Add a cookie to BPF maps to uniquely identify BPF maps for the timespan when the node is up. This is different to comparing a pointer or BPF map id which could get rolled over and reused. Signed-off-by: Daniel Borkmann --- include/linux/bpf.h | 1 + kernel/bpf/syscall.c | 6 ++++++ 2 files changed, 7 insertions(+) diff --git a/include/linux/bpf.h b/include/linux/bpf.h index f9cd2164ed23..308530c8326b 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -310,6 +310,7 @@ struct bpf_map { bool free_after_rcu_gp; atomic64_t sleepable_refcnt; s64 __percpu *elem_count; + u64 cookie; /* write-once */ }; static inline const char *btf_field_type_name(enum btf_field_type type) diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index e63039817af3..7a814e98d5f5 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include @@ -53,6 +54,7 @@ #define BPF_OBJ_FLAG_MASK (BPF_F_RDONLY | BPF_F_WRONLY) DEFINE_PER_CPU(int, bpf_prog_active); +DEFINE_COOKIE(bpf_map_cookie); static DEFINE_IDR(prog_idr); static DEFINE_SPINLOCK(prog_idr_lock); static DEFINE_IDR(map_idr); @@ -1487,6 +1489,10 @@ static int map_create(union bpf_attr *attr, bool kernel) if (err < 0) goto free_map; + preempt_disable(); + map->cookie = gen_cookie_next(&bpf_map_cookie); + preempt_enable(); + atomic64_set(&map->refcnt, 1); atomic64_set(&map->usercnt, 1); mutex_init(&map->freeze_mutex); -- 2.43.0 Given this is only relevant for BPF tail call maps, it is adding up space and penalizing other map types. We also need to extend this with further objects to track / compare to. Therefore, lets move this out into a separate structure and dynamically allocate it only for BPF tail call maps. Signed-off-by: Daniel Borkmann --- include/linux/bpf.h | 36 ++++++++++++++++++++++++------------ kernel/bpf/core.c | 35 ++++++++++++++++++----------------- kernel/bpf/syscall.c | 13 +++++++------ 3 files changed, 49 insertions(+), 35 deletions(-) diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 308530c8326b..a87646cc5398 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -260,6 +260,18 @@ struct bpf_list_node_kern { void *owner; } __attribute__((aligned(8))); +/* 'Ownership' of program-containing map is claimed by the first program + * that is going to use this map or by the first program which FD is + * stored in the map to make sure that all callers and callees have the + * same prog type, JITed flag and xdp_has_frags flag. + */ +struct bpf_map_owner { + enum bpf_prog_type type; + bool jited; + bool xdp_has_frags; + const struct btf_type *attach_func_proto; +}; + struct bpf_map { const struct bpf_map_ops *ops; struct bpf_map *inner_map_meta; @@ -292,18 +304,8 @@ struct bpf_map { struct rcu_head rcu; }; atomic64_t writecnt; - /* 'Ownership' of program-containing map is claimed by the first program - * that is going to use this map or by the first program which FD is - * stored in the map to make sure that all callers and callees have the - * same prog type, JITed flag and xdp_has_frags flag. - */ - struct { - const struct btf_type *attach_func_proto; - spinlock_t lock; - enum bpf_prog_type type; - bool jited; - bool xdp_has_frags; - } owner; + spinlock_t owner_lock; + struct bpf_map_owner *owner; bool bypass_spec_v1; bool frozen; /* write-once; write-protected by freeze_mutex */ bool free_after_mult_rcu_gp; @@ -2109,6 +2111,16 @@ static inline bool bpf_map_flags_access_ok(u32 access_flags) (BPF_F_RDONLY_PROG | BPF_F_WRONLY_PROG); } +static inline struct bpf_map_owner *bpf_map_owner_alloc(struct bpf_map *map) +{ + return kzalloc(sizeof(*map->owner), GFP_ATOMIC); +} + +static inline void bpf_map_owner_free(struct bpf_map *map) +{ + kfree(map->owner); +} + struct bpf_event_entry { struct perf_event *event; struct file *perf_file; diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 61613785bdd0..666d4019d87f 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -2370,28 +2370,29 @@ static bool __bpf_prog_map_compatible(struct bpf_map *map, const struct bpf_prog *fp) { enum bpf_prog_type prog_type = resolve_prog_type(fp); - bool ret; struct bpf_prog_aux *aux = fp->aux; + bool ret = false; if (fp->kprobe_override) - return false; + return ret; - spin_lock(&map->owner.lock); - if (!map->owner.type) { - /* There's no owner yet where we could check for - * compatibility. - */ - map->owner.type = prog_type; - map->owner.jited = fp->jited; - map->owner.xdp_has_frags = aux->xdp_has_frags; - map->owner.attach_func_proto = aux->attach_func_proto; + spin_lock(&map->owner_lock); + /* There's no owner yet where we could check for compatibility. */ + if (!map->owner) { + map->owner = bpf_map_owner_alloc(map); + if (!map->owner) + goto err; + map->owner->type = prog_type; + map->owner->jited = fp->jited; + map->owner->xdp_has_frags = aux->xdp_has_frags; + map->owner->attach_func_proto = aux->attach_func_proto; ret = true; } else { - ret = map->owner.type == prog_type && - map->owner.jited == fp->jited && - map->owner.xdp_has_frags == aux->xdp_has_frags; + ret = map->owner->type == prog_type && + map->owner->jited == fp->jited && + map->owner->xdp_has_frags == aux->xdp_has_frags; if (ret && - map->owner.attach_func_proto != aux->attach_func_proto) { + map->owner->attach_func_proto != aux->attach_func_proto) { switch (prog_type) { case BPF_PROG_TYPE_TRACING: case BPF_PROG_TYPE_LSM: @@ -2404,8 +2405,8 @@ static bool __bpf_prog_map_compatible(struct bpf_map *map, } } } - spin_unlock(&map->owner.lock); - +err: + spin_unlock(&map->owner_lock); return ret; } diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index 7a814e98d5f5..0fbfa8532c39 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -887,6 +887,7 @@ static void bpf_map_free_deferred(struct work_struct *work) security_bpf_map_free(map); bpf_map_release_memcg(map); + bpf_map_owner_free(map); bpf_map_free(map); } @@ -981,12 +982,12 @@ static void bpf_map_show_fdinfo(struct seq_file *m, struct file *filp) struct bpf_map *map = filp->private_data; u32 type = 0, jited = 0; - if (map_type_contains_progs(map)) { - spin_lock(&map->owner.lock); - type = map->owner.type; - jited = map->owner.jited; - spin_unlock(&map->owner.lock); + spin_lock(&map->owner_lock); + if (map->owner) { + type = map->owner->type; + jited = map->owner->jited; } + spin_unlock(&map->owner_lock); seq_printf(m, "map_type:\t%u\n" @@ -1496,7 +1497,7 @@ static int map_create(union bpf_attr *attr, bool kernel) atomic64_set(&map->refcnt, 1); atomic64_set(&map->usercnt, 1); mutex_init(&map->freeze_mutex); - spin_lock_init(&map->owner.lock); + spin_lock_init(&map->owner_lock); if (attr->btf_key_type_id || attr->btf_value_type_id || /* Even the map's value is a kernel's struct, -- 2.43.0 Move them into bpf.h given we also need them in core code, add also for_each_cgroup_storage_type_cond() which we'll be using in a subsequent change. Signed-off-by: Daniel Borkmann --- include/linux/bpf-cgroup.h | 3 --- include/linux/bpf.h | 20 ++++++++++++-------- 2 files changed, 12 insertions(+), 11 deletions(-) diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h index 082ccd8ad96b..d1f01d1168a1 100644 --- a/include/linux/bpf-cgroup.h +++ b/include/linux/bpf-cgroup.h @@ -77,9 +77,6 @@ to_cgroup_bpf_attach_type(enum bpf_attach_type attach_type) extern struct static_key_false cgroup_bpf_enabled_key[MAX_CGROUP_BPF_ATTACH_TYPE]; #define cgroup_bpf_enabled(atype) static_branch_unlikely(&cgroup_bpf_enabled_key[atype]) -#define for_each_cgroup_storage_type(stype) \ - for (stype = 0; stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++) - struct bpf_cgroup_storage_map; struct bpf_storage_buffer { diff --git a/include/linux/bpf.h b/include/linux/bpf.h index a87646cc5398..a41dff574327 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -208,6 +208,18 @@ enum btf_field_type { BPF_RES_SPIN_LOCK = (1 << 12), }; +enum bpf_cgroup_storage_type { + BPF_CGROUP_STORAGE_SHARED, + BPF_CGROUP_STORAGE_PERCPU, + __BPF_CGROUP_STORAGE_MAX +#define MAX_BPF_CGROUP_STORAGE_TYPE __BPF_CGROUP_STORAGE_MAX +}; + +#define for_each_cgroup_storage_type(stype) \ + for (stype = 0; stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++) +#define for_each_cgroup_storage_type_cond(stype, cond) \ + for (stype = 0; (cond) && stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++) + typedef void (*btf_dtor_kfunc_t)(void *); struct btf_field_kptr { @@ -1085,14 +1097,6 @@ struct bpf_prog_offload { u32 jited_len; }; -enum bpf_cgroup_storage_type { - BPF_CGROUP_STORAGE_SHARED, - BPF_CGROUP_STORAGE_PERCPU, - __BPF_CGROUP_STORAGE_MAX -}; - -#define MAX_BPF_CGROUP_STORAGE_TYPE __BPF_CGROUP_STORAGE_MAX - /* The longest tracepoint has 12 args. * See include/trace/bpf_probe.h */ -- 2.43.0 Lonial reported that an out-of-bounds access in cgroup local storage can be crafted via tail calls. Given two programs each utilizing a cgroup local storage with a different value size, and one program doing a tail call into the other. The verifier will validate each of the indivial programs just fine. However, in the runtime context the bpf_cg_run_ctx holds an bpf_prog_array_item which contains the BPF program as well as any cgroup local storage flavor the program uses. Helpers such as bpf_get_local_storage() pick this up from the runtime context: ctx = container_of(current->bpf_ctx, struct bpf_cg_run_ctx, run_ctx); storage = ctx->prog_item->cgroup_storage[stype]; if (stype == BPF_CGROUP_STORAGE_SHARED) ptr = &READ_ONCE(storage->buf)->data[0]; else ptr = this_cpu_ptr(storage->percpu_buf); For the second program which was called from the originally attached one, this means bpf_get_local_storage() will pick up the former program's map, not its own. With mismatching sizes, this can result in an unintended out-of-bounds access. To fix this issue, we need to extend bpf_map_owner with an array of storage_cookie[] to match on i) the exact maps from the original program if the second program was using bpf_get_local_storage(), or ii) allow the tail call combination if the second program was not using any of the cgroup local storage maps. Fixes: 7d9c3427894f ("bpf: Make cgroup storages shared between programs on the same cgroup") Reported-by: Lonial Con Signed-off-by: Daniel Borkmann --- include/linux/bpf.h | 1 + kernel/bpf/core.c | 11 +++++++++++ 2 files changed, 12 insertions(+) diff --git a/include/linux/bpf.h b/include/linux/bpf.h index a41dff574327..63169b1611cc 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -281,6 +281,7 @@ struct bpf_map_owner { enum bpf_prog_type type; bool jited; bool xdp_has_frags; + u64 storage_cookie[MAX_BPF_CGROUP_STORAGE_TYPE]; const struct btf_type *attach_func_proto; }; diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 666d4019d87f..7a8201763936 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -2371,6 +2371,7 @@ static bool __bpf_prog_map_compatible(struct bpf_map *map, { enum bpf_prog_type prog_type = resolve_prog_type(fp); struct bpf_prog_aux *aux = fp->aux; + enum bpf_cgroup_storage_type i; bool ret = false; if (fp->kprobe_override) @@ -2386,11 +2387,21 @@ static bool __bpf_prog_map_compatible(struct bpf_map *map, map->owner->jited = fp->jited; map->owner->xdp_has_frags = aux->xdp_has_frags; map->owner->attach_func_proto = aux->attach_func_proto; + for_each_cgroup_storage_type(i) { + map->owner->storage_cookie[i] = + aux->cgroup_storage[i] ? + aux->cgroup_storage[i]->cookie : 0; + } ret = true; } else { ret = map->owner->type == prog_type && map->owner->jited == fp->jited && map->owner->xdp_has_frags == aux->xdp_has_frags; + for_each_cgroup_storage_type_cond(i, ret) { + u64 cookie = aux->cgroup_storage[i] ? + aux->cgroup_storage[i]->cookie : 0; + ret = map->owner->storage_cookie[i] == cookie || !cookie; + } if (ret && map->owner->attach_func_proto != aux->attach_func_proto) { switch (prog_type) { -- 2.43.0