drm/amdgpu: add ras_controller and err_event_athub interrupt support

Ras controller interrupt and Ras err event athub interrupt are two dedicated
interrupts for RAS support.

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -79,10 +79,14 @@ struct amdgpu_nbio_funcs {
 	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);
 };
 
 struct amdgpu_nbio {
 	const struct nbio_hdp_flush_reg *hdp_flush_reg;
+	struct amdgpu_irq_src ras_controller_irq;
+	struct amdgpu_irq_src ras_err_event_athub_irq;
 	const struct amdgpu_nbio_funcs *funcs;
 };
 

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

@@ -29,6 +29,7 @@
 #include "amdgpu.h"
 #include "amdgpu_ras.h"
 #include "amdgpu_atomfirmware.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
 
 const char *ras_error_string[] = {
 	"none",
@@ -1500,6 +1501,7 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev,
 int amdgpu_ras_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	int r;
 
 	if (con)
 		return 0;
@@ -1527,6 +1529,18 @@ 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);
+		if (r)
+			return r;
+	}
+
+	if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) {
+		r = adev->nbio.funcs->init_ras_err_event_athub_interrupt(adev);
+		if (r)
+			return r;
+	}
+
 	if (amdgpu_ras_recovery_init(adev))
 		goto recovery_out;
 

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

@@ -27,6 +27,7 @@
 #include "nbio/nbio_7_4_offset.h"
 #include "nbio/nbio_7_4_sh_mask.h"
 #include "nbio/nbio_7_4_0_smn.h"
+#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
 #include <uapi/linux/kfd_ioctl.h>
 
 #define smnNBIF_MGCG_CTRL_LCLK	0x1013a21c
@@ -345,6 +346,128 @@ static void nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_d
 	}
 }
 
+
+static int nbio_v7_4_set_ras_controller_irq_state(struct amdgpu_device *adev,
+						  struct amdgpu_irq_src *src,
+						  unsigned type,
+						  enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_intr_cntl;
+
+	bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL);
+	if (state == AMDGPU_IRQ_STATE_ENABLE) {
+		/* set interrupt vector select bit to 0 to select
+		 * vetcor 1 for bare metal case */
+		bif_intr_cntl = REG_SET_FIELD(bif_intr_cntl,
+					      BIF_INTR_CNTL,
+					      RAS_INTR_VEC_SEL, 0);
+		WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL, bif_intr_cntl);
+	}
+
+	return 0;
+}
+
+static int nbio_v7_4_process_ras_controller_irq(struct amdgpu_device *adev,
+						struct amdgpu_irq_src *source,
+						struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for ras_controller_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static int nbio_v7_4_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev,
+						       struct amdgpu_irq_src *src,
+						       unsigned type,
+						       enum amdgpu_interrupt_state state)
+{
+	/* The ras_controller_irq enablement should be done in psp bl when it
+	 * tries to enable ras feature. Driver only need to set the correct interrupt
+	 * vector for bare-metal and sriov use case respectively
+	 */
+	uint32_t bif_intr_cntl;
+
+	bif_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL);
+	if (state == AMDGPU_IRQ_STATE_ENABLE) {
+		/* set interrupt vector select bit to 0 to select
+		 * vetcor 1 for bare metal case */
+		bif_intr_cntl = REG_SET_FIELD(bif_intr_cntl,
+					      BIF_INTR_CNTL,
+					      RAS_INTR_VEC_SEL, 0);
+		WREG32_SOC15(NBIO, 0, mmBIF_INTR_CNTL, bif_intr_cntl);
+	}
+
+	return 0;
+}
+
+static int nbio_v7_4_process_err_event_athub_irq(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *source,
+						 struct amdgpu_iv_entry *entry)
+{
+	/* By design, the ih cookie for err_event_athub_irq should be written
+	 * to BIFring instead of general iv ring. However, due to known bif ring
+	 * hw bug, it has to be disabled. There is no chance the process function
+	 * will be involked. Just left it as a dummy one.
+	 */
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs nbio_v7_4_ras_controller_irq_funcs = {
+	.set = nbio_v7_4_set_ras_controller_irq_state,
+	.process = nbio_v7_4_process_ras_controller_irq,
+};
+
+static const struct amdgpu_irq_src_funcs nbio_v7_4_ras_err_event_athub_irq_funcs = {
+	.set = nbio_v7_4_set_ras_err_event_athub_irq_state,
+	.process = nbio_v7_4_process_err_event_athub_irq,
+};
+
+static int nbio_v7_4_init_ras_controller_interrupt (struct amdgpu_device *adev)
+{
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_controller_irq.funcs =
+		&nbio_v7_4_ras_controller_irq_funcs;
+	adev->nbio.ras_controller_irq.num_types = 1;
+
+	/* register ras controller interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT,
+			      &adev->nbio.ras_controller_irq);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int nbio_v7_4_init_ras_err_event_athub_interrupt (struct amdgpu_device *adev)
+{
+
+	int r;
+
+	/* init the irq funcs */
+	adev->nbio.ras_err_event_athub_irq.funcs =
+		&nbio_v7_4_ras_err_event_athub_irq_funcs;
+	adev->nbio.ras_err_event_athub_irq.num_types = 1;
+
+	/* register ras err event athub interrupt */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF,
+			      NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT,
+			      &adev->nbio.ras_err_event_athub_irq);
+	if (r)
+		return r;
+
+	return 0;
+}
+
 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,
@@ -368,4 +491,6 @@ const struct amdgpu_nbio_funcs nbio_v7_4_funcs = {
 	.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,
 };