From b2197755b2633e164a439682fb05a9b5ea48f706 Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Thu, 29 Oct 2015 14:58:09 +0100 Subject: bpf: add support for persistent maps/progs This work adds support for "persistent" eBPF maps/programs. The term "persistent" is to be understood that maps/programs have a facility that lets them survive process termination. This is desired by various eBPF subsystem users. Just to name one example: tc classifier/action. Whenever tc parses the ELF object, extracts and loads maps/progs into the kernel, these file descriptors will be out of reach after the tc instance exits. So a subsequent tc invocation won't be able to access/relocate on this resource, and therefore maps cannot easily be shared, f.e. between the ingress and egress networking data path. The current workaround is that Unix domain sockets (UDS) need to be instrumented in order to pass the created eBPF map/program file descriptors to a third party management daemon through UDS' socket passing facility. This makes it a bit complicated to deploy shared eBPF maps or programs (programs f.e. for tail calls) among various processes. We've been brainstorming on how we could tackle this issue and various approches have been tried out so far, which can be read up further in the below reference. The architecture we eventually ended up with is a minimal file system that can hold map/prog objects. The file system is a per mount namespace singleton, and the default mount point is /sys/fs/bpf/. Any subsequent mounts within a given namespace will point to the same instance. The file system allows for creating a user-defined directory structure. The objects for maps/progs are created/fetched through bpf(2) with two new commands (BPF_OBJ_PIN/BPF_OBJ_GET). I.e. a bpf file descriptor along with a pathname is being passed to bpf(2) that in turn creates (we call it eBPF object pinning) the file system nodes. Only the pathname is being passed to bpf(2) for getting a new BPF file descriptor to an existing node. The user can use that to access maps and progs later on, through bpf(2). Removal of file system nodes is being managed through normal VFS functions such as unlink(2), etc. The file system code is kept to a very minimum and can be further extended later on. The next step I'm working on is to add dump eBPF map/prog commands to bpf(2), so that a specification from a given file descriptor can be retrieved. This can be used by things like CRIU but also applications can inspect the meta data after calling BPF_OBJ_GET. Big thanks also to Alexei and Hannes who significantly contributed in the design discussion that eventually let us end up with this architecture here. Reference: https://lkml.org/lkml/2015/10/15/925 Signed-off-by: Daniel Borkmann Signed-off-by: Alexei Starovoitov Signed-off-by: Hannes Frederic Sowa Signed-off-by: David S. Miller --- kernel/bpf/inode.c | 387 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 387 insertions(+) create mode 100644 kernel/bpf/inode.c (limited to 'kernel/bpf/inode.c') diff --git a/kernel/bpf/inode.c b/kernel/bpf/inode.c new file mode 100644 index 000000000000..be6d726e31c9 --- /dev/null +++ b/kernel/bpf/inode.c @@ -0,0 +1,387 @@ +/* + * Minimal file system backend for holding eBPF maps and programs, + * used by bpf(2) object pinning. + * + * Authors: + * + * Daniel Borkmann + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * version 2 as published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +enum bpf_type { + BPF_TYPE_UNSPEC = 