Replace deprecated 0-length array in struct bpf_lpm_trie_key with
flexible array. Found with GCC 13:
./kernel/bpf/lpm_trie.c:207:51: warning: array subscript i is outside array bounds of 'const __u8[0]' {aka 'const unsigned char[]'} [-Warray-bounds=]
207 | *(__be16 *)&key->data[i]);
| ^~~~~~~~~~~~~
./include/uapi/linux/swab.h:102:54: note: in definition of macro '__swab16'
102 | #define __swab16(x) (__u16)__builtin_bswap16((__u16)(x))
| ^
./include/linux/byteorder/generic.h:97:21: note: in expansion of macro '__be16_to_cpu'
97 | #define be16_to_cpu __be16_to_cpu
| ^~~~~~~~~~~~~
./kernel/bpf/lpm_trie.c:206:28: note: in expansion of macro 'be16_to_cpu'
206 | u16 diff = be16_to_cpu(*(__be16 *)&node->data[i]
^
| ^~~~~~~~~~~
In file included from ../include/linux/bpf.h:7:
./include/uapi/linux/bpf.h:82:17: note: while referencing 'data'
82 | __u8 data[0]; /* Arbitrary size */
| ^~~~
And found at run-time under CONFIG_FORTIFY_SOURCE:
UBSAN: array-index-out-of-bounds in kernel/bpf/lpm_trie.c:218:49
index 0 is out of range for type '__u8 [*]'
Changing struct bpf_lpm_trie_key is difficult since has been used by
userspace. For example, in Cilium:
struct egress_gw_policy_key {
struct bpf_lpm_trie_key lpm_key;
__u32 saddr;
__u32 daddr;
};
While direct references to the "data" member haven't been found, there
are static initializers what include the final member. For example,
the "{}" here:
struct egress_gw_policy_key in_key = {
.lpm_key = { 32 + 24, {} },
.saddr = CLIENT_IP,
.daddr = EXTERNAL_SVC_IP & 0Xffffff,
};
To avoid the build time and run time warnings seen with a 0-sized
trailing array for struct bpf_lpm_trie_key, introduce a new struct
that correctly uses a flexible array for the trailing bytes,
struct bpf_lpm_trie_key_u8. As part of this, include the "header"
portion (which is just the "prefixlen" member), so it can be used
by anything building a bpf_lpr_trie_key that has trailing members that
aren't a u8 flexible array (like the self-test[1]), which is named
struct bpf_lpm_trie_key_hdr.
Adjust the kernel code to use struct bpf_lpm_trie_key_u8 through-out,
and for the selftest to use struct bpf_lpm_trie_key_hdr. Add a comment
to the UAPI header directing folks to the two new options.
Link: https://lore.kernel.org/all/202206281009.4332AA33@keescook/ [1]
Reported-by: Mark Rutland <[email protected]>
Closes: https://paste.debian.net/hidden/ca500597/
Acked-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Kees Cook <[email protected]>
---
Cc: Daniel Borkmann <[email protected]>
Cc: Alexei Starovoitov <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Martin KaFai Lau <[email protected]>
Cc: Song Liu <[email protected]>
Cc: Yonghong Song <[email protected]>
Cc: John Fastabend <[email protected]>
Cc: KP Singh <[email protected]>
Cc: Stanislav Fomichev <[email protected]>
Cc: Hao Luo <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mykola Lysenko <[email protected]>
Cc: Shuah Khan <[email protected]>
Cc: Haowen Bai <[email protected]>
Cc: [email protected]
Cc: [email protected]
v4- add missing stddef.h include
v3- https://lore.kernel.org/all/[email protected]/
v2- https://lore.kernel.org/lkml/[email protected]/
v1- https://lore.kernel.org/lkml/[email protected]/
---
Documentation/bpf/map_lpm_trie.rst | 2 +-
include/uapi/linux/bpf.h | 15 +++++++++++++-
kernel/bpf/lpm_trie.c | 20 +++++++++----------
samples/bpf/map_perf_test_user.c | 2 +-
samples/bpf/xdp_router_ipv4_user.c | 2 +-
tools/include/uapi/linux/bpf.h | 15 +++++++++++++-
.../selftests/bpf/progs/map_ptr_kern.c | 2 +-
tools/testing/selftests/bpf/test_lpm_map.c | 18 ++++++++---------
8 files changed, 51 insertions(+), 25 deletions(-)
diff --git a/Documentation/bpf/map_lpm_trie.rst b/Documentation/bpf/map_lpm_trie.rst
index 74d64a30f500..f9cd579496c9 100644
--- a/Documentation/bpf/map_lpm_trie.rst
+++ b/Documentation/bpf/map_lpm_trie.rst
@@ -17,7 +17,7 @@ significant byte.
