Re: [PATCH v9 06/10] bpf: Add bpf_lookup_*_key() and bpf_key_put() kfuncs

From: Daniel Borkmann
Date: Tue Aug 09 2022 - 18:53:50 EST


On 8/9/22 3:45 PM, Roberto Sassu wrote:
Add the bpf_lookup_user_key(), bpf_lookup_system_key() and bpf_key_put()
kfuncs, to respectively search a key with a given serial and flags, obtain
a key from a pre-determined ID defined in include/linux/verification.h, and
cleanup.

Signed-off-by: Roberto Sassu <roberto.sassu@xxxxxxxxxx>
---
include/linux/bpf.h | 6 ++
kernel/trace/bpf_trace.c | 151 +++++++++++++++++++++++++++++++++++++++
2 files changed, 157 insertions(+)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 0d56c23cc504..564b9e0b8c16 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -2572,4 +2572,10 @@ static inline void bpf_cgroup_atype_get(u32 attach_btf_id, int cgroup_atype) {}
static inline void bpf_cgroup_atype_put(int cgroup_atype) {}
#endif /* CONFIG_BPF_LSM */
+#ifdef CONFIG_KEYS
+struct bpf_key {
+ struct key *key;
+ bool valid_ptr;
+};
+#endif /* CONFIG_KEYS */
#endif /* _LINUX_BPF_H */
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 68e5cdd24cef..33ca4cfe6e26 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -20,6 +20,7 @@
#include <linux/fprobe.h>
#include <linux/bsearch.h>
#include <linux/sort.h>
+#include <linux/key.h>
#include <net/bpf_sk_storage.h>
@@ -1181,6 +1182,156 @@ static const struct bpf_func_proto bpf_get_func_arg_cnt_proto = {
.arg1_type = ARG_PTR_TO_CTX,
};
+#ifdef CONFIG_KEYS
+__diag_push();
+__diag_ignore_all("-Wmissing-prototypes",
+ "kfuncs which will be used in BPF programs");
+
+/**
+ * bpf_lookup_user_key - lookup a key by its serial
+ * @serial: key serial
+ * @flags: lookup-specific flags
+ *
+ * Search a key with a given *serial* and the provided *flags*. The
+ * returned key, if found, has the reference count incremented by
+ * one, and is stored in a bpf_key structure, returned to the caller.
+ * The bpf_key structure must be passed to bpf_key_put() when done
+ * with it, so that the key reference count is decremented and the
+ * bpf_key structure is freed.
+ *
+ * Permission checks are deferred to the time the key is used by
+ * one of the available key-specific kfuncs.
+ *
+ * Set *flags* with 1, to attempt creating a requested special
+ * keyring (e.g. session keyring), if it doesn't yet exist. Set
+ * *flags* with 2 to lookup a key without waiting for the key
+ * construction, and to retrieve uninstantiated keys (keys without
+ * data attached to them).
+ *
+ * Return: a bpf_key pointer with a valid key pointer if the key is found, a
+ * NULL pointer otherwise.
+ */
+noinline __weak struct bpf_key *bpf_lookup_user_key(u32 serial, u64 flags)

Why the need for noinline and the __weak here and below? (If indeed needed this
should probably be explained in the commit desc.)

+{
+ key_ref_t key_ref;
+ struct bpf_key *bkey;
+
+ /* Keep in sync with include/linux/key.h. */
+ if (flags > (KEY_LOOKUP_PARTIAL << 1) - 1)

Can't we just simplify and test flags & ~(KEY_LOOKUP_CREATE|KEY_LOOKUP_PARTIAL)?

+ return NULL;
+
+ /* Permission check is deferred until actual kfunc using the key. */
+ key_ref = lookup_user_key(serial, flags, KEY_DEFER_PERM_CHECK);
+ if (IS_ERR(key_ref))
+ return NULL;
+
+ bkey = kmalloc(sizeof(*bkey), GFP_KERNEL);
+ if (!bkey) {
+ key_put(key_ref_to_ptr(key_ref));
+ return bkey;

nit: just return NULL probably cleaner

+ }
+
+ bkey->key = key_ref_to_ptr(key_ref);
+ bkey->valid_ptr = true;

nit: I'd probably rename s/valid_ptr/has_ref/.

+ return bkey;
+}
+
+/**
+ * bpf_lookup_system_key - lookup a key by a system-defined ID
+ * @id: key ID
+ *
+ * Obtain a bpf_key structure with a key pointer set to the passed key ID.
+ * The key pointer is marked as invalid, to prevent bpf_key_put() from
+ * attempting to decrement the key reference count on that pointer. The key
+ * pointer set in such way is currently understood only by
+ * verify_pkcs7_signature().
+ *
+ * Set *id* to one of the values defined in include/linux/verification.h:
+ * 0 for the primary keyring (immutable keyring of system keys); 1 for both
+ * the primary and secondary keyring (where keys can be added only if they
+ * are vouched for by existing keys in those keyrings); 2 for the platform
+ * keyring (primarily used by the integrity subsystem to verify a kexec'ed
+ * kerned image and, possibly, the initramfs signature).
+ *
+ * Return: a bpf_key pointer with an invalid key pointer set from the
+ * pre-determined ID on success, a NULL pointer otherwise
+ */
+noinline __weak struct bpf_key *bpf_lookup_system_key(u64 id)
+{
+ struct bpf_key *bkey;
+
+ /* Keep in sync with defs in include/linux/verification.h. */
+ if (id > (unsigned long)VERIFY_USE_PLATFORM_KEYRING)
+ return NULL;
+
+ bkey = kmalloc(sizeof(*bkey), GFP_KERNEL);

nit: Can't this be GFP_ATOMIC? Then bpf_lookup_system_key doesn't need KF_SLEEPABLE
attribute, fwiw. Overall, the bpf_lookup_{system,user}_key() look reasonable.

+ if (!bkey)
+ return bkey;
+
+ bkey->key = (struct key *)(unsigned long)id;
+ bkey->valid_ptr = false;
+
+ return bkey;
+}
+
+/**
+ * bpf_key_put - decrement key reference count if key is valid and free bpf_key
+ * @bkey: bpf_key structure
+ *
+ * Decrement the reference count of the key inside *bkey*, if the pointer
+ * is valid, and free *bkey*.
+ */
+noinline __weak void bpf_key_put(struct bpf_key *bkey)
+{
+ if (bkey->valid_ptr)
+ key_put(bkey->key);
+
+ kfree(bkey);
+}
+
+__diag_pop();
+
+BTF_SET8_START(key_kfunc_set)
+BTF_ID_FLAGS(func, bpf_lookup_user_key, KF_ACQUIRE | KF_RET_NULL | KF_SLEEPABLE)
+BTF_ID_FLAGS(func, bpf_lookup_system_key,
+ KF_ACQUIRE | KF_RET_NULL | KF_SLEEPABLE)
+BTF_ID_FLAGS(func, bpf_key_put, KF_RELEASE)
+BTF_SET8_END(key_kfunc_set)
+
+static const struct btf_kfunc_id_set bpf_key_kfunc_set = {
+ .owner = THIS_MODULE,
+ .set = &key_kfunc_set,
+};
+#endif /* CONFIG_KEYS */
+
+const struct btf_kfunc_id_set *kfunc_sets[] = {
+#ifdef CONFIG_KEYS
+ &bpf_key_kfunc_set,
+#endif /* CONFIG_KEYS */
+};