drm/amdgpu: Register aqua vanjaram vcn poison irq

Register aqua vanjaram vcn poison irq, add vcn poison handle.

Signed-off-by: Stanley.Yang <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

Author: Stanley.Yang

drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c

@@ -169,6 +169,10 @@ static int vcn_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	/* VCN POISON TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+		VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst->ras_poison_irq);
+
 	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
 
 		r = amdgpu_vcn_sw_init(adev, i);
@@ -387,6 +391,9 @@ static int vcn_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block)
 			vinst->set_pg_state(vinst, AMD_PG_STATE_GATE);
 	}
 
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN))
+		amdgpu_irq_put(adev, &adev->vcn.inst->ras_poison_irq, 0);
+
 	return 0;
 }
 
@@ -1814,11 +1821,24 @@ static int vcn_v4_0_3_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static int vcn_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned int type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
 static const struct amdgpu_irq_src_funcs vcn_v4_0_3_irq_funcs = {
 	.set = vcn_v4_0_3_set_interrupt_state,
 	.process = vcn_v4_0_3_process_interrupt,
 };
 
+static const struct amdgpu_irq_src_funcs vcn_v4_0_3_ras_irq_funcs = {
+	.set = vcn_v4_0_3_set_ras_interrupt_state,
+	.process = amdgpu_vcn_process_poison_irq,
+};
+
 /**
  * vcn_v4_0_3_set_irq_funcs - set VCN block interrupt irq functions
  *
@@ -1834,6 +1854,9 @@ static void vcn_v4_0_3_set_irq_funcs(struct amdgpu_device *adev)
 		adev->vcn.inst->irq.num_types++;
 	}
 	adev->vcn.inst->irq.funcs = &vcn_v4_0_3_irq_funcs;
+
+	adev->vcn.inst->ras_poison_irq.num_types = 1;
+	adev->vcn.inst->ras_poison_irq.funcs = &vcn_v4_0_3_ras_irq_funcs;
 }
 
 static void vcn_v4_0_3_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p)
@@ -1981,9 +2004,44 @@ static void vcn_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev)
 		vcn_v4_0_3_inst_reset_ras_error_count(adev, i);
 }
 
+static uint32_t vcn_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev,
+			uint32_t instance, uint32_t sub_block)
+{
+	uint32_t poison_stat = 0, reg_value = 0;
+
+	switch (sub_block) {
+	case AMDGPU_VCN_V4_0_3_VCPU_VCODEC:
+		reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS);
+		poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF);
+		break;
+	default:
+		break;
+	}
+
+	if (poison_stat)
+		dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n",
+			instance, sub_block);
+
+	return poison_stat;
+}
+
+static bool vcn_v4_0_3_query_poison_status(struct amdgpu_device *adev)
+{
+	uint32_t inst, sub;
+	uint32_t poison_stat = 0;
+
+	for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++)
+		for (sub = 0; sub < AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK; sub++)
+			poison_stat +=
+			vcn_v4_0_3_query_poison_by_instance(adev, inst, sub);
+
+	return !!poison_stat;
+}
+
 static const struct amdgpu_ras_block_hw_ops vcn_v4_0_3_ras_hw_ops = {
 	.query_ras_error_count = vcn_v4_0_3_query_ras_error_count,
 	.reset_ras_error_count = vcn_v4_0_3_reset_ras_error_count,
+	.query_poison_status = vcn_v4_0_3_query_poison_status,
 };
 
 static int vcn_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank,
@@ -2059,6 +2117,13 @@ static int vcn_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_commo
 	if (r)
 		return r;
 
+	if (amdgpu_ras_is_supported(adev, ras_block->block) &&
+		adev->vcn.inst->ras_poison_irq.funcs) {
+		r = amdgpu_irq_get(adev, &adev->vcn.inst->ras_poison_irq, 0);
+		if (r)
+			goto late_fini;
+	}
+
 	r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__VCN,
 				&vcn_v4_0_3_aca_info, NULL);
 	if (r)

drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h

@@ -24,6 +24,12 @@
 #ifndef __VCN_V4_0_3_H__
 #define __VCN_V4_0_3_H__
 
+enum amdgpu_vcn_v4_0_3_sub_block {
+	AMDGPU_VCN_V4_0_3_VCPU_VCODEC = 0,
+
+	AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK,
+};
+
 extern const struct amdgpu_ip_block_version vcn_v4_0_3_ip_block;
 
 void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,