drm/amdgpu: refine smu v13.0.6 mca dump driver

refine smu mca driver to support query ras error from pmfw path.
- correct gfx smu bank hwid (from mp5 to smu bank)
- retire unused callback function in amdgpu_mca_smu_funcs{}
- add new mca_bank_set{} structure to collect mca bank
- move enum mca_reg_idx into amdgpu_mca.h header
- add mca status register field decode macro

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

2 files changed, 215 insertions, 10 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
index 5a828c175e3a..3302f4a29387 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
@@ -143,6 +143,46 @@ int amdgpu_mca_mpio_ras_sw_init(struct amdgpu_device *adev)
 	return 0;
 }
 
+void amdgpu_mca_bank_set_init(struct mca_bank_set *mca_set)
+{
+	if (!mca_set)
+		return;
+
+	memset(mca_set, 0, sizeof(*mca_set));
+	INIT_LIST_HEAD(&mca_set->list);
+}
+
+int amdgpu_mca_bank_set_add_entry(struct mca_bank_set *mca_set, struct mca_bank_entry *entry)
+{
+	struct mca_bank_node *node;
+
+	if (!entry)
+		return -EINVAL;
+
+	node = kvzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return -ENOMEM;
+
+	memcpy(&node->entry, entry, sizeof(*entry));
+
+	INIT_LIST_HEAD(&node->node);
+	list_add_tail(&node->node, &mca_set->list);
+
+	mca_set->nr_entries++;
+
+	return 0;
+}
+
+void amdgpu_mca_bank_set_release(struct mca_bank_set *mca_set)
+{
+	struct mca_bank_node *node, *tmp;
+
+	list_for_each_entry_safe(node, tmp, &mca_set->list, node) {
+		list_del(&node->node);
+		kvfree(node);
+	}
+}
+
 void amdgpu_mca_smu_init_funcs(struct amdgpu_device *adev, const struct amdgpu_mca_smu_funcs *mca_funcs)
 {
 	struct amdgpu_mca *mca = &adev->mca;
@@ -160,6 +200,58 @@ int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable)
 	return -EOPNOTSUPP;
 }
 
+static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, struct mca_bank_entry *entry)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(entry->regs); i++)
+		dev_dbg(adev->dev, "mca entry[%02d].regs[%02d]=0x%016llx\n", idx, i, entry->regs[i]);
+}
+
+int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, struct ras_err_data *err_data)
+{
+	struct amdgpu_smuio_mcm_config_info mcm_info;
+	struct mca_bank_set mca_set;
+	struct mca_bank_node *node;
+	struct mca_bank_entry *entry;
+	uint32_t count;
+	int ret, i = 0;
+
+	amdgpu_mca_bank_set_init(&mca_set);
+
+	ret = amdgpu_mca_smu_get_mca_set(adev, blk, type, &mca_set);
+	if (ret)
+		goto out_mca_release;
+
+	list_for_each_entry(node, &mca_set.list, node) {
+		entry = &node->entry;
+
+		amdgpu_mca_smu_mca_bank_dump(adev, i++, entry);
+
+		count = 0;
+		ret = amdgpu_mca_smu_parse_mca_error_count(adev, blk, type, entry, &count);
+		if (ret)
+			goto out_mca_release;
+
+		if (!count)
+			continue;
+
+		mcm_info.socket_id = entry->info.socket_id;
+		mcm_info.die_id = entry->info.aid;
+
+		if (type == AMDGPU_MCA_ERROR_TYPE_UE)
+			amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, (uint64_t)count);
+		else
+			amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, (uint64_t)count);
+	}
+
+out_mca_release:
+	amdgpu_mca_bank_set_release(&mca_set);
+
+	return ret;
+}
+
+
 int amdgpu_mca_smu_get_valid_mca_count(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, uint32_t *count)
 {
 	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
@@ -173,17 +265,77 @@ int amdgpu_mca_smu_get_valid_mca_count(struct amdgpu_device *adev, enum amdgpu_m
 	return -EOPNOTSUPP;
 }
 
