drm/amdgpu: store noretry parameter per driver instance

This will allow us to have different defaults per asic
in a future patch.
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Luben Tuikov <luben.tuikov@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -581,6 +581,13 @@ uint32_t amdgpu_amdkfd_get_asic_rev_id(struct kgd_dev *kgd)
 	return adev->rev_id;
 }
 
+int amdgpu_amdkfd_get_noretry(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	return adev->gmc.noretry;
+}
+
 int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
 				uint32_t vmid, uint64_t gpu_addr,
 				uint32_t *ib_cmd, uint32_t ib_len)

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

@@ -181,6 +181,7 @@ uint64_t amdgpu_amdkfd_get_unique_id(struct kgd_dev *kgd);
 uint64_t amdgpu_amdkfd_get_mmio_remap_phys_addr(struct kgd_dev *kgd);
 uint32_t amdgpu_amdkfd_get_num_gws(struct kgd_dev *kgd);
 uint32_t amdgpu_amdkfd_get_asic_rev_id(struct kgd_dev *kgd);
+int amdgpu_amdkfd_get_noretry(struct kgd_dev *kgd);
 uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct kgd_dev *dst, struct kgd_dev *src);
 
 /* Read user wptr from a specified user address space with page fault

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

@@ -1265,6 +1265,8 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
 		dev_warn(adev->dev, "set kernel compute queue number to 8 due to invalid parameter provided by user\n");
 	}
 
+	amdgpu_gmc_noretry_set(adev);
+
 	return 0;
 }
 

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

@@ -413,6 +413,20 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev)
 	}
 }
 
+/**
+ * amdgpu_noretry_set -- set per asic noretry defaults
+ * @adev: amdgpu_device pointer
+ *
+ * Set a per asic default for the no-retry parameter.
+ *
+ */
+void amdgpu_gmc_noretry_set(struct amdgpu_device *adev)
+{
+	struct amdgpu_gmc *gmc = &adev->gmc;
+
+	gmc->noretry = amdgpu_noretry;
+}
+
 void amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type,
 				   bool enable)
 {

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

@@ -239,6 +239,7 @@ struct amdgpu_gmc {
 
 	struct amdgpu_xgmi xgmi;
 	struct amdgpu_irq_src	ecc_irq;
+	int noretry;
 };
 
 #define amdgpu_gmc_flush_gpu_tlb(adev, vmid, vmhub, type) ((adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid), (vmhub), (type)))
@@ -300,6 +301,7 @@ void amdgpu_gmc_ras_fini(struct amdgpu_device *adev);
 int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev);
 
 extern void amdgpu_gmc_tmz_set(struct amdgpu_device *adev);
+extern void amdgpu_gmc_noretry_set(struct amdgpu_device *adev);
 
 extern void
 amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type,

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

@@ -177,7 +177,7 @@ int amdgpu_driver_load_kms(struct amdgpu_device *adev, unsigned long flags)
 			break;
 		case CHIP_VEGA10:
 			/* turn runpm on if noretry=0 */
-			if (!amdgpu_noretry)
+			if (!adev->gmc.noretry)
 				adev->runpm = true;
 			break;
 		default:

drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c

@@ -2555,14 +2555,14 @@ static void gfx_v9_0_constants_init(struct amdgpu_device *adev)
 			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
 					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
 			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
-					    !!amdgpu_noretry);
+					    !!adev->gmc.noretry);
 			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp);
 			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_BASES, 0);
 		} else {
 			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
 					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
 			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_DISABLE,
-					    !!amdgpu_noretry);
+					    !!adev->gmc.noretry);
 			WREG32_SOC15_RLC(GC, 0, mmSH_MEM_CONFIG, tmp);
 			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
 				(adev->gmc.private_aperture_start >> 48));

drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c

@@ -245,7 +245,7 @@ static void gfxhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL,
 				    i * hub->ctx_distance, tmp);
 		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,

drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c

@@ -313,7 +313,7 @@ static void gfxhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL,
 				    i * hub->ctx_distance, tmp);
 		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,

drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c

@@ -319,7 +319,7 @@ static void gfxhub_v2_1_setup_vmid_config(struct amdgpu_device *adev)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL,
 				    i * hub->ctx_distance, tmp);
 		WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,

drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c

@@ -268,7 +268,7 @@ static void mmhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL,
 				    i * hub->ctx_distance, tmp);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,

drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c

@@ -369,7 +369,7 @@ static void mmhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL,
 				    i * hub->ctx_distance, tmp);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32,

drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c

@@ -330,7 +330,7 @@ static void mmhub_v9_4_setup_vmid_config(struct amdgpu_device *adev, int hubid)
 		/* Send no-retry XNACK on fault to suppress VM fault storm. */
 		tmp = REG_SET_FIELD(tmp, VML2VC0_VM_CONTEXT1_CNTL,
 				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT,
-				    !amdgpu_noretry);
+				    !adev->gmc.noretry);
 		WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2VC0_VM_CONTEXT1_CNTL,
 				    hubid * MMHUB_INSTANCE_REGISTER_OFFSET +
 				    i * hub->ctx_distance, tmp);

drivers/gpu/drm/amd/amdkfd/kfd_device.c

@@ -718,6 +718,8 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 
 	kfd->unique_id = amdgpu_amdkfd_get_unique_id(kfd->kgd);
 
+	kfd->noretry = amdgpu_amdkfd_get_noretry(kfd->kgd);
+
 	if (kfd_interrupt_init(kfd)) {
 		dev_err(kfd_device, "Error initializing interrupts\n");
 		goto kfd_interrupt_error;

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c

@@ -61,7 +61,7 @@ static int update_qpd_v9(struct device_queue_manager *dqm,
 		qpd->sh_mem_config =
 				SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
 					SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
-		if (amdgpu_noretry &&
+		if (dqm->dev->noretry &&
 		    !dqm->dev->use_iommu_v2)
 			qpd->sh_mem_config |=
 				1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT;

drivers/gpu/drm/amd/amdkfd/kfd_priv.h

@@ -317,6 +317,8 @@ struct kfd_dev {
 
 	struct ida doorbell_ida;
 	unsigned int max_doorbell_slices;
+
+	int noretry;
 };
 
 enum kfd_mempool {