[PATCH v3 0/5] Audit: Records for multiple security contexts

[PATCH v3 0/5] Audit: Records for multiple security contexts

[Casey Schaufler]

The Linux audit system includes LSM based security "context" information
in its events. Historically, only one LSM that uses security contexts can
be active on a system. One of the few obsticles to allowing multiple LSM
support is the inability to report more than one security context in an
audit event. This patchset provides a mechanism to provide supplimental
records containing more than one security context for subjects and
objects.

The mechanism for reporting multiple security contexts inspired
considerable discussion. It would have been possible to add multiple
contexts to existing records using sophisticated formatting. This would
have significant backward compatibility issues, and require additional
parsing in user space code. Adding new records for an event that contain
the contexts is more in keeping with the way audit events have been
constructed in the past.

Only audit events associated with system calls have required multiple
records prior to this. Mechanism has been added allowing any event
to be composed of multiple records. This should make it easier to
add information to existing audit events without breaking backward
compatability.

v3:
Rework how security modules identify that they provide security
contexts to the audit system. Maintain a list within the audit
system of the security modules that provide security contexts.
Revert the separate counts of subject and object contexts.
v2:
Maintain separate counts for LSMs using subject contexts and object
contexts. AppArmor uses the former but not the latter.
Correct error handling in object record creation.

https://github.com/cschaufler/lsm-stacking#audit-6.14-rc1-v3

Casey Schaufler (5):
  Audit: Create audit_stamp structure
  LSM: security_lsmblob_to_secctx module selection
  Audit: Add record for multiple task security contexts
  Audit: multiple subject lsm values for netlabel
  Audit: Add record for multiple object contexts

 include/linux/audit.h        |  19 +++
 include/linux/security.h     |   6 +-
 include/uapi/linux/audit.h   |   2 +
 kernel/audit.c               | 255 +++++++++++++++++++++++++++++------
 kernel/audit.h               |  13 +-
 kernel/auditsc.c             |  65 +++------
 net/netlabel/netlabel_user.c |   8 +-
 security/apparmor/lsm.c      |   3 +
 security/security.c          |  13 +-
 security/selinux/hooks.c     |   3 +
 security/smack/smack_lsm.c   |   3 +
 11 files changed, 291 insertions(+), 99 deletions(-)

-- 
2.47.0

[PATCH v3 1/5] Audit: Create audit_stamp structure

[Casey Schaufler]

Replace the timestamp and serial number pair used in audit records
with a structure containing the two elements.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 kernel/audit.c   | 17 +++++++++--------
 kernel/audit.h   | 13 +++++++++----
 kernel/auditsc.c | 22 +++++++++-------------
 3 files changed, 27 insertions(+), 25 deletions(-)

diff --git a/kernel/audit.c b/kernel/audit.c
index 5f5bf85bcc90..2a567f667528 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -1833,11 +1833,11 @@ unsigned int audit_serial(void)
 }
 
 static inline void audit_get_stamp(struct audit_context *ctx,
-				   struct timespec64 *t, unsigned int *serial)
+				   struct audit_stamp *stamp)
 {
-	if (!ctx || !auditsc_get_stamp(ctx, t, serial)) {
-		ktime_get_coarse_real_ts64(t);
-		*serial = audit_serial();
+	if (!ctx || !auditsc_get_stamp(ctx, stamp)) {
+		ktime_get_coarse_real_ts64(&stamp->ctime);
+		stamp->serial = audit_serial();
 	}
 }
 
@@ -1860,8 +1860,7 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 				     int type)
 {
 	struct audit_buffer *ab;
-	struct timespec64 t;
-	unsigned int serial;
+	struct audit_stamp stamp;
 
 	if (audit_initialized != AUDIT_INITIALIZED)
 		return NULL;
@@ -1916,12 +1915,14 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 		return NULL;
 	}
 
-	audit_get_stamp(ab->ctx, &t, &serial);
+	audit_get_stamp(ab->ctx, &stamp);
 	/* cancel dummy context to enable supporting records */
 	if (ctx)
 		ctx->dummy = 0;
 	audit_log_format(ab, "audit(%llu.%03lu:%u): ",
-			 (unsigned long long)t.tv_sec, t.tv_nsec/1000000, serial);
+			 (unsigned long long)stamp.ctime.tv_sec,
+			 stamp.ctime.tv_nsec/1000000,
+			 stamp.serial);
 
 	return ab;
 }
diff --git a/kernel/audit.h b/kernel/audit.h
index 0211cb307d30..4d6dd2588f9b 100644
--- a/kernel/audit.h
+++ b/kernel/audit.h
@@ -99,6 +99,12 @@ struct audit_proctitle {
 	char	*value;	/* the cmdline field */
 };
 
+/* A timestamp/serial pair to identify an event */
+struct audit_stamp {
+	struct timespec64	ctime;	/* time of syscall entry */
+	unsigned int		serial;	/* serial number for record */
+};
+
 /* The per-task audit context. */
 struct audit_context {
 	int		    dummy;	/* must be the first element */
@@ -108,10 +114,9 @@ struct audit_context {
 		AUDIT_CTX_URING,	/* in use by io_uring */
 	} context;
 	enum audit_state    state, current_state;
-	unsigned int	    serial;     /* serial number for record */
+	struct audit_stamp  stamp;	/* event identifier */
 	int		    major;      /* syscall number */
 	int		    uring_op;   /* uring operation */
-	struct timespec64   ctime;      /* time of syscall entry */
 	unsigned long	    argv[4];    /* syscall arguments */
 	long		    return_code;/* syscall return code */
 	u64		    prio;
@@ -263,7 +268,7 @@ extern void audit_put_tty(struct tty_struct *tty);
 extern unsigned int audit_serial(void);
 #ifdef CONFIG_AUDITSYSCALL
 extern int auditsc_get_stamp(struct audit_context *ctx,
-			      struct timespec64 *t, unsigned int *serial);
+			     struct audit_stamp *stamp);
 
 extern void audit_put_watch(struct audit_watch *watch);
 extern void audit_get_watch(struct audit_watch *watch);
@@ -304,7 +309,7 @@ extern void audit_filter_inodes(struct task_struct *tsk,
 				struct audit_context *ctx);
 extern struct list_head *audit_killed_trees(void);
 #else /* CONFIG_AUDITSYSCALL */
-#define auditsc_get_stamp(c, t, s) 0
+#define auditsc_get_stamp(c, s) 0
 #define audit_put_watch(w) do { } while (0)
 #define audit_get_watch(w) do { } while (0)
 #define audit_to_watch(k, p, l, o) (-EINVAL)
diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index 9c853cde9abe..60f2c927afd7 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -994,10 +994,10 @@ static void audit_reset_context(struct audit_context *ctx)
 	 */
 
 	ctx->current_state = ctx->state;
-	ctx->serial = 0;
+	ctx->stamp.serial = 0;
+	ctx->stamp.ctime = (struct timespec64){ .tv_sec = 0, .tv_nsec = 0 };
 	ctx->major = 0;
 	ctx->uring_op = 0;
-	ctx->ctime = (struct timespec64){ .tv_sec = 0, .tv_nsec = 0 };
 	memset(ctx->argv, 0, sizeof(ctx->argv));
 	ctx->return_code = 0;
 	ctx->prio = (ctx->state == AUDIT_STATE_RECORD ? ~0ULL : 0);
@@ -1917,7 +1917,7 @@ void __audit_uring_entry(u8 op)
 
 	ctx->context = AUDIT_CTX_URING;
 	ctx->current_state = ctx->state;
-	ktime_get_coarse_real_ts64(&ctx->ctime);
+	ktime_get_coarse_real_ts64(&ctx->stamp.ctime);
 }
 
 /**
@@ -2039,7 +2039,7 @@ void __audit_syscall_entry(int major, unsigned long a1, unsigned long a2,
 	context->argv[3]    = a4;
 	context->context = AUDIT_CTX_SYSCALL;
 	context->current_state  = state;
-	ktime_get_coarse_real_ts64(&context->ctime);
+	ktime_get_coarse_real_ts64(&context->stamp.ctime);
 }
 
