drm/amdgpu: split nbio callbacks into ras and non-ras ones

nbio ras is not managed by gpu driver when gpu is
connected to cpu through xgmi. split nbio callbacks
into ras and non-ras ones so gpu driver only
initializes nbio ras callbacks when it manages
nbio ras.
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Dennis Li <Dennis.Li@amd.com>
Reviewed-by: John Clements <John.Clements@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -199,13 +199,13 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg)
 	 * ack the interrupt if it is there
 	 */
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF)) {
-		if (adev->nbio.funcs &&
-		    adev->nbio.funcs->handle_ras_controller_intr_no_bifring)
-			adev->nbio.funcs->handle_ras_controller_intr_no_bifring(adev);
+		if (adev->nbio.ras_funcs &&
+		    adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring)
+			adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring(adev);
 
-		if (adev->nbio.funcs &&
-		    adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring)
-			adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring(adev);
+		if (adev->nbio.ras_funcs &&
+		    adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring)
+			adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring(adev);
 	}
 
 	return ret;

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

@@ -47,6 +47,17 @@ struct nbio_hdp_flush_reg {
 	u32 ref_and_mask_sdma7;
 };
 
+struct amdgpu_nbio_ras_funcs {
+	void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev);
+	void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev);
+	int (*init_ras_controller_interrupt)(struct amdgpu_device *adev);
+	int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev);
+	void (*query_ras_error_count)(struct amdgpu_device *adev,
+				      void *ras_error_status);
+	int (*ras_late_init)(struct amdgpu_device *adev);
+	void (*ras_fini)(struct amdgpu_device *adev);
+};
+
 struct amdgpu_nbio_funcs {
 	const struct nbio_hdp_flush_reg *hdp_flush_reg;
 	u32 (*get_hdp_flush_req_offset)(struct amdgpu_device *adev);
@@ -79,13 +90,6 @@ struct amdgpu_nbio_funcs {
 	void (*ih_control)(struct amdgpu_device *adev);
 	void (*init_registers)(struct amdgpu_device *adev);
 	void (*remap_hdp_registers)(struct amdgpu_device *adev);
-	void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev);
-	void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev);
-	int (*init_ras_controller_interrupt)(struct amdgpu_device *adev);
-	int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev);
-	void (*query_ras_error_count)(struct amdgpu_device *adev,
-					void *ras_error_status);
-	int (*ras_late_init)(struct amdgpu_device *adev);
 	void (*enable_aspm)(struct amdgpu_device *adev,
 			    bool enable);
 	void (*program_aspm)(struct amdgpu_device *adev);
@@ -97,6 +101,7 @@ struct amdgpu_nbio {
 	struct amdgpu_irq_src ras_err_event_athub_irq;
 	struct ras_common_if *ras_if;
 	const struct amdgpu_nbio_funcs *funcs;
+	const struct amdgpu_nbio_ras_funcs *ras_funcs;
 };
 
 int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev);

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

@@ -804,8 +804,9 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev,
 			adev->mmhub.funcs->query_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__PCIE_BIF:
-		if (adev->nbio.funcs->query_ras_error_count)
-			adev->nbio.funcs->query_ras_error_count(adev, &err_data);
+		if (adev->nbio.ras_funcs &&
+		    adev->nbio.ras_funcs->query_ras_error_count)
+			adev->nbio.ras_funcs->query_ras_error_count(adev, &err_data);
 		break;
 	case AMDGPU_RAS_BLOCK__XGMI_WAFL:
 		amdgpu_xgmi_query_ras_error_count(adev, &err_data);
@@ -2030,14 +2031,31 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 	/* Might need get this flag from vbios. */
 	con->flags = RAS_DEFAULT_FLAGS;
 
-	if (adev->nbio.funcs->init_ras_controller_interrupt) {
-		r = adev->nbio.funcs->init_ras_controller_interrupt(adev);
+	/* initialize nbio ras function ahead of any other
+	 * ras functions so hardware fatal error interrupt
+	 * can be enabled as early as possible */
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+	case CHIP_ARCTURUS:
+	case CHIP_ALDEBARAN:
+		if (!adev->gmc.xgmi.connected_to_cpu)
+			adev->nbio.ras_funcs = &nbio_v7_4_ras_funcs;
+		break;
+	default:
+		/* nbio ras is not available */
+		break;
+	}
+
+	if (adev->nbio.ras_funcs &&
+	    adev->nbio.ras_funcs->init_ras_controller_interrupt) {
+		r = adev->nbio.ras_funcs->init_ras_controller_interrupt(adev);
 		if (r)
 			goto release_con;
 	}
 
-	if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) {
-		r = adev->nbio.funcs->init_ras_err_event_athub_interrupt(adev);
+	if (adev->nbio.ras_funcs &&
+	    adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) {
+		r = adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt(adev);
 		if (r)
 			goto release_con;
 	}

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

@@ -557,6 +557,16 @@ static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev,
 		       DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1);
 }
 
+const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs = {
+	.handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring,
+	.handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring,
+	.init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt,
+	.init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt,
+	.query_ras_error_count = nbio_v7_4_query_ras_error_count,
+	.ras_late_init = amdgpu_nbio_ras_late_init,
+	.ras_fini = amdgpu_nbio_ras_fini,
+};
+
 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,
@@ -577,10 +587,4 @@ const struct amdgpu_nbio_funcs nbio_v7_4_funcs = {
 	.ih_control = nbio_v7_4_ih_control,
 	.init_registers = nbio_v7_4_init_registers,
 	.remap_hdp_registers = nbio_v7_4_remap_hdp_registers,
-	.handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring,
-	.handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring,
-	.init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt,
-	.init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt,
-	.query_ras_error_count = nbio_v7_4_query_ras_error_count,
-	.ras_late_init = amdgpu_nbio_ras_late_init,
 };

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

@@ -28,5 +28,6 @@
 
 extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg;
 extern const struct amdgpu_nbio_funcs nbio_v7_4_funcs;
+extern const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs;
 
 #endif

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

@@ -1523,8 +1523,9 @@ static int soc15_common_late_init(void *handle)
 	if (adev->hdp.funcs->reset_ras_error_count)
 		adev->hdp.funcs->reset_ras_error_count(adev);
 
-	if (adev->nbio.funcs->ras_late_init)
-		r = adev->nbio.funcs->ras_late_init(adev);
+	if (adev->nbio.ras_funcs &&
+	    adev->nbio.ras_funcs->ras_late_init)
+		r = adev->nbio.ras_funcs->ras_late_init(adev);
 
 	return r;
 }
@@ -1545,7 +1546,9 @@ static int soc15_common_sw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	amdgpu_nbio_ras_fini(adev);
+	if (adev->nbio.ras_funcs &&
+	    adev->nbio.ras_funcs->ras_fini)
+		adev->nbio.ras_funcs->ras_fini(adev);
 	adev->df.funcs->sw_fini(adev);
 	return 0;
 }
@@ -1609,9 +1612,11 @@ static int soc15_common_hw_fini(void *handle)
 
 	if (adev->nbio.ras_if &&
 	    amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) {
-		if (adev->nbio.funcs->init_ras_controller_interrupt)
+		if (adev->nbio.ras_funcs &&
+		    adev->nbio.ras_funcs->init_ras_controller_interrupt)
 			amdgpu_irq_put(adev, &adev->nbio.ras_controller_irq, 0);
-		if (adev->nbio.funcs->init_ras_err_event_athub_interrupt)
+		if (adev->nbio.ras_funcs &&
+		    adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt)
 			amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0);
 	}