linux/security/keys/big_key.c

278 lines
6.4 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/* Large capacity key type
*
* Copyright (C) 2017-2020 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
* Copyright (C) 2013 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*/
KEYS: Sort out big_key initialisation big_key has two separate initialisation functions, one that registers the key type and one that registers the crypto. If the key type fails to register, there's no problem if the crypto registers successfully because there's no way to reach the crypto except through the key type. However, if the key type registers successfully but the crypto does not, big_key_rng and big_key_blkcipher may end up set to NULL - but the code neither checks for this nor unregisters the big key key type. Furthermore, since the key type is registered before the crypto, it is theoretically possible for the kernel to try adding a big_key before the crypto is set up, leading to the same effect. Fix this by merging big_key_crypto_init() and big_key_init() and calling the resulting function late. If they're going to be encrypted, we shouldn't be creating big_keys before we have the facilities to do the encryption available. The key type registration is also moved after the crypto initialisation. The fix also includes message printing on failure. If the big_key type isn't correctly set up, simply doing: dd if=/dev/zero bs=4096 count=1 | keyctl padd big_key a @s ought to cause an oops. Fixes: 13100a72f40f5748a04017e0ab3df4cf27c809ef ('Security: Keys: Big keys stored encrypted') Signed-off-by: David Howells <dhowells@redhat.com> cc: Peter Hlavaty <zer0mem@yahoo.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com> cc: Artem Savkov <asavkov@redhat.com> cc: stable@vger.kernel.org Signed-off-by: James Morris <james.l.morris@oracle.com>
2016-10-26 22:02:01 +08:00
#define pr_fmt(fmt) "big_key: "fmt
#include <linux/init.h>
#include <linux/seq_file.h>
#include <linux/file.h>
#include <linux/shmem_fs.h>
#include <linux/err.h>
security/keys: rewrite all of big_key crypto This started out as just replacing the use of crypto/rng with get_random_bytes_wait, so that we wouldn't use bad randomness at boot time. But, upon looking further, it appears that there were even deeper underlying cryptographic problems, and that this seems to have been committed with very little crypto review. So, I rewrote the whole thing, trying to keep to the conventions introduced by the previous author, to fix these cryptographic flaws. It makes no sense to seed crypto/rng at boot time and then keep using it like this, when in fact there's already get_random_bytes_wait, which can ensure there's enough entropy and be a much more standard way of generating keys. Since this sensitive material is being stored untrusted, using ECB and no authentication is simply not okay at all. I find it surprising and a bit horrifying that this code even made it past basic crypto review, which perhaps points to some larger issues. This patch moves from using AES-ECB to using AES-GCM. Since keys are uniquely generated each time, we can set the nonce to zero. There was also a race condition in which the same key would be reused at the same time in different threads. A mutex fixes this issue now. So, to summarize, this commit fixes the following vulnerabilities: * Low entropy key generation, allowing an attacker to potentially guess or predict keys. * Unauthenticated encryption, allowing an attacker to modify the cipher text in particular ways in order to manipulate the plaintext, which is is even more frightening considering the next point. * Use of ECB mode, allowing an attacker to trivially swap blocks or compare identical plaintext blocks. * Key re-use. * Faulty memory zeroing. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Reviewed-by: Eric Biggers <ebiggers3@gmail.com> Signed-off-by: David Howells <dhowells@redhat.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Kirill Marinushkin <k.marinushkin@gmail.com> Cc: security@kernel.org Cc: stable@vger.kernel.org
2017-09-20 22:58:39 +08:00
#include <linux/random.h>
#include <keys/user-type.h>
#include <keys/big_key-type.h>
#include <crypto/chacha20poly1305.h>
KEYS: Use individual pages in big_key for crypto buffers kmalloc() can't always allocate large enough buffers for big_key to use for crypto (1MB + some metadata) so we cannot use that to allocate the buffer. Further, vmalloc'd pages can't be passed to sg_init_one() and the aead crypto accessors cannot be called progressively and must be passed all the data in one go (which means we can't pass the data in one block at a time). Fix this by allocating the buffer pages individually and passing them through a multientry scatterlist to the crypto layer. This has the bonus advantage that we don't have to allocate a contiguous series of pages. We then vmap() the page list and pass that through to the VFS read/write routines. This can trigger a warning: WARNING: CPU: 0 PID: 60912 at mm/page_alloc.c:3883 __alloc_pages_nodemask+0xb7c/0x15f8 ([<00000000002acbb6>] __alloc_pages_nodemask+0x1ee/0x15f8) [<00000000002dd356>] kmalloc_order+0x46/0x90 [<00000000002dd3e0>] kmalloc_order_trace+0x40/0x1f8 [<0000000000326a10>] __kmalloc+0x430/0x4c0 [<00000000004343e4>] big_key_preparse+0x7c/0x210 [<000000000042c040>] key_create_or_update+0x128/0x420 [<000000000042e52c>] SyS_add_key+0x124/0x220 [<00000000007bba2c>] system_call+0xc4/0x2b0 from the keyctl/padd/useradd test of the keyutils testsuite on s390x. Note that it might be better to shovel data through in page-sized lumps instead as there's no particular need to use a monolithic buffer unless the kernel itself wants to access the data. Fixes: 13100a72f40f ("Security: Keys: Big keys stored encrypted") Reported-by: Paul Bunyan <pbunyan@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com>
2018-02-22 22:38:34 +08:00
/*
* Layout of key payload words.
*/
enum {
big_key_data,
big_key_path,
big_key_path_2nd_part,
big_key_len,
};
/*
* If the data is under this limit, there's no point creating a shm file to
* hold it as the permanently resident metadata for the shmem fs will be at
* least as large as the data.