0, + BPF_TYPE_PROG, + BPF_TYPE_MAP, +}; + +static void *bpf_any_get(void *raw, enum bpf_type type) +{ + switch (type) { + case BPF_TYPE_PROG: + atomic_inc(&((struct bpf_prog *)raw)->aux->refcnt); + break; + case BPF_TYPE_MAP: + atomic_inc(&((struct bpf_map *)raw)->refcnt); + break; + default: + WARN_ON_ONCE(1); + break; + } + + return raw; +} + +static void bpf_any_put(void *raw, enum bpf_type type) +{ + switch (type) { + case BPF_TYPE_PROG: + bpf_prog_put(raw); + break; + case BPF_TYPE_MAP: + bpf_map_put(raw); + break; + default: + WARN_ON_ONCE(1); + break; + } +} + +static void *bpf_fd_probe_obj(u32 ufd, enum bpf_type *type) +{ + void *raw; + + *type = BPF_TYPE_MAP; + raw = bpf_map_get(ufd); + if (IS_ERR(raw)) { + *type = BPF_TYPE_PROG; + raw = bpf_prog_get(ufd); + } + + return raw; +} + +static const struct inode_operations bpf_dir_iops; + +static const struct inode_operations bpf_prog_iops = { }; +static const struct inode_operations bpf_map_iops = { }; + +static struct inode *bpf_get_inode(struct super_block *sb, + const struct inode *dir, + umode_t mode) +{ + struct inode *inode; + + switch (mode & S_IFMT) { + case S_IFDIR: + case S_IFREG: + break; + default: + return ERR_PTR(-EINVAL); + } + + inode = new_inode(sb); + if (!inode) + return ERR_PTR(-ENOSPC); + + inode->i_ino = get_next_ino(); + inode->i_atime = CURRENT_TIME; + inode->i_mtime = inode->i_atime; + inode->i_ctime = inode->i_atime; + + inode_init_owner(inode, dir, mode); + + return inode; +} + +static int bpf_inode_type(const struct inode *inode, enum bpf_type *type) +{ + *type = BPF_TYPE_UNSPEC; + if (inode->i_op == &bpf_prog_iops) + *type = BPF_TYPE_PROG; + else if (inode->i_op == &bpf_map_iops) + *type = BPF_TYPE_MAP; + else + return -EACCES; + + return 0; +} + +static bool bpf_dname_reserved(const struct dentry *dentry) +{ + return strchr(dentry->d_name.name, '.'); +} + +static int bpf_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode) +{ + struct inode *inode; + + if (bpf_dname_reserved(dentry)) + return -EPERM; + + inode = bpf_get_inode(dir->i_sb, dir, mode | S_IFDIR); + if (IS_ERR(inode)) + return PTR_ERR(inode); + + inode->i_op = &bpf_dir_iops; + inode->i_fop = &simple_dir_operations; + + inc_nlink(inode); + inc_nlink(dir); + + d_instantiate(dentry, inode); + dget(dentry); + + return 0; +} + +static int bpf_mkobj_ops(struct inode *dir, struct dentry *dentry, + umode_t mode, const struct inode_operations *iops) +{ + struct inode *inode; + + if (bpf_dname_reserved(dentry)) + return -EPERM; + + inode = bpf_get_inode(dir->i_sb, dir, mode | S_IFREG); + if (IS_ERR(inode)) + return PTR_ERR(inode); + + inode->i_op = iops; + inode->i_private = dentry->d_fsdata; + + d_instantiate(dentry, inode); + dget(dentry); + + return 0; +} + +static int bpf_mkobj(struct inode *dir, struct dentry *dentry, umode_t mode, + dev_t devt) +{ + enum bpf_type type = MINOR(devt); + + if (MAJOR(devt) != UNNAMED_MAJOR || !S_ISREG(mode) || + dentry->d_fsdata == NULL) + return -EPERM; + + switch (type) { + case BPF_TYPE_PROG: + return bpf_mkobj_ops(dir, dentry, mode, &bpf_prog_iops); + case BPF_TYPE_MAP: + return bpf_mkobj_ops(dir, dentry, mode, &bpf_map_iops); + default: + return -EPERM; + } +} + +static const struct inode_operations bpf_dir_iops = { + .lookup = simple_lookup, + .mknod = bpf_mkobj, + .mkdir = bpf_mkdir, + .rmdir = simple_rmdir, + .unlink = simple_unlink, +}; + +static int bpf_obj_do_pin(const struct filename *pathname, void *raw, + enum bpf_type type) +{ + struct dentry *dentry; + struct inode *dir; + struct