diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 92 |
1 files changed, 83 insertions, 9 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 6db772ecfee4..ff44190d7d98 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -47,7 +47,7 @@ struct amdgpu_iv_entry; #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, 14, 13) #define AMDGPU_RAS_GPU_ERR_DATA_ABORT(x) AMDGPU_GET_REG_FIELD(x, 29, 29) -#define AMDGPU_RAS_GPU_ERR_UNKNOWN(x) AMDGPU_GET_REG_FIELD(x, 30, 30) +#define AMDGPU_RAS_GPU_ERR_GENERIC(x) AMDGPU_GET_REG_FIELD(x, 30, 30) #define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT 100 #define AMDGPU_RAS_BOOT_STEADY_STATUS 0xBA @@ -65,7 +65,7 @@ struct amdgpu_iv_entry; /* Reserve 8 physical dram row for possible retirement. * In worst cases, it will lose 8 * 2MB memory in vram domain */ -#define AMDGPU_RAS_RESERVED_VRAM_SIZE (16ULL << 20) +#define AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT (16ULL << 20) /* The high three bits indicates socketid */ #define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK) @@ -98,8 +98,10 @@ enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__JPEG, AMDGPU_RAS_BLOCK__IH, AMDGPU_RAS_BLOCK__MPIO, + AMDGPU_RAS_BLOCK__MMSCH, - AMDGPU_RAS_BLOCK__LAST + AMDGPU_RAS_BLOCK__LAST, + AMDGPU_RAS_BLOCK__ANY = -1 }; enum amdgpu_ras_mca_block { @@ -482,17 +484,53 @@ struct ras_ecc_err { uint64_t ipid; uint64_t addr; uint64_t pa_pfn; + /* save global channel index across all UMC instances */ + uint32_t channel_idx; struct ras_err_pages err_pages; }; struct ras_ecc_log_info { struct mutex lock; struct radix_tree_root de_page_tree; - uint64_t de_queried_count; - uint64_t prev_de_queried_count; + uint64_t de_queried_count; + uint64_t consumption_q_count; +}; + +struct ras_critical_region { + struct list_head node; + struct amdgpu_bo *bo; + uint64_t start; + uint64_t size; +}; + +struct ras_eeprom_table_version { + uint32_t minor : 16; + uint32_t major : 16; +}; + +struct ras_eeprom_smu_funcs { + int (*get_ras_table_version)(struct amdgpu_device *adev, + uint32_t *table_version); + int (*get_badpage_count)(struct amdgpu_device *adev, uint32_t *count, uint32_t timeout); + int (*get_badpage_mca_addr)(struct amdgpu_device *adev, uint16_t index, uint64_t *mca_addr); + int (*set_timestamp)(struct amdgpu_device *adev, uint64_t timestamp); + int (*get_timestamp)(struct amdgpu_device *adev, + uint16_t index, uint64_t *timestamp); + int (*get_badpage_ipid)(struct amdgpu_device *adev, uint16_t index, uint64_t *ipid); + int (*erase_ras_table)(struct amdgpu_device *adev, uint32_t *result); +}; + +enum ras_smu_feature_flags { + RAS_SMU_FEATURE_BIT__RAS_EEPROM = BIT_ULL(0), +}; + +struct ras_smu_drv { + const struct ras_eeprom_smu_funcs *smu_eeprom_funcs; + void (*ras_smu_feature_flags)(struct amdgpu_device *adev, uint64_t *flags); }; struct amdgpu_ras { + void *ras_mgr; /* ras infrastructure */ /* for ras itself. */ uint32_t features; @@ -511,6 +549,7 @@ struct amdgpu_ras { /* gpu recovery */ struct work_struct recovery_work; atomic_t in_recovery; + atomic_t rma_in_recovery; struct amdgpu_device *adev; /* error handler data */ struct ras_err_handler_data *eh_data; @@ -553,19 +592,35 @@ struct amdgpu_ras { struct mutex page_retirement_lock; atomic_t page_retirement_req_cnt; atomic_t poison_creation_count; + atomic_t poison_consumption_count; 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 errors detected */ + unsigned long ras_err_state; /* RAS event manager */ struct ras_event_manager __event_mgr; struct ras_event_manager *event_mgr; uint64_t reserved_pages_in_bytes; + + pid_t init_task_pid; + char init_task_comm[TASK_COMM_LEN]; + + int bad_page_num; + + struct list_head critical_region_head; + struct mutex critical_region_lock; + + /* Protect poison injection */ + struct mutex poison_lock; + + /* Disable/Enable uniras switch */ + bool uniras_enabled; + const struct ras_smu_drv *ras_smu_drv; }; struct ras_fs_data { @@ -604,6 +659,7 @@ struct ras_err_handler_data { struct eeprom_table_record *bps; /* the count of entries */ int count; + int count_saved; /* the space can place new entries */ int space_left; }; @@ -750,7 +806,7 @@ int amdgpu_ras_query_error_count(struct amdgpu_device *adev, /* error handling functions */ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, - struct eeprom_table_record *bps, int pages); + struct eeprom_table_record *bps, int pages, bool from_rom); int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev, unsigned long *new_cnt); @@ -792,6 +848,12 @@ amdgpu_ras_block_to_ta(enum amdgpu_ras_block block) { return TA_RAS_BLOCK__VCN; case AMDGPU_RAS_BLOCK__JPEG: return TA_RAS_BLOCK__JPEG; + case AMDGPU_RAS_BLOCK__IH: + return TA_RAS_BLOCK__IH; + case AMDGPU_RAS_BLOCK__MPIO: + return TA_RAS_BLOCK__MPIO; + case AMDGPU_RAS_BLOCK__MMSCH: + return TA_RAS_BLOCK__MMSCH; default: WARN_ONCE(1, "RAS ERROR: unexpected block id %d\n", block); return TA_RAS_BLOCK__UMC; @@ -878,7 +940,7 @@ static inline void amdgpu_ras_intr_cleared(void) atomic_set(&amdgpu_ras_in_intr, 0); } -void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); +int amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready); @@ -952,6 +1014,10 @@ ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *a void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status); bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev); +void amdgpu_ras_set_err_poison(struct amdgpu_device *adev, + enum amdgpu_ras_block block); +void amdgpu_ras_clear_err_state(struct amdgpu_device *adev); +bool amdgpu_ras_is_err_state(struct amdgpu_device *adev, int block); u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type type); int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_type type, @@ -959,6 +1025,9 @@ int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_ int amdgpu_ras_reserve_page(struct amdgpu_device *adev, uint64_t pfn); +int amdgpu_ras_add_critical_region(struct amdgpu_device *adev, struct amdgpu_bo *bo); +bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr); + 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); @@ -970,4 +1039,9 @@ void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id, const char *fmt, ...); bool amdgpu_ras_is_rma(struct amdgpu_device *adev); + +void amdgpu_ras_pre_reset(struct amdgpu_device *adev, + struct list_head *device_list); +void amdgpu_ras_post_reset(struct amdgpu_device *adev, + struct list_head *device_list); #endif |