Hi
I think that touching the internals of the key structure and duplicating
the function user_key_payload in dm-crypt is not quite right. Duplicating
the code is bad and duplicating it in a different module is even worse -
after long time it becomes problematic to find all the duplicates.
I suggest to keep using user_key_payload() and change rcu_read_lock() to
down_read(&key->sem) and rcu_read_unlock() to up_read(&key->sem) - this is
not performance-sensitive code, so the locks won't hurt.
BTW. Dawid Howells, the maintaner of keyring said that he could eventually
provide a function to access the key payload from RCU read section. When
he does, we can revert back to RCU.
Mikulas
On Tue, 31 Jan 2017, Ondrej Kozina wrote:
> This fixes following lockdep splat emerging on
> table load with a key in kernel keyring service.
>
> In fact it hints a bug in RCU usage in dm-crypt since
> kernel keyring fn user_key_payload() is in fact a wrapper
> for rcu_dereference_protected() which must not be used
> with only rcu_read_lock() mark.
>
> ===============================
> [ INFO: suspicious RCU usage. ]
> 4.10.0-rc5 #2 Not tainted
> -------------------------------
> ./include/keys/user-type.h:53 suspicious usage!
> other info that might help us debug this:
> rcu_scheduler_active = 2, debug_locks = 1
> 2 locks held by cryptsetup/5555:
> #0: (&md->type_lock){+.+.+.}, at: [<ffffffffa02472a2>] dm_lock_md_type+0x12/0x20 [dm_mod]
> #1: (rcu_read_lock){......}, at: [<ffffffffa026b2f8>] crypt_set_key+0x1d8/0x4b0 [dm_crypt]
> stack backtrace:
> CPU: 1 PID: 5555 Comm: cryptsetup Not tainted 4.10.0-rc5 #2
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.1-1.fc24 04/01/2014
> Call Trace:
> dump_stack+0x67/0x92
> lockdep_rcu_suspicious+0xc5/0x100
> crypt_set_key+0x351/0x4b0 [dm_crypt]
> ? crypt_set_key+0x1d8/0x4b0 [dm_crypt]
> crypt_ctr+0x341/0xa53 [dm_crypt]
> dm_table_add_target+0x147/0x330 [dm_mod]
> table_load+0x111/0x350 [dm_mod]
> ? retrieve_status+0x1c0/0x1c0 [dm_mod]
> ctl_ioctl+0x1f5/0x510 [dm_mod]
> dm_ctl_ioctl+0xe/0x20 [dm_mod]
> do_vfs_ioctl+0x8e/0x690
> ? task_work_run+0x7e/0xa0
> ? trace_hardirqs_on_caller+0x122/0x1b0
> SyS_ioctl+0x3c/0x70
> entry_SYSCALL_64_fastpath+0x18/0xad
> RIP: 0033:0x7f1be07ceec7
> RSP: 002b:00007fff38c094d8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> RAX: ffffffffffffffda RBX: 000000000d4dbae9 RCX: 00007f1be07ceec7
> RDX: 0000000001768100 RSI: 00000000c138fd09 RDI: 0000000000000005
> RBP: 0000000000000006 R08: 00000000ffffffff R09: 00000000017623f0
> R10: 2a28205d34383336 R11: 0000000000000246 R12: 0000000000000001
> R13: 00007fff38c0987c R14: 00007fff38c097ec R15: 0000000000000000
>
> Reported-by: Milan Broz <mbroz@xxxxxxxxxx>
> Fixes: c538f6ec9f56 (dm crypt: add ability to use keys from the kernel key retention service)
> Signed-off-by: Ondrej Kozina <okozina@xxxxxxxxxx>
> ---
> drivers/md/dm-crypt.c | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
> index 7c6c572..37d5027 100644
> --- a/drivers/md/dm-crypt.c
> +++ b/drivers/md/dm-crypt.c
> @@ -31,6 +31,7 @@
> #include <crypto/md5.h>
> #include <crypto/algapi.h>
> #include <crypto/skcipher.h>
> +#include <linux/rcupdate.h>
> #include <keys/user-type.h>
>
> #include <linux/device-mapper.h>
> @@ -1498,6 +1499,12 @@ static bool contains_whitespace(const char *str)
> return false;
> }
>
> +/* workaround for missing RCU read key payload wrapper */
> +static const struct user_key_payload *_user_key_payload(const struct key *key)
> +{
> + return (const struct user_key_payload *) rcu_dereference(key->payload.rcu_data0);
> +}
> +
> static int crypt_set_keyring_key(struct crypt_config *cc, const char *key_string)
> {
> char *new_key_string, *key_desc;
> @@ -1536,7 +1543,7 @@ static int crypt_set_keyring_key(struct crypt_config *cc, const char *key_string
>
> rcu_read_lock();
>
> - ukp = user_key_payload(key);
> + ukp = _user_key_payload(key);
> if (!ukp) {
> rcu_read_unlock();
> key_put(key);
> --
> 2.7.4
>
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