-int amdgpu_mca_smu_get_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
-				   enum amdgpu_mca_error_type type, uint32_t *count)
+int amdgpu_mca_smu_get_mca_set_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					    enum amdgpu_mca_error_type type, uint32_t *total)
 {
 	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
-	if (!count)
+	struct mca_bank_set mca_set;
+	struct mca_bank_node *node;
+	struct mca_bank_entry *entry;
+	uint32_t count;
+	int ret;
+
+	if (!total)
 		return -EINVAL;
 
-	if (mca_funcs && mca_funcs->mca_get_error_count)
-		return mca_funcs->mca_get_error_count(adev, blk, type, count);
+	if (!mca_funcs)
+		return -EOPNOTSUPP;
 
-	return -EOPNOTSUPP;
+	if (!mca_funcs->mca_get_ras_mca_set || !mca_funcs->mca_get_valid_mca_count)
+		return -EOPNOTSUPP;
+
+	amdgpu_mca_bank_set_init(&mca_set);
+
+	ret = mca_funcs->mca_get_ras_mca_set(adev, blk, type, &mca_set);
+	if (ret)
+		goto err_mca_set_release;
+
+	*total = 0;
+	list_for_each_entry(node, &mca_set.list, node) {
+		entry = &node->entry;
+
+		count = 0;
+		ret = mca_funcs->mca_parse_mca_error_count(adev, blk, type, entry, &count);
+		if (ret)
+			goto err_mca_set_release;
+
+		*total += count;
+	}
+
+err_mca_set_release:
+	amdgpu_mca_bank_set_release(&mca_set);
+
+	return ret;
+}
+
+int amdgpu_mca_smu_parse_mca_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					 enum amdgpu_mca_error_type type, struct mca_bank_entry *entry, uint32_t *count)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+	if (!count || !entry)
+		return -EINVAL;
+
+	if (!mca_funcs || !mca_funcs->mca_parse_mca_error_count)
+		return -EOPNOTSUPP;
+
+
+	return mca_funcs->mca_parse_mca_error_count(adev, blk, type, entry, count);
+}
+
+int amdgpu_mca_smu_get_mca_set(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+			       enum amdgpu_mca_error_type type, struct mca_bank_set *mca_set)
+{
+	const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+
+	if (!mca_set)
+		return -EINVAL;
+
+	if (!mca_funcs || !mca_funcs->mca_get_ras_mca_set)
+		return -EOPNOTSUPP;
+
+	WARN_ON(!list_empty(&mca_set->list));
+
+	return mca_funcs->mca_get_ras_mca_set(adev, blk, type, mca_set);
 }
 
 int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
index 28ad463cf5c9..45f90edf2283 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
@@ -25,6 +25,27 @@
 
 #define MCA_MAX_REGS_COUNT	(16)
 
+#define MCA_REG_FIELD(x, h, l)			(((x) & GENMASK_ULL(h, l)) >> l)
+#define MCA_REG__STATUS__VAL(x)			MCA_REG_FIELD(x, 63, 63)
+#define MCA_REG__STATUS__OVERFLOW(x)		MCA_REG_FIELD(x, 62, 62)
+#define MCA_REG__STATUS__UC(x)			MCA_REG_FIELD(x, 61, 61)
+#define MCA_REG__STATUS__EN(x)			MCA_REG_FIELD(x, 60, 60)
+#define MCA_REG__STATUS__MISCV(x)		MCA_REG_FIELD(x, 59, 59)
+#define MCA_REG__STATUS__ADDRV(x)		MCA_REG_FIELD(x, 58, 58)
+#define MCA_REG__STATUS__PCC(x)			MCA_REG_FIELD(x, 57, 57)
+#define MCA_REG__STATUS__ERRCOREIDVAL(x)	MCA_REG_FIELD(x, 56, 56)
+#define MCA_REG__STATUS__TCC(x)			MCA_REG_FIELD(x, 55, 55)
+#define MCA_REG__STATUS__SYNDV(x)		MCA_REG_FIELD(x, 53, 53)
+#define MCA_REG__STATUS__CECC(x)		MCA_REG_FIELD(x, 46, 46)
+#define MCA_REG__STATUS__UECC(x)		MCA_REG_FIELD(x, 45, 45)
+#define MCA_REG__STATUS__DEFERRED(x)		MCA_REG_FIELD(x, 44, 44)
+#define MCA_REG__STATUS__POISON(x)		MCA_REG_FIELD(x, 43, 43)
+#define MCA_REG__STATUS__SCRUB(x)		MCA_REG_FIELD(x, 40, 40)
+#define MCA_REG__STATUS__ERRCOREID(x)		MCA_REG_FIELD(x, 37, 32)
+#define MCA_REG__STATUS__ADDRLSB(x)		MCA_REG_FIELD(x, 29, 24)
+#define MCA_REG__STATUS__ERRORCODEEXT(x)	MCA_REG_FIELD(x, 21, 16)
+#define MCA_REG__STATUS__ERRORCODE(x)		MCA_REG_FIELD(x, 15, 0)
+
 enum amdgpu_mca_ip {
 	AMDGPU_MCA_IP_UNKNOW = -1,
 	AMDGPU_MCA_IP_PSP = 0,
@@ -57,6 +78,17 @@ struct amdgpu_mca {
 	const struct amdgpu_mca_smu_funcs *mca_funcs;
 };
 