LPM tries may be created with a maximum prefix length that is a multiple
of 8, in the range from 8 to 2048. The key used for lookup and update
-operations is a ``struct bpf_lpm_trie_key``, extended by
+operations is a ``struct bpf_lpm_trie_key_u8``, extended by
``max_prefixlen/8`` bytes.
- For IPv4 addresses the data length is 4 bytes
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 754e68ca8744..31e9bdd4641e 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -8,6 +8,7 @@
#ifndef _UAPI__LINUX_BPF_H__
#define _UAPI__LINUX_BPF_H__
+#include <linux/stddef.h>
#include <linux/types.h>
#include <linux/bpf_common.h>
@@ -77,12 +78,24 @@ struct bpf_insn {
__s32 imm; /* signed immediate constant */
};
-/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */
+/* Deprecated: use struct bpf_lpm_trie_key_u8 (when the "data" member is needed for
+ * byte access) or struct bpf_lpm_trie_key_hdr (when using an alternative type for
+ * the trailing flexible array member) instead.
+ */
struct bpf_lpm_trie_key {
__u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */
__u8 data[0]; /* Arbitrary size */
};
+/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry, with trailing byte array. */
+struct bpf_lpm_trie_key_u8 {
+ __struct_group(bpf_lpm_trie_key_hdr, hdr, /* no attrs */,
+ /* up to 32 for AF_INET, 128 for AF_INET6 */
+ __u32 prefixlen;
+ );
+ __u8 data[]; /* Arbitrary size */
+};
+
struct bpf_cgroup_storage_key {
__u64 cgroup_inode_id; /* cgroup inode id */
__u32 attach_type; /* program attach type (enum bpf_attach_type) */
diff --git a/kernel/bpf/lpm_trie.c b/kernel/bpf/lpm_trie.c
index b32be680da6c..050fe1ebf0f7 100644
--- a/kernel/bpf/lpm_trie.c
+++ b/kernel/bpf/lpm_trie.c
@@ -164,13 +164,13 @@ static inline int extract_bit(const u8 *data, size_t index)
*/
static size_t longest_prefix_match(const struct lpm_trie *trie,
const struct lpm_trie_node *node,
- const struct bpf_lpm_trie_key *key)
+ const struct bpf_lpm_trie_key_u8 *key)
{
u32 limit = min(node->prefixlen, key->prefixlen);
u32 prefixlen = 0, i = 0;
BUILD_BUG_ON(offsetof(struct lpm_trie_node, data) % sizeof(u32));
- BUILD_BUG_ON(offsetof(struct bpf_lpm_trie_key, data) % sizeof(u32));
+ BUILD_BUG_ON(offsetof(struct bpf_lpm_trie_key_u8, data) % sizeof(u32));
#if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && defined(CONFIG_64BIT)
@@ -229,7 +229,7 @@ static void *trie_lookup_elem(struct bpf_map *map, void *_key)
{
struct lpm_trie *trie = container_of(map, struct lpm_trie, map);
struct lpm_trie_node *node, *found = NULL;
- struct bpf_lpm_trie_key *key = _key;
+ struct bpf_lpm_trie_key_u8 *key = _key;
if (key->prefixlen > trie->max_prefixlen)
