drm/amdgpu: read back register after written for VCN v4.0.5

commit ee7360f upstream.

On VCN v4.0.5 there is a race condition where the WPTR is not
updated after starting from idle when doorbell is used. Adding
register read-back after written at function end is to ensure
all register writes are done before they can be used.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12528
Signed-off-by: David (Ming Qiang) Wu <[email protected]>
Reviewed-by: Mario Limonciello <[email protected]>
Tested-by: Mario Limonciello <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Ruijing Dong <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
(cherry picked from commit 07c9db090b86e5211188e1b351303fbc673378cf)
Cc: [email protected]
(cherry picked from commit ee7360f)
Hand modified for contextual changes where there is a for loop
in 6.12 that was dropped later on.
Signed-off-by: Mario Limonciello <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
(cherry picked from commit cef081c8231b52b4a652bbd022e03d80441118be)
Signed-off-by: Jack Vogel <[email protected]>

Author: davidwuAMD

drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c

@@ -985,6 +985,10 @@ static int vcn_v4_0_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, b
 			ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
 			VCN_RB1_DB_CTRL__EN_MASK);
 
+	/* Keeping one read-back to ensure all register writes are done, otherwise
+	 * it may introduce race conditions */
+	RREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL);
+
 	return 0;
 }
 
@@ -1167,6 +1171,10 @@ static int vcn_v4_0_5_start(struct amdgpu_device *adev)
 		tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
 		WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
 		fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+		/* Keeping one read-back to ensure all register writes are done, otherwise
+		 * it may introduce race conditions */
+		RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
 	}
 
 	return 0;