drm/amdgpu: Enable IH retry CAM on GFX9

This patch enables the IH retry CAM on GFX9 series cards. This
retry filter is used to prevent sending lots of retry interrupts
in a short span of time and overflowing the IH ring buffer. This
will also help reduce CPU interrupt workload.

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

4 files changed, 62 insertions, 48 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index e9f2c11ea416..be243adf3e65 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -98,6 +98,8 @@ struct amdgpu_irq {
 	struct irq_domain		*domain; /* GPU irq controller domain */
 	unsigned			virq[AMDGPU_MAX_IRQ_SRC_ID];
 	uint32_t                        srbm_soft_reset;
+	u32                             retry_cam_doorbell_index;
+	bool                            retry_cam_enabled;
 };
 
 void amdgpu_irq_disable_all(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 67ad92097a65..7ec70666c648 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -555,32 +555,49 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 	const char *mmhub_cid;
 	const char *hub_name;
 	u64 addr;
+	uint32_t cam_index = 0;
+	int ret;
 
 	addr = (u64)entry->src_data[0] << 12;
 	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
 
 	if (retry_fault) {
-		/* Returning 1 here also prevents sending the IV to the KFD */
+		if (adev->irq.retry_cam_enabled) {
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
+
+			cam_index = entry->src_data[2] & 0x3ff;
 
-		/* Process it onyl if it's the first fault for this address */
-		if (entry->ih != &adev->irq.ih_soft &&
-		    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
+			ret = amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault);
+			WDOORBELL32(adev->irq.retry_cam_doorbell_index, cam_index);
+			if (ret)
+				return 1;
+		} else {
+			/* Process it onyl if it's the first fault for this address */
+			if (entry->ih != &adev->irq.ih_soft &&
+			    amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid,
 					     entry->timestamp))
-			return 1;
+				return 1;
 
-		/* Delegate it to a different ring if the hardware hasn't
-		 * already done it.
-		 */
-		if (entry->ih == &adev->irq.ih) {
-			amdgpu_irq_delegate(adev, entry, 8);
-			return 1;
-		}
+			/* Delegate it to a different ring if the hardware hasn't
+			 * already done it.
+			 */
+			if (entry->ih == &adev->irq.ih) {
+				amdgpu_irq_delegate(adev, entry, 8);
+				return 1;
+			}
 
-		/* Try to handle the recoverable page faults by filling page
-		 * tables
-		 */
-		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
-			return 1;
+			/* Try to handle the recoverable page faults by filling page
+			 * tables
+			 */
+			if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
+				return 1;
+		}
 	}
 
 	if (!printk_ratelimit())
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
index 19455a725939..685abf57ffdd 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -238,7 +238,7 @@ static void nbio_v7_4_ih_doorbell_range(struct amdgpu_device *adev,
 
 	if (use_doorbell) {
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
-		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 4);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 8);
 	} else
 		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
index 827e2768f867..536128447b71 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c
@@ -38,6 +38,11 @@
 #define mmIH_CHICKEN_ALDEBARAN			0x18d
 #define mmIH_CHICKEN_ALDEBARAN_BASE_IDX		0
 
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN		0x00ea
+#define mmIH_RETRY_INT_CAM_CNTL_ALDEBARAN_BASE_IDX	0
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE__SHIFT	0x10
+#define IH_RETRY_INT_CAM_CNTL_ALDEBARAN__ENABLE_MASK	0x00010000L
+
 static void vega20_ih_set_interrupt_funcs(struct amdgpu_device *adev);
 
 /**
@@ -251,36 +256,14 @@ static int vega20_ih_enable_ring(struct amdgpu_device *adev,
 	return 0;
 }
 
-/**
- * vega20_ih_reroute_ih - reroute VMC/UTCL2 ih to an ih ring
- *
- * @adev: amdgpu_device pointer
- *
- * Reroute VMC and UMC interrupts on primary ih ring to
- * ih ring 1 so they won't lose when bunches of page faults
- * interrupts overwhelms the interrupt handler(VEGA20)
- */
-static void vega20_ih_reroute_ih(struct amdgpu_device *adev)
+static uint32_t vega20_setup_retry_doorbell(u32 doorbell_index)
 {
-	uint32_t tmp;
+	u32 val = 0;
 
-	/* vega20 ih reroute will go through psp this
-	 * function is used for newer asics starting arcturus
-	 */
-	if (adev->ip_versions[OSSSYS_HWIP][0] >= IP_VERSION(4, 2, 1)) {
-		/* Reroute to IH ring 1 for VMC */
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x12);
-		tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, CLIENT_TYPE, 1);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA, tmp);
-
-		/* Reroute IH ring 1 for UTCL2 */
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x1B);
-		tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA);
-		tmp = REG_SET_FIELD(tmp, IH_CLIENT_CFG_DATA, RING_ID, 1);
-		WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA, tmp);
-	}
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, OFFSET, doorbell_index);
+	val = REG_SET_FIELD(val, IH_DOORBELL_RPTR, ENABLE, 1);
+
+	return val;
 }
 
 /**
@@ -333,8 +316,6 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev)
 
 	for (i = 0; i < ARRAY_SIZE(ih); i++) {
 		if (ih[i]->ring_size) {
-			if (i == 1)
-				vega20_ih_reroute_ih(adev);
 			ret = vega20_ih_enable_ring(adev, ih[i]);
 			if (ret)
 				return ret;
@@ -347,6 +328,20 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev)
 
 	pci_set_master(adev->pdev);
 
+	/* Allocate the doorbell for IH Retry CAM */
+	adev->irq.retry_cam_doorbell_index = (adev->doorbell_index.ih + 3) << 1;
+	WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RETRY_CAM,
+		vega20_setup_retry_doorbell(adev->irq.retry_cam_doorbell_index));
+
+	/* Enable IH Retry CAM */
+	if (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 0))
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL_ALDEBARAN,
+			       ENABLE, 1);
+	else
+		WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL, ENABLE, 1);
+
+	adev->irq.retry_cam_enabled = true;
+
 	/* enable interrupts */
 	ret = vega20_ih_toggle_interrupts(adev, true);
 	if (ret)