return NULL;
@@ -309,7 +309,7 @@ static long trie_update_elem(struct bpf_map *map,
struct lpm_trie *trie = container_of(map, struct lpm_trie, map);
struct lpm_trie_node *node, *im_node = NULL, *new_node = NULL;
struct lpm_trie_node __rcu **slot;
- struct bpf_lpm_trie_key *key = _key;
+ struct bpf_lpm_trie_key_u8 *key = _key;
unsigned long irq_flags;
unsigned int next_bit;
size_t matchlen = 0;
@@ -437,7 +437,7 @@ static long trie_update_elem(struct bpf_map *map,
static long trie_delete_elem(struct bpf_map *map, void *_key)
{
struct lpm_trie *trie = container_of(map, struct lpm_trie, map);
- struct bpf_lpm_trie_key *key = _key;
+ struct bpf_lpm_trie_key_u8 *key = _key;
struct lpm_trie_node __rcu **trim, **trim2;
struct lpm_trie_node *node, *parent;
unsigned long irq_flags;
@@ -536,7 +536,7 @@ static long trie_delete_elem(struct bpf_map *map, void *_key)
sizeof(struct lpm_trie_node))
#define LPM_VAL_SIZE_MIN 1
-#define LPM_KEY_SIZE(X) (sizeof(struct bpf_lpm_trie_key) + (X))
+#define LPM_KEY_SIZE(X) (sizeof(struct bpf_lpm_trie_key_u8) + (X))
#define LPM_KEY_SIZE_MAX LPM_KEY_SIZE(LPM_DATA_SIZE_MAX)
#define LPM_KEY_SIZE_MIN LPM_KEY_SIZE(LPM_DATA_SIZE_MIN)
@@ -565,7 +565,7 @@ static struct bpf_map *trie_alloc(union bpf_attr *attr)
/* copy mandatory map attributes */
bpf_map_init_from_attr(&trie->map, attr);
trie->data_size = attr->key_size -
- offsetof(struct bpf_lpm_trie_key, data);
+ offsetof(struct bpf_lpm_trie_key_u8, data);
trie->max_prefixlen = trie->data_size * 8;
spin_lock_init(&trie->lock);
@@ -616,7 +616,7 @@ static int trie_get_next_key(struct bpf_map *map, void *_key, void *_next_key)
{
struct lpm_trie_node *node, *next_node = NULL, *parent, *search_root;
struct lpm_trie *trie = container_of(map, struct lpm_trie, map);
- struct bpf_lpm_trie_key *key = _key, *next_key = _next_key;
+ struct bpf_lpm_trie_key_u8 *key = _key, *next_key = _next_key;
struct lpm_trie_node **node_stack = NULL;
int err = 0, stack_ptr = -1;
unsigned int next_bit;
@@ -703,7 +703,7 @@ static int trie_get_next_key(struct bpf_map *map, void *_key, void *_next_key)
}
do_copy:
next_key->prefixlen = next_node->prefixlen;
- memcpy((void *)next_key + offsetof(struct bpf_lpm_trie_key, data),
+ memcpy((void *)next_key + offsetof(struct bpf_lpm_trie_key_u8, data),
next_node->data, trie->data_size);
free_stack:
kfree(node_stack);
@@ -715,7 +715,7 @@ static int trie_check_btf(const struct bpf_map *map,
const struct btf_type *key_type,
const struct btf_type *value_type)
{
- /* Keys must have struct bpf_lpm_trie_key embedded. */
+ /* Keys must have struct bpf_lpm_trie_key_u8 embedded. */
return BTF_INFO_KIND(key_type->info) != BTF_KIND_STRUCT ?