path path; + umode_t mode; + dev_t devt; + int ret; + + dentry = kern_path_create(AT_FDCWD, pathname->name, &path, 0); + if (IS_ERR(dentry)) + return PTR_ERR(dentry); + + mode = S_IFREG | ((S_IRUSR | S_IWUSR) & ~current_umask()); + devt = MKDEV(UNNAMED_MAJOR, type); + + ret = security_path_mknod(&path, dentry, mode, devt); + if (ret) + goto out; + + dir = d_inode(path.dentry); + if (dir->i_op != &bpf_dir_iops) { + ret = -EPERM; + goto out; + } + + dentry->d_fsdata = raw; + ret = vfs_mknod(dir, dentry, mode, devt); + dentry->d_fsdata = NULL; +out: + done_path_create(&path, dentry); + return ret; +} + +int bpf_obj_pin_user(u32 ufd, const char __user *pathname) +{ + struct filename *pname; + enum bpf_type type; + void *raw; + int ret; + + pname = getname(pathname); + if (IS_ERR(pname)) + return PTR_ERR(pname); + + raw = bpf_fd_probe_obj(ufd, &type); + if (IS_ERR(raw)) { + ret = PTR_ERR(raw); + goto out; + } + + ret = bpf_obj_do_pin(pname, raw, type); + if (ret != 0) + bpf_any_put(raw, type); +out: + putname(pname); + return ret; +} + +static void *bpf_obj_do_get(const struct filename *pathname, + enum bpf_type *type) +{ + struct inode *inode; + struct path path; + void *raw; + int ret; + + ret = kern_path(pathname->name, LOOKUP_FOLLOW, &path); + if (ret) + return ERR_PTR(ret); + + inode = d_backing_inode(path.dentry); + ret = inode_permission(inode, MAY_WRITE); + if (ret) + goto out; + + ret = bpf_inode_type(inode, type); + if (ret) + goto out; + + raw = bpf_any_get(inode->i_private, *type); + touch_atime(&path); + + path_put(&path); + return raw; +out: + path_put(&path); + return ERR_PTR(ret); +} + +int bpf_obj_get_user(const char __user *pathname) +{ + enum bpf_type type = BPF_TYPE_UNSPEC; + struct filename *pname; + int ret = -ENOENT; + void *raw; + + pname = getname(pathname); + if (IS_ERR(pname)) + return PTR_ERR(pname); + + raw = bpf_obj_do_get(pname, &type); + if (IS_ERR(raw)) { + ret = PTR_ERR(raw); + goto out; + } + + if (type == BPF_TYPE_PROG) + ret = bpf_prog_new_fd(raw); + else if (type == BPF_TYPE_MAP) + ret = bpf_map_new_fd(raw); + else + goto out; + + if (ret < 0) + bpf_any_put(raw, type); +out: + putname(pname); + return ret; +} + +static void bpf_evict_inode(struct inode *inode) +{ + enum bpf_type type; + + truncate_inode_pages_final(&inode->i_data); + clear_inode(inode); + + if (!bpf_inode_type(inode, &type)) + bpf_any_put(inode->i_private, type); +} + +static const struct super_operations bpf_super_ops = { + .statfs = simple_statfs, + .drop_inode = generic_delete_inode, + .evict_inode = bpf_evict_inode, +}; + +static int bpf_fill_super(struct super_block *sb, void *data, int silent) +{ + static struct tree_descr bpf_rfiles[] = { { "" } }; + struct inode *inode; + int ret; + + ret = simple_fill_super(sb, BPF_FS_MAGIC, bpf_rfiles); + if (ret) + return ret; + + sb->s_op = &bpf_super_ops; + + inode = sb->s_root->d_inode; + inode->i_op = &bpf_dir_iops; + inode->i_mode &= ~S_IALLUGO; + inode->i_mode |= S_ISVTX | S_IRWXUGO; + + return 0; +} + +static struct dentry *bpf_mount(struct file_system_type *type, int flags, + const char *dev_name, void *data) +{ + return mount_ns(type, flags, current->nsproxy->mnt_ns, bpf_fill_super); +} + +static struct file_system_type bpf_fs_type = { + .owner = THIS_MODULE, + .name = "bpf", + .mount = bpf_mount, + .kill_sb = kill_litter_super, + .fs_flags = FS_USERNS_MOUNT, +}; + +MODULE_ALIAS_FS("bpf"); + +static int __init bpf_init(void) +{ + int ret; + + ret = sysfs_create_mount_point(fs_kobj, "bpf"); + if (ret) + return