5 files changed, 53 insertions, 3 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index cf9a49f49d3a..c1e4d44d6137 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -467,6 +467,8 @@ struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
 	memset(&kfd->doorbell_available_index, 0,
 		sizeof(kfd->doorbell_available_index));
 
+	atomic_set(&kfd->sram_ecc_flag, 0);
+
 	return kfd;
 }
 
@@ -492,9 +494,9 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 {
 	unsigned int size;
 
-	kfd->mec_fw_version = kfd->kfd2kgd->get_fw_version(kfd->kgd,
+	kfd->mec_fw_version = amdgpu_amdkfd_get_fw_version(kfd->kgd,
 			KGD_ENGINE_MEC1);
-	kfd->sdma_fw_version = kfd->kfd2kgd->get_fw_version(kfd->kgd,
+	kfd->sdma_fw_version = amdgpu_amdkfd_get_fw_version(kfd->kgd,
 			KGD_ENGINE_SDMA1);
 	kfd->shared_resources = *gpu_resources;
 
@@ -662,6 +664,9 @@ int kgd2kfd_post_reset(struct kfd_dev *kfd)
 		return ret;
 	count = atomic_dec_return(&kfd_locked);
 	WARN_ONCE(count != 0, "KFD reset ref. error");
+
+	atomic_set(&kfd->sram_ecc_flag, 0);
+
 	return 0;
 }
 
@@ -1025,6 +1030,12 @@ int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj)
 	return 0;
 }
 
+void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd)
+{
+	if (kfd)
+		atomic_inc(&kfd->sram_ecc_flag);
+}
+
 #if defined(CONFIG_DEBUG_FS)
 
 /* This function will send a package to HIQ to hang the HWS
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
index e9f0e0a1b41c..6e1d41c5bf86 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@@ -1011,25 +1011,41 @@ void kfd_signal_vm_fault_event(struct kfd_dev *dev, unsigned int pasid,
 void kfd_signal_reset_event(struct kfd_dev *dev)
 {
 	struct kfd_hsa_hw_exception_data hw_exception_data;
+	struct kfd_hsa_memory_exception_data memory_exception_data;
 	struct kfd_process *p;
 	struct kfd_event *ev;
 	unsigned int temp;
 	uint32_t id, idx;
+	int reset_cause = atomic_read(&dev->sram_ecc_flag) ?
+			KFD_HW_EXCEPTION_ECC :
+			KFD_HW_EXCEPTION_GPU_HANG;
 
 	/* Whole gpu reset caused by GPU hang and memory is lost */
 	memset(&hw_exception_data, 0, sizeof(hw_exception_data));
 	hw_exception_data.gpu_id = dev->id;
 	hw_exception_data.memory_lost = 1;
+	hw_exception_data.reset_cause = reset_cause;
+
+	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+	memory_exception_data.ErrorType = KFD_MEM_ERR_SRAM_ECC;
+	memory_exception_data.gpu_id = dev->id;
+	memory_exception_data.failure.imprecise = true;
 
 	idx = srcu_read_lock(&kfd_processes_srcu);
 	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
 		mutex_lock(&p->event_mutex);
 		id = KFD_FIRST_NONSIGNAL_EVENT_ID;
-		idr_for_each_entry_continue(&p->event_idr, ev, id)
+		idr_for_each_entry_continue(&p->event_idr, ev, id) {
 			if (ev->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
 				ev->hw_exception_data = hw_exception_data;
 				set_event(ev);
 			}
+			if (ev->type == KFD_EVENT_TYPE_MEMORY &&
+			    reset_cause == KFD_HW_EXCEPTION_ECC) {
+				ev->memory_exception_data = memory_exception_data;
+				set_event(ev);
+			}
+		}
 		mutex_unlock(&p->event_mutex);
 	}
 	srcu_read_unlock(&kfd_processes_srcu, idx);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 0eeee3c6d6dc..9e0230965675 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -276,6 +276,9 @@ struct kfd_dev {
 	uint64_t hive_id;
 
 	bool pci_atomic_requested;
+
+	/* SRAM ECC flag */
+	atomic_t sram_ecc_flag;
 };
 
 enum kfd_mempool {
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index 09da91644f9f..2cb09e088dce 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -37,6 +37,7 @@
 #include "kfd_device_queue_manager.h"
 #include "kfd_iommu.h"
 #include "amdgpu_amdkfd.h"
+#include "amdgpu_ras.h"
 
 /* topology_device_list - Master list of all topology devices */
 static struct list_head topology_device_list;
@@ -1197,6 +1198,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 	void *crat_image = NULL;
 	size_t image_size = 0;
 	int proximity_domain;
+	struct amdgpu_ras *ctx;
 
 	INIT_LIST_HEAD(&temp_topology_device_list);
 
@@ -1328,6 +1330,20 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 		dev->node_props.capability |= HSA_CAP_ATS_PRESENT;
 	}
 
+	ctx = amdgpu_ras_get_context((struct amdgpu_device *)(dev->gpu->kgd));
+	if (ctx) {
+		/* kfd only concerns sram ecc on GFX/SDMA and HBM ecc on UMC */
+		dev->node_props.capability |=
+			(((ctx->features & BIT(AMDGPU_RAS_BLOCK__SDMA)) != 0) ||
+			 ((ctx->features & BIT(AMDGPU_RAS_BLOCK__GFX)) != 0)) ?
+			HSA_CAP_SRAM_EDCSUPPORTED : 0;
+		dev->node_props.capability |= ((ctx->features & BIT(AMDGPU_RAS_BLOCK__UMC)) != 0) ?
+			HSA_CAP_MEM_EDCSUPPORTED : 0;
+
+		dev->node_props.capability |= (ctx->features != 0) ?
+			HSA_CAP_RASEVENTNOTIFY : 0;
+	}
+
 	kfd_debug_print_topology();
 
 	if (!res)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
index 92a19be07344..84710cfd23c2 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
@@ -48,6 +48,10 @@
 #define HSA_CAP_DOORBELL_TYPE_2_0		0x2
 #define HSA_CAP_AQL_QUEUE_DOUBLE_MAP		0x00004000
 
+#define HSA_CAP_SRAM_EDCSUPPORTED		0x00080000
+#define HSA_CAP_MEM_EDCSUPPORTED		0x00100000
+#define HSA_CAP_RASEVENTNOTIFY			0x00200000
+
 struct kfd_node_properties {
 	uint64_t hive_id;
 	uint32_t cpu_cores_count;