diff options
Diffstat (limited to 'kernel/bpf/syscall.c')
| -rw-r--r-- | kernel/bpf/syscall.c | 410 |
1 files changed, 342 insertions, 68 deletions
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index 7626b8024471..bb1ab7da6103 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -42,6 +42,8 @@ static DEFINE_IDR(prog_idr); static DEFINE_SPINLOCK(prog_idr_lock); static DEFINE_IDR(map_idr); static DEFINE_SPINLOCK(map_idr_lock); +static DEFINE_IDR(link_idr); +static DEFINE_SPINLOCK(link_idr_lock); int sysctl_unprivileged_bpf_disabled __read_mostly; @@ -49,9 +51,11 @@ static const struct bpf_map_ops * const bpf_map_types[] = { #define BPF_PROG_TYPE(_id, _name, prog_ctx_type, kern_ctx_type) #define BPF_MAP_TYPE(_id, _ops) \ [_id] = &_ops, +#define BPF_LINK_TYPE(_id, _name) #include <linux/bpf_types.h> #undef BPF_PROG_TYPE #undef BPF_MAP_TYPE +#undef BPF_LINK_TYPE }; /* @@ -1546,9 +1550,11 @@ static const struct bpf_prog_ops * const bpf_prog_types[] = { #define BPF_PROG_TYPE(_id, _name, prog_ctx_type, kern_ctx_type) \ [_id] = & _name ## _prog_ops, #define BPF_MAP_TYPE(_id, _ops) +#define BPF_LINK_TYPE(_id, _name) #include <linux/bpf_types.h> #undef BPF_PROG_TYPE #undef BPF_MAP_TYPE +#undef BPF_LINK_TYPE }; static int find_prog_type(enum bpf_prog_type type, struct bpf_prog *prog) @@ -2181,25 +2187,39 @@ static int bpf_obj_get(const union bpf_attr *attr) attr->file_flags); } -void bpf_link_init(struct bpf_link *link, const struct bpf_link_ops *ops, - struct bpf_prog *prog) +void bpf_link_init(struct bpf_link *link, enum bpf_link_type type, + const struct bpf_link_ops *ops, struct bpf_prog *prog) { atomic64_set(&link->refcnt, 1); + link->type = type; + link->id = 0; link->ops = ops; link->prog = prog; } +static void bpf_link_free_id(int id) +{ + if (!id) + return; + + spin_lock_bh(&link_idr_lock); + idr_remove(&link_idr, id); + spin_unlock_bh(&link_idr_lock); +} + /* Clean up bpf_link and corresponding anon_inode file and FD. After * anon_inode is created, bpf_link can't be just kfree()'d due to deferred - * anon_inode's release() call. This helper manages marking bpf_link as - * defunct, releases anon_inode file and puts reserved FD. + * anon_inode's release() call. This helper marksbpf_link as + * defunct, releases anon_inode file and puts reserved FD. bpf_prog's refcnt + * is not decremented, it's the responsibility of a calling code that failed + * to complete bpf_link initialization. */ -void bpf_link_cleanup(struct bpf_link *link, struct file *link_file, - int link_fd) +void bpf_link_cleanup(struct bpf_link_primer *primer) { - link->prog = NULL; - fput(link_file); - put_unused_fd(link_fd); + primer->link->prog = NULL; + bpf_link_free_id(primer->id); + fput(primer->file); + put_unused_fd(primer->fd); } void bpf_link_inc(struct bpf_link *link) @@ -2210,6 +2230,7 @@ void bpf_link_inc(struct bpf_link *link) /* bpf_link_free is guaranteed to be called from process context */ static void bpf_link_free(struct bpf_link *link) { + bpf_link_free_id(link->id); if (link->prog) { /* detach BPF program, clean up used resources */ link->ops->release(link); @@ -2251,35 +2272,35 @@ static int bpf_link_release(struct inode *inode, struct file *filp) } #ifdef CONFIG_PROC_FS -static const struct bpf_link_ops bpf_raw_tp_lops; -static const struct bpf_link_ops bpf_tracing_link_lops; +#define BPF_PROG_TYPE(_id, _name, prog_ctx_type, kern_ctx_type) +#define