 /**
@@ -2510,21 +2510,17 @@ EXPORT_SYMBOL_GPL(__audit_inode_child);
 /**
  * auditsc_get_stamp - get local copies of audit_context values
  * @ctx: audit_context for the task
- * @t: timespec64 to store time recorded in the audit_context
- * @serial: serial value that is recorded in the audit_context
+ * @stamp: timestamp to record
  *
  * Also sets the context as auditable.
  */
-int auditsc_get_stamp(struct audit_context *ctx,
-		       struct timespec64 *t, unsigned int *serial)
+int auditsc_get_stamp(struct audit_context *ctx, struct audit_stamp *stamp)
 {
 	if (ctx->context == AUDIT_CTX_UNUSED)
 		return 0;
-	if (!ctx->serial)
-		ctx->serial = audit_serial();
-	t->tv_sec  = ctx->ctime.tv_sec;
-	t->tv_nsec = ctx->ctime.tv_nsec;
-	*serial    = ctx->serial;
+	if (!ctx->stamp.serial)
+		ctx->stamp.serial = audit_serial();
+	*stamp = ctx->stamp;
 	if (!ctx->prio) {
 		ctx->prio = 1;
 		ctx->current_state = AUDIT_STATE_RECORD;
-- 
2.47.0

Re: [PATCH v3 1/5] Audit: Create audit_stamp structure

[Paul Moore]

On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> 
> Replace the timestamp and serial number pair used in audit records
> with a structure containing the two elements.
> 
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> ---
>  kernel/audit.c   | 17 +++++++++--------
>  kernel/audit.h   | 13 +++++++++----
>  kernel/auditsc.c | 22 +++++++++-------------
>  3 files changed, 27 insertions(+), 25 deletions(-)

Looks good to me, thanks for moving the timestamp and serial number
closer together.

--
paul-moore.com

[PATCH v3 2/5] LSM: security_lsmblob_to_secctx module selection

[Casey Schaufler]

Add a parameter lsmid to security_lsmblob_to_secctx() to identify which
of the security modules that may be active should provide the security
context. If the value of lsmid is LSM_ID_UNDEF the first LSM providing
a hook is used. security_secid_to_secctx() is unchanged, and will
always report the first LSM providing a hook.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 include/linux/security.h     |  6 ++++--
 kernel/audit.c               |  4 ++--
 kernel/auditsc.c             |  8 +++++---
 net/netlabel/netlabel_user.c |  3 ++-
 security/security.c          | 13 +++++++++++--
 5 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/include/linux/security.h b/include/linux/security.h
index 980b6c207cad..540894695c4b 100644
--- a/include/linux/security.h
+++ b/include/linux/security.h
@@ -566,7 +566,8 @@ int security_setprocattr(int lsmid, const char *name, void *value, size_t size);
 int security_netlink_send(struct sock *sk, struct sk_buff *skb);
 int security_ismaclabel(const char *name);
 int security_secid_to_secctx(u32 secid, struct lsm_context *cp);
-int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp);
+int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp,
+			       int lsmid);
 int security_secctx_to_secid(const char *secdata, u32 seclen, u32 *secid);
 void security_release_secctx(struct lsm_context *cp);
 void security_inode_invalidate_secctx(struct inode *inode);
@@ -1543,7 +1544,8 @@ static inline int security_secid_to_secctx(u32 secid, struct lsm_context *cp)
 }
 
 static inline int security_lsmprop_to_secctx(struct lsm_prop *prop,
-					     struct lsm_context *cp)
+					     struct lsm_context *cp,
+					     int lsmid)
 {
 	return -EOPNOTSUPP;
 }
diff --git a/kernel/audit.c b/kernel/audit.c
index 2a567f667528..6bbadb605ca3 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -1473,7 +1473,7 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh,
 	case AUDIT_SIGNAL_INFO:
 		if (lsmprop_is_set(&audit_sig_lsm)) {
 			err = security_lsmprop_to_secctx(&audit_sig_lsm,
-							 &lsmctx);
+							 &lsmctx, LSM_ID_UNDEF);
 			if (err < 0)
 				return err;
 		}
@@ -2188,7 +2188,7 @@ int audit_log_task_context(struct audit_buffer *ab)
 	if (!lsmprop_is_set(&prop))
 		return 0;
 
-	error = security_lsmprop_to_secctx(&prop, &ctx);
+	error = security_lsmprop_to_secctx(&prop, &ctx, LSM_ID_UNDEF);
 	if (error < 0) {
 		if (error != -EINVAL)
 			goto error_path;
diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index 60f2c927afd7..dc3f7e9666f2 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -1109,7 +1109,7 @@ static int audit_log_pid_context(struct audit_context *context, pid_t pid,
 			 from_kuid(&init_user_ns, auid),
 			 from_kuid(&init_user_ns, uid), sessionid);
 	if (lsmprop_is_set(prop)) {
-		if (security_lsmprop_to_secctx(prop, &ctx) < 0) {
+		if (security_lsmprop_to_secctx(prop, &ctx, LSM_ID_UNDEF) < 0) {
 			audit_log_format(ab, " obj=(none)");
 			rc = 1;
 		} else {
@@ -1395,7 +1395,8 @@ static void show_special(struct audit_context *context, int *call_panic)
 			struct lsm_context lsmctx;
 
 			if (security_lsmprop_to_secctx(&context->ipc.oprop,
-						       &lsmctx) < 0) {
+						       &lsmctx,
+						       LSM_ID_UNDEF) < 0) {
 				*call_panic = 1;
 			} else {
 				audit_log_format(ab, " obj=%s", lsmctx.context);
@@ -1560,7 +1561,8 @@ static void audit_log_name(struct audit_context *context, struct audit_names *n,
 	if (lsmprop_is_set(&n->oprop)) {
 		struct lsm_context ctx;
 
-		if (security_lsmprop_to_secctx(&n->oprop, &ctx) < 0) {
+		if (security_lsmprop_to_secctx(&n->oprop, &ctx,
+					       LSM_ID_UNDEF) < 0) {
 			if (call_panic)
 				*call_panic = 2;
 		} else {
diff --git a/net/netlabel/netlabel_user.c b/net/netlabel/netlabel_user.c
index 0d04d23aafe7..6d6545297ee3 100644
--- a/net/netlabel/netlabel_user.c
+++ b/net/netlabel/netlabel_user.c
@@ -98,7 +98,8 @@ struct audit_buffer *netlbl_audit_start_common(int type,
 			 audit_info->sessionid);
 
 	if (lsmprop_is_set(&audit_info->prop) &&
-	    security_lsmprop_to_secctx(&audit_info->prop, &ctx) > 0) {
+	    security_lsmprop_to_secctx(&audit_info->prop, &ctx,
+				       LSM_ID_UNDEF) > 0) {
 		audit_log_format(audit_buf, " subj=%s", ctx.context);
 		security_release_secctx(&ctx);
 	}
diff --git a/security/security.c b/security/security.c
index 143561ebc3e8..55f9c7ad3f89 100644
--- a/security/security.c
+++ b/security/security.c
@@ -4312,6 +4312,7 @@ EXPORT_SYMBOL(security_ismaclabel);
  * security_secid_to_secctx() - Convert a secid to a secctx
  * @secid: secid
  * @cp: the LSM context
+ * @lsmid: which security module to report
  *
  * Convert secid to security context.  If @cp is NULL the length of the
  * result will be returned, but no data will be returned.  This
@@ -4338,9 +4339,17 @@ EXPORT_SYMBOL(security_secid_to_secctx);
  *
  * Return: Return length of data on success, error on failure.
  */
-int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp)
+int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp,
+			       int lsmid)
 {
-	return call_int_hook(lsmprop_to_secctx, prop, cp);
+	struct lsm_static_call *scall;
+
+	lsm_for_each_hook(scall, lsmprop_to_secctx) {
+		if (lsmid != 0 && lsmid != scall->hl->lsmid->id)
+			continue;
+		return scall->hl->hook.lsmprop_to_secctx(prop, cp);
+	}
+	return LSM_RET_DEFAULT(lsmprop_to_secctx);
 }
 EXPORT_SYMBOL(security_lsmprop_to_secctx);
 
-- 
2.47.0

Re: [PATCH v3 2/5] LSM: security_lsmblob_to_secctx module selection

[Fan Wu]

