drm/amdgpu/soc15: disable doorbell interrupt as part of BACO entry sequence

Workaround to make RAS recovery work in BACO reset.
Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h

@@ -67,6 +67,8 @@ struct amdgpu_nbio_funcs {
 						  bool enable);
 	void (*ih_doorbell_range)(struct amdgpu_device *adev,
 				  bool use_doorbell, int doorbell_index);
+	void (*enable_doorbell_interrupt)(struct amdgpu_device *adev,
+					  bool enable);
 	void (*update_medium_grain_clock_gating)(struct amdgpu_device *adev,
 						 bool enable);
 	void (*update_medium_grain_light_sleep)(struct amdgpu_device *adev,

drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c

@@ -502,6 +502,13 @@ static void nbio_v7_4_query_ras_error_count(struct amdgpu_device *adev,
 	}
 }
 
+static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD15(NBIO, 0, BIF_DOORBELL_INT_CNTL,
+		       DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1);
+}
+
 const struct amdgpu_nbio_funcs nbio_v7_4_funcs = {
 	.get_hdp_flush_req_offset = nbio_v7_4_get_hdp_flush_req_offset,
 	.get_hdp_flush_done_offset = nbio_v7_4_get_hdp_flush_done_offset,
@@ -516,6 +523,7 @@ const struct amdgpu_nbio_funcs nbio_v7_4_funcs = {
 	.enable_doorbell_aperture = nbio_v7_4_enable_doorbell_aperture,
 	.enable_doorbell_selfring_aperture = nbio_v7_4_enable_doorbell_selfring_aperture,
 	.ih_doorbell_range = nbio_v7_4_ih_doorbell_range,
+	.enable_doorbell_interrupt = nbio_v7_4_enable_doorbell_interrupt,
 	.update_medium_grain_clock_gating = nbio_v7_4_update_medium_grain_clock_gating,
 	.update_medium_grain_light_sleep = nbio_v7_4_update_medium_grain_light_sleep,
 	.get_clockgating_state = nbio_v7_4_get_clockgating_state,

drivers/gpu/drm/amd/amdgpu/soc15.c

@@ -493,10 +493,15 @@ static int soc15_asic_baco_reset(struct amdgpu_device *adev)
 {
 	void *pp_handle = adev->powerplay.pp_handle;
 	const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
+	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 
 	if (!pp_funcs ||!pp_funcs->get_asic_baco_state ||!pp_funcs->set_asic_baco_state)
 		return -ENOENT;
 
+	/* avoid NBIF got stuck when do RAS recovery in BACO reset */
+	if (ras && ras->supported)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, false);
+
 	/* enter BACO state */
 	if (pp_funcs->set_asic_baco_state(pp_handle, 1))
 		return -EIO;
@@ -505,6 +510,10 @@ static int soc15_asic_baco_reset(struct amdgpu_device *adev)
 	if (pp_funcs->set_asic_baco_state(pp_handle, 0))
 		return -EIO;
 
+	/* re-enable doorbell interrupt after BACO exit */
+	if (ras && ras->supported)
+		adev->nbio.funcs->enable_doorbell_interrupt(adev, true);
+
 	dev_info(adev->dev, "GPU BACO reset\n");
 
 	adev->in_baco_reset = 1;