*/
#define BIG_KEY_FILE_THRESHOLD (sizeof(struct inode) + sizeof(struct dentry))
/*
* big_key defined keys take an arbitrary string as the description and an
* arbitrary blob of data as the payload
*/
struct key_type key_type_big_key = {
.name = "big_key",
.preparse = big_key_preparse,
.free_preparse = big_key_free_preparse,
.instantiate = generic_key_instantiate,
.revoke = big_key_revoke,
.destroy = big_key_destroy,
.describe = big_key_describe,
.read = big_key_read,
/* no ->update(); don't add it without changing chacha20poly1305's nonce */
};
/*
* Preparse a big key
*/
int big_key_preparse(struct key_preparsed_payload *prep)
{
struct path *path = (struct path *)&prep->payload.data[big_key_path];
struct file *file;
u8 *buf, *enckey;
ssize_t written;
size_t datalen = prep->datalen;
size_t enclen = datalen + CHACHA20POLY1305_AUTHTAG_SIZE;
int ret;
if (datalen <= 0 || datalen > 1024 * 1024 || !prep->data)
KEYS: Use individual pages in big_key for crypto buffers kmalloc() can't always allocate large enough buffers for big_key to use for crypto (1MB + some metadata) so we cannot use that to allocate the buffer. Further, vmalloc'd pages can't be passed to sg_init_one() and the aead crypto accessors cannot be called progressively and must be passed all the data in one go (which means we can't pass the data in one block at a time). Fix this by allocating the buffer pages individually and passing them through a multientry scatterlist to the crypto layer. This has the bonus advantage that we don't have to allocate a contiguous series of pages. We then vmap() the page list and pass that through to the VFS read/write routines. This can trigger a warning: WARNING: CPU: 0 PID: 60912 at mm/page_alloc.c:3883 __alloc_pages_nodemask+0xb7c/0x15f8 ([<00000000002acbb6>] __alloc_pages_nodemask+0x1ee/0x15f8) [<00000000002dd356>] kmalloc_order+0x46/0x90 [<00000000002dd3e0>] kmalloc_order_trace+0x40/0x1f8 [<0000000000326a10>] __kmalloc+0x430/0x4c0 [<00000000004343e4>] big_key_preparse+0x7c/0x210 [<000000000042c040>] key_create_or_update+0x128/0x420 [<000000000042e52c>] SyS_add_key+0x124/0x220 [<00000000007bba2c>] system_call+0xc4/0x2b0 from the keyctl/padd/useradd test of the keyutils testsuite on s390x. Note that it might be better to shovel data through in page-sized lumps instead as there's no particular need to use a monolithic buffer unless the kernel itself wants to access the data. Fixes: 13100a72f40f ("Security: Keys: Big keys stored encrypted") Reported-by: Paul Bunyan <pbunyan@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com>
2018-02-22 22:38:34 +08:00
return -EINVAL;
/* Set an arbitrary quota */
prep->quotalen = 16;
prep->payload.data[big_key_len] = (void *)(unsigned long)datalen;
if (datalen > BIG_KEY_FILE_THRESHOLD) {
/* Create a shmem file to store the data in. This will permit the data
* to be swapped out if needed.
*
* File content is stored encrypted with randomly generated key.
* Since the key is random for each file, we can set the nonce
* to zero, provided we never define a ->update() call.
*/
loff_t pos = 0;
buf = kvmalloc(enclen, GFP_KERNEL);
KEYS: Use individual pages in big_key for crypto buffers kmalloc() can't always allocate large enough buffers for big_key to use for crypto (1MB + some metadata) so we cannot use that to allocate the buffer. Further, vmalloc'd pages can't be passed to sg_init_one() and the aead crypto accessors cannot be called progressively and must be passed all the data in one go (which means we can't pass the data in one block at a time). Fix this by allocating the buffer pages individually and passing them through a multientry scatterlist to the crypto layer. This has the bonus advantage that we don't have to allocate a contiguous series of pages. We then vmap() the page list and pass that through to the VFS read/write routines. This can trigger a warning: WARNING: CPU: 0 PID: 60912 at mm/page_alloc.c:3883 __alloc_pages_nodemask+0xb7c/0x15f8 ([<00000000002acbb6>] __alloc_pages_nodemask+0x1ee/0x15f8) [<00000000002dd356>] kmalloc_order+0x46/0x90 [<00000000002dd3e0>] kmalloc_order_trace+0x40/0x1f8 [<0000000000326a10>] __kmalloc+0x430/0x4c0 [<00000000004343e4>] big_key_preparse+0x7c/0x210 [<000000000042c040>] key_create_or_update+0x128/0x420 [<000000000042e52c>] SyS_add_key+0x124/0x220 [<00000000007bba2c>] system_call+0xc4/0x2b0 from the keyctl/padd/useradd test of the keyutils testsuite on s390x. Note that it might be better to shovel data through in page-sized lumps instead as there's no particular need to use a monolithic buffer unless the kernel itself wants to access the data. Fixes: 13100a72f40f ("Security: Keys: Big keys stored encrypted") Reported-by: Paul Bunyan <pbunyan@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com>
2018-02-22 22:38:34 +08:00
if (!buf)
return -ENOMEM;
/* generate random key */
enckey = kmalloc(CHACHA20POLY1305_KEY_SIZE, GFP_KERNEL);
if (!enckey) {
ret = -ENOMEM;
goto error;
}
ret = get_random_bytes_wait(enckey, CHACHA20POLY1305_KEY_SIZE);