+enum mca_reg_idx {
+	MCA_REG_IDX_CONTROL		= 0,
+	MCA_REG_IDX_STATUS		= 1,
+	MCA_REG_IDX_ADDR		= 2,
+	MCA_REG_IDX_MISC0		= 3,
+	MCA_REG_IDX_CONFIG		= 4,
+	MCA_REG_IDX_IPID		= 5,
+	MCA_REG_IDX_SYND		= 6,
+	MCA_REG_IDX_COUNT		= 16,
+};
+
 struct mca_bank_info {
 	int socket_id;
 	int aid;
@@ -72,18 +104,28 @@ struct mca_bank_entry {
 	uint64_t regs[MCA_MAX_REGS_COUNT];
 };
 
+struct mca_bank_node {
+	struct mca_bank_entry entry;
+	struct list_head node;
+};
+
+struct mca_bank_set {
+	int nr_entries;
+	struct list_head list;
+};
+
 struct amdgpu_mca_smu_funcs {
 	int max_ue_count;
 	int max_ce_count;
 	int (*mca_set_debug_mode)(struct amdgpu_device *adev, bool enable);
-	int (*mca_get_error_count)(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
-				   enum amdgpu_mca_error_type type, uint32_t *count);
+	int (*mca_get_ras_mca_set)(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+				   struct mca_bank_set *mca_set);
+	int (*mca_parse_mca_error_count)(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type,
+					 struct mca_bank_entry *entry, uint32_t *count);
 	int (*mca_get_valid_mca_count)(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
 				       uint32_t *count);
 	int (*mca_get_mca_entry)(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
 				 int idx, struct mca_bank_entry *entry);
-	int (*mca_get_ras_mca_idx_array)(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
-					 enum amdgpu_mca_error_type type, int *idx_array, int *idx_array_size);
 };
 
 void amdgpu_mca_query_correctable_error_count(struct amdgpu_device *adev,
@@ -107,11 +149,22 @@ int amdgpu_mca_mpio_ras_sw_init(struct amdgpu_device *adev);
 void amdgpu_mca_smu_init_funcs(struct amdgpu_device *adev, const struct amdgpu_mca_smu_funcs *mca_funcs);
 int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable);
 int amdgpu_mca_smu_get_valid_mca_count(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, uint32_t *count);
+int amdgpu_mca_smu_get_mca_set_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					   enum amdgpu_mca_error_type type, uint32_t *total);
 int amdgpu_mca_smu_get_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
 				   enum amdgpu_mca_error_type type, uint32_t *count);
+int amdgpu_mca_smu_parse_mca_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+					 enum amdgpu_mca_error_type type, struct mca_bank_entry *entry, uint32_t *count);
+int amdgpu_mca_smu_get_mca_set(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+			       enum amdgpu_mca_error_type type, struct mca_bank_set *mca_set);
 int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev, enum amdgpu_mca_error_type type,
 				 int idx, struct mca_bank_entry *entry);
 
 void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root);
 
+void amdgpu_mca_bank_set_init(struct mca_bank_set *mca_set);
+int amdgpu_mca_bank_set_add_entry(struct mca_bank_set *mca_set, struct mca_bank_entry *entry);
+void amdgpu_mca_bank_set_release(struct mca_bank_set *mca_set);
+int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, struct ras_err_data *err_data);
+
 #endif