diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 164 |
1 files changed, 159 insertions, 5 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 665414c22ca9..c8980d5f6540 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -26,12 +26,34 @@ #include <linux/debugfs.h> #include <linux/list.h> +#include <linux/kfifo.h> +#include <linux/radix-tree.h> +#include <linux/siphash.h> #include "ta_ras_if.h" #include "amdgpu_ras_eeprom.h" #include "amdgpu_smuio.h" +#include "amdgpu_aca.h" struct amdgpu_iv_entry; +#define AMDGPU_RAS_GPU_ERR_MEM_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 0, 0) +#define AMDGPU_RAS_GPU_ERR_FW_LOAD(x) AMDGPU_GET_REG_FIELD(x, 1, 1) +#define AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 2, 2) +#define AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 3, 3) +#define AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 4, 4) +#define AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 5, 5) +#define AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(x) AMDGPU_GET_REG_FIELD(x, 6, 6) +#define AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(x) AMDGPU_GET_REG_FIELD(x, 7, 7) +#define AMDGPU_RAS_GPU_ERR_SOCKET_ID(x) AMDGPU_GET_REG_FIELD(x, 10, 8) +#define AMDGPU_RAS_GPU_ERR_AID_ID(x) AMDGPU_GET_REG_FIELD(x, 12, 11) +#define AMDGPU_RAS_GPU_ERR_HBM_ID(x) AMDGPU_GET_REG_FIELD(x, 13, 13) +#define AMDGPU_RAS_GPU_ERR_BOOT_STATUS(x) AMDGPU_GET_REG_FIELD(x, 31, 31) + +#define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT 1000 +#define AMDGPU_RAS_BOOT_STEADY_STATUS 0xBA +#define AMDGPU_RAS_BOOT_STATUS_MASK 0xFF +#define AMDGPU_RAS_BOOT_SUCEESS 0x80000000 + #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) /* position of instance value in sub_block_index of * ta_ras_trigger_error_input, the sub block uses lower 12 bits @@ -39,6 +61,20 @@ struct amdgpu_iv_entry; #define AMDGPU_RAS_INST_MASK 0xfffff000 #define AMDGPU_RAS_INST_SHIFT 0xc +#define AMDGPU_RAS_FEATURES_SOCKETID_SHIFT 29 +#define AMDGPU_RAS_FEATURES_SOCKETID_MASK 0xe0000000 + +/* The high three bits indicates socketid */ +#define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK) + +#define RAS_EVENT_LOG(_adev, _id, _fmt, ...) \ +do { \ + if (amdgpu_ras_event_id_is_valid((_adev), (_id))) \ + dev_info((_adev)->dev, "{%llu}" _fmt, (_id), ##__VA_ARGS__); \ + else \ + dev_info((_adev)->dev, _fmt, ##__VA_ARGS__); \ +} while (0) + enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__UMC = 0, AMDGPU_RAS_BLOCK__SDMA, @@ -57,6 +93,8 @@ enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__MCA, AMDGPU_RAS_BLOCK__VCN, AMDGPU_RAS_BLOCK__JPEG, + AMDGPU_RAS_BLOCK__IH, + AMDGPU_RAS_BLOCK__MPIO, AMDGPU_RAS_BLOCK__LAST }; @@ -320,6 +358,12 @@ enum amdgpu_ras_ret { AMDGPU_RAS_PT, }; +enum amdgpu_ras_error_query_mode { + AMDGPU_RAS_INVALID_ERROR_QUERY = 0, + AMDGPU_RAS_DIRECT_ERROR_QUERY = 1, + AMDGPU_RAS_FIRMWARE_ERROR_QUERY = 2, +}; + /* ras error status reisger fields */ #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG__SHIFT 0x0 #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG_MASK 0x00000001L @@ -386,6 +430,52 @@ struct umc_ecc_info { int record_ce_addr_supported; }; +enum ras_event_type { + RAS_EVENT_TYPE_INVALID = -1, + RAS_EVENT_TYPE_ISR = 0, + RAS_EVENT_TYPE_COUNT, +}; + +struct ras_event_manager { + atomic64_t seqnos[RAS_EVENT_TYPE_COUNT]; +}; + +struct ras_query_context { + enum ras_event_type type; + u64 event_id; +}; + +typedef int (*pasid_notify)(struct amdgpu_device *adev, + uint16_t pasid, void *data); + +struct ras_poison_msg { + enum amdgpu_ras_block block; + uint16_t pasid; + uint32_t reset; + pasid_notify pasid_fn; + void *data; +}; + +struct ras_err_pages { + uint32_t count; + uint64_t *pfn; +}; + +struct ras_ecc_err { + u64 hash_index; + uint64_t status; + uint64_t ipid; + uint64_t addr; + struct ras_err_pages err_pages; +}; + +struct ras_ecc_log_info { + struct mutex lock; + siphash_key_t