summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c68
1 files changed, 44 insertions, 24 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 4a3726bb6..31823a30d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -305,11 +305,13 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f,
return -EINVAL;
data->head.block = block_id;
- /* only ue and ce errors are supported */
+ /* only ue, ce and poison errors are supported */
if (!memcmp("ue", err, 2))
data->head.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
else if (!memcmp("ce", err, 2))
data->head.type = AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE;
+ else if (!memcmp("poison", err, 6))
+ data->head.type = AMDGPU_RAS_ERROR__POISON;
else
return -EINVAL;
@@ -431,9 +433,10 @@ static void amdgpu_ras_instance_mask_check(struct amdgpu_device *adev,
* The block is one of: umc, sdma, gfx, etc.
* see ras_block_string[] for details
*
- * The error type is one of: ue, ce, where,
+ * The error type is one of: ue, ce and poison where,
* ue is multi-uncorrectable
* ce is single-correctable
+ * poison is poison
*
* The sub-block is a the sub-block index, pass 0 if there is no sub-block.
* The address and value are hexadecimal numbers, leading 0x is optional.
@@ -1067,8 +1070,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
mcm_info = &err_info->mcm_info;
if (err_info->ce_count) {
dev_info(adev->dev, "socket: %d, die: %d, "
- "%lld new correctable hardware errors detected in %s block, "
- "no user action is needed\n",
+ "%lld new correctable hardware errors detected in %s block\n",
mcm_info->socket_id,
mcm_info->die_id,
err_info->ce_count,
@@ -1080,8 +1082,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
err_info = &err_node->err_info;
mcm_info = &err_info->mcm_info;
dev_info(adev->dev, "socket: %d, die: %d, "
- "%lld correctable hardware errors detected in total in %s block, "
- "no user action is needed\n",
+ "%lld correctable hardware errors detected in total in %s block\n",
mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, blk_name);
}
}
@@ -1108,16 +1109,14 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
adev->smuio.funcs->get_die_id) {
dev_info(adev->dev, "socket: %d, die: %d "
"%ld correctable hardware errors "
- "detected in %s block, no user "
- "action is needed.\n",
+ "detected in %s block\n",
adev->smuio.funcs->get_socket_id(adev),
adev->smuio.funcs->get_die_id(adev),
ras_mgr->err_data.ce_count,
blk_name);
} else {
dev_info(adev->dev, "%ld correctable hardware errors "
- "detected in %s block, no user "
- "action is needed.\n",
+ "detected in %s block\n",
ras_mgr->err_data.ce_count,
blk_name);
}
@@ -1156,8 +1155,10 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s
for_each_ras_error(err_node, err_data) {
err_info = &err_node->err_info;
- amdgpu_ras_error_statistic_ce_count(&obj->err_data, &err_info->mcm_info, err_info->ce_count);
- amdgpu_ras_error_statistic_ue_count(&obj->err_data, &err_info->mcm_info, err_info->ue_count);
+ amdgpu_ras_error_statistic_ce_count(&obj->err_data,
+ &err_info->mcm_info, NULL, err_info->ce_count);
+ amdgpu_ras_error_statistic_ue_count(&obj->err_data,
+ &err_info->mcm_info, NULL, err_info->ue_count);
}
} else {
/* for legacy asic path which doesn't has error source info */
@@ -1918,7 +1919,7 @@ static void amdgpu_ras_interrupt_poison_creation_handler(struct ras_manager *obj
struct amdgpu_iv_entry *entry)
{
dev_info(obj->adev->dev,
- "Poison is created, no user action is needed.\n");
+ "Poison is created\n");
}
static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj,
@@ -2541,7 +2542,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
return 0;
data = &con->eh_data;
- *data = kmalloc(sizeof(**data), GFP_KERNEL | __GFP_ZERO);
+ *data = kzalloc(sizeof(**data), GFP_KERNEL);
if (!*data) {
ret = -ENOMEM;
goto out;
@@ -2828,10 +2829,10 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
if (con)
return 0;
- con = kmalloc(sizeof(struct amdgpu_ras) +
+ con = kzalloc(sizeof(*con) +
sizeof(struct ras_manager) * AMDGPU_RAS_BLOCK_COUNT +
sizeof(struct ras_manager) * AMDGPU_RAS_MCA_BLOCK_COUNT,
- GFP_KERNEL|__GFP_ZERO);
+ GFP_KERNEL);
if (!con)
return -ENOMEM;
@@ -2918,6 +2919,11 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
amdgpu_ras_query_poison_mode(adev);
+ /* Packed socket_id to ras feature mask bits[31:29] */
+ if (adev->smuio.funcs &&
+ adev->smuio.funcs->get_socket_id)
+ con->features |= ((adev->smuio.funcs->get_socket_id(adev)) << 29);
+
/* Get RAS schema for particular SOC */
con->schema = amdgpu_get_ras_schema(adev);
@@ -3136,6 +3142,8 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev)
if (amdgpu_sriov_vf(adev))
return 0;
+ amdgpu_ras_set_mca_debug_mode(adev, false);
+
list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
if (!node->ras_obj) {
dev_warn(adev->dev, "Warning: abnormal ras list node.\n");
@@ -3409,12 +3417,18 @@ int amdgpu_ras_reset_gpu(struct amdgpu_device *adev)
return 0;
}
-void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable)
+int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ int ret = 0;
- if (con)
- con->is_mca_debug_mode = enable;
+ if (con) {
+ ret = amdgpu_mca_smu_set_debug_mode(adev, enable);
+ if (!ret)
+ con->is_mca_debug_mode = enable;
+ }
+
+ return ret;
}
bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev)
@@ -3685,7 +3699,8 @@ static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct
}
static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data,
- struct amdgpu_smuio_mcm_config_info *mcm_info)
+ struct amdgpu_smuio_mcm_config_info *mcm_info,
+ struct ras_err_addr *err_addr)
{
struct ras_err_node *err_node;
@@ -3699,6 +3714,9 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d
memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info));
+ if (err_addr)
+ memcpy(&err_node->err_info.err_addr, err_addr, sizeof(*err_addr));
+
err_data->err_list_count++;
list_add_tail(&err_node->node, &err_data->err_node_list);
list_sort(NULL, &err_data->err_node_list, ras_err_info_cmp);
@@ -3707,7 +3725,8 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d
}
int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
- struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count)
+ struct amdgpu_smuio_mcm_config_info *mcm_info,
+ struct ras_err_addr *err_addr, u64 count)
{
struct ras_err_info *err_info;
@@ -3717,7 +3736,7 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
if (!count)
return 0;
- err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+ err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr);
if (!err_info)
return -EINVAL;
@@ -3728,7 +3747,8 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
}
int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
- struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count)
+ struct amdgpu_smuio_mcm_config_info *mcm_info,
+ struct ras_err_addr *err_addr, u64 count)
{
struct ras_err_info *err_info;
@@ -3738,7 +3758,7 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
if (!count)
return 0;
- err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+ err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr);
if (!err_info)
return -EINVAL;
generated by cgit v1.2.3 (git 2.39.1) at 2024-08-02 05:26:40 +0000