On Wed, Mar 19, 2025 at 7:50 PM Casey Schaufler <casey@schaufler-ca.com> wrote:
>
> Add a parameter lsmid to security_lsmblob_to_secctx() to identify which
> of the security modules that may be active should provide the security
> context. If the value of lsmid is LSM_ID_UNDEF the first LSM providing
> a hook is used. security_secid_to_secctx() is unchanged, and will
> always report the first LSM providing a hook.
>
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
...
> diff --git a/security/security.c b/security/security.c
> index 143561ebc3e8..55f9c7ad3f89 100644
> --- a/security/security.c
> +++ b/security/security.c
> @@ -4312,6 +4312,7 @@ EXPORT_SYMBOL(security_ismaclabel);
>   * security_secid_to_secctx() - Convert a secid to a secctx
>   * @secid: secid
>   * @cp: the LSM context
> + * @lsmid: which security module to report
>   *
>   * Convert secid to security context.  If @cp is NULL the length of the
>   * result will be returned, but no data will be returned.  This
> @@ -4338,9 +4339,17 @@ EXPORT_SYMBOL(security_secid_to_secctx);
>   *
>   * Return: Return length of data on success, error on failure.
>   */
> -int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp)
> +int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp,
> +                              int lsmid)
>  {
> -       return call_int_hook(lsmprop_to_secctx, prop, cp);
> +       struct lsm_static_call *scall;
> +
> +       lsm_for_each_hook(scall, lsmprop_to_secctx) {
> +               if (lsmid != 0 && lsmid != scall->hl->lsmid->id)

It took me some time to figure out why if LSM_ID_UNDEF is passed the
first LSM providing a hook is used, might be better to change it to:

               if (lsmid != LSM_ID_UNDEF && lsmid != scall->hl->lsmid->id)

Otherwise, it works as described. I'm working on adding a new IPE
property based on SELinux file labels, and this just came up as I
needed it. Thank you.

Tested-by: Fan Wu <wufan@kernel.org>

> +                       continue;
> +               return scall->hl->hook.lsmprop_to_secctx(prop, cp);
> +       }
> +       return LSM_RET_DEFAULT(lsmprop_to_secctx);
>  }
>  EXPORT_SYMBOL(security_lsmprop_to_secctx);
>
> --
> 2.47.0
>
>

Re: [PATCH v3 2/5] LSM: security_lsmblob_to_secctx module selection

[Casey Schaufler]

On 3/25/2025 4:44 PM, Fan Wu wrote:
> On Wed, Mar 19, 2025 at 7:50 PM Casey Schaufler <casey@schaufler-ca.com> wrote:
>> Add a parameter lsmid to security_lsmblob_to_secctx() to identify which
>> of the security modules that may be active should provide the security
>> context. If the value of lsmid is LSM_ID_UNDEF the first LSM providing
>> a hook is used. security_secid_to_secctx() is unchanged, and will
>> always report the first LSM providing a hook.
>>
>> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> ..
>> diff --git a/security/security.c b/security/security.c
>> index 143561ebc3e8..55f9c7ad3f89 100644
>> --- a/security/security.c
>> +++ b/security/security.c
>> @@ -4312,6 +4312,7 @@ EXPORT_SYMBOL(security_ismaclabel);
>>   * security_secid_to_secctx() - Convert a secid to a secctx
>>   * @secid: secid
>>   * @cp: the LSM context
>> + * @lsmid: which security module to report
>>   *
>>   * Convert secid to security context.  If @cp is NULL the length of the
>>   * result will be returned, but no data will be returned.  This
>> @@ -4338,9 +4339,17 @@ EXPORT_SYMBOL(security_secid_to_secctx);
>>   *
>>   * Return: Return length of data on success, error on failure.
>>   */
>> -int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp)
>> +int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp,
>> +                              int lsmid)
>>  {
>> -       return call_int_hook(lsmprop_to_secctx, prop, cp);
>> +       struct lsm_static_call *scall;
>> +
>> +       lsm_for_each_hook(scall, lsmprop_to_secctx) {
>> +               if (lsmid != 0 && lsmid != scall->hl->lsmid->id)
> It took me some time to figure out why if LSM_ID_UNDEF is passed the
> first LSM providing a hook is used, might be better to change it to:
>
>                if (lsmid != LSM_ID_UNDEF && lsmid != scall->hl->lsmid->id)

Thank you. That change will be in v4.

>
> Otherwise, it works as described. I'm working on adding a new IPE
> property based on SELinux file labels, and this just came up as I
> needed it. Thank you.
>
> Tested-by: Fan Wu <wufan@kernel.org>
>
>> +                       continue;
>> +               return scall->hl->hook.lsmprop_to_secctx(prop, cp);
>> +       }
>> +       return LSM_RET_DEFAULT(lsmprop_to_secctx);
>>  }
>>  EXPORT_SYMBOL(security_lsmprop_to_secctx);
>>
>> --
>> 2.47.0
>>
>>

Re: [PATCH v3 2/5] LSM: security_lsmblob_to_secctx module selection

[Paul Moore]

On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> 
> Add a parameter lsmid to security_lsmblob_to_secctx() to identify which
> of the security modules that may be active should provide the security
> context. If the value of lsmid is LSM_ID_UNDEF the first LSM providing
> a hook is used. security_secid_to_secctx() is unchanged, and will
> always report the first LSM providing a hook.
> 
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> Tested-by: Fan Wu <wufan@kernel.org>
> ---
>  include/linux/security.h     |  6 ++++--
>  kernel/audit.c               |  4 ++--
>  kernel/auditsc.c             |  8 +++++---
>  net/netlabel/netlabel_user.c |  3 ++-
>  security/security.c          | 13 +++++++++++--
>  5 files changed, 24 insertions(+), 10 deletions(-)

...

> diff --git a/security/security.c b/security/security.c
> index 143561ebc3e8..55f9c7ad3f89 100644
> --- a/security/security.c
> +++ b/security/security.c
> @@ -4312,6 +4312,7 @@ EXPORT_SYMBOL(security_ismaclabel);
>   * security_secid_to_secctx() - Convert a secid to a secctx
>   * @secid: secid
>   * @cp: the LSM context
> + * @lsmid: which security module to report
>   *
>   * Convert secid to security context.  If @cp is NULL the length of the
>   * result will be returned, but no data will be returned.  This
> @@ -4338,9 +4339,17 @@ EXPORT_SYMBOL(security_secid_to_secctx);
>   *
>   * Return: Return length of data on success, error on failure.
>   */
> -int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp)
> +int security_lsmprop_to_secctx(struct lsm_prop *prop, struct lsm_context *cp,
> +			       int lsmid)
>  {
> -	return call_int_hook(lsmprop_to_secctx, prop, cp);
> +	struct lsm_static_call *scall;
> +
> +	lsm_for_each_hook(scall, lsmprop_to_secctx) {
> +		if (lsmid != 0 && lsmid != scall->hl->lsmid->id)
> +			continue;

As mentioned in the v2 review:

 "Let's use LSM_ID_UNDEF instead of 0 here to add some clarity on
  how an undefined ID is handled.  The function header comment
  should also explain the special handling when LSM_ID_UNDEF is
  specified."

https://lore.kernel.org/audit/5838489ecd5186900315f8f6c6e02f22@paul-moore.com/

> +		return scall->hl->hook.lsmprop_to_secctx(prop, cp);
> +	}
> +	return LSM_RET_DEFAULT(lsmprop_to_secctx);
>  }
>  EXPORT_SYMBOL(security_lsmprop_to_secctx);
>  
> -- 
> 2.47.0

--
paul-moore.com

[PATCH v3 3/5] Audit: Add record for multiple task security contexts

[Casey Schaufler]

Replace the single skb pointer in an audit_buffer with a list of
skb pointers. Add the audit_stamp information to the audit_buffer as
there's no guarantee that there will be an audit_context containing
the stamp associated with the event. At audit_log_end() time create
auxiliary records (none are currently defined) as have been added to the
list. Functions are created to manage the skb list in the audit_buffer.

Create a new audit record AUDIT_MAC_TASK_CONTEXTS.
An example of the MAC_TASK_CONTEXTS (1423) record is:

    type=MAC_TASK_CONTEXTS[1423]
    msg=audit(1600880931.832:113)
    subj_apparmor=unconfined
    subj_smack=_

When an audit event includes a AUDIT_MAC_TASK_CONTEXTS record the
"subj=" field in other records in the event will be "subj=?".
An AUDIT_MAC_TASK_CONTEXTS record is supplied when the system has
multiple security modules that may make access decisions based on a
subject security context.

