From 1f415a74b0ca64b5bfacbb12d71ed2ec050a8cfb Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 2 Aug 2016 16:12:14 +0100 Subject: bpf: fix method of PTR_TO_PACKET reg id generation Using per-register incrementing ID can lead to find_good_pkt_pointers() confusing registers which have completely different values. Consider example: 0: (bf) r6 = r1 1: (61) r8 = *(u32 *)(r6 +76) 2: (61) r0 = *(u32 *)(r6 +80) 3: (bf) r7 = r8 4: (07) r8 += 32 5: (2d) if r8 > r0 goto pc+9 R0=pkt_end R1=ctx R6=ctx R7=pkt(id=0,off=0,r=32) R8=pkt(id=0,off=32,r=32) R10=fp 6: (bf) r8 = r7 7: (bf) r9 = r7 8: (71) r1 = *(u8 *)(r7 +0) 9: (0f) r8 += r1 10: (71) r1 = *(u8 *)(r7 +1) 11: (0f) r9 += r1 12: (07) r8 += 32 13: (2d) if r8 > r0 goto pc+1 R0=pkt_end R1=inv56 R6=ctx R7=pkt(id=0,off=0,r=32) R8=pkt(id=1,off=32,r=32) R9=pkt(id=1,off=0,r=32) R10=fp 14: (71) r1 = *(u8 *)(r9 +16) 15: (b7) r7 = 0 16: (bf) r0 = r7 17: (95) exit We need to get a UNKNOWN_VALUE with imm to force id generation so lines 0-5 make r7 a valid packet pointer. We then read two different bytes from the packet and add them to copies of the constructed packet pointer. r8 (line 9) and r9 (line 11) will get the same id of 1, independently. When either of them is validated (line 13) - find_good_pkt_pointers() will also mark the other as safe. This leads to access on line 14 being mistakenly considered safe. Fixes: 969bf05eb3ce ("bpf: direct packet access") Signed-off-by: Jakub Kicinski Acked-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/verifier.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index f72f23b8fdab..7094c69ac199 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -194,6 +194,7 @@ struct verifier_env { struct verifier_state_list **explored_states; /* search pruning optimization */ struct bpf_map *used_maps[MAX_USED_MAPS]; /* array of map's used by eBPF program */ u32 used_map_cnt; /* number of used maps */ + u32 id_gen; /* used to generate unique reg IDs */ bool allow_ptr_leaks; }; @@ -1301,7 +1302,7 @@ add_imm: /* dst_reg stays as pkt_ptr type and since some positive * integer value was added to the pointer, increment its 'id' */ - dst_reg->id++; + dst_reg->id = ++env->id_gen; /* something was added to pkt_ptr, set range and off to zero */ dst_reg->off = 0; -- cgit v1.2.3-55-g7522 From a6ed3ea65d9868fdf9eff84e6fe4f666b8d14b02 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Fri, 5 Aug 2016 14:01:27 -0700 Subject: bpf: restore behavior of bpf_map_update_elem The introduction of pre-allocated hash elements inadvertently broke the behavior of bpf hash maps where users expected to call bpf_map_update_elem() without considering that the map can be full. Some programs do: old_value = bpf_map_lookup_elem(map, key); if (old_value) { ... prepare new_value on stack ... bpf_map_update_elem(map, key, new_value); } Before pre-alloc the update() for existing element would work even in 'map full' condition. Restore this behavior. The above program could have updated old_value in place instead of update() which would be faster and most programs use that approach, but sometimes the values are large and the programs use update() helper to do atomic replacement of the element. Note we cannot simply update element's value in-place like percpu hash map does and have to allocate extra num_possible_cpu elements and use this extra reserve when the map is full. Fixes: 6c9059817432 ("bpf: pre-allocate hash map elements") Signed-off-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/hashtab.c | 84 +++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 73 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/bpf/hashtab.c b/kernel/bpf/hashtab.c index fff3650d52fc..570eeca7bdfa 100644 --- a/kernel/bpf/hashtab.c +++ b/kernel/bpf/hashtab.c @@ -26,11 +26,18 @@ struct bpf_htab { struct bucket *buckets; void *elems; struct pcpu_freelist freelist; + void __percpu *extra_elems; atomic_t count; /* number of elements in this hashtable */ u32 n_buckets; /* number of hash buckets */ u32 elem_size; /* size of each element in bytes */ }; +enum extra_elem_state { + HTAB_NOT_AN_EXTRA_ELEM = 0, + HTAB_EXTRA_ELEM_FREE, + HTAB_EXTRA_ELEM_USED +}; + /* each htab element is struct htab_elem + key + value */ struct htab_elem { union { @@ -38,7 +45,10 @@ struct htab_elem { struct bpf_htab *htab; struct pcpu_freelist_node fnode; }; - struct rcu_head rcu; + union { + struct rcu_head rcu; + enum extra_elem_state state; + }; u32 hash; char key[0] __aligned(8); }; @@ -113,6 +123,23 @@ free_elems: return err; } +static int alloc_extra_elems(struct bpf_htab *htab) +{ + void __percpu *pptr; + int cpu; + + pptr = __alloc_percpu_gfp(htab->elem_size, 8, GFP_USER | __GFP_NOWARN); + if (!pptr) + return -ENOMEM; + + for_each_possible_cpu(cpu) { + ((struct htab_elem *)per_cpu_ptr(pptr, cpu))->state = + HTAB_EXTRA_ELEM_FREE; + } + htab->extra_elems = pptr; + return 0; +} + /* Called from syscall */ static struct bpf_map *htab_map_alloc(union bpf_attr *attr) { @@ -185,6 +212,8 @@ static struct bpf_map *htab_map_alloc(union bpf_attr *attr) if (percpu) cost += (u64) round_up(htab->map.value_size, 8) * num_possible_cpus() * htab->map.max_entries; + else + cost += (u64) htab->elem_size * num_possible_cpus(); if (cost >= U32_MAX - PAGE_SIZE) /* make sure page count doesn't overflow */ @@ -212,14 +241,22 @@ static struct bpf_map *htab_map_alloc(union bpf_attr *attr) raw_spin_lock_init(&htab->buckets[i].lock); } + if (!percpu) { + err = alloc_extra_elems(htab); + if (err) + goto free_buckets; + } + if (!(attr->map_flags & BPF_F_NO_PREALLOC)) { err = prealloc_elems_and_freelist(htab); if (err) - goto free_buckets; + goto free_extra_elems; } return &htab->map; +free_extra_elems: + free_percpu(htab->extra_elems); free_buckets: kvfree(htab->buckets); free_htab: @@ -349,7 +386,6 @@ static void htab_elem_free(struct bpf_htab *htab, struct htab_elem *l) if (htab->map.map_type == BPF_MAP_TYPE_PERCPU_HASH) free_percpu(htab_elem_get_ptr(l, htab->map.key_size)); kfree(l); - } static void htab_elem_free_rcu(struct rcu_head *head) @@ -370,6 +406,11 @@ static void htab_elem_free_rcu(struct rcu_head *head) static void free_htab_elem(struct bpf_htab *htab, struct htab_elem *l) { + if (l->state == HTAB_EXTRA_ELEM_USED) { + l->state = HTAB_EXTRA_ELEM_FREE; + return; + } + if (!(htab->map.map_flags & BPF_F_NO_PREALLOC)) { pcpu_freelist_push(&htab->freelist, &l->fnode); } else { @@ -381,25 +422,44 @@ static void free_htab_elem(struct bpf_htab *htab, struct htab_elem *l) static struct htab_elem *alloc_htab_elem(struct bpf_htab *htab, void *key, void *value, u32 key_size, u32 hash, - bool percpu, bool onallcpus) + bool percpu, bool onallcpus, + bool old_elem_exists) { u32 size = htab->map.value_size; bool prealloc = !