30 09:57:53 UTC 2023


GFX doorbell range should be set after flr otherwise the gfx doorbell
range will be overlap with MEC.

v2: remove "amdgpu_sriov_vf" and "amdgpu_in_reset" check, and add grbm
select for the case of 2 gfx rings.

Signed-off-by: Lin.Cao <lincao12 at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index d9ccacd06fba..c9f4e8252070 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -6467,6 +6467,13 @@ static int gfx_v10_0_gfx_init_queue(struct amdgpu_ring *ring)
 		if (adev->gfx.me.mqd_backup[mqd_idx])
 			memcpy(adev->gfx.me.mqd_backup[mqd_idx], mqd, sizeof(*mqd));
 	} else {
+		mutex_lock(&adev->srbm_mutex);
+		nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		if (ring->doorbell_index == adev->doorbell_index.gfx_ring0 << 1)
+			gfx_v10_0_cp_gfx_set_doorbell(adev, ring);
+
+		nv_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
 		/* restore mqd with the backup copy */
 		if (adev->gfx.me.mqd_backup[mqd_idx])
 			memcpy(mqd, adev->gfx.me.mqd_backup[mqd_idx], sizeof(*mqd));
-- 
2.25.1



More information about the amd-gfx mailing list