lkml.org 
[lkml]   [2022]   [Mar]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v8 5/5] certs: Allow root user to append signed hashes to the blacklist keyring

On 09/03/2022 17:01, Jarkko Sakkinen wrote:
> On Tue, Mar 08, 2022 at 05:02:23PM +0100, Mickaël Salaün wrote:
>>
>> On 08/03/2022 14:19, Jarkko Sakkinen wrote:
>>> On Tue, Mar 08, 2022 at 01:18:28PM +0100, Mickaël Salaün wrote:
>>>>
>>>> On 08/03/2022 12:53, Jarkko Sakkinen wrote:
>>>>> On Mon, Jul 12, 2021 at 07:03:13PM +0200, Mickaël Salaün wrote:
>>>>>> From: Mickaël Salaün <mic@linux.microsoft.com>
>>>>>>
>>>>>> Add a kernel option SYSTEM_BLACKLIST_AUTH_UPDATE to enable the root user
>>>>>> to dynamically add new keys to the blacklist keyring. This enables to
>>>>>> invalidate new certificates, either from being loaded in a keyring, or
>>>>>> from being trusted in a PKCS#7 certificate chain. This also enables to
>>>>>> add new file hashes to be denied by the integrity infrastructure.
>>>>>>
>>>>>> Being able to untrust a certificate which could have normaly been
>>>>>> trusted is a sensitive operation. This is why adding new hashes to the
>>>>>> blacklist keyring is only allowed when these hashes are signed and
>>>>>> vouched by the builtin trusted keyring. A blacklist hash is stored as a
>>>>>> key description. The PKCS#7 signature of this description must be
>>>>>> provided as the key payload.
>>>>>>
>>>>>> Marking a certificate as untrusted should be enforced while the system
>>>>>> is running. It is then forbiden to remove such blacklist keys.
>>>>>>
>>>>>> Update blacklist keyring, blacklist key and revoked certificate access rights:
>>>>>> * allows the root user to search for a specific blacklisted hash, which
>>>>>> make sense because the descriptions are already viewable;
>>>>>> * forbids key update (blacklist and asymmetric ones);
>>>>>> * restricts kernel rights on the blacklist keyring to align with the
>>>>>> root user rights.
>>>>>>
>>>>>> See help in tools/certs/print-cert-tbs-hash.sh .
>>>>>>
>>>>>> Cc: David Howells <dhowells@redhat.com>
>>>>>> Cc: David Woodhouse <dwmw2@infradead.org>
>>>>>> Cc: Eric Snowberg <eric.snowberg@oracle.com>
>>>>>> Cc: Jarkko Sakkinen <jarkko@kernel.org>
>>>>>> Signed-off-by: Mickaël Salaün <mic@linux.microsoft.com>
>>>>>> Link: https://lore.kernel.org/r/20210712170313.884724-6-mic@digikod.net
>>>>>> ---
>>>>>>
>>>>>> Changes since v6:
>>>>>> * Rebase on keys-cve-2020-26541-v3: commit ebd9c2ae369a ("integrity:
>>>>>> Load mokx variables into the blacklist keyring").
>>>>>>
>>>>>> Changes since v5:
>>>>>> * Rebase on keys-next, fix Kconfig conflict, and update the asymmetric
>>>>>> key rights added to the blacklist keyring by the new
>>>>>> add_key_to_revocation_list(): align with blacklist key rights by
>>>>>> removing KEY_POS_WRITE as a safeguard, and add
>>>>>> KEY_ALLOC_BYPASS_RESTRICTION to not be subject to
>>>>>> restrict_link_for_blacklist() that only allows blacklist key types to
>>>>>> be added to the keyring.
>>>>>> * Change the return code for restrict_link_for_blacklist() from -EPERM
>>>>>> to -EOPNOTSUPP to align with asymmetric key keyrings.
>>>>>>
>>>>>> Changes since v3:
>>>>>> * Update commit message for print-cert-tbs-hash.sh .
>>>>>>
>>>>>> Changes since v2:
>>>>>> * Add comment for blacklist_key_instantiate().
>>>>>> ---
>>>>>> certs/Kconfig | 10 +++++
>>>>>> certs/blacklist.c | 96 ++++++++++++++++++++++++++++++++++++-----------
>>>>>> 2 files changed, 85 insertions(+), 21 deletions(-)
>>>>>>
>>>>>> diff --git a/certs/Kconfig b/certs/Kconfig
>>>>>> index 0fbe184ceca5..e0e524b7eff9 100644
>>>>>> --- a/certs/Kconfig
>>>>>> +++ b/certs/Kconfig
>>>>>> @@ -103,4 +103,14 @@ config SYSTEM_REVOCATION_KEYS
>>>>>> containing X.509 certificates to be included in the default blacklist
>>>>>> keyring.
>>>>>> +config SYSTEM_BLACKLIST_AUTH_UPDATE
>>>>>> + bool "Allow root to add signed blacklist keys"
>>>>>> + depends on SYSTEM_BLACKLIST_KEYRING
>>>>>> + depends on SYSTEM_DATA_VERIFICATION