BPF_MAP_TYPE(_id, _ops) +#define BPF_LINK_TYPE(_id, _name) [_id] = #_name, +static const char *bpf_link_type_strs[] = { + [BPF_LINK_TYPE_UNSPEC] = "<invalid>", +#include <linux/bpf_types.h> +}; +#undef BPF_PROG_TYPE +#undef BPF_MAP_TYPE +#undef BPF_LINK_TYPE static void bpf_link_show_fdinfo(struct seq_file *m, struct file *filp) { const struct bpf_link *link = filp->private_data; const struct bpf_prog *prog = link->prog; char prog_tag[sizeof(prog->tag) * 2 + 1] = { }; - const char *link_type; - - if (link->ops == &bpf_raw_tp_lops) - link_type = "raw_tracepoint"; - else if (link->ops == &bpf_tracing_link_lops) - link_type = "tracing"; -#ifdef CONFIG_CGROUP_BPF - else if (link->ops == &bpf_cgroup_link_lops) - link_type = "cgroup"; -#endif - else - link_type = "unknown"; bin2hex(prog_tag, prog->tag, sizeof(prog->tag)); seq_printf(m, "link_type:\t%s\n" + "link_id:\t%u\n" "prog_tag:\t%s\n" "prog_id:\t%u\n", - link_type, + bpf_link_type_strs[link->type], + link->id, prog_tag, prog->aux->id); + if (link->ops->show_fdinfo) + link->ops->show_fdinfo(link, m); } #endif @@ -2292,36 +2313,77 @@ static const struct file_operations bpf_link_fops = { .write = bpf_dummy_write, }; -int bpf_link_new_fd(struct bpf_link *link) +static int bpf_link_alloc_id(struct bpf_link *link) { - return anon_inode_getfd("bpf-link", &bpf_link_fops, link, O_CLOEXEC); -} + int id; -/* Similar to bpf_link_new_fd, create anon_inode for given bpf_link, but - * instead of immediately installing fd in fdtable, just reserve it and - * return. Caller then need to either install it with fd_install(fd, file) or - * release with put_unused_fd(fd). - * This is useful for cases when bpf_link attachment/detachment are - * complicated and expensive operations and should be delayed until all the fd - * reservation and anon_inode creation succeeds. + idr_preload(GFP_KERNEL); + spin_lock_bh(&link_idr_lock); + id = idr_alloc_cyclic(&link_idr, link, 1, INT_MAX, GFP_ATOMIC); + spin_unlock_bh(&link_idr_lock); + idr_preload_end(); + + return id; +} + +/* Prepare bpf_link to be exposed to user-space by allocating anon_inode file, + * reserving unused FD and allocating ID from link_idr. This is to be paired + * with bpf_link_settle() to install FD and ID and expose bpf_link to + * user-space, if bpf_link is successfully attached. If not, bpf_link and + * pre-allocated resources are to be freed with bpf_cleanup() call. All the + * transient state is passed around in struct bpf_link_primer. + * This is preferred way to create and initialize bpf_link, especially when + * there are complicated and expensive operations inbetween creating bpf_link + * itself and attaching it to BPF hook. By using bpf_link_prime() and + * bpf_link_settle() kernel code using bpf_link doesn't have to perform + * expensive (and potentially failing) roll back operations in a rare case + * that file, FD, or ID can't be allocated. */ -struct file *bpf_link_new_file(struct bpf_link *link, int *reserved_fd) +int bpf_link_prime(struct bpf_link *link, struct bpf_link_primer *primer) { struct file *file; - int fd; + int fd, id; fd = get_unused_fd_flags(O_CLOEXEC); if (fd < 0) - return ERR_PTR(fd); + return fd; + + + id = bpf_link_alloc_id(link); + if (id < 0) { + put_unused_fd(fd); + return id; + } file = anon_inode_getfile("bpf_link", &bpf_link_fops, link, O_CLOEXEC); if (IS_ERR(file)) { + bpf_link_free_id(id); put_unused_fd(fd); - return