diff --git a/include/linux/bpf.h b/include/linux/bpf.h index d7b43e73fe87fd84533a3eb9b59053d3217bdcf0..9adfef694a25877cc5a2bede2c06a3db7832f41b 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -13,6 +13,7 @@ #include #include +struct perf_event; struct bpf_map; /* map is generic key/value storage optionally accesible by eBPF programs */ @@ -166,8 +167,16 @@ struct bpf_array { void __percpu *pptrs[0] __aligned(8); }; }; + #define MAX_TAIL_CALL_CNT 32 +struct bpf_event_entry { + struct perf_event *event; + struct file *perf_file; + struct file *map_file; + struct rcu_head rcu; +}; + u64 bpf_tail_call(u64 ctx, u64 r2, u64 index, u64 r4, u64 r5); u64 bpf_get_stackid(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5); diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index bfedcbdb4d84f191311825e87f45f9da191564dd..5af30732697ba1b7022c40c60a4544b1a657ceb3 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -427,59 +427,105 @@ static int __init register_prog_array_map(void) } late_initcall(register_prog_array_map); -static void perf_event_array_map_free(struct bpf_map *map) +static struct bpf_event_entry *bpf_event_entry_gen(struct file *perf_file, + struct file *map_file) { - bpf_fd_array_map_clear(map); - fd_array_map_free(map); + struct bpf_event_entry *ee; + + ee = kzalloc(sizeof(*ee), GFP_KERNEL); + if (ee) { + ee->event = perf_file->private_data; + ee->perf_file = perf_file; + ee->map_file = map_file; + } + + return ee; +} + +static void __bpf_event_entry_free(struct rcu_head *rcu) +{ + struct bpf_event_entry *ee; + + ee = container_of(rcu, struct bpf_event_entry, rcu); + fput(ee->perf_file); + kfree(ee); +} + +static void bpf_event_entry_free_rcu(struct bpf_event_entry *ee) +{ + call_rcu(&ee->rcu, __bpf_event_entry_free); } static void *perf_event_fd_array_get_ptr(struct bpf_map *map, struct file *map_file, int fd) { - struct perf_event *event; const struct perf_event_attr *attr; - struct file *file; + struct bpf_event_entry *ee; + struct perf_event *event; + struct file *perf_file; - file = perf_event_get(fd); - if (IS_ERR(file)) - return file; + perf_file = perf_event_get(fd); + if (IS_ERR(perf_file)) + return perf_file; - event = file->private_data; + event = perf_file->private_data; + ee = ERR_PTR(-EINVAL); attr = perf_event_attrs(event); - if (IS_ERR(attr)) - goto err; - - if (attr->inherit) - goto err; - - if (attr->type == PERF_TYPE_RAW) - return file; - - if (attr->type == PERF_TYPE_HARDWARE) - return file; + if (IS_ERR(attr) || attr->inherit) + goto err_out; + + switch (attr->type) { + case PERF_TYPE_SOFTWARE: + if (attr->config != PERF_COUNT_SW_BPF_OUTPUT) + goto err_out; + /* fall-through */ + case PERF_TYPE_RAW: + case PERF_TYPE_HARDWARE: + ee = bpf_event_entry_gen(perf_file, map_file); + if (ee) + return ee; + ee = ERR_PTR(-ENOMEM); + /* fall-through */ + default: + break; + } - if (attr->type == PERF_TYPE_SOFTWARE && - attr->config == PERF_COUNT_SW_BPF_OUTPUT) - return file; -err: - fput(file); - return ERR_PTR(-EINVAL); +err_out: + fput(perf_file); + return ee; } static void perf_event_fd_array_put_ptr(void *ptr) { - fput((struct file *)ptr); + bpf_event_entry_free_rcu(ptr); +} + +static void perf_event_fd_array_release(struct bpf_map *map, + struct file *map_file) +{ + struct bpf_array *array = container_of(map, struct bpf_array, map); + struct bpf_event_entry *ee; + int i; + + rcu_read_lock(); + for (i = 0; i < array->map.max_entries; i++) { + ee = READ_ONCE(array->ptrs[i]); + if (ee && ee->map_file == map_file) + fd_array_map_delete_elem(map, &i); + } + rcu_read_unlock(); } static const struct bpf_map_ops perf_event_array_ops = { .map_alloc = fd_array_map_alloc, - .map_free = perf_event_array_map_free, + .map_free = fd_array_map_free, .map_get_next_key = array_map_get_next_key, .map_lookup_elem = fd_array_map_lookup_elem, .map_delete_elem = fd_array_map_delete_elem, .map_fd_get_ptr = perf_event_fd_array_get_ptr, .map_fd_put_ptr = perf_event_fd_array_put_ptr, + .map_release = perf_event_fd_array_release, }; static struct bpf_map_type_list perf_event_array_type __read_mostly = { diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 720b7bb01d43ae5f49761da3a34d8fce4e55dbc2..037ea6ea3cb26d5ddf38f0aaf529d2f189412fc1 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -192,18 +192,17 @@ static u64 bpf_perf_event_read(u64 r1, u64 index, u64 r3, u64 r4, u64 r5) { struct bpf_map *map = (struct bpf_map *) (unsigned long) r1; struct bpf_array *array = container_of(map, struct bpf_array, map); + struct bpf_event_entry *ee; struct perf_event *event; - struct file *file; if (unlikely(index >= array->map.max_entries)) return -E2BIG; - file = READ_ONCE(array->ptrs[index]); - if (unlikely(!file)) + ee = READ_ONCE(array->ptrs[index]); + if (unlikely(!ee)) return -ENOENT; - event = file->private_data; - + event = ee->event; /* make sure event is local and doesn't have pmu::count */ if (event->oncpu != smp_processor_id() || event->pmu->count) @@ -233,8 +232,8 @@ static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size) u64 index = flags & BPF_F_INDEX_MASK; void *data = (void *) (long) r4; struct perf_sample_data sample_data; + struct bpf_event_entry *ee; struct perf_event *event; - struct file *file; struct perf_raw_record raw = { .size = size, .data = data, @@ -247,12 +246,11 @@ static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size) if (unlikely(index >= array->map.max_entries)) return -E2BIG; - file = READ_ONCE(array->ptrs[index]); - if (unlikely(!file)) + ee = READ_ONCE(array->ptrs[index]); + if (unlikely(!ee)) return -ENOENT; - event = file->private_data; - + event = ee->event; if (unlikely(event->attr.type != PERF_TYPE_SOFTWARE || event->attr.config != PERF_COUNT_SW_BPF_OUTPUT)) return -EINVAL;