ret; + + ret = register_filesystem(&bpf_fs_type); + if (ret) + sysfs_remove_mount_point(fs_kobj, "bpf"); + + return ret; +} +fs_initcall(bpf_init); -- cgit v1.2.3 From c9da161c6517ba12154059d3b965c2cbaf16f90f Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Tue, 24 Nov 2015 21:28:15 +0100 Subject: bpf: fix clearing on persistent program array maps Currently, when having map file descriptors pointing to program arrays, there's still the issue that we unconditionally flush program array contents via bpf_fd_array_map_clear() in bpf_map_release(). This happens when such a file descriptor is released and is independent of the map's refcount. Having this flush independent of the refcount is for a reason: there can be arbitrary complex dependency chains among tail calls, also circular ones (direct or indirect, nesting limit determined during runtime), and we need to make sure that the map drops all references to eBPF programs it holds, so that the map's refcount can eventually drop to zero and initiate its freeing. Btw, a walk of the whole dependency graph would not be possible for various reasons, one being complexity and another one inconsistency, i.e. new programs can be added to parts of the graph at any time, so there's no guaranteed consistent state for the time of such a walk. Now, the program array pinning itself works, but the issue is that each derived file descriptor on close would nevertheless call unconditionally into bpf_fd_array_map_clear(). Instead, keep track of users and postpone this flush until the last reference to a user is dropped. As this only concerns a subset of references (f.e. a prog array could hold a program that itself has reference on the prog array holding it, etc), we need to track them separately. Short analysis on the refcounting: on map creation time usercnt will be one, so there's no change in behaviour for bpf_map_release(), if unpinned. If we already fail in map_create(), we are immediately freed, and no file descriptor has been made public yet. In bpf_obj_pin_user(), we need to probe for a possible map in bpf_fd_probe_obj() already with a usercnt reference, so before we drop the reference on the fd with fdput(). Therefore, if actual pinning fails, we need to drop that reference again in bpf_any_put(), otherwise we keep holding it. When last reference drops on the inode, the bpf_any_put() in bpf_evict_inode() will take care of dropping the usercnt again. In the bpf_obj_get_user() case, the bpf_any_get() will grab a reference on the usercnt, still at a time when we have the reference on the path. Should we later on fail to grab a new file descriptor, bpf_any_put() will drop it, otherwise we hold it until bpf_map_release() time. Joint work with Alexei. Fixes: b2197755b263 ("bpf: add support for persistent maps/progs") Signed-off-by: Daniel Borkmann Signed-off-by: Alexei Starovoitov Signed-off-by: David S. Miller --- include/linux/bpf.h | 5 ++++- kernel/bpf/inode.c | 6 +++--- kernel/bpf/syscall.c | 36 +++++++++++++++++++++++++----------- kernel/bpf/verifier.c | 3 +-- 4 files changed, 33 insertions(+), 17 deletions(-) (limited to 'kernel/bpf/inode.c') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index de464e6683b6..83d1926c61e4 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -40,6 +40,7 @@ struct bpf_map { struct user_struct *user; const struct bpf_map_ops *ops; struct work_struct work; + atomic_t usercnt; }; struct bpf_map_type_list { @@ -167,8 +168,10 @@ struct bpf_prog *bpf_prog_get(u32 ufd); void bpf_prog_put(struct