diff options
Diffstat (limited to 'include/uapi/linux/kfd_ioctl.h')
| -rw-r--r-- | include/uapi/linux/kfd_ioctl.h | 135 |
1 files changed, 117 insertions, 18 deletions
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h index 9ce46edc62a5..1e59344c5673 100644 --- a/include/uapi/linux/kfd_ioctl.h +++ b/include/uapi/linux/kfd_ioctl.h @@ -41,9 +41,12 @@ * - 1.13 - Add debugger API * - 1.14 - Update kfd_event_data * - 1.15 - Enable managing mappings in compute VMs with GEM_VA ioctl + * - 1.16 - Add contiguous VRAM allocation flag + * - 1.17 - Add SDMA queue creation with target SDMA engine ID + * - 1.18 - Rename pad in set_memory_policy_args to misc_process_flag */ #define KFD_IOCTL_MAJOR_VERSION 1 -#define KFD_IOCTL_MINOR_VERSION 15 +#define KFD_IOCTL_MINOR_VERSION 18 struct kfd_ioctl_get_version_args { __u32 major_version; /* from KFD */ @@ -55,10 +58,13 @@ struct kfd_ioctl_get_version_args { #define KFD_IOC_QUEUE_TYPE_SDMA 0x1 #define KFD_IOC_QUEUE_TYPE_COMPUTE_AQL 0x2 #define KFD_IOC_QUEUE_TYPE_SDMA_XGMI 0x3 +#define KFD_IOC_QUEUE_TYPE_SDMA_BY_ENG_ID 0x4 #define KFD_MAX_QUEUE_PERCENTAGE 100 #define KFD_MAX_QUEUE_PRIORITY 15 +#define KFD_MIN_QUEUE_RING_SIZE 1024 + struct kfd_ioctl_create_queue_args { __u64 ring_base_address; /* to KFD */ __u64 write_pointer_address; /* from KFD */ @@ -77,6 +83,8 @@ struct kfd_ioctl_create_queue_args { __u64 ctx_save_restore_address; /* to KFD */ __u32 ctx_save_restore_size; /* to KFD */ __u32 ctl_stack_size; /* to KFD */ + __u32 sdma_engine_id; /* to KFD */ + __u32 pad; }; struct kfd_ioctl_destroy_queue_args { @@ -143,6 +151,9 @@ struct kfd_dbg_device_info_entry { #define KFD_IOC_CACHE_POLICY_COHERENT 0 #define KFD_IOC_CACHE_POLICY_NONCOHERENT 1 +/* Misc. per process flags */ +#define KFD_PROC_FLAG_MFMA_HIGH_PRECISION (1 << 0) + struct kfd_ioctl_set_memory_policy_args { __u64 alternate_aperture_base; /* to KFD */ __u64 alternate_aperture_size; /* to KFD */ @@ -150,7 +161,7 @@ struct kfd_ioctl_set_memory_policy_args { __u32 gpu_id; /* to KFD */ __u32 default_policy; /* to KFD */ __u32 alternate_policy; /* to KFD */ - __u32 pad; + __u32 misc_process_flag; /* to KFD */ }; /* @@ -407,6 +418,7 @@ struct kfd_ioctl_acquire_vm_args { #define KFD_IOC_ALLOC_MEM_FLAGS_COHERENT (1 << 26) #define KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED (1 << 25) #define KFD_IOC_ALLOC_MEM_FLAGS_EXT_COHERENT (1 << 24) +#define KFD_IOC_ALLOC_MEM_FLAGS_CONTIGUOUS (1 << 23) /* Allocate memory for later SVM (shared virtual memory) mapping. * @@ -534,26 +546,29 @@ enum kfd_smi_event { KFD_SMI_EVENT_ALL_PROCESS = 64 }; +/* The reason of the page migration event */ enum KFD_MIGRATE_TRIGGERS { - KFD_MIGRATE_TRIGGER_PREFETCH, - KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, - KFD_MIGRATE_TRIGGER_PAGEFAULT_CPU, - KFD_MIGRATE_TRIGGER_TTM_EVICTION + KFD_MIGRATE_TRIGGER_PREFETCH, /* Prefetch to GPU VRAM or system memory */ + KFD_MIGRATE_TRIGGER_PAGEFAULT_GPU, /* GPU page fault recover */ + KFD_MIGRATE_TRIGGER_PAGEFAULT_CPU, /* CPU page fault recover */ + KFD_MIGRATE_TRIGGER_TTM_EVICTION /* TTM eviction */ }; +/* The reason of user queue evition event */ enum KFD_QUEUE_EVICTION_TRIGGERS { - KFD_QUEUE_EVICTION_TRIGGER_SVM, - KFD_QUEUE_EVICTION_TRIGGER_USERPTR, - KFD_QUEUE_EVICTION_TRIGGER_TTM, - KFD_QUEUE_EVICTION_TRIGGER_SUSPEND, - KFD_QUEUE_EVICTION_CRIU_CHECKPOINT, - KFD_QUEUE_EVICTION_CRIU_RESTORE + KFD_QUEUE_EVICTION_TRIGGER_SVM, /* SVM buffer migration */ + KFD_QUEUE_EVICTION_TRIGGER_USERPTR, /* userptr movement */ + KFD_QUEUE_EVICTION_TRIGGER_TTM, /* TTM move buffer */ + KFD_QUEUE_EVICTION_TRIGGER_SUSPEND, /* GPU suspend */ + KFD_QUEUE_EVICTION_CRIU_CHECKPOINT, /* CRIU checkpoint */ + KFD_QUEUE_EVICTION_CRIU_RESTORE /* CRIU restore */ }; +/* The reason of unmap buffer from GPU event */ enum KFD_SVM_UNMAP_TRIGGERS { - KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY, - KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY_MIGRATE, - KFD_SVM_UNMAP_TRIGGER_UNMAP_FROM_CPU + KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY, /* MMU notifier CPU buffer movement */ + KFD_SVM_UNMAP_TRIGGER_MMU_NOTIFY_MIGRATE,/* MMU notifier page migration */ + KFD_SVM_UNMAP_TRIGGER_UNMAP_FROM_CPU /* Unmap to free the buffer */ }; #define KFD_SMI_EVENT_MASK_FROM_INDEX(i) (1ULL << ((i) - 1)) @@ -564,6 +579,78 @@ struct kfd_ioctl_smi_events_args { __u32 anon_fd; /* from KFD */ }; +/* + * SVM event tracing via SMI system management interface + * + * Open event file descriptor + * use ioctl AMDKFD_IOC_SMI_EVENTS, pass in gpuid and return a anonymous file + * descriptor to receive SMI events. + * If calling with sudo permission, then file descriptor can be used to receive + * SVM events from all processes, otherwise, to only receive SVM events of same + * process. + * + * To enable the SVM event + * Write event file descriptor with KFD_SMI_EVENT_MASK_FROM_INDEX(event) bitmap + * mask to start record the event to the kfifo, use bitmap mask combination + * for multiple events. New event mask will overwrite the previous event mask. + * KFD_SMI_EVENT_MASK_FROM_INDEX(KFD_SMI_EVENT_ALL_PROCESS) bit requires sudo + * permisson to receive SVM events from all process. + * + * To receive the event + * Application can poll file descriptor to wait for the events, then read event + * from the file into a buffer. Each event is one line string message, starting + * with the event id, then the event specific information. + * + * To decode event information + * The following event format string macro can be used with sscanf to decode + * the specific event information. + * event triggers: the reason to generate the event, defined as enum for unmap, + * eviction and migrate events. + * node, from, to, prefetch_loc, preferred_loc: GPU ID, or 0 for system memory. + * addr: user mode address, in pages + * size: in pages + * pid: the process ID to generate the event + * ns: timestamp in nanosecond-resolution, starts at system boot time but + * stops during suspend + * migrate_update: GPU page fault is recovered by 'M' for migrate, 'U' for update + * rw: 'W' for write page fault, 'R' for read page fault + * rescheduled: 'R' if the queue restore failed and rescheduled to try again + * error_code: migrate failure error code, 0 if no error + */ +#define KFD_EVENT_FMT_UPDATE_GPU_RESET(reset_seq_num, reset_cause)\ + "%x %s\n", (reset_seq_num), (reset_cause) + +#define KFD_EVENT_FMT_THERMAL_THROTTLING(bitmask, counter)\ + "%llx:%llx\n", (bitmask), (counter) + +#define