security/keys: rewrite all of big_key crypto This started out as just replacing the use of crypto/rng with get_random_bytes_wait, so that we wouldn't use bad randomness at boot time. But, upon looking further, it appears that there were even deeper underlying cryptographic problems, and that this seems to have been committed with very little crypto review. So, I rewrote the whole thing, trying to keep to the conventions introduced by the previous author, to fix these cryptographic flaws. It makes no sense to seed crypto/rng at boot time and then keep using it like this, when in fact there's already get_random_bytes_wait, which can ensure there's enough entropy and be a much more standard way of generating keys. Since this sensitive material is being stored untrusted, using ECB and no authentication is simply not okay at all. I find it surprising and a bit horrifying that this code even made it past basic crypto review, which perhaps points to some larger issues. This patch moves from using AES-ECB to using AES-GCM. Since keys are uniquely generated each time, we can set the nonce to zero. There was also a race condition in which the same key would be reused at the same time in different threads. A mutex fixes this issue now. So, to summarize, this commit fixes the following vulnerabilities: * Low entropy key generation, allowing an attacker to potentially guess or predict keys. * Unauthenticated encryption, allowing an attacker to modify the cipher text in particular ways in order to manipulate the plaintext, which is is even more frightening considering the next point. * Use of ECB mode, allowing an attacker to trivially swap blocks or compare identical plaintext blocks. * Key re-use. * Faulty memory zeroing. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Reviewed-by: Eric Biggers <ebiggers3@gmail.com> Signed-off-by: David Howells <dhowells@redhat.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Kirill Marinushkin <k.marinushkin@gmail.com> Cc: security@kernel.org Cc: stable@vger.kernel.org
2017-09-20 22:58:39 +08:00
if (unlikely(ret))
goto err_enckey;
/* encrypt data */
chacha20poly1305_encrypt(buf, prep->data, datalen, NULL, 0,
0, enckey);
/* save aligned data to file */
file = shmem_kernel_file_setup("", enclen, 0);
if (IS_ERR(file)) {
ret = PTR_ERR(file);
goto err_enckey;
}
written = kernel_write(file, buf, enclen, &pos);
if (written != enclen) {
ret = written;
if (written >= 0)
ret = -EIO;
goto err_fput;
}
/* Pin the mount and dentry to the key so that we can open it again
* later
*/
prep->payload.data[big_key_data] = enckey;
*path = file->f_path;
path_get(path);
fput(file);
memzero_explicit(buf, enclen);
kvfree(buf);
} else {
/* Just store the data in a buffer */
void *data = kmalloc(datalen, GFP_KERNEL);
if (!data)
return -ENOMEM;
prep->payload.data[big_key_data] = data;
memcpy(data, prep->data, prep->datalen);
}
return 0;
err_fput:
fput(file);
err_enckey:
kzfree(enckey);
error:
memzero_explicit(buf, enclen);
kvfree(buf);
return ret;
}
/*
* Clear preparsement.
*/
void big_key_free_preparse(struct key_preparsed_payload *prep)
{
if (prep->datalen > BIG_KEY_FILE_THRESHOLD) {
struct path *path = (struct path *)&prep->payload.data[big_key_path];
path_put(path);
}
kzfree(prep->payload.data[big_key_data]);
}
/*
* dispose of the links from a revoked keyring
* - called with the key sem write-locked
*/
void big_key_revoke(struct key *key)
{
struct path *path = (struct path *)&key->payload.data[big_key_path];
/* clear the quota */
key_payload_reserve(key, 0);
KEYS: Fix race between updating and finding a negative key Consolidate KEY_FLAG_INSTANTIATED, KEY_FLAG_NEGATIVE and the rejection error into one field such that: (1) The instantiation state can be modified/read atomically. (2) The error can be accessed atomically with the state. (3) The error isn't stored unioned with the payload pointers. This deals with the problem that the state is spread over three different objects (two bits and a separate variable) and reading or updating them atomically isn't practical, given that not only can uninstantiated keys change into instantiated or rejected keys, but rejected keys can also turn into instantiated keys - and someone accessing the key might not be using any locking. The main side effect of this problem is that what was held in the payload may change, depending on the state. For instance, you might observe the key to be in the rejected state. You then read the cached error, but if the key semaphore wasn't locked, the key might've become instantiated between the two reads - and you might now have something in hand that isn't actually an error code. The state is now KEY_IS_UNINSTANTIATED, KEY_IS_POSITIVE or a negative error code if the key is negatively instantiated. The key_is_instantiated() function is replaced with key_is_positive() to avoid confusion as negative keys are also 'instantiated'. Additionally, barriering is included: (1) Order payload-set before state-set during instantiation. (2) Order state-read before payload-read when using the key. Further separate barriering is necessary if RCU is being used to access the payload content after reading the payload pointers. Fixes: 146aa8b1453b ("KEYS: Merge the type-specific data with the payload data") Cc: stable@vger.kernel.org # v4.4+ Reported-by: Eric Biggers <ebiggers@google.com> Signed-off-by: David Howells <dhowells@redhat.com> Reviewed-by: Eric Biggers <ebiggers@google.com>