-EINVAL : 0;
}
diff --git a/samples/bpf/map_perf_test_user.c b/samples/bpf/map_perf_test_user.c
index d2fbcf963cdf..07ff471ed6ae 100644
--- a/samples/bpf/map_perf_test_user.c
+++ b/samples/bpf/map_perf_test_user.c
@@ -370,7 +370,7 @@ static void run_perf_test(int tasks)
static void fill_lpm_trie(void)
{
- struct bpf_lpm_trie_key *key;
+ struct bpf_lpm_trie_key_u8 *key;
unsigned long value = 0;
unsigned int i;
int r;
diff --git a/samples/bpf/xdp_router_ipv4_user.c b/samples/bpf/xdp_router_ipv4_user.c
index 9d41db09c480..266fdd0b025d 100644
--- a/samples/bpf/xdp_router_ipv4_user.c
+++ b/samples/bpf/xdp_router_ipv4_user.c
@@ -91,7 +91,7 @@ static int recv_msg(struct sockaddr_nl sock_addr, int sock)
static void read_route(struct nlmsghdr *nh, int nll)
{
char dsts[24], gws[24], ifs[16], dsts_len[24], metrics[24];
- struct bpf_lpm_trie_key *prefix_key;
+ struct bpf_lpm_trie_key_u8 *prefix_key;
struct rtattr *rt_attr;
struct rtmsg *rt_msg;
int rtm_family;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 7f24d898efbb..76b1b41c0dce 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -8,6 +8,7 @@
#ifndef _UAPI__LINUX_BPF_H__
#define _UAPI__LINUX_BPF_H__
+#include <linux/stddef.h>
#include <linux/types.h>
#include <linux/bpf_common.h>
@@ -77,12 +78,24 @@ struct bpf_insn {
__s32 imm; /* signed immediate constant */
};
-/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */
+/* Deprecated: use struct bpf_lpm_trie_key_u8 (when the "data" member is needed for
+ * byte access) or struct bpf_lpm_trie_key_hdr (when using an alternative type for
+ * the trailing flexible array member) instead.
+ */
struct bpf_lpm_trie_key {
__u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */
__u8 data[0]; /* Arbitrary size */
};
+/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry, with trailing byte array. */
+struct bpf_lpm_trie_key_u8 {
+ __struct_group(bpf_lpm_trie_key_hdr, hdr, /* no attrs */,
+ /* up to 32 for AF_INET, 128 for AF_INET6 */
+ __u32 prefixlen;
+ );
+ __u8 data[]; /* Arbitrary size */
+};
+
struct bpf_cgroup_storage_key {
__u64 cgroup_inode_id; /* cgroup inode id */
__u32 attach_type; /* program attach type (enum bpf_attach_type) */
diff --git a/tools/testing/selftests/bpf/progs/map_ptr_kern.c b/tools/testing/selftests/bpf/progs/map_ptr_kern.c
index 3325da17ec81..efaf622c28dd 100644
--- a/tools/testing/selftests/bpf/progs/map_ptr_kern.c
+++ b/tools/testing/selftests/bpf/progs/map_ptr_kern.c
@@ -316,7 +316,7 @@ struct lpm_trie {
} __attribute__((preserve_access_index));
struct lpm_key {
- struct bpf_lpm_trie_key trie_key;
+ struct bpf_lpm_trie_key_hdr trie_key;
__u32 data;
};
diff --git a/tools/testing/selftests/bpf/test_lpm_map.c b/tools/testing/selftests/bpf/test_lpm_map.c
index c028d621c744..d98c72dc563e 100644
--- a/tools/testing/selftests/bpf/test_lpm_map.c
+++ b/tools/testing/selftests/bpf/test_lpm_map.c
@@ -211,7 +211,7 @@ static void test_lpm_map(int keysize)
volatile size_t n_matches, n_matches_after_delete;
size_t i, j, n_nodes, n_lookups;
struct tlpm_node *t, *list = NULL;
- struct bpf_lpm_trie_key *key;
+ struct bpf_lpm_trie_key_u8 *key;
uint8_t *data, *value;
int r, map;
@@ -331,8 +331,8 @@ static void test_lpm_map(int keysize)
static void test_lpm_ipaddr(void)
{