(htab->map.map_flags & BPF_F_NO_PREALLOC); struct htab_elem *l_new; void __percpu *pptr; + int err = 0; if (prealloc) { l_new = (struct htab_elem *)pcpu_freelist_pop(&htab->freelist); if (!l_new) - return ERR_PTR(-E2BIG); + err = -E2BIG; } else { if (atomic_inc_return(&htab->count) > htab->map.max_entries) { atomic_dec(&htab->count); - return ERR_PTR(-E2BIG); + err = -E2BIG; + } else { + l_new = kmalloc(htab->elem_size, + GFP_ATOMIC | __GFP_NOWARN); + if (!l_new) + return ERR_PTR(-ENOMEM); } - l_new = kmalloc(htab->elem_size, GFP_ATOMIC | __GFP_NOWARN); - if (!l_new) - return ERR_PTR(-ENOMEM); + } + + if (err) { + if (!old_elem_exists) + return ERR_PTR(err); + + /* if we're updating the existing element and the hash table + * is full, use per-cpu extra elems + */ + l_new = this_cpu_ptr(htab->extra_elems); + if (l_new->state != HTAB_EXTRA_ELEM_FREE) + return ERR_PTR(-E2BIG); + l_new->state = HTAB_EXTRA_ELEM_USED; + } else { + l_new->state = HTAB_NOT_AN_EXTRA_ELEM; } memcpy(l_new->key, key, key_size); @@ -489,7 +549,8 @@ static int htab_map_update_elem(struct bpf_map *map, void *key, void *value, if (ret) goto err; - l_new = alloc_htab_elem(htab, key, value, key_size, hash, false, false); + l_new = alloc_htab_elem(htab, key, value, key_size, hash, false, false, + !!l_old); if (IS_ERR(l_new)) { /* all pre-allocated elements are in use or memory exhausted */ ret = PTR_ERR(l_new); @@ -563,7 +624,7 @@ static int __htab_percpu_map_update_elem(struct bpf_map *map, void *key, } } else { l_new = alloc_htab_elem(htab, key, value, key_size, - hash, true, onallcpus); + hash, true, onallcpus, false); if (IS_ERR(l_new)) { ret = PTR_ERR(l_new); goto err; @@ -652,6 +713,7 @@ static void htab_map_free(struct bpf_map *map) htab_free_elems(htab); pcpu_freelist_destroy(&htab->freelist); } + free_percpu(htab->extra_elems); kvfree(htab->buckets); kfree(htab); } -- cgit v1.2.3-55-g7522 From 747ea55e4f78fd980350c39570a986b8c1c3e4aa Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Fri, 12 Aug 2016 22:17:17 +0200 Subject: bpf: fix bpf_skb_in_cgroup helper naming While hashing out BPF's current_task_under_cgroup helper bits, it came to discussion that the skb_in_cgroup helper name was suboptimally chosen. Tejun says: So, I think in_cgroup should mean that the object is in that particular cgroup while under_cgroup in the subhierarchy of that cgroup. Let's rename the other subhierarchy test to under too. I think that'd be a lot less confusing going forward. [...] It's more intuitive and gives us the room to implement the real "in" test if ever necessary in the future. Since this touches uapi bits, we need to change this as long as v4.8 is not yet officially released. Thus, change the helper enum and rename related bits. Fixes: 4a482f34afcc ("cgroup: bpf: Add bpf_skb_in_cgroup_proto") Reference: http://patchwork.ozlabs.org/patch/658500/ Suggested-by: Sargun Dhillon Suggested-by: Tejun Heo Signed-off-by: Daniel Borkmann Acked-by: Alexei Starovoitov --- include/uapi/linux/bpf.h | 4 ++-- kernel/bpf/verifier.c | 4 ++-- net/core/filter.c | 10 +++++----- samples/bpf/bpf_helpers.h | 4 ++-- samples/bpf/test_cgrp2_tc_kern.c | 2 +- 5 files changed, 12 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index da218fec6056..9e5fc168c8a3 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -339,7 +339,7 @@ enum bpf_func_id { BPF_FUNC_skb_change_type, /** - * bpf_skb_in_cgroup(skb, map, index) - Check cgroup2 membership of skb + * bpf_skb_under_cgroup(skb, map, index) - Check cgroup2 membership of skb * @skb: pointer to skb * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type * @index: index of the cgroup in the bpf_map @@ -348,7 +348,7 @@ enum bpf_func_id { * == 1 skb succeeded the cgroup2 descendant test * < 0 error */ - BPF_FUNC_skb_in_cgroup, + BPF_FUNC_skb_under_cgroup, /** * bpf_get_hash_recalc(skb) diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 7094c69ac199..daea765d72e6 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -1053,7 +1053,7 @@ static int check_map_func_compatibility(struct bpf_map *map, int func_id) goto error; break; case BPF_MAP_TYPE_CGROUP_ARRAY: - if (func_id != BPF_FUNC_skb_in_cgroup) + if (func_id != BPF_FUNC_skb_under_cgroup) goto error; break; default: @@ -1075,7 +1075,7 @@ static int check_map_func_compatibility(struct bpf_map *map, int func_id) if (map->map_type != BPF_MAP_TYPE_STACK_TRACE) goto error; break; - case BPF_FUNC_skb_in_cgroup: + case BPF_FUNC_skb_under_cgroup: if (map->map_type != BPF_MAP_TYPE_CGROUP_ARRAY) goto error; break; diff --git a/net/core/filter.c b/net/core/filter.c index b5add4ef0d1d..bd9bf2e5fafa 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -2317,7 +2317,7 @@ bpf_get_skb_set_tunnel_proto(enum bpf_func_id which) } #ifdef CONFIG_SOCK_CGROUP_DATA -static u64 bpf_skb_in_cgroup(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5) +static u64 bpf_skb_under_cgroup(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5) { struct sk_buff *skb = (struct sk_buff *)(long)r1; struct bpf_map *map = (struct bpf_map *)(long)r2; @@ -2340,8 +2340,8 @@ static u64 bpf_skb_in_cgroup(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5) return cgroup_is_descendant(sock_cgroup_ptr(&sk->sk_cgrp_data), cgrp); } -static const struct bpf_func_proto bpf_skb_in_cgroup_proto = { - .func = bpf_skb_in_cgroup, +static const struct bpf_func_proto bpf_skb_under_cgroup_proto = { + .func = bpf_skb_under_cgroup, .gpl_only = false, .ret_type = RET_INTEGER, .arg1_type = ARG_PTR_TO_CTX, @@ -2421,8 +2421,8 @@ tc_cls_act_func_proto(enum bpf_func_id func_id) case BPF_FUNC_get_smp_processor_id: return &bpf_get_smp_processor_id_proto; #ifdef CONFIG_SOCK_CGROUP_DATA - case BPF_FUNC_skb_in_cgroup: - return &bpf_skb_in_cgroup_proto; + case BPF_FUNC_skb_under_cgroup: + return &bpf_skb_under_cgroup_proto; #endif default: return sk_filter_func_proto(func_id); diff --git a/samples/bpf/bpf_helpers.h b/samples/bpf/bpf_helpers.h index 217c8d507f2e..7927a090fa0d 100644 --- a/samples/bpf/bpf_helpers.h +++ b/samples/bpf/bpf_helpers.h @@ -72,8 +72,8 @@ static int (*bpf_l3_csum_replace)(void *ctx, int off, int from, int to, int flag (void *) BPF_FUNC_l3_csum_replace; static int (*bpf_l4_csum_replace)(void *ctx, int off, int from, int to, int flags) = (void *) BPF_FUNC_l4_csum_replace; -static int (*bpf_skb_in_cgroup)(void *ctx, void *map, int index) = - (void *) BPF_FUNC_skb_in_cgroup; +static int (*bpf_skb_under_cgroup)(void *ctx, void *map, int index) = + (void *) BPF_FUNC_skb_under_cgroup; #if defined(__x86_64__) diff --git a/samples/bpf/test_cgrp2_tc_kern.c b/samples/bpf/test_cgrp2_tc_kern.c index 2732c37c8d5b..10ff73404e3a 100644 --- a/samples/bpf/test_cgrp2_tc_kern.c +++ b/samples/bpf/test_cgrp2_tc_kern.c @@ -57,7 +57,7 @@ int handle_egress(struct __sk_buff *skb) bpf_trace_printk(dont_care_msg, sizeof(dont_care_msg), eth->h_proto, ip6h->nexthdr); return TC_ACT_OK; - } else if (bpf_skb_in_cgroup(skb, &test_cgrp2_array_pin, 0) != 1) { + } else if (bpf_skb_under_cgroup(skb, &test_cgrp2_array_pin, 0) != 1) { bpf_trace_printk(pass_msg, sizeof(pass_msg)); return TC_ACT_OK; } else { -- cgit v1.2.3-55-g7522