2017-10-04 23:43:25 +08:00
if (key_is_positive(key) &&
(size_t)key->payload.data[big_key_len] > BIG_KEY_FILE_THRESHOLD)
vfs_truncate(path, 0);
}
/*
* dispose of the data dangling from the corpse of a big_key key
*/
void big_key_destroy(struct key *key)
{
size_t datalen = (size_t)key->payload.data[big_key_len];
if (datalen > BIG_KEY_FILE_THRESHOLD) {
struct path *path = (struct path *)&key->payload.data[big_key_path];
path_put(path);
path->mnt = NULL;
path->dentry = NULL;
}
kzfree(key->payload.data[big_key_data]);
key->payload.data[big_key_data] = NULL;
}
/*
* describe the big_key key
*/
void big_key_describe(const struct key *key, struct seq_file *m)
{
size_t datalen = (size_t)key->payload.data[big_key_len];
seq_puts(m, key->description);
KEYS: Fix race between updating and finding a negative key Consolidate KEY_FLAG_INSTANTIATED, KEY_FLAG_NEGATIVE and the rejection error into one field such that: (1) The instantiation state can be modified/read atomically. (2) The error can be accessed atomically with the state. (3) The error isn't stored unioned with the payload pointers. This deals with the problem that the state is spread over three different objects (two bits and a separate variable) and reading or updating them atomically isn't practical, given that not only can uninstantiated keys change into instantiated or rejected keys, but rejected keys can also turn into instantiated keys - and someone accessing the key might not be using any locking. The main side effect of this problem is that what was held in the payload may change, depending on the state. For instance, you might observe the key to be in the rejected state. You then read the cached error, but if the key semaphore wasn't locked, the key might've become instantiated between the two reads - and you might now have something in hand that isn't actually an error code. The state is now KEY_IS_UNINSTANTIATED, KEY_IS_POSITIVE or a negative error code if the key is negatively instantiated. The key_is_instantiated() function is replaced with key_is_positive() to avoid confusion as negative keys are also 'instantiated'. Additionally, barriering is included: (1) Order payload-set before state-set during instantiation. (2) Order state-read before payload-read when using the key. Further separate barriering is necessary if RCU is being used to access the payload content after reading the payload pointers. Fixes: 146aa8b1453b ("KEYS: Merge the type-specific data with the payload data") Cc: stable@vger.kernel.org # v4.4+ Reported-by: Eric Biggers <ebiggers@google.com> Signed-off-by: David Howells <dhowells@redhat.com> Reviewed-by: Eric Biggers <ebiggers@google.com>
2017-10-04 23:43:25 +08:00
if (key_is_positive(key))
seq_printf(m, ": %zu [%s]",
datalen,
datalen > BIG_KEY_FILE_THRESHOLD ? "file" : "buff");
}
/*
* read the key data
* - the key's semaphore is read-locked
*/
KEYS: Don't write out to userspace while holding key semaphore A lockdep circular locking dependency report was seen when running a keyutils test: [12537.027242] ====================================================== [12537.059309] WARNING: possible circular locking dependency detected [12537.088148] 4.18.0-147.7.1.el8_1.x86_64+debug #1 Tainted: G OE --------- - - [12537.125253] ------------------------------------------------------ [12537.153189] keyctl/25598 is trying to acquire lock: [12537.175087] 000000007c39f96c (&mm->mmap_sem){++++}, at: __might_fault+0xc4/0x1b0 [12537.208365] [12537.208365] but task is already holding lock: [12537.234507] 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12537.270476] [12537.270476] which lock already depends on the new lock. [12537.270476] [12537.307209] [12537.307209] the existing dependency chain (in reverse order) is: [12537.340754] [12537.340754] -> #3 (&type->lock_class){++++}: [12537.367434] down_write+0x4d/0x110 [12537.385202] __key_link_begin+0x87/0x280 [12537.405232] request_key_and_link+0x483/0xf70 [12537.427221] request_key+0x3c/0x80 [12537.444839] dns_query+0x1db/0x5a5 [dns_resolver] [12537.468445] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.496731] cifs_reconnect+0xe04/0x2500 [cifs] [12537.519418] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.546263] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.573551] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.601045] kthread+0x30c/0x3d0 [12537.617906] ret_from_fork+0x3a/0x50 [12537.636225] [12537.636225] -> #2 (root_key_user.cons_lock){+.+.}: [12537.664525] __mutex_lock+0x105/0x11f0 [12537.683734] request_key_and_link+0x35a/0xf70 [12537.705640] request_key+0x3c/0x80 [12537.723304] dns_query+0x1db/0x5a5 [dns_resolver] [12537.746773] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.775607] cifs_reconnect+0xe04/0x2500 [cifs] [12537.798322] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.823369] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.847262] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.873477] kthread+0x30c/0x3d0 [12537.890281] ret_from_fork+0x3a/0x50 [12537.908649] [12537.908649] -> #1 (&tcp_ses->srv_mutex){+.