>>>>>> + help
>>>>>> + If set, provide the ability to load new blacklist keys at run time if
>>>>>> + they are signed and vouched by a certificate from the builtin trusted
>>>>>> + keyring. The PKCS#7 signature of the description is set in the key
>>>>>> + payload. Blacklist keys cannot be removed.
>>>>>> +
>>>>>> endmenu
>>>>>> diff --git a/certs/blacklist.c b/certs/blacklist.c
>>>>>> index b254c87ceb3a..486ce0dd8e9c 100644
>>>>>> --- a/certs/blacklist.c
>>>>>> +++ b/certs/blacklist.c
>>>>>> @@ -15,6 +15,7 @@
>>>>>> #include <linux/err.h>
>>>>>> #include <linux/seq_file.h>
>>>>>> #include <linux/uidgid.h>
>>>>>> +#include <linux/verification.h>
>>>>>> #include <keys/system_keyring.h>
>>>>>> #include "blacklist.h"
>>>>>> #include "common.h"
>>>>>> @@ -26,6 +27,9 @@
>>>>>> */
>>>>>> #define MAX_HASH_LEN 128
>>>>>> +#define BLACKLIST_KEY_PERM (KEY_POS_SEARCH | KEY_POS_VIEW | \
>>>>>> + KEY_USR_SEARCH | KEY_USR_VIEW)
>>>>>> +
>>>>>> static const char tbs_prefix[] = "tbs";
>>>>>> static const char bin_prefix[] = "bin";
>>>>>> @@ -80,19 +84,51 @@ static int blacklist_vet_description(const char *desc)
>>>>>> return 0;
>>>>>> }
>>>>>> -/*
>>>>>> - * The hash to be blacklisted is expected to be in the description. There will
>>>>>> - * be no payload.
>>>>>> - */
>>>>>> -static int blacklist_preparse(struct key_preparsed_payload *prep)
>>>>>> +static int blacklist_key_instantiate(struct key *key,
>>>>>> + struct key_preparsed_payload *prep)
>>>>>> {
>>>>>> - if (prep->datalen > 0)
>>>>>> - return -EINVAL;
>>>>>> - return 0;
>>>>>> +#ifdef CONFIG_SYSTEM_BLACKLIST_AUTH_UPDATE
>>>>>> + int err;
>>>>>> +#endif
>>>>>> +
>>>>>> + /* Sets safe default permissions for keys loaded by user space. */
>>>>>> + key->perm = BLACKLIST_KEY_PERM;
>>>>>> +
>>>>>> + /*
>>>>>> + * Skips the authentication step for builtin hashes, they are not
>>>>>> + * signed but still trusted.
>>>>>> + */
>>>>>> + if (key->flags & (1 << KEY_FLAG_BUILTIN))
>>>>>> + goto out;
>>>>>> +
>>>>>> +#ifdef CONFIG_SYSTEM_BLACKLIST_AUTH_UPDATE
>>>>>> + /*
>>>>>> + * Verifies the description's PKCS#7 signature against the builtin
>>>>>> + * trusted keyring.
>>>>>> + */
>>>>>> + err = verify_pkcs7_signature(key->description,
>>>>>> + strlen(key->description), prep->data, prep->datalen,
>>>>>> + NULL, VERIFYING_UNSPECIFIED_SIGNATURE, NULL, NULL);
>>>>>> + if (err)
>>>>>> + return err;
>>>>>> +#else
>>>>>> + /*
>>>>>> + * It should not be possible to come here because the keyring doesn't
>>>>>> + * have KEY_USR_WRITE and the only other way to call this function is
>>>>>> + * for builtin hashes.
>>>>>> + */
>>>>>> + WARN_ON_ONCE(1);
>>>>>> + return -EPERM;
>>>>>> +#endif
>>>>>> +
>>>>>> +out:
>>>>>> + return generic_key_instantiate(key, prep);
>>>>>> }
>>>>>> -static void blacklist_free_preparse(struct key_preparsed_payload *prep)
>>>>>> +static int blacklist_key_update(struct key *key,
>>>>>> + struct key_preparsed_payload *prep)
>>>>>> {
>>>>>> + return -EPERM;
>>>>>> }
>>>>>> static void blacklist_describe(const struct key *key, struct seq_file *m)
>>>>>> @@ -103,9 +139,8 @@ static void blacklist_describe(const struct key *key, struct seq_file *m)
>>>>>> static struct key_type key_type_blacklist = {
>>>>>> .name = "blacklist",
>>>>>> .vet_description = blacklist_vet_description,
>>>>>> - .preparse = blacklist_preparse,
>>>>>> - .free_preparse = blacklist_free_preparse,
>>>>>> - .instantiate = generic_key_instantiate,
>>>>>> + .instantiate = blacklist_key_instantiate,
>>>>>> + .update = blacklist_key_update,
>>>>>> .describe = blacklist_describe,
>>>>>> };
>>>>>> @@ -154,8 +189,7 @@ static int mark_raw_hash_blacklisted(const char *hash)
>>>>>> hash,
>>>>>> NULL,
>>>>>> 0,
>>>>>> - ((KEY_POS_ALL & ~KEY_POS_SETATTR) |
>>>>>> - KEY_USR_VIEW),
>>>>>> + BLACKLIST_KEY_PERM,
>>>>>> KEY_ALLOC_NOT_IN_QUOTA |
>>>>>> KEY_ALLOC_BUILT_IN);
>>>>>> if (IS_ERR(key)) {
>>>>>> @@ -232,8 +266,10 @@ int add_key_to_revocation_list(const char *data, size_t size)
>>>>>> NULL,
>>>>>> data,