Suggested-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 include/linux/audit.h      |   6 ++
 include/uapi/linux/audit.h |   1 +
 kernel/audit.c             | 171 ++++++++++++++++++++++++++++++-------
 security/apparmor/lsm.c    |   3 +
 security/selinux/hooks.c   |   3 +
 security/smack/smack_lsm.c |   3 +
 6 files changed, 158 insertions(+), 29 deletions(-)

diff --git a/include/linux/audit.h b/include/linux/audit.h
index 0050ef288ab3..b493ca5976cf 100644
--- a/include/linux/audit.h
+++ b/include/linux/audit.h
@@ -37,6 +37,7 @@ struct audit_watch;
 struct audit_tree;
 struct sk_buff;
 struct kern_ipc_perm;
+struct lsm_id;
 
 struct audit_krule {
 	u32			pflags;
@@ -210,6 +211,8 @@ extern u32 audit_enabled;
 
 extern int audit_signal_info(int sig, struct task_struct *t);
 
+extern void audit_lsm_secctx(const struct lsm_id *lsmid);
+
 #else /* CONFIG_AUDIT */
 static inline __printf(4, 5)
 void audit_log(struct audit_context *ctx, gfp_t gfp_mask, int type,
@@ -269,6 +272,9 @@ static inline int audit_signal_info(int sig, struct task_struct *t)
 	return 0;
 }
 
+static inline void audit_lsm_secctx(const struct lsm_id *lsmid)
+{ }
+
 #endif /* CONFIG_AUDIT */
 
 #ifdef CONFIG_AUDIT_COMPAT_GENERIC
diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h
index d9a069b4a775..5ebb5d80363d 100644
--- a/include/uapi/linux/audit.h
+++ b/include/uapi/linux/audit.h
@@ -146,6 +146,7 @@
 #define AUDIT_IPE_ACCESS	1420	/* IPE denial or grant */
 #define AUDIT_IPE_CONFIG_CHANGE	1421	/* IPE config change */
 #define AUDIT_IPE_POLICY_LOAD	1422	/* IPE policy load */
+#define AUDIT_MAC_TASK_CONTEXTS	1423	/* Multiple LSM task contexts */
 
 #define AUDIT_FIRST_KERN_ANOM_MSG   1700
 #define AUDIT_LAST_KERN_ANOM_MSG    1799
diff --git a/kernel/audit.c b/kernel/audit.c
index 6bbadb605ca3..7ec3919ae925 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -54,6 +54,7 @@
 #include <net/netlink.h>
 #include <linux/skbuff.h>
 #include <linux/security.h>
+#include <linux/lsm_hooks.h>
 #include <linux/freezer.h>
 #include <linux/pid_namespace.h>
 #include <net/netns/generic.h>
@@ -81,6 +82,11 @@ static u32	audit_failure = AUDIT_FAIL_PRINTK;
 /* private audit network namespace index */
 static unsigned int audit_net_id;
 
+/* Number of modules that provide a security context.
+   List of lsms that provide a security context */
+static u32 audit_secctx_cnt = 0;
+static const struct lsm_id *audit_lsms[MAX_LSM_COUNT];
+
 /**
  * struct audit_net - audit private network namespace data
  * @sk: communication socket
@@ -195,8 +201,10 @@ static struct audit_ctl_mutex {
  * to place it on a transmit queue.  Multiple audit_buffers can be in
  * use simultaneously. */
 struct audit_buffer {
-	struct sk_buff       *skb;	/* formatted skb ready to send */
+	struct sk_buff       *skb;	/* the skb for audit_log functions */
+	struct sk_buff_head  skb_list;	/* formatted skbs, ready to send */
 	struct audit_context *ctx;	/* NULL or associated context */
+	struct audit_stamp   stamp;	/* audit stamp for these records */
 	gfp_t		     gfp_mask;
 };
 
@@ -278,6 +286,18 @@ static pid_t auditd_pid_vnr(void)
 	return pid;
 }
 
+/**
+ * audit_lsm_secctx - Identify a security module as providing a secctx.
+ * @lsmid - LSM identity
+ *
+ * Description:
+ * Increments the count of the security modules providing a secctx.
+ */
+void audit_lsm_secctx(const struct lsm_id *lsmid)
+{
+	audit_lsms[audit_secctx_cnt++] = lsmid;
+}
+
 /**
  * audit_get_sk - Return the audit socket for the given network namespace
  * @net: the destination network namespace
@@ -1776,10 +1796,13 @@ __setup("audit_backlog_limit=", audit_backlog_limit_set);
 
 static void audit_buffer_free(struct audit_buffer *ab)
 {
+	struct sk_buff *skb;
+
 	if (!ab)
 		return;
 
-	kfree_skb(ab->skb);
+	while ((skb = skb_dequeue(&ab->skb_list)))
+		kfree_skb(skb);
 	kmem_cache_free(audit_buffer_cache, ab);
 }
 
@@ -1795,6 +1818,10 @@ static struct audit_buffer *audit_buffer_alloc(struct audit_context *ctx,
 	ab->skb = nlmsg_new(AUDIT_BUFSIZ, gfp_mask);
 	if (!ab->skb)
 		goto err;
+
+	skb_queue_head_init(&ab->skb_list);
+	skb_queue_tail(&ab->skb_list, ab->skb);
+
 	if (!nlmsg_put(ab->skb, 0, 0, type, 0, 0))
 		goto err;
 
@@ -1860,7 +1887,6 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 				     int type)
 {
 	struct audit_buffer *ab;
-	struct audit_stamp stamp;
 
 	if (audit_initialized != AUDIT_INITIALIZED)
 		return NULL;
@@ -1915,14 +1941,14 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 		return NULL;
 	}
 
-	audit_get_stamp(ab->ctx, &stamp);
+	audit_get_stamp(ab->ctx, &ab->stamp);
 	/* cancel dummy context to enable supporting records */
 	if (ctx)
 		ctx->dummy = 0;
 	audit_log_format(ab, "audit(%llu.%03lu:%u): ",
-			 (unsigned long long)stamp.ctime.tv_sec,
-			 stamp.ctime.tv_nsec/1000000,
-			 stamp.serial);
+			 (unsigned long long)ab->stamp.ctime.tv_sec,
+			 ab->stamp.ctime.tv_nsec/1000000,
+			 ab->stamp.serial);
 
 	return ab;
 }
@@ -2178,25 +2204,104 @@ void audit_log_key(struct audit_buffer *ab, char *key)
 		audit_log_format(ab, "(null)");
 }
 
+/**
+ * audit_buffer_aux_new - Add an aux record buffer to the skb list
+ * @ab: audit_buffer
+ * @type: message type
+ *
+ * Aux records are allocated and added to the skb list of
+ * the "main" record. The ab->skb is reset to point to the
+ * aux record on its creation. When the aux record in complete
+ * ab->skb has to be reset to point to the "main" record.
+ * This allows the audit_log_ functions to be ignorant of
+ * which kind of record it is logging to. It also avoids adding
+ * special data for aux records.
+ *
+ * On success ab->skb will point to the new aux record.
+ * Returns 0 on success, -ENOMEM should allocation fail.
+ */
+static int audit_buffer_aux_new(struct audit_buffer *ab, int type)
+{
+	WARN_ON(ab->skb != skb_peek(&ab->skb_list));
+
+	ab->skb = nlmsg_new(AUDIT_BUFSIZ, ab->gfp_mask);
+	if (!ab->skb)
+		goto err;
+	if (!nlmsg_put(ab->skb, 0, 0, type, 0, 0))
+		goto err;
+	skb_queue_tail(&ab->skb_list, ab->skb);
+
+	audit_log_format(ab, "audit(%llu.%03lu:%u): ",
+			 (unsigned long long)ab->stamp.ctime.tv_sec,
+			 ab->stamp.ctime.tv_nsec/1000000,
+			 ab->stamp.serial);
+
+	return 0;
+
+err:
+	kfree_skb(ab->skb);
+	ab->skb = skb_peek(&ab->skb_list);
+	return -ENOMEM;
+}
+
+/**
+ * audit_buffer_aux_end - Switch back to the "main" record from an aux record
+ * @ab: audit_buffer
+ *
+ * Restores the "main" audit record to ab->skb.
+ */
+static void audit_buffer_aux_end(struct audit_buffer *ab)
+{
+	ab->skb = skb_peek(&ab->skb_list);
+}
+
 int audit_log_task_context(struct audit_buffer *ab)
 {
 	struct lsm_prop prop;
 	struct lsm_context ctx;
+	bool space = false;
 	int error;
+	int i;
 
 	security_current_getlsmprop_subj(&prop);
 	if (!lsmprop_is_set(&prop))
 		return 0;
 