+.}: [12537.935225] __mutex_lock+0x105/0x11f0 [12537.954450] cifs_call_async+0x102/0x7f0 [cifs] [12537.977250] smb2_async_readv+0x6c3/0xc90 [cifs] [12538.000659] cifs_readpages+0x120a/0x1e50 [cifs] [12538.023920] read_pages+0xf5/0x560 [12538.041583] __do_page_cache_readahead+0x41d/0x4b0 [12538.067047] ondemand_readahead+0x44c/0xc10 [12538.092069] filemap_fault+0xec1/0x1830 [12538.111637] __do_fault+0x82/0x260 [12538.129216] do_fault+0x419/0xfb0 [12538.146390] __handle_mm_fault+0x862/0xdf0 [12538.167408] handle_mm_fault+0x154/0x550 [12538.187401] __do_page_fault+0x42f/0xa60 [12538.207395] do_page_fault+0x38/0x5e0 [12538.225777] page_fault+0x1e/0x30 [12538.243010] [12538.243010] -> #0 (&mm->mmap_sem){++++}: [12538.267875] lock_acquire+0x14c/0x420 [12538.286848] __might_fault+0x119/0x1b0 [12538.306006] keyring_read_iterator+0x7e/0x170 [12538.327936] assoc_array_subtree_iterate+0x97/0x280 [12538.352154] keyring_read+0xe9/0x110 [12538.370558] keyctl_read_key+0x1b9/0x220 [12538.391470] do_syscall_64+0xa5/0x4b0 [12538.410511] entry_SYSCALL_64_after_hwframe+0x6a/0xdf [12538.435535] [12538.435535] other info that might help us debug this: [12538.435535] [12538.472829] Chain exists of: [12538.472829] &mm->mmap_sem --> root_key_user.cons_lock --> &type->lock_class [12538.472829] [12538.524820] Possible unsafe locking scenario: [12538.524820] [12538.551431] CPU0 CPU1 [12538.572654] ---- ---- [12538.595865] lock(&type->lock_class); [12538.613737] lock(root_key_user.cons_lock); [12538.644234] lock(&type->lock_class); [12538.672410] lock(&mm->mmap_sem); [12538.687758] [12538.687758] *** DEADLOCK *** [12538.687758] [12538.714455] 1 lock held by keyctl/25598: [12538.732097] #0: 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12538.770573] [12538.770573] stack backtrace: [12538.790136] CPU: 2 PID: 25598 Comm: keyctl Kdump: loaded Tainted: G [12538.844855] Hardware name: HP ProLiant DL360 Gen9/ProLiant DL360 Gen9, BIOS P89 12/27/2015 [12538.881963] Call Trace: [12538.892897] dump_stack+0x9a/0xf0 [12538.907908] print_circular_bug.isra.25.cold.50+0x1bc/0x279 [12538.932891] ? save_trace+0xd6/0x250 [12538.948979] check_prev_add.constprop.32+0xc36/0x14f0 [12538.971643] ? keyring_compare_object+0x104/0x190 [12538.992738] ? check_usage+0x550/0x550 [12539.009845] ? sched_clock+0x5/0x10 [12539.025484] ? sched_clock_cpu+0x18/0x1e0 [12539.043555] __lock_acquire+0x1f12/0x38d0 [12539.061551] ? trace_hardirqs_on+0x10/0x10 [12539.080554] lock_acquire+0x14c/0x420 [12539.100330] ? __might_fault+0xc4/0x1b0 [12539.119079] __might_fault+0x119/0x1b0 [12539.135869] ? __might_fault+0xc4/0x1b0 [12539.153234] keyring_read_iterator+0x7e/0x170 [12539.172787] ? keyring_read+0x110/0x110 [12539.190059] assoc_array_subtree_iterate+0x97/0x280 [12539.211526] keyring_read+0xe9/0x110 [12539.227561] ? keyring_gc_check_iterator+0xc0/0xc0 [12539.249076] keyctl_read_key+0x1b9/0x220 [12539.266660] do_syscall_64+0xa5/0x4b0 [12539.283091] entry_SYSCALL_64_after_hwframe+0x6a/0xdf One way to prevent this deadlock scenario from happening is to not allow writing to userspace while holding the key semaphore. Instead, an internal buffer is allocated for getting the keys out from the read method first before copying them out to userspace without holding the lock. That requires taking out the __user modifier from all the relevant read methods as well as additional changes to not use any userspace write helpers. That is, 1) The put_user() call is replaced by a direct copy. 2) The copy_to_user() call is replaced by memcpy(). 3) All the fault handling code is removed. Compiling on a x86-64 system, the size of the rxrpc_read() function is reduced from 3795 bytes to 2384 bytes with this patch. Fixes: ^1da177e4c3f4 ("Linux-2.6.12-rc2") Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com> Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com>
2020-03-22 09:11:24 +08:00
long big_key_read(const struct key *key, char *buffer, size_t buflen)
{
size_t datalen = (size_t)key->payload.data[big_key_len];
long ret;
if (!buffer || buflen < datalen)
return datalen;
if (datalen > BIG_KEY_FILE_THRESHOLD) {
struct path *path = (struct path *)&key->payload.data[big_key_path];
struct file *file;
u8 *buf, *enckey = (u8 *)key->payload.data[big_key_data];
size_t enclen = datalen + CHACHA20POLY1305_AUTHTAG_SIZE;
loff_t pos = 0;
buf = kvmalloc(enclen, GFP_KERNEL);