KFD_EVENT_FMT_VMFAULT(pid, task_name)\ + "%x:%s\n", (pid), (task_name) + +#define KFD_EVENT_FMT_PAGEFAULT_START(ns, pid, addr, node, rw)\ + "%lld -%d @%lx(%x) %c\n", (ns), (pid), (addr), (node), (rw) + +#define KFD_EVENT_FMT_PAGEFAULT_END(ns, pid, addr, node, migrate_update)\ + "%lld -%d @%lx(%x) %c\n", (ns), (pid), (addr), (node), (migrate_update) + +#define KFD_EVENT_FMT_MIGRATE_START(ns, pid, start, size, from, to, prefetch_loc,\ + preferred_loc, migrate_trigger)\ + "%lld -%d @%lx(%lx) %x->%x %x:%x %d\n", (ns), (pid), (start), (size),\ + (from), (to), (prefetch_loc), (preferred_loc), (migrate_trigger) + +#define KFD_EVENT_FMT_MIGRATE_END(ns, pid, start, size, from, to, migrate_trigger, error_code) \ + "%lld -%d @%lx(%lx) %x->%x %d %d\n", (ns), (pid), (start), (size),\ + (from), (to), (migrate_trigger), (error_code) + +#define KFD_EVENT_FMT_QUEUE_EVICTION(ns, pid, node, evict_trigger)\ + "%lld -%d %x %d\n", (ns), (pid), (node), (evict_trigger) + +#define KFD_EVENT_FMT_QUEUE_RESTORE(ns, pid, node, rescheduled)\ + "%lld -%d %x %c\n", (ns), (pid), (node), (rescheduled) + +#define KFD_EVENT_FMT_UNMAP_FROM_GPU(ns, pid, addr, size, node, unmap_trigger)\ + "%lld -%d @%lx(%lx) %x %d\n", (ns), (pid), (addr), (size),\ + (node), (unmap_trigger) + /************************************************************************************************** * CRIU IOCTLs (Checkpoint Restore In Userspace) * @@ -852,6 +939,7 @@ enum kfd_dbg_trap_address_watch_mode { /* Additional wave settings */ enum kfd_dbg_trap_flags { KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP = 1, + KFD_DBG_TRAP_FLAG_SINGLE_ALU_OP = 2, }; /* Trap exceptions */ @@ -913,14 +1001,25 @@ enum kfd_dbg_trap_exception_code { KFD_EC_MASK(EC_DEVICE_NEW)) #define KFD_EC_MASK_PROCESS (KFD_EC_MASK(EC_PROCESS_RUNTIME) | \ KFD_EC_MASK(EC_PROCESS_DEVICE_REMOVE)) +#define KFD_EC_MASK_PACKET (KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_DIM_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_GROUP_SEGMENT_SIZE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_CODE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_RESERVED) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_UNSUPPORTED) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_WORK_GROUP_SIZE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_REGISTER_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_VENDOR_UNSUPPORTED)) /* Checks for exception code types for KFD search */ +#define KFD_DBG_EC_IS_VALID(ecode) (ecode > EC_NONE && ecode < EC_MAX) #define KFD_DBG_EC_TYPE_IS_QUEUE(ecode) \ - (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_QUEUE)) + (KFD_DBG_EC_IS_VALID(ecode) && !!(KFD_EC_MASK(ecode) & KFD_EC_MASK_QUEUE)) #define KFD_DBG_EC_TYPE_IS_DEVICE(ecode) \ - (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_DEVICE)) + (KFD_DBG_EC_IS_VALID(ecode) && !!(KFD_EC_MASK(ecode) & KFD_EC_MASK_DEVICE)) #define KFD_DBG_EC_TYPE_IS_PROCESS(ecode) \ - (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_PROCESS)) + (KFD_DBG_EC_IS_VALID(ecode) && !!(KFD_EC_MASK(ecode) & KFD_EC_MASK_PROCESS)) +#define KFD_DBG_EC_TYPE_IS_PACKET(ecode) \ + (KFD_DBG_EC_IS_VALID(ecode) && !!(KFD_EC_MASK(ecode) & KFD_EC_MASK_PACKET)) /* Runtime enable states */ |