[PATCH v3 3/9] KEYS: Added KEYRING_CHECK policy for key measurement

[Mimi Zohar]

On Wed, 2019-10-30 at 18:19 -0700, Lakshmi Ramasubramanian wrote:
> An IMA policy to manage measurement of keys is not supported.
> A new IMA policy is needed to manage the measurement of keys.
> A policy option is also needed to allow measurement of keys
> linked to a given set of keyrings only.
> 
> This patch defines KEYRING_CHECK and keyrings in IMA policy
> for this purpose. 

"KEYRING_CHECK" and "keyrings" are not related.   One is a "func"
name, while the other is an IMA policy option.  This should be broken
up into two different patches.  When defining a new policy option, the
only code in that patch should be the new policy option.

> 
> KEYRING_CHECK can be added in the IMA policy to measure keys.
> keyrings can be, optionally, set to only measure keys
> added or updated to a given set of keyrings. If keyrings is not
> specified for KEYRING_CHECK, keys added or updated in
> all keyrings are measured.
> 
> Signed-off-by: Lakshmi Ramasubramanian <nramas at linux.microsoft.com>
> ---
>  Documentation/ABI/testing/ima_policy | 15 +++++++++++++++
>  security/integrity/ima/ima.h         |  1 +
>  security/integrity/ima/ima_main.c    |  2 +-
>  security/integrity/ima/ima_policy.c  |  2 +-
>  security/integrity/ima/ima_queue.c   |  2 +-
>  5 files changed, 19 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/ABI/testing/ima_policy b/Documentation/ABI/testing/ima_policy
> index fc376a323908..757faf1a1a27 100644
> --- a/Documentation/ABI/testing/ima_policy
> +++ b/Documentation/ABI/testing/ima_policy
> @@ -25,10 +25,12 @@ Description:
>  			lsm:	[[subj_user=] [subj_role=] [subj_type=]
>  				 [obj_user=] [obj_role=] [obj_type=]]
>  			option:	[[appraise_type=]] [template=] [permit_directio]
> +				[keyrings=]
>  		base: 	func:= [BPRM_CHECK][MMAP_CHECK][CREDS_CHECK][FILE_CHECK][MODULE_CHECK]
>  				[FIRMWARE_CHECK]
>  				[KEXEC_KERNEL_CHECK] [KEXEC_INITRAMFS_CHECK]
>  				[KEXEC_CMDLINE]
> +				[KEYRING_CHECK]

This patch is measuring keys, not keyrings.


>  			mask:= [[^]MAY_READ] [[^]MAY_WRITE] [[^]MAY_APPEND]
>  			       [[^]MAY_EXEC]
>  			fsmagic:= hex value
> @@ -38,6 +40,9 @@ Description:
>  			fowner:= decimal value
>  		lsm:  	are LSM specific
>  		option:	appraise_type:= [imasig]
> +			keyrings: = list of keyrings to measure
> +			(eg, .builtin_trusted_keys|.ima). Only valid
> +			when action is "measure" and func is KEYRING_CHECK.
>  			template:= name of a defined IMA template type
>  			(eg, ima-ng). Only valid when action is "measure".
>  			pcr:= decimal value
> @@ -105,3 +110,13 @@ Description:
>  
>  			measure func=KEXEC_KERNEL_CHECK pcr=4
>  			measure func=KEXEC_INITRAMFS_CHECK pcr=5
> +
> +		Example of measure rules using KEYRING_CHECK
> +			To measure keys added to
> +			.builtin_trusted_keys or .ima keyring:
> +
> +			measure func=KEYRING_CHECK keyrings=.builtin_trusted_keys|.ima
> +
> +			To measure keys added to all keyrings:
> +
> +			measure func=KEYRING_CHECK

The patch that introduces the new IMA "func" should document the new
IMA "func".  The patch that introduces the new "keyring=" policy
option should document the new IMA policy option.  Examples could be
included in each of the patches descriptions.


> diff --git a/security/integrity/ima/ima.h b/security/integrity/ima/ima.h
> index b9600070e415..12e9ec6847b5 100644
> --- a/security/integrity/ima/ima.h
> +++ b/security/integrity/ima/ima.h
> @@ -191,6 +191,7 @@ static inline unsigned long ima_hash_key(u8 *digest)
>  	hook(KEXEC_INITRAMFS_CHECK)	\
>  	hook(POLICY_CHECK)		\
>  	hook(KEXEC_CMDLINE)		\
> +	hook(KEYRING_CHECK)		\
>  	hook(MAX_CHECK)
>  #define __ima_hook_enumify(ENUM)	ENUM,
>  
> diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c
> index 18e1bc105be7..72ae0878ec5d 100644
> --- a/security/integrity/ima/ima_main.c
> +++ b/security/integrity/ima/ima_main.c
> @@ -718,7 +718,7 @@ void ima_post_key_create_or_update(struct key *keyring, struct key *key,
>  	pk = key->payload.data[asym_crypto];
>  	process_buffer_measurement(pk->key, pk->keylen,
>  				   keyring->description,
> -				   NONE, 0);
> +				   KEYRING_CHECK, 0);
>  }
>  
>  static int __init init_ima(void)
> diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
> index 6df7f641ff66..0cc49f2d5233 100644
> --- a/security/integrity/ima/ima_policy.c
> +++ b/security/integrity/ima/ima_policy.c
> @@ -370,7 +370,7 @@ static bool ima_match_rules(struct ima_rule_entry *rule, struct inode *inode,
>  {
>  	int i;
>  
> -	if (func == KEXEC_CMDLINE) {
> +	if ((func == KEXEC_CMDLINE) || (func == KEYRING_CHECK)) {
>  		if ((rule->flags & IMA_FUNC) && (rule->func == func))
>  			return true;
>  		return false;
> diff --git a/security/integrity/ima/ima_queue.c b/security/integrity/ima/ima_queue.c
> index f2503f10abf4..5625381c5a97 100644
> --- a/security/integrity/ima/ima_queue.c
> +++ b/security/integrity/ima/ima_queue.c
> @@ -317,7 +317,7 @@ void ima_measure_queued_keys(void)
>  		process_buffer_measurement(entry->public_key,
>  					   entry->public_key_len,
>  					   entry->keyring_name,
> -					   NONE, 0);
> +					   KEYRING_CHECK, 0);

Changing a newly defined call should be an indication that the patch
ordering is wrong.  If the new "func" was defined prior or with the
new IMA hook, then this change wouldn't be needed.

Mimi


>  		list_del(&entry->list);
>  		ima_free_measure_key_entry(entry);
>  	}