file; + return PTR_ERR(file); } - *reserved_fd = fd; - return file; + primer->link = link; + primer->file = file; + primer->fd = fd; + primer->id = id; + return 0; +} + +int bpf_link_settle(struct bpf_link_primer *primer) +{ + /* make bpf_link fetchable by ID */ + spin_lock_bh(&link_idr_lock); + primer->link->id = primer->id; + spin_unlock_bh(&link_idr_lock); + /* make bpf_link fetchable by FD */ + fd_install(primer->fd, primer->file); + /* pass through installed FD */ + return primer->fd; +} + +int bpf_link_new_fd(struct bpf_link *link) +{ + return anon_inode_getfd("bpf-link", &bpf_link_fops, link, O_CLOEXEC); } struct bpf_link *bpf_link_get_from_fd(u32 ufd) @@ -2345,6 +2407,7 @@ struct bpf_link *bpf_link_get_from_fd(u32 ufd) struct bpf_tracing_link { struct bpf_link link; + enum bpf_attach_type attach_type; }; static void bpf_tracing_link_release(struct bpf_link *link) @@ -2360,16 +2423,40 @@ static void bpf_tracing_link_dealloc(struct bpf_link *link) kfree(tr_link); } +static void bpf_tracing_link_show_fdinfo(const struct bpf_link *link, + struct seq_file *seq) +{ + struct bpf_tracing_link *tr_link = + container_of(link, struct bpf_tracing_link, link); + + seq_printf(seq, + "attach_type:\t%d\n", + tr_link->attach_type); +} + +static int bpf_tracing_link_fill_link_info(const struct bpf_link *link, + struct bpf_link_info *info) +{ + struct bpf_tracing_link *tr_link = + container_of(link, struct bpf_tracing_link, link); + + info->tracing.attach_type = tr_link->attach_type; + + return 0; +} + static const struct bpf_link_ops bpf_tracing_link_lops = { .release = bpf_tracing_link_release, .dealloc = bpf_tracing_link_dealloc, + .show_fdinfo = bpf_tracing_link_show_fdinfo, + .fill_link_info = bpf_tracing_link_fill_link_info, }; static int bpf_tracing_prog_attach(struct bpf_prog *prog) { + struct bpf_link_primer link_primer; struct bpf_tracing_link *link; - struct file *link_file; - int link_fd, err; + int err; switch (prog->type) { case BPF_PROG_TYPE_TRACING: @@ -2402,24 +2489,23 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog) err = -ENOMEM; goto out_put_prog; } - bpf_link_init(&link->link, &bpf_tracing_link_lops, prog); + bpf_link_init(&link->link, BPF_LINK_TYPE_TRACING, + &bpf_tracing_link_lops, prog); + link->attach_type = prog->expected_attach_type; - link_file = bpf_link_new_file(&link->link, &link_fd); - if (IS_ERR(link_file)) { + err = bpf_link_prime(&link->link, &link_primer); + if (err) { kfree(link); - err = PTR_ERR(link_file); goto out_put_prog; } err = bpf_trampoline_link_prog(prog); if (err) { - bpf_link_cleanup(&link->link, link_file, link_fd); + bpf_link_cleanup(&link_primer); goto out_put_prog; } - fd_install(link_fd, link_file); - return link_fd; - + return bpf_link_settle(&link_primer); out_put_prog: bpf_prog_put(prog); return err; @@ -2447,22 +2533,69 @@ static void bpf_raw_tp_link_dealloc(struct bpf_link *link) kfree(raw_tp); } -static const struct bpf_link_ops bpf_raw_tp_lops = { +static void bpf_raw_tp_link_show_fdinfo(const struct bpf_link *link, + struct seq_file *seq) +{ + struct bpf_raw_tp_link *raw_tp_link = + container_of(link, struct bpf_raw_tp_link, link); + + seq_printf(seq, + "tp_name:\t%s\n", + raw_tp_link->btp->tp->name); +} + +static int bpf_raw_tp_link_fill_link_info(const struct bpf_link *link, + struct bpf_link_info *info) +{ + struct bpf_raw_tp_link *raw_tp_link = + container_of(link, struct