-	error = security_lsmprop_to_secctx(&prop, &ctx, LSM_ID_UNDEF);
-	if (error < 0) {
-		if (error != -EINVAL)
-			goto error_path;
+	if (audit_secctx_cnt < 2) {
+		error = security_lsmprop_to_secctx(&prop, &ctx, LSM_ID_UNDEF);
+		if (error < 0) {
+			if (error != -EINVAL)
+				goto error_path;
+			return 0;
+		}
+		audit_log_format(ab, " subj=%s", ctx.context);
+		security_release_secctx(&ctx);
 		return 0;
 	}
-
-	audit_log_format(ab, " subj=%s", ctx.context);
-	security_release_secctx(&ctx);
+	/* Multiple LSMs provide contexts. Include an aux record. */
+	audit_log_format(ab, " subj=?");
+	error = audit_buffer_aux_new(ab, AUDIT_MAC_TASK_CONTEXTS);
+	if (error)
+		goto error_path;
+
+	for (i = 0; i < audit_secctx_cnt; i++) {
+		error = security_lsmprop_to_secctx(&prop, &ctx,
+						   audit_lsms[i]->id);
+		if (error < 0) {
+			if (error == -EOPNOTSUPP)
+				continue;
+			audit_log_format(ab, "%ssubj_%s=?", space ? " " : "",
+					 audit_lsms[i]->name);
+			if (error != -EINVAL)
+				audit_panic("error in audit_log_task_context");
+		} else {
+			audit_log_format(ab, "%ssubj_%s=%s", space ? " " : "",
+					 audit_lsms[i]->name, ctx.context);
+			security_release_secctx(&ctx);
+		}
+		space = true;
+	}
+	audit_buffer_aux_end(ab);
 	return 0;
 
 error_path:
@@ -2412,26 +2517,14 @@ int audit_signal_info(int sig, struct task_struct *t)
 }
 
 /**
- * audit_log_end - end one audit record
- * @ab: the audit_buffer
- *
- * We can not do a netlink send inside an irq context because it blocks (last
- * arg, flags, is not set to MSG_DONTWAIT), so the audit buffer is placed on a
- * queue and a kthread is scheduled to remove them from the queue outside the
- * irq context.  May be called in any context.
+ * __audit_log_end - enqueue one audit record
+ * @skb: the buffer to send
  */
-void audit_log_end(struct audit_buffer *ab)
+static void __audit_log_end(struct sk_buff *skb)
 {
-	struct sk_buff *skb;
 	struct nlmsghdr *nlh;
 
-	if (!ab)
-		return;
-
 	if (audit_rate_check()) {
-		skb = ab->skb;
-		ab->skb = NULL;
-
 		/* setup the netlink header, see the comments in
 		 * kauditd_send_multicast_skb() for length quirks */
 		nlh = nlmsg_hdr(skb);
@@ -2442,6 +2535,26 @@ void audit_log_end(struct audit_buffer *ab)
 		wake_up_interruptible(&kauditd_wait);
 	} else
 		audit_log_lost("rate limit exceeded");
+}
+
+/**
+ * audit_log_end - end one audit record
+ * @ab: the audit_buffer
+ *
+ * We can not do a netlink send inside an irq context because it blocks (last
+ * arg, flags, is not set to MSG_DONTWAIT), so the audit buffer is placed on a
+ * queue and a kthread is scheduled to remove them from the queue outside the
+ * irq context.  May be called in any context.
+ */
+void audit_log_end(struct audit_buffer *ab)
+{
+	struct sk_buff *skb;
+
+	if (!ab)
+		return;
+
+	while ((skb = skb_dequeue(&ab->skb_list)))
+		__audit_log_end(skb);
 
 	audit_buffer_free(ab);
 }
diff --git a/security/apparmor/lsm.c b/security/apparmor/lsm.c
index 9b6c2f157f83..50242210670a 100644
--- a/security/apparmor/lsm.c
+++ b/security/apparmor/lsm.c
@@ -2250,6 +2250,9 @@ static int __init apparmor_init(void)
 	security_add_hooks(apparmor_hooks, ARRAY_SIZE(apparmor_hooks),
 				&apparmor_lsmid);
 
+	/* Inform the audit system that secctx is used */
+	audit_lsm_secctx(&apparmor_lsmid);
+
 	/* Report that AppArmor successfully initialized */
 	apparmor_initialized = 1;
 	if (aa_g_profile_mode == APPARMOR_COMPLAIN)
diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c
index 7b867dfec88b..0772e9dc0414 100644
--- a/security/selinux/hooks.c
+++ b/security/selinux/hooks.c
@@ -7456,6 +7456,9 @@ static __init int selinux_init(void)
 	/* Set the security state for the initial task. */
 	cred_init_security();
 
+	/* Inform the audit system that secctx is used */
+	audit_lsm_secctx(&selinux_lsmid);
+
 	default_noexec = !(VM_DATA_DEFAULT_FLAGS & VM_EXEC);
 	if (!default_noexec)
 		pr_notice("SELinux:  virtual memory is executable by default\n");
diff --git a/security/smack/smack_lsm.c b/security/smack/smack_lsm.c
index 239773cdcdcf..214989d2146b 100644
--- a/security/smack/smack_lsm.c
+++ b/security/smack/smack_lsm.c
@@ -5290,6 +5290,9 @@ static __init int smack_init(void)
 	/* initialize the smack_known_list */
 	init_smack_known_list();
 
+	/* Inform the audit system that secctx is used */
+	audit_lsm_secctx(&smack_lsmid);
+
 	return 0;
 }
 
-- 
2.47.0

Re: [PATCH v3 3/5] Audit: Add record for multiple task security contexts

[Paul Moore]

On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> 
> Replace the single skb pointer in an audit_buffer with a list of
> skb pointers. Add the audit_stamp information to the audit_buffer as
> there's no guarantee that there will be an audit_context containing
> the stamp associated with the event. At audit_log_end() time create
> auxiliary records (none are currently defined) as have been added to the
> list. Functions are created to manage the skb list in the audit_buffer.
> 
> Create a new audit record AUDIT_MAC_TASK_CONTEXTS.
> An example of the MAC_TASK_CONTEXTS (1423) record is:
> 
>     type=MAC_TASK_CONTEXTS[1423]
>     msg=audit(1600880931.832:113)
>     subj_apparmor=unconfined
>     subj_smack=_
> 
> When an audit event includes a AUDIT_MAC_TASK_CONTEXTS record the
> "subj=" field in other records in the event will be "subj=?".
> An AUDIT_MAC_TASK_CONTEXTS record is supplied when the system has
> multiple security modules that may make access decisions based on a
> subject security context.
> 
> Suggested-by: Paul Moore <paul@paul-moore.com>
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> ---
>  include/linux/audit.h      |   6 ++
>  include/uapi/linux/audit.h |   1 +
>  kernel/audit.c             | 171 ++++++++++++++++++++++++++++++-------
>  security/apparmor/lsm.c    |   3 +
>  security/selinux/hooks.c   |   3 +
>  security/smack/smack_lsm.c |   3 +
>  6 files changed, 158 insertions(+), 29 deletions(-)
> 
> diff --git a/include/linux/audit.h b/include/linux/audit.h
> index 0050ef288ab3..b493ca5976cf 100644
> --- a/include/linux/audit.h
> +++ b/include/linux/audit.h
> @@ -37,6 +37,7 @@ struct audit_watch;
>  struct audit_tree;
>  struct sk_buff;
>  struct kern_ipc_perm;
> +struct lsm_id;
>  
>  struct audit_krule {
>  	u32			pflags;
> @@ -210,6 +211,8 @@ extern u32 audit_enabled;
>  
>  extern int audit_signal_info(int sig, struct task_struct *t);
>  
> +extern void audit_lsm_secctx(const struct lsm_id *lsmid);
> +
>  #else /* CONFIG_AUDIT */
>  static inline __printf(4, 5)
>  void audit_log(struct audit_context *ctx, gfp_t gfp_mask, int type,
> @@ -269,6 +272,9 @@ static inline int audit_signal_info(int sig, struct task_struct *t)
>  	return 0;
>  }
>  
> +static inline void audit_lsm_secctx(const struct lsm_id *lsmid)
> +{ }
> +
>  #endif /* CONFIG_AUDIT */
>  
>  #ifdef CONFIG_AUDIT_COMPAT_GENERIC
> diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h
> index d9a069b4a775..5ebb5d80363d 100644
> --- a/include/uapi/linux/audit.h
> +++ b/include/uapi/linux/audit.h
> @@ -146,6 +146,7 @@
>  #define AUDIT_IPE_ACCESS	1420	/* IPE denial or grant */
>  #define AUDIT_IPE_CONFIG_CHANGE	1421	/* IPE config change */
>  #define AUDIT_IPE_POLICY_LOAD	1422	/* IPE policy load */
> +#define AUDIT_MAC_TASK_CONTEXTS	1423	/* Multiple LSM task contexts */
>  
>  #define AUDIT_FIRST_KERN_ANOM_MSG   1700
>  #define AUDIT_LAST_KERN_ANOM_MSG    1799
> diff --git a/kernel/audit.c b/kernel/audit.c
> index 6bbadb605ca3..7ec3919ae925 100644
> --- a/kernel/audit.c
> +++ b/kernel/audit.c
> @@ -54,6 +54,7 @@
>  #include <net/netlink.h>
>  #include <linux/skbuff.h>
>  #include <linux/security.h>
> +#include <linux/lsm_hooks.h>
>  #include <linux/freezer.h>
>  #include <linux/pid_namespace.h>
>  #include <net/netns/generic.h>
> @@ -81,6 +82,11 @@ static u32	audit_failure = AUDIT_FAIL_PRINTK;
>  /* private audit network namespace index */
>  static unsigned int audit_net_id;
>  
> +/* Number of modules that provide a security context.
> +   List of lsms that provide a security context */
> +static u32 audit_secctx_cnt = 0;
> +static const struct lsm_id *audit_lsms[MAX_LSM_COUNT];