LIBBPF_OPTS(bpf_map_create_opts, opts, .map_flags = BPF_F_NO_PREALLOC);
- struct bpf_lpm_trie_key *key_ipv4;
- struct bpf_lpm_trie_key *key_ipv6;
+ struct bpf_lpm_trie_key_u8 *key_ipv4;
+ struct bpf_lpm_trie_key_u8 *key_ipv6;
size_t key_size_ipv4;
size_t key_size_ipv6;
int map_fd_ipv4;
@@ -423,7 +423,7 @@ static void test_lpm_ipaddr(void)
static void test_lpm_delete(void)
{
LIBBPF_OPTS(bpf_map_create_opts, opts, .map_flags = BPF_F_NO_PREALLOC);
- struct bpf_lpm_trie_key *key;
+ struct bpf_lpm_trie_key_u8 *key;
size_t key_size;
int map_fd;
__u64 value;
@@ -532,7 +532,7 @@ static void test_lpm_delete(void)
static void test_lpm_get_next_key(void)
{
LIBBPF_OPTS(bpf_map_create_opts, opts, .map_flags = BPF_F_NO_PREALLOC);
- struct bpf_lpm_trie_key *key_p, *next_key_p;
+ struct bpf_lpm_trie_key_u8 *key_p, *next_key_p;
size_t key_size;
__u32 value = 0;
int map_fd;
@@ -693,9 +693,9 @@ static void *lpm_test_command(void *arg)
{
int i, j, ret, iter, key_size;
struct lpm_mt_test_info *info = arg;
- struct bpf_lpm_trie_key *key_p;
+ struct bpf_lpm_trie_key_u8 *key_p;
- key_size = sizeof(struct bpf_lpm_trie_key) + sizeof(__u32);
+ key_size = sizeof(*key_p) + sizeof(__u32);
key_p = alloca(key_size);
for (iter = 0; iter < info->iter; iter++)
for (i = 0; i < MAX_TEST_KEYS; i++) {
@@ -717,7 +717,7 @@ static void *lpm_test_command(void *arg)
ret = bpf_map_lookup_elem(info->map_fd, key_p, &value);
assert(ret == 0 || errno == ENOENT);
} else {
- struct bpf_lpm_trie_key *next_key_p = alloca(key_size);
+ struct bpf_lpm_trie_key_u8 *next_key_p = alloca(key_size);
ret = bpf_map_get_next_key(info->map_fd, key_p, next_key_p);
assert(ret == 0 || errno == ENOENT || errno == ENOMEM);
}
@@ -752,7 +752,7 @@ static void test_lpm_multi_thread(void)
/* create a trie */
value_size = sizeof(__u32);
- key_size = sizeof(struct bpf_lpm_trie_key) + value_size;
+ key_size = sizeof(struct bpf_lpm_trie_key_hdr) + value_size;
map_fd = bpf_map_create(BPF_MAP_TYPE_LPM_TRIE, NULL, key_size, value_size, 100, &opts);
/* create 4 threads to test update, delete, lookup and get_next_key */
--
2.34.1
On 2/20/24 7:54 PM, Kees Cook wrote:
> Replace deprecated 0-length array in struct bpf_lpm_trie_key with
> flexible array. Found with GCC 13:
>
> ../kernel/bpf/lpm_trie.c:207:51: warning: array subscript i is outside array bounds of 'const __u8[0]' {aka 'const unsigned char[]'} [-Warray-bounds=]
> 207 | *(__be16 *)&key->data[i]);
> | ^~~~~~~~~~~~~
> ../include/uapi/linux/swab.h:102:54: note: in definition of macro '__swab16'
> 102 | #define __swab16(x) (__u16)__builtin_bswap16((__u16)(x))
> | ^
> ../include/linux/byteorder/generic.h:97:21: note: in expansion of macro '__be16_to_cpu'
> 97 | #define be16_to_cpu __be16_to_cpu
> | ^~~~~~~~~~~~~
> ../kernel/bpf/lpm_trie.c:206:28: note: in expansion of macro 'be16_to_cpu'
> 206 | u16 diff = be16_to_cpu(*(__be16 *)&node->data[i]
> ^
> | ^~~~~~~~~~~
> In file included from ../include/linux/bpf.h:7:
> ../include/uapi/linux/bpf.h:82:17: note: while referencing 'data'
> 82 | __u8 data[0]; /* Arbitrary size */
> | ^~~~
>
> And found at run-time under CONFIG_FORTIFY_SOURCE:
>
> UBSAN: array-index-out-of-bounds in kernel/bpf/lpm_trie.c:218:49
> index 0 is out of range for type '__u8 [*]'
>
> Changing struct bpf_lpm_trie_key is difficult since has been used by