bpf_prog *prog); void bpf_prog_put_rcu(struct bpf_prog *prog); -struct bpf_map *bpf_map_get(u32 ufd); +struct bpf_map *bpf_map_get_with_uref(u32 ufd); struct bpf_map *__bpf_map_get(struct fd f); +void bpf_map_inc(struct bpf_map *map, bool uref); +void bpf_map_put_with_uref(struct bpf_map *map); void bpf_map_put(struct bpf_map *map); extern int sysctl_unprivileged_bpf_disabled; diff --git a/kernel/bpf/inode.c b/kernel/bpf/inode.c index be6d726e31c9..5a8a797d50b7 100644 --- a/kernel/bpf/inode.c +++ b/kernel/bpf/inode.c @@ -34,7 +34,7 @@ static void *bpf_any_get(void *raw, enum bpf_type type) atomic_inc(&((struct bpf_prog *)raw)->aux->refcnt); break; case BPF_TYPE_MAP: - atomic_inc(&((struct bpf_map *)raw)->refcnt); + bpf_map_inc(raw, true); break; default: WARN_ON_ONCE(1); @@ -51,7 +51,7 @@ static void bpf_any_put(void *raw, enum bpf_type type) bpf_prog_put(raw); break; case BPF_TYPE_MAP: - bpf_map_put(raw); + bpf_map_put_with_uref(raw); break; default: WARN_ON_ONCE(1); @@ -64,7 +64,7 @@ static void *bpf_fd_probe_obj(u32 ufd, enum bpf_type *type) void *raw; *type = BPF_TYPE_MAP; - raw = bpf_map_get(ufd); + raw = bpf_map_get_with_uref(ufd); if (IS_ERR(raw)) { *type = BPF_TYPE_PROG; raw = bpf_prog_get(ufd); diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index 0d3313d02a7e..4a8f3c1d7da6 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -82,6 +82,14 @@ static void bpf_map_free_deferred(struct work_struct *work) map->ops->map_free(map); } +static void bpf_map_put_uref(struct bpf_map *map) +{ + if (atomic_dec_and_test(&map->usercnt)) { + if (map->map_type == BPF_MAP_TYPE_PROG_ARRAY) + bpf_fd_array_map_clear(map); + } +} + /* decrement map refcnt and schedule it for freeing via workqueue * (unrelying map implementation ops->map_free() might sleep) */ @@ -93,17 +101,15 @@ void bpf_map_put(struct bpf_map *map) } } -static int bpf_map_release(struct inode *inode, struct file *filp) +void bpf_map_put_with_uref(struct bpf_map *map) { - struct bpf_map *map = filp->private_data; - - if (map->map_type == BPF_MAP_TYPE_PROG_ARRAY) - /* prog_array stores refcnt-ed bpf_prog pointers - * release them all when user space closes prog_array_fd - */ - bpf_fd_array_map_clear(map); - + bpf_map_put_uref(map); bpf_map_put(map); +} + +static int bpf_map_release(struct inode *inode, struct file *filp) +{ + bpf_map_put_with_uref(filp->private_data); return 0; } @@ -142,6 +148,7 @@ static int map_create(union bpf_attr *attr) return PTR_ERR(map); atomic_set(&map->refcnt, 1); + atomic_set(&map->usercnt, 1); err = bpf_map_charge_memlock(map); if (err) @@ -174,7 +181,14 @@ struct bpf_map *__bpf_map_get(struct fd f) return f.file->private_data; } -struct bpf_map *bpf_map_get(u32 ufd) +void bpf_map_inc(struct bpf_map *map, bool uref) +{ + atomic_inc(&map->refcnt); + if (uref) + atomic_inc(&map->usercnt); +} + +struct bpf_map *bpf_map_get_with_uref(u32 ufd) { struct fd f = fdget(ufd); struct bpf_map *map; @@ -183,7 +197,7 @@ struct bpf_map *bpf_map_get(u32 ufd) if (IS_ERR(map)) return map; - atomic_inc(&map->refcnt); + bpf_map_inc(map, true); fdput(f); return map; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index c6073056badf..a7945d10b378 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -2021,8 +2021,7 @@ static int replace_map_fd_with_map_ptr(struct verifier_env *env) * will be used by the valid program until it's unloaded * and all maps are released in free_bpf_prog_info() */ - atomic_inc(&map->refcnt); - + bpf_map_inc(map, false); fdput(f); next_insn: insn++; -- cgit v1.2.3