We've already talked about this in other threads, offline, etc., but
for the sake of others, this should be adjusted to use the counts
provided in the LSM initialization code rework.

https://lore.kernel.org/linux-security-module/20250409185019.238841-31-paul@paul-moore.com/

> @@ -2412,26 +2517,14 @@ int audit_signal_info(int sig, struct task_struct *t)
>  }
>  
>  /**
> - * audit_log_end - end one audit record
> - * @ab: the audit_buffer
> - *
> - * We can not do a netlink send inside an irq context because it blocks (last
> - * arg, flags, is not set to MSG_DONTWAIT), so the audit buffer is placed on a
> - * queue and a kthread is scheduled to remove them from the queue outside the
> - * irq context.  May be called in any context.
> + * __audit_log_end - enqueue one audit record
> + * @skb: the buffer to send
>   */
> -void audit_log_end(struct audit_buffer *ab)
> +static void __audit_log_end(struct sk_buff *skb)
>  {
> -	struct sk_buff *skb;
>  	struct nlmsghdr *nlh;
>  
> -	if (!ab)
> -		return;
> -
>  	if (audit_rate_check()) {
> -		skb = ab->skb;
> -		ab->skb = NULL;
> -
>  		/* setup the netlink header, see the comments in
>  		 * kauditd_send_multicast_skb() for length quirks */
>  		nlh = nlmsg_hdr(skb);
> @@ -2442,6 +2535,26 @@ void audit_log_end(struct audit_buffer *ab)
>  		wake_up_interruptible(&kauditd_wait);
>  	} else
>  		audit_log_lost("rate limit exceeded");
> +}

Okay, this is twice now in one patchset ... as I mentioned in the v2
review:

 "We should probably move the kauditd thread wake into
  audit_log_end() so we don't end up poking the scheduler
  multiple times."

https://lore.kernel.org/audit/69ee16ce82a564e09b2060d46fa2be0d@paul-moore.com/

> +/**
> + * audit_log_end - end one audit record
> + * @ab: the audit_buffer
> + *
> + * We can not do a netlink send inside an irq context because it blocks (last
> + * arg, flags, is not set to MSG_DONTWAIT), so the audit buffer is placed on a
> + * queue and a kthread is scheduled to remove them from the queue outside the
> + * irq context.  May be called in any context.
> + */
> +void audit_log_end(struct audit_buffer *ab)
> +{
> +	struct sk_buff *skb;
> +
> +	if (!ab)
> +		return;
> +
> +	while ((skb = skb_dequeue(&ab->skb_list)))
> +		__audit_log_end(skb);

The wakeup should go here.

>  	audit_buffer_free(ab);
>  }

--
paul-moore.com

[PATCH v3 4/5] Audit: multiple subject lsm values for netlabel

[Casey Schaufler]

Refactor audit_log_task_context(), creating a new audit_log_subj_ctx().
This is used in netlabel auditing to provide multiple subject security
contexts as necessary.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 include/linux/audit.h        |  7 +++++++
 kernel/audit.c               | 28 +++++++++++++++++++++-------
 net/netlabel/netlabel_user.c |  9 +--------
 3 files changed, 29 insertions(+), 15 deletions(-)

diff --git a/include/linux/audit.h b/include/linux/audit.h
index b493ca5976cf..3402e3ca43c6 100644
--- a/include/linux/audit.h
+++ b/include/linux/audit.h
@@ -38,6 +38,7 @@ struct audit_tree;
 struct sk_buff;
 struct kern_ipc_perm;
 struct lsm_id;
+struct lsm_prop;
 
 struct audit_krule {
 	u32			pflags;
@@ -186,6 +187,7 @@ extern void		    audit_log_path_denied(int type,
 						  const char *operation);
 extern void		    audit_log_lost(const char *message);
 
+extern int audit_log_subj_ctx(struct audit_buffer *ab, struct lsm_prop *prop);
 extern int audit_log_task_context(struct audit_buffer *ab);
 extern void audit_log_task_info(struct audit_buffer *ab);
 
@@ -248,6 +250,11 @@ static inline void audit_log_key(struct audit_buffer *ab, char *key)
 { }
 static inline void audit_log_path_denied(int type, const char *operation)
 { }
+static inline int audit_log_subj_ctx(struct audit_buffer *ab,
+				     struct lsm_prop *prop)
+{
+	return 0;
+}
 static inline int audit_log_task_context(struct audit_buffer *ab)
 {
 	return 0;
diff --git a/kernel/audit.c b/kernel/audit.c
index 7ec3919ae925..8ce453f6dc7d 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -2255,20 +2255,25 @@ static void audit_buffer_aux_end(struct audit_buffer *ab)
 	ab->skb = skb_peek(&ab->skb_list);
 }
 
-int audit_log_task_context(struct audit_buffer *ab)
+/**
+ * audit_log_subj_ctx - Add LSM subject information
+ * @ab: audit_buffer
+ * @prop: LSM subject properties.
+ *
+ * Add a subj= field and, if necessary, a AUDIT_MAC_TASK_CONTEXTS record.
+ */
+int audit_log_subj_ctx(struct audit_buffer *ab, struct lsm_prop *prop)
 {
-	struct lsm_prop prop;
 	struct lsm_context ctx;
 	bool space = false;
 	int error;
 	int i;
 
-	security_current_getlsmprop_subj(&prop);
-	if (!lsmprop_is_set(&prop))
+	if (!lsmprop_is_set(prop))
 		return 0;
 
 	if (audit_secctx_cnt < 2) {
-		error = security_lsmprop_to_secctx(&prop, &ctx, LSM_ID_UNDEF);
+		error = security_lsmprop_to_secctx(prop, &ctx, LSM_ID_UNDEF);
 		if (error < 0) {
 			if (error != -EINVAL)
 				goto error_path;
@@ -2285,7 +2290,7 @@ int audit_log_task_context(struct audit_buffer *ab)
 		goto error_path;
 
 	for (i = 0; i < audit_secctx_cnt; i++) {
-		error = security_lsmprop_to_secctx(&prop, &ctx,
+		error = security_lsmprop_to_secctx(prop, &ctx,
 						   audit_lsms[i]->id);
 		if (error < 0) {
 			if (error == -EOPNOTSUPP)
@@ -2305,9 +2310,18 @@ int audit_log_task_context(struct audit_buffer *ab)
 	return 0;
 
 error_path:
-	audit_panic("error in audit_log_task_context");
+	audit_panic("error in audit_log_subj_ctx");
 	return error;
 }
+EXPORT_SYMBOL(audit_log_subj_ctx);
+
+int audit_log_task_context(struct audit_buffer *ab)
+{
+	struct lsm_prop prop;
+
+	security_current_getlsmprop_subj(&prop);
+	return audit_log_subj_ctx(ab, &prop);
+}
 EXPORT_SYMBOL(audit_log_task_context);
 
 void audit_log_d_path_exe(struct audit_buffer *ab,
diff --git a/net/netlabel/netlabel_user.c b/net/netlabel/netlabel_user.c
index 6d6545297ee3..0da652844dd6 100644
--- a/net/netlabel/netlabel_user.c
+++ b/net/netlabel/netlabel_user.c
@@ -84,7 +84,6 @@ struct audit_buffer *netlbl_audit_start_common(int type,
 					       struct netlbl_audit *audit_info)
 {
 	struct audit_buffer *audit_buf;
-	struct lsm_context ctx;
 
 	if (audit_enabled == AUDIT_OFF)
 		return NULL;
@@ -96,13 +95,7 @@ struct audit_buffer *netlbl_audit_start_common(int type,
 	audit_log_format(audit_buf, "netlabel: auid=%u ses=%u",
 			 from_kuid(&init_user_ns, audit_info->loginuid),
 			 audit_info->sessionid);
-
-	if (lsmprop_is_set(&audit_info->prop) &&
-	    security_lsmprop_to_secctx(&audit_info->prop, &ctx,
-				       LSM_ID_UNDEF) > 0) {
-		audit_log_format(audit_buf, " subj=%s", ctx.context);
-		security_release_secctx(&ctx);
-	}
+	audit_log_subj_ctx(audit_buf, &audit_info->prop);
 
 	return audit_buf;
 }
-- 
2.47.0

Re: [PATCH v3 4/5] Audit: multiple subject lsm values for netlabel

[Paul Moore]