> userspace. For example, in Cilium:
>
> struct egress_gw_policy_key {
> struct bpf_lpm_trie_key lpm_key;
> __u32 saddr;
> __u32 daddr;
> };
>
> While direct references to the "data" member haven't been found, there
> are static initializers what include the final member. For example,
> the "{}" here:
>
> struct egress_gw_policy_key in_key = {
> .lpm_key = { 32 + 24, {} },
> .saddr = CLIENT_IP,
> .daddr = EXTERNAL_SVC_IP & 0Xffffff,
> };
>
> To avoid the build time and run time warnings seen with a 0-sized
> trailing array for struct bpf_lpm_trie_key, introduce a new struct
> that correctly uses a flexible array for the trailing bytes,
> struct bpf_lpm_trie_key_u8. As part of this, include the "header"
> portion (which is just the "prefixlen" member), so it can be used
> by anything building a bpf_lpr_trie_key that has trailing members that
> aren't a u8 flexible array (like the self-test[1]), which is named
> struct bpf_lpm_trie_key_hdr.
>
> Adjust the kernel code to use struct bpf_lpm_trie_key_u8 through-out,
> and for the selftest to use struct bpf_lpm_trie_key_hdr. Add a comment
> to the UAPI header directing folks to the two new options.
>
> Link: https://lore.kernel.org/all/202206281009.4332AA33@keescook/ [1]
> Reported-by: Mark Rutland <[email protected]>
> Closes: https://paste.debian.net/hidden/ca500597/
> Acked-by: Gustavo A. R. Silva <[email protected]>
> Signed-off-by: Kees Cook <[email protected]>
[...]
The build in BPF CI is still broken, did you try to build selftests?
https://github.com/kernel-patches/bpf/actions/runs/7978647641
[...]
GEN-SKEL [test_progs] linked_funcs.skel.h
LINK-BPF [test_progs] test_usdt.bpf.o
GEN-SKEL [test_progs-no_alu32] profiler1.skel.h
GEN-SKEL [test_progs] test_usdt.skel.h
In file included from /tmp/work/bpf/bpf/tools/include/uapi/linux/bpf.h:11,
from test_cpp.cpp:4:
/tmp/work/bpf/bpf/tools/include/uapi/linux/bpf.h:92:17: error: ‘struct bpf_lpm_trie_key_u8::<unnamed union>::bpf_lpm_trie_key_hdr’ invalid; an anonymous union may only have public non-static data members [-fpermissive]
92 | __struct_group(bpf_lpm_trie_key_hdr, hdr, /* no attrs */,
| ^~~~~~~~~~~~~~~~~~~~
/tmp/work/bpf/bpf/tools/include/uapi/linux/stddef.h:29:10: note: in definition of macro ‘__struct_group’
29 | struct TAG { MEMBERS } ATTRS NAME; \
| ^~~
BINARY bench
make: *** [Makefile:703: /tmp/work/bpf/bpf/tools/testing/selftests/bpf/test_cpp] Error 1
make: *** Waiting for unfinished jobs....
make: Leaving directory '/tmp/work/bpf/bpf/tools/testing/selftests/bpf'
Error: Process completed with exit code 2.
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index 754e68ca8744..31e9bdd4641e 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -8,6 +8,7 @@
> #ifndef _UAPI__LINUX_BPF_H__
> #define _UAPI__LINUX_BPF_H__
>
> +#include <linux/stddef.h>
> #include <linux/types.h>
> #include <linux/bpf_common.h>
>
> @@ -77,12 +78,24 @@ struct bpf_insn {
> __s32 imm; /* signed immediate constant */
> };
>
> -/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */
> +/* Deprecated: use struct bpf_lpm_trie_key_u8 (when the "data" member is needed for
> + * byte access) or struct bpf_lpm_trie_key_hdr (when using an alternative type for
> + * the trailing flexible array member) instead.