KEYS: Use individual pages in big_key for crypto buffers kmalloc() can't always allocate large enough buffers for big_key to use for crypto (1MB + some metadata) so we cannot use that to allocate the buffer. Further, vmalloc'd pages can't be passed to sg_init_one() and the aead crypto accessors cannot be called progressively and must be passed all the data in one go (which means we can't pass the data in one block at a time). Fix this by allocating the buffer pages individually and passing them through a multientry scatterlist to the crypto layer. This has the bonus advantage that we don't have to allocate a contiguous series of pages. We then vmap() the page list and pass that through to the VFS read/write routines. This can trigger a warning: WARNING: CPU: 0 PID: 60912 at mm/page_alloc.c:3883 __alloc_pages_nodemask+0xb7c/0x15f8 ([<00000000002acbb6>] __alloc_pages_nodemask+0x1ee/0x15f8) [<00000000002dd356>] kmalloc_order+0x46/0x90 [<00000000002dd3e0>] kmalloc_order_trace+0x40/0x1f8 [<0000000000326a10>] __kmalloc+0x430/0x4c0 [<00000000004343e4>] big_key_preparse+0x7c/0x210 [<000000000042c040>] key_create_or_update+0x128/0x420 [<000000000042e52c>] SyS_add_key+0x124/0x220 [<00000000007bba2c>] system_call+0xc4/0x2b0 from the keyctl/padd/useradd test of the keyutils testsuite on s390x. Note that it might be better to shovel data through in page-sized lumps instead as there's no particular need to use a monolithic buffer unless the kernel itself wants to access the data. Fixes: 13100a72f40f ("Security: Keys: Big keys stored encrypted") Reported-by: Paul Bunyan <pbunyan@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com>
2018-02-22 22:38:34 +08:00
if (!buf)
return -ENOMEM;
file = dentry_open(path, O_RDONLY, current_cred());
if (IS_ERR(file)) {
ret = PTR_ERR(file);
goto error;
}
/* read file to kernel and decrypt */
ret = kernel_read(file, buf, enclen, &pos);
if (ret != enclen) {
if (ret >= 0)
ret = -EIO;
goto err_fput;
}
ret = chacha20poly1305_decrypt(buf, buf, enclen, NULL, 0, 0,
enckey) ? 0 : -EBADMSG;
if (unlikely(ret))
goto err_fput;
ret = datalen;
KEYS: Don't write out to userspace while holding key semaphore A lockdep circular locking dependency report was seen when running a keyutils test: [12537.027242] ====================================================== [12537.059309] WARNING: possible circular locking dependency detected [12537.088148] 4.18.0-147.7.1.el8_1.x86_64+debug #1 Tainted: G OE --------- - - [12537.125253] ------------------------------------------------------ [12537.153189] keyctl/25598 is trying to acquire lock: [12537.175087] 000000007c39f96c (&mm->mmap_sem){++++}, at: __might_fault+0xc4/0x1b0 [12537.208365] [12537.208365] but task is already holding lock: [12537.234507] 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12537.270476] [12537.270476] which lock already depends on the new lock. [12537.270476] [12537.307209] [12537.307209] the existing dependency chain (in reverse order) is: [12537.340754] [12537.340754] -> #3 (&type->lock_class){++++}: [12537.367434] down_write+0x4d/0x110 [12537.385202] __key_link_begin+0x87/0x280 [12537.405232] request_key_and_link+0x483/0xf70 [12537.427221] request_key+0x3c/0x80 [12537.444839] dns_query+0x1db/0x5a5 [dns_resolver] [12537.468445] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.496731] cifs_reconnect+0xe04/0x2500 [cifs] [12537.519418] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.546263] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.573551] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.601045] kthread+0x30c/0x3d0 [12537.617906] ret_from_fork+0x3a/0x50 [12537.636225] [12537.636225] -> #2 (root_key_user.cons_lock){+.+.}: [12537.664525] __mutex_lock+0x105/0x11f0 [12537.683734] request_key_and_link+0x35a/0xf70 [12537.705640] request_key+0x3c/0x80 [12537.723304] dns_query+0x1db/0x5a5 [dns_resolver] [12537.746773] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.775607] cifs_reconnect+0xe04/0x2500 [cifs] [12537.798322] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.823369] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.847262] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.873477] kthread+0x30c/0x3d0 [12537.890281] ret_from_fork+0x3a/0x50 [12537.908649] [12537.908649] -> #1 (&tcp_ses->srv_mutex){+.+.}: [12537.935225] __mutex_lock+0x105/0x11f0 [12537.954450] cifs_call_async+0x102/0x7f0 [cifs] [12537.977250] smb2_async_readv+0x6c3/0xc90 [cifs] [12538.000659] cifs_readpages+0x120a/0x1e50 [cifs] [12538.023920] read_pages+0xf5/0x560 [12538.041583] __do_page_cache_readahead+0x41d/0x4b0 [12538.067047] ondemand_readahead+0x44c/0xc10 [12538.092069] filemap_fault+0xec1/0x1830 [12538.111637] __do_fault+0x82/0x260 [12538.129216] do_fault+0x419/0xfb0 [12538.146390] __handle_mm_fault+0x862/0xdf0 [12538.167408] handle_mm_fault+0x154/0x550 [12538.187401] __do_page_fault+0x42f/0xa60 [12538.207395] do_page_fault+0x38/0x5e0 [12538.225777] page_fault+0x1e/0x30 [12538.243010] [12538.243010] -> #0 (&mm->mmap_sem){++++}: [12538.267875] lock_acquire+0x14c/0x420 [12538.286848] __might_fault+0x119/0x1b0 [12538.306006] keyring_read_iterator+0x7e/0x170 [12538.327936] assoc_array_subtree_iterate+0x97/0x280 [12538.352154] keyring_read+0xe9/0x110 [12538.370558] keyctl_read_key+0x1b9/0x220 [12538.391470] do_syscall_64+0xa5/0x4b0 [12538.410511] entry_SYSCALL_64_after_hwframe+0x6a/0xdf [12538.435535] [12538.435535] other info that might help us debug this: [12538.435535] [12538.472829] Chain exists of: [12538.472829] &mm->mmap_sem --> root_key_user.cons_lock --> &type->lock_class [12538.472829] [12538.524820] Possible unsafe locking scenario: [12538.524820] [12538.551431] CPU0 CPU1 [12538.572654] ---- ---- [12538.595865] lock(&type->lock_class); [12538.613737] lock(root_key_user.cons_lock); [12538.644234] lock(&type->lock_class); [12538.672410] lock(&mm->mmap_sem); [12538.687758] [12538.687758] *** DEADLOCK *** [12538.687758] [12538.714455] 1 lock held by keyctl/25598: [12538.732097] #0: 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12538.770573] [12538.770573] stack backtrace: [12538.790136] CPU: 2 PID: 25598 Comm: keyctl Kdump: loaded Tainted: G [12538.844855] Hardware name: HP ProLiant DL360 Gen9/ProLiant DL360 Gen9, BIOS P89 12/27/2015 [12538.881963] Call Trace: [12538.892897] dump_stack+0x9a/0xf0 [12538.907908] print_circular_bug.isra.25.cold.50+0x1bc/0x279 [12538.932891] ? save_trace+0xd6/0x250 [12538.948979] check_prev_add.constprop.32+0xc36/0x14f0 [12538.971643] ? keyring_compare_object+0x104/0x190 [12538.992738] ? check_usage+0x550/0x550 [12539.009845] ? sched_clock+0x5/0x10 [12539.025484] ? sched_clock_cpu+0x18/0x1e0 [12539.043555] __lock_acquire+0x1f12/0x38d0 [12539.061551] ? trace_hardirqs_on+0x10/0x10 [12539.080554] lock_acquire+0x14c/0x420 [12539.100330] ? __might_fault+0xc4/0x1b0 [12539.119079] __might_fault+0x119/0x1b0 [12539.135869] ? __might_fault+0xc4/0x1b0 [12539.153234] keyring_read_iterator+0x7e/0x170 [12539.172787] ? keyring_read+0x110/0x110 [12539.190059] assoc_array_subtree_iterate+0x97/0x280 [12539.211526] keyring_read+0xe9/0x110 [12539.227561] ? keyring_gc_check_iterator+0xc0/0xc0 [12539.249076] keyctl_read_key+0x1b9/0x220 [12539.266660] do_syscall_64+0xa5/0x4b0 [12539.283091] entry_SYSCALL_64_after_hwframe+0x6a/0xdf One way to prevent this deadlock scenario from happening is to not allow writing to userspace while holding the key semaphore. Instead, an internal buffer is allocated for getting the keys out from the read method first before copying them out to userspace without holding the lock. That requires taking out the __user modifier from all the relevant read methods as well as additional changes to not use any userspace write helpers. That is, 1) The put_user() call is replaced by a direct copy. 2) The copy_to_user() call is replaced by memcpy(). 3) All the fault handling code is removed. Compiling on a x86-64 system, the size of the rxrpc_read() function is reduced from 3795 bytes to 2384 bytes with this patch. Fixes: ^1da177e4c3f4 ("Linux-2.6.12-rc2") Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com> Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com>
2020-03-22 09:11:24 +08:00
/* copy out decrypted data */
memcpy(buffer, buf, datalen);
err_fput:
fput(file);
error:
memzero_explicit(buf, enclen);
kvfree(buf);
} else {
ret = datalen;
KEYS: Don't write out to userspace while holding key semaphore A lockdep circular locking dependency report was seen when running a keyutils test: [12537.027242] ====================================================== [12537.059309] WARNING: possible circular locking dependency detected [12537.088148] 4.18.0-147.7.1.el8_1.x86_64+debug #1 Tainted: G OE --------- - - [12537.125253] ------------------------------------------------------ [12537.153189] keyctl/25598 is trying to acquire lock: [12537.175087] 000000007c39f96c (&mm->mmap_sem){++++}, at: __might_fault+0xc4/0x1b0 [12537.208365] [12537.208365] but task is already holding lock: [12537.234507] 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12537.270476] [12537.270476] which lock already depends on the new lock. [12537.270476] [12537.307209] [12537.307209] the existing dependency chain (in reverse order) is: [12537.340754] [12537.340754] -> #3 (&type->lock_class){++++}: [12537.367434] down_write+0x4d/0x110 [12537.385202] __key_link_begin+0x87/0x280 [12537.405232] request_key_and_link+0x483/0xf70 [12537.427221] request_key+0x3c/0x80 [12537.444839] dns_query+0x1db/0x5a5 [dns_resolver] [12537.468445] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.496731] cifs_reconnect+0xe04/0x2500 [cifs] [12537.519418] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.546263] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.573551] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.601045] kthread+0x30c/0x3d0 [12537.617906] ret_from_fork+0x3a/0x50 [12537.636225] [12537.636225] -> #2 (root_key_user.cons_lock){+.+.}: [12537.664525] __mutex_lock+0x105/0x11f0 [12537.683734] request_key_and_link+0x35a/0xf70 [12537.705640] request_key+0x3c/0x80 [12537.723304] dns_query+0x1db/0x5a5 [dns_resolver] [12537.746773] dns_resolve_server_name_to_ip+0x1e1/0x4d0 [cifs] [12537.775607] cifs_reconnect+0xe04/0x2500 [cifs] [12537.798322] cifs_readv_from_socket+0x461/0x690 [cifs] [12537.823369] cifs_read_from_socket+0xa0/0xe0 [cifs] [12537.847262] cifs_demultiplex_thread+0x311/0x2db0 [cifs] [12537.873477] kthread+0x30c/0x3d0 [12537.890281] ret_from_fork+0x3a/0x50 [12537.908649] [12537.908649] -> #1 (&tcp_ses->srv_mutex){+.+.