bpf_raw_tp_link, link); + char __user *ubuf = u64_to_user_ptr(info->raw_tracepoint.tp_name); + const char *tp_name = raw_tp_link->btp->tp->name; + u32 ulen = info->raw_tracepoint.tp_name_len; + size_t tp_len = strlen(tp_name); + + if (ulen && !ubuf) + return -EINVAL; + + info->raw_tracepoint.tp_name_len = tp_len + 1; + + if (!ubuf) + return 0; + + if (ulen >= tp_len + 1) { + if (copy_to_user(ubuf, tp_name, tp_len + 1)) + return -EFAULT; + } else { + char zero = '\0'; + + if (copy_to_user(ubuf, tp_name, ulen - 1)) + return -EFAULT; + if (put_user(zero, ubuf + ulen - 1)) + return -EFAULT; + return -ENOSPC; + } + + return 0; +} + +static const struct bpf_link_ops bpf_raw_tp_link_lops = { .release = bpf_raw_tp_link_release, .dealloc = bpf_raw_tp_link_dealloc, + .show_fdinfo = bpf_raw_tp_link_show_fdinfo, + .fill_link_info = bpf_raw_tp_link_fill_link_info, }; #define BPF_RAW_TRACEPOINT_OPEN_LAST_FIELD raw_tracepoint.prog_fd static int bpf_raw_tracepoint_open(const union bpf_attr *attr) { + struct bpf_link_primer link_primer; struct bpf_raw_tp_link *link; struct bpf_raw_event_map *btp; - struct file *link_file; struct bpf_prog *prog; const char *tp_name; char buf[128]; - int link_fd, err; + int err; if (CHECK_ATTR(BPF_RAW_TRACEPOINT_OPEN)) return -EINVAL; @@ -2515,24 +2648,23 @@ static int bpf_raw_tracepoint_open(const union bpf_attr *attr) err = -ENOMEM; goto out_put_btp; } - bpf_link_init(&link->link, &bpf_raw_tp_lops, prog); + bpf_link_init(&link->link, BPF_LINK_TYPE_RAW_TRACEPOINT, + &bpf_raw_tp_link_lops, prog); link->btp = btp; - link_file = bpf_link_new_file(&link->link, &link_fd); - if (IS_ERR(link_file)) { + err = bpf_link_prime(&link->link, &link_primer); + if (err) { kfree(link); - err = PTR_ERR(link_file); goto out_put_btp; } err = bpf_probe_register(link->btp, prog); if (err) { - bpf_link_cleanup(&link->link, link_file, link_fd); + bpf_link_cleanup(&link_primer); goto out_put_btp; } - fd_install(link_fd, link_file); - return link_fd; + return bpf_link_settle(&link_primer); out_put_btp: bpf_put_raw_tracepoint(btp); @@ -3313,6 +3445,42 @@ static int bpf_btf_get_info_by_fd(struct btf *btf, return btf_get_info_by_fd(btf, attr, uattr); } +static int bpf_link_get_info_by_fd(struct bpf_link *link, + const union bpf_attr *attr, + union bpf_attr __user *uattr) +{ + struct bpf_link_info __user *uinfo = u64_to_user_ptr(attr->info.info); + struct bpf_link_info info; + u32 info_len = attr->info.info_len; + int err; + + err = bpf_check_uarg_tail_zero(uinfo, sizeof(info), info_len); + if (err) + return err; + info_len = min_t(u32, sizeof(info), info_len); + + memset(&info, 0, sizeof(info)); + if (copy_from_user(&info, uinfo, info_len)) + return -EFAULT; + + info.type = link->type; + info.id = link->id; + info.prog_id = link->prog->aux->id; + + if (link->ops->fill_link_info) { + err = link->ops->fill_link_info(link, &info); + if (err) + return err; + } + + if (copy_to_user(uinfo, &info, info_len) || + put_user(info_len, &uattr->info.info_len)) + return -EFAULT; + + return 0; +} + + #define BPF_OBJ_GET_INFO_BY_FD_LAST_FIELD info.info static int bpf_obj_get_info_by_fd(const union bpf_attr *attr, @@ -3337,6 +3505,9 @@ static int bpf_obj_get_info_by_fd(const union bpf_attr *attr, uattr); else if (f.file->f_op == &btf_fops) err = bpf_btf_get_info_by_fd(f.file->private_data, attr, uattr); + else if (f.file->f_op == &bpf_link_fops) + err = bpf_link_get_info_by_fd(f.file->private_data, + attr, uattr); else err = -EINVAL; @@ -3464,7 +3635,7 @@ static int bpf_task_fd_query(const union bpf_attr *attr, if (file->f_op == &bpf_link_fops) { struct bpf_link *link = file->private_data; - if (link->ops == &bpf_raw_tp_lops) { + if (link->ops == &bpf_raw_tp_link_lops) { struct bpf_raw_tp_link *raw_tp = container_of(link, struct bpf_raw_tp_link, link); struct bpf_raw_event_map *btp = raw_tp->btp; @@ -3645,13 +3816,10 @@ static int link_update(union bpf_attr *attr) goto out_put_progs; } -#ifdef CONFIG_CGROUP_BPF - if (link->ops == &bpf_cgroup_link_lops) { - ret = cgroup_bpf_replace(link, old_prog, new_prog); - goto out_put_progs; - } -#endif - ret = -EINVAL; + if (link->ops->update_prog) + ret = link->ops->update_prog(link, new_prog, old_prog); + else + ret = EINVAL; out_put_progs: if (old_prog) @@ -3663,6 +3831,102 @@ out_put_link: return ret; } +static int bpf_link_inc_not_zero(struct bpf_link *link) +{ + return atomic64_fetch_add_unless(&link->refcnt, 1, 0) ? 0 : -ENOENT; +} + +#define BPF_LINK_GET_FD_BY_ID_LAST_FIELD link_id + +static int bpf_link_get_fd_by_id(const union bpf_attr *attr) +{ + struct bpf_link *link; + u32 id = attr->link_id; + int fd, err; + + if (CHECK_ATTR(BPF_LINK_GET_FD_BY_ID)) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + spin_lock_bh(&link_idr_lock); + link = idr_find(&link_idr, id); + /* before link is "settled", ID is 0, pretend it doesn't exist yet */ + if (link) { + if (link->id) + err = bpf_link_inc_not_zero(link); + else + err = -EAGAIN; + } else { + err = -ENOENT; + } + spin_unlock_bh(&link_idr_lock); + + if (err) + return err; + + fd = bpf_link_new_fd(link); + if (fd < 0) + bpf_link_put(link); + + return fd; +} + +DEFINE_MUTEX(bpf_stats_enabled_mutex); + +static int bpf_stats_release(struct inode *inode, struct file *file) +{ + mutex_lock(&bpf_stats_enabled_mutex); + static_key_slow_dec(&bpf_stats_enabled_key.key); + mutex_unlock(&bpf_stats_enabled_mutex); + return 0; +} + +static const struct file_operations bpf_stats_fops = { + .release = bpf_stats_release, +}; + +static int bpf_enable_runtime_stats(void) +{ + int fd; + + mutex_lock(&bpf_stats_enabled_mutex); + + /* Set a very high limit to avoid overflow */ + if (static_key_count(&bpf_stats_enabled_key.key) > INT_MAX / 2) { + mutex_unlock(&bpf_stats_enabled_mutex); + return -EBUSY; + } + + fd = anon_inode_getfd("bpf-stats", &bpf_stats_fops, NULL, O_CLOEXEC); + if (fd >= 0) + static_key_slow_inc(&bpf_stats_enabled_key.key); + + mutex_unlock(&bpf_stats_enabled_mutex); + return fd; +} + +#define BPF_ENABLE_STATS_LAST_FIELD enable_stats.type + +static int bpf_enable_stats(union bpf_attr *attr) +{ + + if (CHECK_ATTR(BPF_ENABLE_STATS)) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + switch (attr->enable_stats.type) { + case BPF_STATS_RUN_TIME: + return bpf_enable_runtime_stats(); + default: + break; + } + return -EINVAL; +} + SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, size) { union bpf_attr attr; @@ -3780,6 +4044,16 @@ SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, siz case BPF_LINK_UPDATE: err = link_update(&attr); break; + case BPF_LINK_GET_FD_BY_ID: + err = bpf_link_get_fd_by_id(&attr); + break; + case BPF_LINK_GET_NEXT_ID: + err = bpf_obj_get_next_id(&attr, uattr, + &link_idr, &link_idr_lock); + break; + case BPF_ENABLE_STATS: + err = bpf_enable_stats(&attr); + break; default: err = -EINVAL; break; |