> + */
> struct bpf_lpm_trie_key {
> __u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */
> __u8 data[0]; /* Arbitrary size */
> };
>
> +/* Key of an a BPF_MAP_TYPE_LPM_TRIE entry, with trailing byte array. */
> +struct bpf_lpm_trie_key_u8 {
> + __struct_group(bpf_lpm_trie_key_hdr, hdr, /* no attrs */,
> + /* up to 32 for AF_INET, 128 for AF_INET6 */
> + __u32 prefixlen;
> + );
> + __u8 data[]; /* Arbitrary size */
> +};
> +
> struct bpf_cgroup_storage_key {
> __u64 cgroup_inode_id; /* cgroup inode id */
> __u32 attach_type; /* program attach type (enum bpf_attach_type) */
On Wed, Feb 21, 2024 at 05:39:55PM +0100, Daniel Borkmann wrote:
> On 2/20/24 7:54 PM, Kees Cook wrote:
> > Replace deprecated 0-length array in struct bpf_lpm_trie_key with
> > flexible array. Found with GCC 13:
> >
> > ../kernel/bpf/lpm_trie.c:207:51: warning: array subscript i is outside array bounds of 'const __u8[0]' {aka 'const unsigned char[]'} [-Warray-bounds=]
> > 207 | *(__be16 *)&key->data[i]);
> > | ^~~~~~~~~~~~~
> > ../include/uapi/linux/swab.h:102:54: note: in definition of macro '__swab16'
> > 102 | #define __swab16(x) (__u16)__builtin_bswap16((__u16)(x))
> > | ^
> > ../include/linux/byteorder/generic.h:97:21: note: in expansion of macro '__be16_to_cpu'
> > 97 | #define be16_to_cpu __be16_to_cpu
> > | ^~~~~~~~~~~~~
> > ../kernel/bpf/lpm_trie.c:206:28: note: in expansion of macro 'be16_to_cpu'
> > 206 | u16 diff = be16_to_cpu(*(__be16 *)&node->data[i]
> > ^
> > | ^~~~~~~~~~~
> > In file included from ../include/linux/bpf.h:7:
> > ../include/uapi/linux/bpf.h:82:17: note: while referencing 'data'
> > 82 | __u8 data[0]; /* Arbitrary size */
> > | ^~~~
> >
> > And found at run-time under CONFIG_FORTIFY_SOURCE:
> >
> > UBSAN: array-index-out-of-bounds in kernel/bpf/lpm_trie.c:218:49
> > index 0 is out of range for type '__u8 [*]'
> >
> > Changing struct bpf_lpm_trie_key is difficult since has been used by
> > userspace. For example, in Cilium:
> >
> > struct egress_gw_policy_key {
> > struct bpf_lpm_trie_key lpm_key;
> > __u32 saddr;
> > __u32 daddr;
> > };
> >
> > While direct references to the "data" member haven't been found, there
> > are static initializers what include the final member. For example,
> > the "{}" here:
> >
> > struct egress_gw_policy_key in_key = {
> > .lpm_key = { 32 + 24, {} },
> > .saddr = CLIENT_IP,
> > .daddr = EXTERNAL_SVC_IP & 0Xffffff,
> > };
> >
> > To avoid the build time and run time warnings seen with a 0-sized
> > trailing array for struct bpf_lpm_trie_key, introduce a new struct
> > that correctly uses a flexible array for the trailing bytes,
> > struct bpf_lpm_trie_key_u8. As part of this, include the "header"
> > portion (which is just the "prefixlen" member), so it can be used
> > by anything building a bpf_lpr_trie_key that has trailing members that
> > aren't a u8 flexible array (like the self-test[1]), which is named
> > struct bpf_lpm_trie_key_hdr.
> >
> > Adjust the kernel code to use struct bpf_lpm_trie_key_u8 through-out,
> > and for the selftest to use struct bpf_lpm_trie_key_hdr. Add a comment
> > to the UAPI header directing folks to the two new options.