ecc_key; + struct radix_tree_root de_page_tree; + bool de_updated; +}; + struct amdgpu_ras { /* ras infrastructure */ /* for ras itself. */ @@ -435,10 +525,27 @@ struct amdgpu_ras { /* Indicates smu whether need update bad channel info */ bool update_channel_flag; /* Record status of smu mca debug mode */ - bool is_mca_debug_mode; + bool is_aca_debug_mode; /* Record special requirements of gpu reset caller */ uint32_t gpu_reset_flags; + + struct task_struct *page_retirement_thread; + wait_queue_head_t page_retirement_wq; + struct mutex page_retirement_lock; + atomic_t page_retirement_req_cnt; + struct mutex page_rsv_lock; + DECLARE_KFIFO(poison_fifo, struct ras_poison_msg, 128); + struct ras_ecc_log_info umc_ecc_log; + struct delayed_work page_retirement_dwork; + + /* Fatal error detected flag */ + atomic_t fed; + + /* RAS event manager */ + struct ras_event_manager __event_mgr; + struct ras_event_manager *event_mgr; + }; struct ras_fs_data { @@ -446,10 +553,19 @@ struct ras_fs_data { char debugfs_name[32]; }; +struct ras_err_addr { + struct list_head node; + uint64_t err_status; + uint64_t err_ipid; + uint64_t err_addr; +}; + struct ras_err_info { struct amdgpu_smuio_mcm_config_info mcm_info; u64 ce_count; u64 ue_count; + u64 de_count; + struct list_head err_addr_list; }; struct ras_err_node { @@ -460,8 +576,10 @@ struct ras_err_node { struct ras_err_data { unsigned long ue_count; unsigned long ce_count; + unsigned long de_count; unsigned long err_addr_cnt; struct eeprom_table_record *err_addr; + unsigned long err_addr_len; u32 err_list_count; struct list_head err_node_list; }; @@ -516,6 +634,8 @@ struct ras_manager { struct ras_ih_data ih_data; struct ras_err_data err_data; + + struct aca_handle aca_handle; }; struct ras_badpage { @@ -535,6 +655,7 @@ struct ras_query_if { struct ras_common_if head; unsigned long ue_count; unsigned long ce_count; + unsigned long de_count; }; struct ras_inject_if { @@ -767,8 +888,11 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); -void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); -bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); +int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); +int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable); +bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev); +bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, + unsigned int *mode); int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, struct amdgpu_ras_block_object *ras_block_obj); @@ -798,8 +922,38 @@ void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev, int amdgpu_ras_error_data_init(struct ras_err_data *err_data); void amdgpu_ras_error_data_fini(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); 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); +int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count); +void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances); +int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, + const struct aca_info *aca_info, void *data); +int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk); + +ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, + struct aca_handle *handle, char *buf, void *data); + +void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, + struct ras_err_addr *err_addr); + +void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, + struct ras_err_addr *mca_err_addr); + +void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status); +bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev); + +bool amdgpu_ras_event_id_is_valid(struct amdgpu_device *adev, u64 id); +u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type); + +int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn); + +int amdgpu_ras_put_poison_req(struct amdgpu_device *adev, + enum amdgpu_ras_block block, uint16_t pasid, + pasid_notify pasid_fn, void *data, uint32_t reset); #endif |