From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.remote-ip=100.103.45.18 ARC-Seal: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; cv=none; t=1793209577; b=IDkrQ5h9NXMmS4khxnBtNMU5eJYOa3G0y3u8J3bNWjBs3EemJVBg0Wz9JfUSndjVGuVg Z1WNEVA/meauSz3bCoJU5uZZelzABCSdxm7lXCAEWxu6S/7fWP54i7r/8/UZ593jkx6QB AxXBe5pzkQjB51+4SKF+JBtdjVIxROllCfBbwTdf0DwnVVSF67tdyN/BD95wO8ko5RN+l nTVh0PYouBwLrHn/vy6JJ0xwYIH4SC/s6i+M4ixXixPamrO5B+tgrdE2PsbAV22AOjrFr SF7Z2OYtCp+sdzJA8GKzOFjKR3qHmQM9zC0g4zfG6nBlzKmo+cqpKe+oye5byYJSvNA== ARC-Message-Signature: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; c=relaxed/relaxed; t=1793209577; h=Received:DKIM-Signature:From:To:Cc:Subject:Date:Message-ID:X-Mailer: In-Reply-To:References:User-Agent:X-stable:X-Patchwork-Hint: MIME-Version:Content-Transfer-Encoding; bh=1kHO8MiZfAr1r2OMTkLt3S3cMmd+dnlNOxDr+TjKRPk=; b=J7EQNIRG5PKOc2CIyZLsJolf0si1T3v1+gX38xPlp2SDAZtNZktSSVyAuEjhHDMrP3O7 ++Ph9XYfmlxmISEGAKiJ/3NAUZ5SEmSrUSebtg5Q6i1xySPYJV6RmaE+pxcBnzXVUW0CO onca8dl5m8q3CEvds4hskBN5IYW0av+9V5ZvQW/8Pc0ufzjzscOUcwF37mL/iVfSnC4gS cLPAqd61If209728aFQxF9F4gQPf4dXCbpUJ5f9kuhlV9tvXSV7oZccEztmwy6X3IeIHb B/qgYZYnae+3w8H/pBqO4gxp59fcvqnbsqDi5uOCYU0lTSlRyYMCKrbqC+bv6jsMilw== ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dkim=pass header.d=linuxfoundation.org header.i=@linuxfoundation.org header.a=rsa-sha256 header.s=korg header.b=BT5z5cmA; arc=none smtp.remote-ip=100.103.45.18 Received: from smtp.kernel.org (quasi.space.kernel.org [100.103.45.18]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (prime256v1)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id F16DF606A31D; Wed, 28 Oct 2026 17:46:16 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key, unprotected) header.d=linuxfoundation.org header.i=@linuxfoundation.org header.a=rsa-sha256 header.s=korg header.b=BT5z5cmA Received: by smtp.kernel.org (Postfix) with ESMTPSA id 4B7421F01825; Wed, 28 Oct 2026 17:46:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linuxfoundation.org; s=korg; t=1793209576; bh=1kHO8MiZfAr1r2OMTkLt3S3cMmd+dnlNOxDr+TjKRPk=; h=From:To:Cc:Subject:Date:In-Reply-To:References; b=BT5z5cmAYFBdTRVVgg6WmRoGWz6miciqUZlTlsV+0jcagHM5TGFj2mT7Im0HDFUTx dPPp8RW4bgm5yZGclRUVZ0p99/EUSPvWmut4tj79VdSarM/pXTxQg3wJY448bd6dTP PtaghtIb/ugDp4S4O87vrTQ6djseT+jaVQAFQ+LY= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, Kyle Zeng , Jarkko Sakkinen Subject: [PATCH 5.10 513/516] keys: make keyring indexes independent of kernel addresses Date: Wed, 28 Oct 2026 18:01:05 +0100 Message-ID: <20261028170015.628760936@linuxfoundation.org> X-Mailer: git-send-email 2.56.0 In-Reply-To: <20261028170003.522232767@linuxfoundation.org> References: <20261028170003.522232767@linuxfoundation.org> User-Agent: quilt/0.69 X-stable: review X-Patchwork-Hint: ignore X-Mailing-List: patches@lists.linux.dev List-Id: List-Post: List-Help: List-Subscribe: List-Unsubscribe: List-Owner: Precedence: list MIME-Version: 1.0 Content-Transfer-Encoding: 8bit 5.10-stable review patch. If anyone has any objections, please let me know. ------------------ From: Kyle Zeng commit de891c73c1cf41ac8de12c07fa2c84d6a912ad0c upstream. KEYCTL_READ returns the serial numbers in a keyring in associative-array traversal order. The first word of the index is a public, deterministic hash of the type pointer, domain pointer and description. An unprivileged user can therefore insert keys with known descriptions, predict their order for each possible KASLR slide, and recover the kernel image address. On an x86-64 build with 16 MiB alignment, 45 user keys suffice to distinguish the candidate slides. Remove the addresses from the whole index, not just its first hash word. Assign each key type and domain a nonzero 64-bit ID and hash those IDs and the description with a once-initialized SipHash key. Use the full IDs in the later index chunks as well. Truncating or hashing the IDs there would lose the guarantee that distinct type/domain/description tuples have a differing index bit. Allocate IDs lazily since private key types need not be registered. Cache them in the index key so tree walks and copies of an existing index do not need to dereference a type descriptor that may have been unregistered. Keep the 32-bit chunk layout and diff_objects() bit offsets consistent, and retain the separate root slot for keyrings used by recursive search. The key comparison, permissions and userspace data format are unchanged; only the address-dependent in-memory ordering is replaced. Cc: stable@vger.kernel.org Fixes: b2a4df200d57 ("KEYS: Expand the capacity of a keyring") Assisted-by: Codex:gpt-6-astra Signed-off-by: Kyle Zeng Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/20261013231251.56170-1-kylebot@openai.