>>>>>> size,
>>>>>> - ((KEY_POS_ALL & ~KEY_POS_SETATTR) | KEY_USR_VIEW),
>>>>>> - KEY_ALLOC_NOT_IN_QUOTA | KEY_ALLOC_BUILT_IN);
>>>>>> + KEY_POS_VIEW | KEY_POS_READ | KEY_POS_SEARCH
>>>>>> + | KEY_USR_VIEW,
>>>>>> + KEY_ALLOC_NOT_IN_QUOTA | KEY_ALLOC_BUILT_IN
>>>>>> + | KEY_ALLOC_BYPASS_RESTRICTION);
>>>>>> if (IS_ERR(key)) {
>>>>>> pr_err("Problem with revocation key (%ld)\n", PTR_ERR(key));
>>>>>> @@ -260,25 +296,43 @@ int is_key_on_revocation_list(struct pkcs7_message *pkcs7)
>>>>>> }
>>>>>> #endif
>>>>>> +static int restrict_link_for_blacklist(struct key *dest_keyring,
>>>>>> + const struct key_type *type, const union key_payload *payload,
>>>>>> + struct key *restrict_key)
>>>>>> +{
>>>>>> + if (type == &key_type_blacklist)
>>>>>> + return 0;
>>>>>> + return -EOPNOTSUPP;
>>>>>> +}
>>>>>> +
>>>>>> /*
>>>>>> * Initialise the blacklist
>>>>>> */
>>>>>> static int __init blacklist_init(void)
>>>>>> {
>>>>>> const char *const *bl;
>>>>>> + struct key_restriction *restriction;
>>>>>> if (register_key_type(&key_type_blacklist) < 0)
>>>>>> panic("Can't allocate system blacklist key type\n");
>>>>>> + restriction = kzalloc(sizeof(*restriction), GFP_KERNEL);
>>>>>> + if (!restriction)
>>>>>> + panic("Can't allocate blacklist keyring restriction\n");
>>>>>
>>>>>
>>>>> This prevents me from taking this to my pull request. In moderns standards,
>>>>> no new BUG_ON(), panic() etc. should never added to the kernel.
>>>>>
>>>>> I missed this in my review.
>>>>>
>>>>> This should rather be e.g.
>>>>>
>>>>> restriction = kzalloc(sizeof(*restriction), GFP_KERNEL);
>>>>> if (!restriction) {
>>>>> pr_err("Can't allocate blacklist keyring restriction\n");
>>>>> return 0;
>>>>> }
>>>>>
>>>>> Unfortunately I need to drop this patch set, because adding new panic()
>>>>> is simply a no-go.
>>>>
>>>> I agree that panic() is not great in general, but I followed the other part
>>>> of the code (just above) that do the same. This part of the kernel should
>>>> failed if critical memory allocation failed at boot time (only). It doesn't
>>>> impact the kernel once it is running. I don't think that just ignoring this
>>>> error with return 0 is fine, after all it's a critical error right?
>>>
>>> It's not good reason enough to crash the whole kernel, even if it is a
>>> critical error (e.g. run-time foresincs). Even WARN() is not recommended
>>> these days [*].
>>
>> I think that what Greg said in this email is that WARN*() should only be
>> used for cases that should never happen, it is definitely not deprecated,
>> but WARN_ON_ONCE() may be a better idea though. WARN*() helps detect such
>> thought-to-be-impossible cases, that can happen e.g. with code refactoring.
>>
>> A lot of initialization/boot code (e.g. without user space nor external
>> interactions, mostly __init functions) do panic if there is unexpected and
>> unrecoverable errors like failed memory allocations. I think handling such
>> errors otherwise would be more complex for no benefit. Moreover, delegating
>> such error handling to user space could create new (silent) issues.
>
> To crash the whole kernel, you should be able to clearly explain why it
> makes sense in the situation.

If there is no enough memory to allocate 24 bytes (struct
key_restriction) during early boot, I really doubt the kernel can do much.

>
>>>
>>> For the existing panic()-statements: I'm happy to review patches that
>>> render them out. >
>>> Not sure tho, if this fails should it be then "everything blacklisted".
>>> Just one thing to consider.
>>
>> Well, if it fail it will be "nothing will work afterwards". Do you have a
>> working and useful scenario for this kind of error?
>
> So you have zero chances to get a shell without blacklist just to do
> kernel forensics?

Right, I don't think the kernel can launch any process (nor continue
early boot) if it cannot allocate 24 bytes.

\
 
 \ /
  Last update: 2022-03-09 19:38    [W:0.246 / U:0.800 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site