Re: [PATCH v3 3/4] ima: Do not audit if CONFIG_INTEGRITY_AUDIT is not set

From: Paul Moore
Date: Mon Jun 04 2018 - 20:16:57 EST


On Mon, Jun 4, 2018 at 4:54 PM, Stefan Berger
<stefanb@xxxxxxxxxxxxxxxxxx> wrote:
> If Integrity is not auditing, IMA shouldn't audit, either.
>
> Signed-off-by: Stefan Berger <stefanb@xxxxxxxxxxxxxxxxxx>
> ---
> security/integrity/ima/Kconfig | 1 +
> security/integrity/ima/ima_policy.c | 6 +++++-
> security/integrity/integrity.h | 15 +++++++++++++++
> 3 files changed, 21 insertions(+), 1 deletion(-)
>
> diff --git a/security/integrity/ima/Kconfig b/security/integrity/ima/Kconfig
> index 6a8f67714c83..94c2151331aa 100644
> --- a/security/integrity/ima/Kconfig
> +++ b/security/integrity/ima/Kconfig
> @@ -12,6 +12,7 @@ config IMA
> select TCG_TIS if TCG_TPM && X86
> select TCG_CRB if TCG_TPM && ACPI
> select TCG_IBMVTPM if TCG_TPM && PPC_PSERIES
> + select INTEGRITY_AUDIT if AUDIT
> help
> The Trusted Computing Group(TCG) runtime Integrity
> Measurement Architecture(IMA) maintains a list of hash
> diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
> index 3fcf0935468c..bc99713dfe57 100644
> --- a/security/integrity/ima/ima_policy.c
> +++ b/security/integrity/ima/ima_policy.c
> @@ -628,6 +628,9 @@ static int ima_lsm_rule_init(struct ima_rule_entry *entry,
> static void ima_log_string_op(struct audit_buffer *ab, char *key, char *value,
> bool (*rule_operator)(kuid_t, kuid_t))
> {
> + if (!ab)
> + return;
> +
> if (rule_operator == &uid_gt)
> audit_log_format(ab, "%s>", key);
> else if (rule_operator == &uid_lt)
> @@ -649,7 +652,8 @@ static int ima_parse_rule(char *rule, struct ima_rule_entry *entry)
> bool uid_token;
> int result = 0;
>
> - ab = audit_log_start(NULL, GFP_KERNEL, AUDIT_INTEGRITY_RULE);
> + ab = integrity_audit_log_start(NULL, GFP_KERNEL,
> + AUDIT_INTEGRITY_RULE);

There was a lot of confusion here, so this is understandable, but you
should pass "audit_context()"[1] as the first parameter instead of
NULL. Other than that this patch looks fine.

[1] In Linus' tree at the moment you would need to use
current->audit_context, but the audit PR heading to Linus during this
merge window will introduce the "audit_context()" function which is
preferable as we may need to change things around a bit in the near
future.

> entry->uid = INVALID_UID;
> entry->fowner = INVALID_UID;
> diff --git a/security/integrity/integrity.h b/security/integrity/integrity.h
> index 0bb372eed62a..e60473b13a8d 100644
> --- a/security/integrity/integrity.h
> +++ b/security/integrity/integrity.h
> @@ -15,6 +15,7 @@
> #include <linux/integrity.h>
> #include <crypto/sha.h>
> #include <linux/key.h>
> +#include <linux/audit.h>
>
> /* iint action cache flags */
> #define IMA_MEASURE 0x00000001
> @@ -199,6 +200,13 @@ static inline void evm_load_x509(void)
> void integrity_audit_msg(int audit_msgno, struct inode *inode,
> const unsigned char *fname, const char *op,
> const char *cause, int result, int info);
> +
> +static inline struct audit_buffer *
> +integrity_audit_log_start(struct audit_context *ctx, gfp_t gfp_mask, int type)
> +{
> + return audit_log_start(ctx, gfp_mask, type);
> +}
> +
> #else
> static inline void integrity_audit_msg(int audit_msgno, struct inode *inode,
> const unsigned char *fname,
> @@ -206,4 +214,11 @@ static inline void integrity_audit_msg(int audit_msgno, struct inode *inode,
> int result, int info)
> {
> }
> +
> +static inline struct audit_buffer *
> +integrity_audit_log_start(struct audit_context *ctx, gfp_t gfp_mask, int type)
> +{
> + return NULL;
> +}
> +
> #endif
> --
> 2.13.6

--
paul moore
www.paul-moore.com