On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> 
> Refactor audit_log_task_context(), creating a new audit_log_subj_ctx().
> This is used in netlabel auditing to provide multiple subject security
> contexts as necessary.
> 
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> ---
>  include/linux/audit.h        |  7 +++++++
>  kernel/audit.c               | 28 +++++++++++++++++++++-------
>  net/netlabel/netlabel_user.c |  9 +--------
>  3 files changed, 29 insertions(+), 15 deletions(-)

Other than moving to the subject count supplied by the LSM
initialization patchset previously mentioned, this looks fine to me.

--
paul-moore.com

Re: [PATCH v3 4/5] Audit: multiple subject lsm values for netlabel

[Casey Schaufler]

On 4/24/2025 3:18 PM, Paul Moore wrote:
> On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
>> Refactor audit_log_task_context(), creating a new audit_log_subj_ctx().
>> This is used in netlabel auditing to provide multiple subject security
>> contexts as necessary.
>>
>> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
>> ---
>>  include/linux/audit.h        |  7 +++++++
>>  kernel/audit.c               | 28 +++++++++++++++++++++-------
>>  net/netlabel/netlabel_user.c |  9 +--------
>>  3 files changed, 29 insertions(+), 15 deletions(-)
> Other than moving to the subject count supplied by the LSM
> initialization patchset previously mentioned, this looks fine to me.

I'm perfectly willing to switch once the LSM initialization patch set
moves past RFC.

>
> --
> paul-moore.com
>

Re: [PATCH v3 4/5] Audit: multiple subject lsm values for netlabel

[Paul Moore]

On Wed, Apr 30, 2025 at 12:25 PM Casey Schaufler <casey@schaufler-ca.com> wrote:
> On 4/24/2025 3:18 PM, Paul Moore wrote:
> > On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> >> Refactor audit_log_task_context(), creating a new audit_log_subj_ctx().
> >> This is used in netlabel auditing to provide multiple subject security
> >> contexts as necessary.
> >>
> >> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> >> ---
> >>  include/linux/audit.h        |  7 +++++++
> >>  kernel/audit.c               | 28 +++++++++++++++++++++-------
> >>  net/netlabel/netlabel_user.c |  9 +--------
> >>  3 files changed, 29 insertions(+), 15 deletions(-)
> > Other than moving to the subject count supplied by the LSM
> > initialization patchset previously mentioned, this looks fine to me.
>
> I'm perfectly willing to switch once the LSM initialization patch set
> moves past RFC.

It's obviously your choice as to if/when you switch, but I'm trying to
let you know that acceptance into the LSM tree is going to be
dependent on that switch happening.

The initialization patchset is still very much alive, and the next
revision will not be an RFC.  I'm simply waiting on some additional
LSM specific reviews before posting the next revision so as to not
burn out people from looking at multiple iterations.  I've been told
privately by at least one LSM maintainer that reviewing the changes in
their code is on their todo list, but they have been slammed with
other work at their job and haven't had the time to look at that
patchset yet.  I realize you don't have those issues anymore, but I
suspect you are still sympathetic to those problems.

If you're really anxious to continue work on this RIGHT NOW, you can
simply base your patchset on top of the initialization patchset.  Just
make sure you mention in the cover letter what you are using as a base
for the patchset.

If that still doesn't offer any satisfaction, you can always
incorporate the feedback that I made in v2 that was ignored in your v3
posting :-P

-- 
paul-moore.com

Re: [PATCH v3 4/5] Audit: multiple subject lsm values for netlabel

[Casey Schaufler]

On 4/30/2025 11:51 AM, Paul Moore wrote:
> On Wed, Apr 30, 2025 at 12:25 PM Casey Schaufler <casey@schaufler-ca.com> wrote:
>> On 4/24/2025 3:18 PM, Paul Moore wrote:
>>> On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
>>>> Refactor audit_log_task_context(), creating a new audit_log_subj_ctx().
>>>> This is used in netlabel auditing to provide multiple subject security
>>>> contexts as necessary.
>>>>
>>>> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
>>>> ---
>>>>  include/linux/audit.h        |  7 +++++++
>>>>  kernel/audit.c               | 28 +++++++++++++++++++++-------
>>>>  net/netlabel/netlabel_user.c |  9 +--------
>>>>  3 files changed, 29 insertions(+), 15 deletions(-)
>>> Other than moving to the subject count supplied by the LSM
>>> initialization patchset previously mentioned, this looks fine to me.
>> I'm perfectly willing to switch once the LSM initialization patch set
>> moves past RFC.
> It's obviously your choice as to if/when you switch, but I'm trying to
> let you know that acceptance into the LSM tree is going to be
> dependent on that switch happening.

Not a problem. Obviously, I'd prefer this patch going in before the
LSM initialization work, but it is your call.

> The initialization patchset is still very much alive, and the next
> revision will not be an RFC.  I'm simply waiting on some additional
> LSM specific reviews before posting the next revision so as to not
> burn out people from looking at multiple iterations.  I've been told
> privately by at least one LSM maintainer that reviewing the changes in
> their code is on their todo list, but they have been slammed with
> other work at their job and haven't had the time to look at that
> patchset yet.  I realize you don't have those issues anymore, but I
> suspect you are still sympathetic to those problems.

Of course. Waiting on reviews can be frustrating.

> If you're really anxious to continue work on this RIGHT NOW, you can
> simply base your patchset on top of the initialization patchset.  Just
> make sure you mention in the cover letter what you are using as a base
> for the patchset.

As I don't anticipate serious changes to your patch set this makes sense.

> If that still doesn't offer any satisfaction, you can always
> incorporate the feedback that I made in v2 that was ignored in your v3
> posting :-P

Yeah, oops on that.

[PATCH v3 5/5] Audit: Add record for multiple object contexts

[Casey Schaufler]

Create a new audit record AUDIT_MAC_OBJ_CONTEXTS.
An example of the MAC_OBJ_CONTEXTS (1424) record is:

    type=MAC_OBJ_CONTEXTS[1424]
    msg=audit(1601152467.009:1050):
    obj_selinux=unconfined_u:object_r:user_home_t:s0

When an audit event includes a AUDIT_MAC_OBJ_CONTEXTS record
the "obj=" field in other records in the event will be "obj=?".
An AUDIT_MAC_OBJ_CONTEXTS record is supplied when the system has
multiple security modules that may make access decisions based
on an object security context.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 include/linux/audit.h      |  6 +++++
 include/uapi/linux/audit.h |  1 +
 kernel/audit.c             | 51 +++++++++++++++++++++++++++++++++++++-
 kernel/auditsc.c           | 45 ++++++++-------------------------
 4 files changed, 68 insertions(+), 35 deletions(-)

diff --git a/include/linux/audit.h b/include/linux/audit.h
index 3402e3ca43c6..8fdfa2721273 100644
--- a/include/linux/audit.h
+++ b/include/linux/audit.h
@@ -188,6 +188,7 @@ extern void		    audit_log_path_denied(int type,
 extern void		    audit_log_lost(const char *message);
 
 extern int audit_log_subj_ctx(struct audit_buffer *ab, struct lsm_prop *prop);
+extern int audit_log_obj_ctx(struct audit_buffer *ab, struct lsm_prop *prop);
 extern int audit_log_task_context(struct audit_buffer *ab);
 extern void audit_log_task_info(struct audit_buffer *ab);
 