}: [12537.935225] __mutex_lock+0x105/0x11f0 [12537.954450] cifs_call_async+0x102/0x7f0 [cifs] [12537.977250] smb2_async_readv+0x6c3/0xc90 [cifs] [12538.000659] cifs_readpages+0x120a/0x1e50 [cifs] [12538.023920] read_pages+0xf5/0x560 [12538.041583] __do_page_cache_readahead+0x41d/0x4b0 [12538.067047] ondemand_readahead+0x44c/0xc10 [12538.092069] filemap_fault+0xec1/0x1830 [12538.111637] __do_fault+0x82/0x260 [12538.129216] do_fault+0x419/0xfb0 [12538.146390] __handle_mm_fault+0x862/0xdf0 [12538.167408] handle_mm_fault+0x154/0x550 [12538.187401] __do_page_fault+0x42f/0xa60 [12538.207395] do_page_fault+0x38/0x5e0 [12538.225777] page_fault+0x1e/0x30 [12538.243010] [12538.243010] -> #0 (&mm->mmap_sem){++++}: [12538.267875] lock_acquire+0x14c/0x420 [12538.286848] __might_fault+0x119/0x1b0 [12538.306006] keyring_read_iterator+0x7e/0x170 [12538.327936] assoc_array_subtree_iterate+0x97/0x280 [12538.352154] keyring_read+0xe9/0x110 [12538.370558] keyctl_read_key+0x1b9/0x220 [12538.391470] do_syscall_64+0xa5/0x4b0 [12538.410511] entry_SYSCALL_64_after_hwframe+0x6a/0xdf [12538.435535] [12538.435535] other info that might help us debug this: [12538.435535] [12538.472829] Chain exists of: [12538.472829] &mm->mmap_sem --> root_key_user.cons_lock --> &type->lock_class [12538.472829] [12538.524820] Possible unsafe locking scenario: [12538.524820] [12538.551431] CPU0 CPU1 [12538.572654] ---- ---- [12538.595865] lock(&type->lock_class); [12538.613737] lock(root_key_user.cons_lock); [12538.644234] lock(&type->lock_class); [12538.672410] lock(&mm->mmap_sem); [12538.687758] [12538.687758] *** DEADLOCK *** [12538.687758] [12538.714455] 1 lock held by keyctl/25598: [12538.732097] #0: 000000003de5b58d (&type->lock_class){++++}, at: keyctl_read_key+0x15a/0x220 [12538.770573] [12538.770573] stack backtrace: [12538.790136] CPU: 2 PID: 25598 Comm: keyctl Kdump: loaded Tainted: G [12538.844855] Hardware name: HP ProLiant DL360 Gen9/ProLiant DL360 Gen9, BIOS P89 12/27/2015 [12538.881963] Call Trace: [12538.892897] dump_stack+0x9a/0xf0 [12538.907908] print_circular_bug.isra.25.cold.50+0x1bc/0x279 [12538.932891] ? save_trace+0xd6/0x250 [12538.948979] check_prev_add.constprop.32+0xc36/0x14f0 [12538.971643] ? keyring_compare_object+0x104/0x190 [12538.992738] ? check_usage+0x550/0x550 [12539.009845] ? sched_clock+0x5/0x10 [12539.025484] ? sched_clock_cpu+0x18/0x1e0 [12539.043555] __lock_acquire+0x1f12/0x38d0 [12539.061551] ? trace_hardirqs_on+0x10/0x10 [12539.080554] lock_acquire+0x14c/0x420 [12539.100330] ? __might_fault+0xc4/0x1b0 [12539.119079] __might_fault+0x119/0x1b0 [12539.135869] ? __might_fault+0xc4/0x1b0 [12539.153234] keyring_read_iterator+0x7e/0x170 [12539.172787] ? keyring_read+0x110/0x110 [12539.190059] assoc_array_subtree_iterate+0x97/0x280 [12539.211526] keyring_read+0xe9/0x110 [12539.227561] ? keyring_gc_check_iterator+0xc0/0xc0 [12539.249076] keyctl_read_key+0x1b9/0x220 [12539.266660] do_syscall_64+0xa5/0x4b0 [12539.283091] entry_SYSCALL_64_after_hwframe+0x6a/0xdf One way to prevent this deadlock scenario from happening is to not allow writing to userspace while holding the key semaphore. Instead, an internal buffer is allocated for getting the keys out from the read method first before copying them out to userspace without holding the lock. That requires taking out the __user modifier from all the relevant read methods as well as additional changes to not use any userspace write helpers. That is, 1) The put_user() call is replaced by a direct copy. 2) The copy_to_user() call is replaced by memcpy(). 3) All the fault handling code is removed. Compiling on a x86-64 system, the size of the rxrpc_read() function is reduced from 3795 bytes to 2384 bytes with this patch. Fixes: ^1da177e4c3f4 ("Linux-2.6.12-rc2") Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com> Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com>
2020-03-22 09:11:24 +08:00
memcpy(buffer, key->payload.data[big_key_data], datalen);
}
return ret;
}
/*
* Register key type
*/
static int __init big_key_init(void)
{
return register_key_type(&key_type_big_key);
}
KEYS: Sort out big_key initialisation big_key has two separate initialisation functions, one that registers the key type and one that registers the crypto. If the key type fails to register, there's no problem if the crypto registers successfully because there's no way to reach the crypto except through the key type. However, if the key type registers successfully but the crypto does not, big_key_rng and big_key_blkcipher may end up set to NULL - but the code neither checks for this nor unregisters the big key key type. Furthermore, since the key type is registered before the crypto, it is theoretically possible for the kernel to try adding a big_key before the crypto is set up, leading to the same effect. Fix this by merging big_key_crypto_init() and big_key_init() and calling the resulting function late. If they're going to be encrypted, we shouldn't be creating big_keys before we have the facilities to do the encryption available. The key type registration is also moved after the crypto initialisation. The fix also includes message printing on failure. If the big_key type isn't correctly set up, simply doing: dd if=/dev/zero bs=4096 count=1 | keyctl padd big_key a @s ought to cause an oops. Fixes: 13100a72f40f5748a04017e0ab3df4cf27c809ef ('Security: Keys: Big keys stored encrypted') Signed-off-by: David Howells <dhowells@redhat.com> cc: Peter Hlavaty <zer0mem@yahoo.com> cc: Kirill Marinushkin <k.marinushkin@gmail.com> cc: Artem Savkov <asavkov@redhat.com> cc: stable@vger.kernel.org Signed-off-by: James Morris <james.l.morris@oracle.com>
2016-10-26 22:02:01 +08:00
late_initcall(big_key_init);