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman --- include/linux/key-type.h | 1 include/linux/key.h | 5 + security/keys/keyring.c | 150 +++++++++++++++++++++++++---------------------- 3 files changed, 86 insertions(+), 70 deletions(-) --- a/include/linux/key-type.h +++ b/include/linux/key-type.h @@ -157,6 +157,7 @@ struct key_type { const void *in, const void *in2); /* internal fields */ + atomic64_t index_id; /* Opaque keyring index ID */ struct list_head link; /* link in types list */ struct lock_class_key lock_class; /* key->sem lock class */ } __randomize_layout; --- a/include/linux/key.h +++ b/include/linux/key.h @@ -101,6 +101,7 @@ struct keyring_name; struct key_tag { struct rcu_head rcu; + atomic64_t index_id; /* Opaque keyring index ID */ refcount_t usage; bool removed; /* T when subject removed */ }; @@ -123,6 +124,8 @@ struct keyring_index_key { struct key_type *type; struct key_tag *domain_tag; /* Domain of operation */ const char *description; + u64 type_id; /* Cached index IDs */ + u64 domain_id; }; union key_payload { @@ -244,6 +247,8 @@ struct key { struct key_type *type; /* type of key */ struct key_tag *domain_tag; /* Domain of operation */ char *description; + u64 type_id; + u64 domain_id; }; }; --- a/security/keys/keyring.c +++ b/security/keys/keyring.c @@ -14,6 +14,9 @@ #include #include #include +#include +#include +#include #include #include #include @@ -147,49 +150,46 @@ static int keyring_instantiate(struct ke } /* - * Multiply 64-bits by 32-bits to 96-bits and fold back to 64-bit. Ideally we'd - * fold the carry back too, but that requires inline asm. + * Give key types and domains address-independent identities. Zero means that + * an ID has not yet been allocated. As with inode sequence numbers used by + * futexes, this relies on the 64-bit counter not wrapping in a machine's + * lifetime. */ -static u64 mult_64x32_and_fold(u64 x, u32 y) +static u64 key_index_id(atomic64_t *id) { - u64 hi = (u64)(u32)(x >> 32) * y; - u64 lo = (u64)(u32)(x) * y; - return lo + ((u64)(u32)hi << 32) + (u32)(hi >> 32); + static atomic64_t next_id; + u64 old, new; + + old = atomic64_read(id); + if (likely(old)) + return old; + + do { + new = atomic64_inc_return(&next_id); + } while (WARN_ON_ONCE(!new)); + + old = 0; + if (!atomic64_try_cmpxchg_relaxed(id, &old, new)) + return old; + return new; } /* - * Hash a key type and description. + * Hash a key type, domain and description. KEYCTL_READ exposes the tree's + * traversal order, so no part of the index may contain kernel addresses. */ static void hash_key_type_and_desc(struct keyring_index_key *index_key) { const unsigned level_shift = ASSOC_ARRAY_LEVEL_STEP; const unsigned long fan_mask = ASSOC_ARRAY_FAN_MASK; - const char *description = index_key->description; - unsigned long hash, type; - u32 piece; + static siphash_key_t hash_key __read_mostly; + unsigned long hash; u64 acc; - int n, desc_len = index_key->desc_len; - type = (unsigned long)index_key->type; - acc = mult_64x32_and_fold(type, desc_len + 13); - acc = mult_64x32_and_fold(acc, 9207); - piece = (unsigned long)index_key->domain_tag; - acc = mult_64x32_and_fold(acc, piece); - acc = mult_64x32_and_fold(acc, 9207); - - for (;;) { - n = desc_len; - if (n <= 0) - break; - if (n > 4) - n = 4; - piece = 0; - memcpy(&piece, description, n); - description += n; - desc_len -= n; - acc = mult_64x32_and_fold(acc, piece); - acc = mult_64x32_and_fold(acc, 9207); - } + get_random_once(&hash_key, sizeof(hash_key)); + acc = siphash(index_key->description, index_key->desc_len, &hash_key); + acc = siphash_3u64(index_key->type_id, index_key->domain_id, acc, + &hash_key); /* Fold the hash down to 32 bits if need be. */ hash = acc; @@ -209,7 +209,7 @@ static void hash_key_type_and_desc(struc /* * Finalise an index key to include a part of the description actually in the - * index key, to set the domain tag and to calculate the hash. + * index key, to set the domain tag and index IDs, and to calculate the hash. */ void key_set_index_key(struct keyring_index_key *index_key) { @@ -225,6 +225,14 @@ void key_set_index_key(struct keyring_in index_key->domain_tag = &default_domain_tag; } + /* Preserve the IDs when copying an existing key's index. In + * particular, its type may already have been unregistered. + */ + if (!index_key->type_id) + index_key->type_id = key_index_id(&index_key->type->index_id); + if (!index_key->domain_id) + index_key->domain_id = key_index_id(&index_key->domain_tag->index_id); + hash_key_type_and_desc(index_key); } @@ -263,15 +271,19 @@ void key_remove_domain(struct key_tag *d /* * Build the next index key chunk. * + * The hash and inline description are followed by the 64-bit type and domain + * IDs, low word first, and then by the rest of the description. * We return it one word-sized chunk at a time. */ static unsigned long keyring_get_key_chunk(const void *data, int level) { + const unsigned int id_chunks = sizeof(u64) / sizeof(unsigned long); const struct keyring_index_key *index_key = data; + int desc_len = index_key->desc_len; unsigned long chunk = 0; - const u8 *d; - int desc_len = index_key->desc_len, n = sizeof(chunk); + int n = sizeof(chunk); unsigned int offset; + const u8 *d; level /= ASSOC_ARRAY_KEY_CHUNK_SIZE; switch (level) { @@ -279,27 +291,30 @@ static unsigned long keyring_get_key_chu return index_key->hash; case 1: return index_key->x; - case 2: - return (unsigned long)index_key->type; - case 3: - return (unsigned long)index_key->domain_tag; - default: - level -= 4; - offset = sizeof(index_key->desc) + level * sizeof(long); - if (desc_len <= offset) - return 0; - - d = index_key->description + offset; - desc_len -= offset; - if (desc_len > n) - desc_len = n; - d += desc_len; - do { - chunk <<= 8; - chunk |= *--d; - } while (--desc_len > 0); - return chunk; } + + level -= 2; + if (level < id_chunks) + return index_key->type_id >> (level * ASSOC_ARRAY_KEY_CHUNK_SIZE); + level -= id_chunks; + if (level < id_chunks) + return index_key->domain_id >> (level * ASSOC_ARRAY_KEY_CHUNK_SIZE); + level -= id_chunks; + + offset = sizeof(index_key->desc) + level * sizeof(long); + if (desc_len <= offset) + return 0; + + d = index_key->description + offset; + desc_len -= offset; + if (desc_len > n) + desc_len = n; + d += desc_len; + do { + chunk <<= 8; + chunk |= *--d; + } while (--desc_len > 0); + return chunk; } static unsigned long keyring_get_object_key_chunk(const void *object, int level) @@ -331,6 +346,7 @@ static int keyring_diff_objects(const vo const struct keyring_index_key *b = data; unsigned long seg_a, seg_b; int level, i; + u64 diff; level = 0; seg_a = a->hash; @@ -339,28 +355,22 @@ static int keyring_diff_objects(const vo goto differ; level += ASSOC_ARRAY_KEY_CHUNK_SIZE / 8; - /* The number of bits contributed by the hash is controlled by a - * constant in the assoc_array headers. Everything else thereafter we - * can deal with as being machine word-size dependent. - */ + /* The inline description contributes one machine-word-sized chunk. */ seg_a = a->x; seg_b = b->x; if ((seg_a ^ seg_b) != 0) goto differ; level += sizeof(unsigned long); - /* The next bit may not work on big endian */ - seg_a = (unsigned long)a->type; - seg_b = (unsigned long)b->type; - if ((seg_a ^ seg_b) != 0) - goto differ; - level += sizeof(unsigned long); - - seg_a = (unsigned long)a->domain_tag; - seg_b = (unsigned long)b->domain_tag; - if ((seg_a ^ seg_b) != 0) - goto differ; - level += sizeof(unsigned long); + diff = a->type_id ^ b->type_id; + if (diff) + return level * 8 + __ffs64(diff); + level += sizeof(u64); + + diff = a->domain_id ^ b->domain_id; + if (diff) + return level * 8 + __ffs64(diff); + level += sizeof(u64); i = sizeof(a->desc); if (a->desc_len <= i)