@@ -255,6 +256,11 @@ static inline int audit_log_subj_ctx(struct audit_buffer *ab,
 {
 	return 0;
 }
+static inline int audit_log_obj_ctx(struct audit_buffer *ab,
+				    struct lsm_prop *prop)
+{
+	return 0;
+}
 static inline int audit_log_task_context(struct audit_buffer *ab)
 {
 	return 0;
diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h
index 5ebb5d80363d..8ca58144bcc6 100644
--- a/include/uapi/linux/audit.h
+++ b/include/uapi/linux/audit.h
@@ -147,6 +147,7 @@
 #define AUDIT_IPE_CONFIG_CHANGE	1421	/* IPE config change */
 #define AUDIT_IPE_POLICY_LOAD	1422	/* IPE policy load */
 #define AUDIT_MAC_TASK_CONTEXTS	1423	/* Multiple LSM task contexts */
+#define AUDIT_MAC_OBJ_CONTEXTS	1424	/* Multiple LSM objext contexts */
 
 #define AUDIT_FIRST_KERN_ANOM_MSG   1700
 #define AUDIT_LAST_KERN_ANOM_MSG    1799
diff --git a/kernel/audit.c b/kernel/audit.c
index 8ce453f6dc7d..69db0ee09a3f 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -1133,7 +1133,6 @@ static int is_audit_feature_set(int i)
 	return af.features & AUDIT_FEATURE_TO_MASK(i);
 }
 
-
 static int audit_get_feature(struct sk_buff *skb)
 {
 	u32 seq;
@@ -2324,6 +2323,56 @@ int audit_log_task_context(struct audit_buffer *ab)
 }
 EXPORT_SYMBOL(audit_log_task_context);
 
+int audit_log_obj_ctx(struct audit_buffer *ab, struct lsm_prop *prop)
+{
+	int i;
+	int rc;
+	int error = 0;
+	char *space = "";
+	struct lsm_context context;
+
+	if (audit_secctx_cnt < 2) {
+		error = security_lsmprop_to_secctx(prop, &context,
+						   LSM_ID_UNDEF);
+		if (error < 0) {
+			if (error != -EINVAL)
+				goto error_path;
+			return error;
+		}
+		audit_log_format(ab, " obj=%s", context.context);
+		security_release_secctx(&context);
+		return 0;
+	}
+	audit_log_format(ab, " obj=?");
+	error = audit_buffer_aux_new(ab, AUDIT_MAC_OBJ_CONTEXTS);
+	if (error)
+		goto error_path;
+
+	for (i = 0; i < audit_secctx_cnt; i++) {
+		rc = security_lsmprop_to_secctx(prop, &context,
+						audit_lsms[i]->id);
+		if (rc < 0) {
+			audit_log_format(ab, "%sobj_%s=?", space,
+					 audit_lsms[i]->name);
+			if (rc != -EINVAL)
+				audit_panic("error in audit_log_obj_ctx");
+			error = rc;
+		} else {
+			audit_log_format(ab, "%sobj_%s=%s", space,
+					 audit_lsms[i]->name, context.context);
+			security_release_secctx(&context);
+		}
+		space = " ";
+	}
+
+	audit_buffer_aux_end(ab);
+	return error;
+
+error_path:
+	audit_panic("error in audit_log_obj_ctx");
+	return error;
+}
+
 void audit_log_d_path_exe(struct audit_buffer *ab,
 			  struct mm_struct *mm)
 {
diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index dc3f7e9666f2..e39d7be20c29 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -1098,7 +1098,6 @@ static int audit_log_pid_context(struct audit_context *context, pid_t pid,
 				 char *comm)
 {
 	struct audit_buffer *ab;
-	struct lsm_context ctx;
 	int rc = 0;
 
 	ab = audit_log_start(context, GFP_KERNEL, AUDIT_OBJ_PID);
@@ -1108,15 +1107,9 @@ static int audit_log_pid_context(struct audit_context *context, pid_t pid,
 	audit_log_format(ab, "opid=%d oauid=%d ouid=%d oses=%d", pid,
 			 from_kuid(&init_user_ns, auid),
 			 from_kuid(&init_user_ns, uid), sessionid);
-	if (lsmprop_is_set(prop)) {
-		if (security_lsmprop_to_secctx(prop, &ctx, LSM_ID_UNDEF) < 0) {
-			audit_log_format(ab, " obj=(none)");
-			rc = 1;
-		} else {
-			audit_log_format(ab, " obj=%s", ctx.context);
-			security_release_secctx(&ctx);
-		}
-	}
+	if (lsmprop_is_set(prop) && audit_log_obj_ctx(ab, prop))
+		rc = 1;
+
 	audit_log_format(ab, " ocomm=");
 	audit_log_untrustedstring(ab, comm);
 	audit_log_end(ab);
@@ -1392,16 +1385,8 @@ static void show_special(struct audit_context *context, int *call_panic)
 				 from_kgid(&init_user_ns, context->ipc.gid),
 				 context->ipc.mode);
 		if (lsmprop_is_set(&context->ipc.oprop)) {
-			struct lsm_context lsmctx;
-
-			if (security_lsmprop_to_secctx(&context->ipc.oprop,
-						       &lsmctx,
-						       LSM_ID_UNDEF) < 0) {
+			if (audit_log_obj_ctx(ab, &context->ipc.oprop))
 				*call_panic = 1;
-			} else {
-				audit_log_format(ab, " obj=%s", lsmctx.context);
-				security_release_secctx(&lsmctx);
-			}
 		}
 		if (context->ipc.has_perm) {
 			audit_log_end(ab);
@@ -1558,18 +1543,9 @@ static void audit_log_name(struct audit_context *context, struct audit_names *n,
 				 from_kgid(&init_user_ns, n->gid),
 				 MAJOR(n->rdev),
 				 MINOR(n->rdev));
-	if (lsmprop_is_set(&n->oprop)) {
-		struct lsm_context ctx;
-
-		if (security_lsmprop_to_secctx(&n->oprop, &ctx,
-					       LSM_ID_UNDEF) < 0) {
-			if (call_panic)
-				*call_panic = 2;
-		} else {
-			audit_log_format(ab, " obj=%s", ctx.context);
-			security_release_secctx(&ctx);
-		}
-	}
+	if (lsmprop_is_set(&n->oprop) &&
+	    audit_log_obj_ctx(ab, &n->oprop))
+		*call_panic = 2;
 
 	/* log the audit_names record type */
 	switch (n->type) {
@@ -1780,15 +1756,16 @@ static void audit_log_exit(void)
 						  axs->target_sessionid[i],
 						  &axs->target_ref[i],
 						  axs->target_comm[i]))
-				call_panic = 1;
+			call_panic = 1;
 	}
 
 	if (context->target_pid &&
 	    audit_log_pid_context(context, context->target_pid,
 				  context->target_auid, context->target_uid,
 				  context->target_sessionid,
-				  &context->target_ref, context->target_comm))
-			call_panic = 1;
+				  &context->target_ref,
+				  context->target_comm))
+		call_panic = 1;
 
 	if (context->pwd.dentry && context->pwd.mnt) {
 		ab = audit_log_start(context, GFP_KERNEL, AUDIT_CWD);
-- 
2.47.0

Re: [PATCH v3 5/5] Audit: Add record for multiple object contexts

[Paul Moore]

On Mar 19, 2025 Casey Schaufler <casey@schaufler-ca.com> wrote:
> 
> Create a new audit record AUDIT_MAC_OBJ_CONTEXTS.
> An example of the MAC_OBJ_CONTEXTS (1424) record is:
> 
>     type=MAC_OBJ_CONTEXTS[1424]
>     msg=audit(1601152467.009:1050):
>     obj_selinux=unconfined_u:object_r:user_home_t:s0
> 
> When an audit event includes a AUDIT_MAC_OBJ_CONTEXTS record
> the "obj=" field in other records in the event will be "obj=?".
> An AUDIT_MAC_OBJ_CONTEXTS record is supplied when the system has
> multiple security modules that may make access decisions based
> on an object security context.
> 
> Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
> ---
>  include/linux/audit.h      |  6 +++++
>  include/uapi/linux/audit.h |  1 +
>  kernel/audit.c             | 51 +++++++++++++++++++++++++++++++++++++-
>  kernel/auditsc.c           | 45 ++++++++-------------------------
>  4 files changed, 68 insertions(+), 35 deletions(-)

Similar to patch 4/5, this looks fine modulo the obj count changes.

Related, you changed to a single subj/obj count in v3, is it no longer
important to distinguish between the two?

--
paul-moore.com