> >
> > Link: https://lore.kernel.org/all/202206281009.4332AA33@keescook/ [1]
> > Reported-by: Mark Rutland <[email protected]>
> > Closes: https://paste.debian.net/hidden/ca500597/
> > Acked-by: Gustavo A. R. Silva <[email protected]>
> > Signed-off-by: Kees Cook <[email protected]>
> [...]
>
> The build in BPF CI is still broken, did you try to build selftests?
I did! I didn't have this error. :(
> https://github.com/kernel-patches/bpf/actions/runs/7978647641
Thanks for the pointer. It took a bit of digging, but I found this:
https://github.com/libbpf/ci/blob/main/build-selftests/build_selftests.sh
which is much more involved than just "make -C tools/testing/selftests/bpf"
>
> [...]
> GEN-SKEL [test_progs] linked_funcs.skel.h
> LINK-BPF [test_progs] test_usdt.bpf.o
> GEN-SKEL [test_progs-no_alu32] profiler1.skel.h
> GEN-SKEL [test_progs] test_usdt.skel.h
> In file included from /tmp/work/bpf/bpf/tools/include/uapi/linux/bpf.h:11,
> from test_cpp.cpp:4:
> /tmp/work/bpf/bpf/tools/include/uapi/linux/bpf.h:92:17: error: ‘struct bpf_lpm_trie_key_u8::<unnamed union>::bpf_lpm_trie_key_hdr’ invalid; an anonymous union may only have public non-static data members [-fpermissive]
> 92 | __struct_group(bpf_lpm_trie_key_hdr, hdr, /* no attrs */,
> | ^~~~~~~~~~~~~~~~~~~~
> /tmp/work/bpf/bpf/tools/include/uapi/linux/stddef.h:29:10: note: in definition of macro ‘__struct_group’
> 29 | struct TAG { MEMBERS } ATTRS NAME; \
> | ^~~
I'll see if I can figure out where this is coming from. This sounds like
something is being built with an unexpectedly strict option. (The above
report seems weird -- this isn't coming from -fpermissive, and is
actually an _error_ not a warning, which is the opposite of what
-fpermissive is supposed to do.) Also the mention of "public" is scary
here... that implies a C++ compiler is involved? Maybe that's why my
builds didn't catch this?
> make: *** [Makefile:703: /tmp/work/bpf/bpf/tools/testing/selftests/bpf/test_cpp] Error 1
Ah yes, cpp. Fun. I will try to reproduce this failure.
--
Kees Cook
On Wed, Feb 21, 2024 at 05:39:55PM +0100, Daniel Borkmann wrote:
> The build in BPF CI is still broken, did you try to build selftests?
Okay, I give up. How is a mortal supposed to build these?
If I try to follow what I see in
https://github.com/libbpf/ci/blob/main/build-selftests/build_selftests.sh
I just get more and more kinds of errors:
In file included from progs/cb_refs.c:5:
progs/../bpf_testmod/bpf_testmod_kfunc.h:29:8: error: redefinition of 'prog_test_pass1'
29 | struct prog_test_pass1 {
| ^
/srv/code/tools/testing/selftests/bpf/tools/include/vmlinux.h:106850:8: note: previous definition is
here
106850 | struct prog_test_pass1 {
| ^
Messing around with deleting vmlinux.h seems to get me further, but later:
/srv/code/tools/testing/selftests/bpf/bpf_testmod/bpf_testmod.c: In function 'bpf_testmod_ops_is_valid_access':
/srv/code/tools/testing/selftests/bpf/bpf_testmod/bpf_testmod.c:535:16: error: implicit declaration of function 'bpf_tracing_btf_ctx_access' [-Werror=implicit-function-declaration]
535 | return bpf_tracing_btf_ctx_access(off, size, type, prog, info);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
and then I'm stuck. It looks like the build isn't actually using
KBUILD_OUTPUT for finding includes. If I try to add -I flags to the
Makefile I just drown in new errors.
--
Kees Cook