Skip to contentSkip to conversation

ima: add gid support

PATCH In review Alex Henrie 3 participants 4 messages Oct 5, 2021 – Oct 7, 2021

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 ima: add gid support

From: Curtis Veit <[email protected]>

IMA currently supports the concept of rules based on uid where the rule is based on the uid of the file owner or the uid of the user accessing the file. It is useful to have similar rules based on gid. This patch provides that ability.

---
v2: Trivial changes that Mimi requested

---
 Documentation/ABI/testing/ima_policy |   8 +-
 security/integrity/ima/ima_policy.c  | 201 +++++++++++++++++++++++----
 2 files changed, 180 insertions(+), 29 deletions(-)
Documentation/ABI/testing/ima_policy +6 -2
index 5c2798534950..e1a04bd3b9e5 100644
@@ -22,8 +22,9 @@ Description:
 		  action: measure | dont_measure | appraise | dont_appraise |
 			  audit | hash | dont_hash
 		  condition:= base | lsm  [option]
-			base:	[[func=] [mask=] [fsmagic=] [fsuuid=] [uid=]
-				[euid=] [fowner=] [fsname=]]
+			base:	[[func=] [mask=] [fsmagic=] [fsuuid=] [fsname=]
+				[uid=] [euid=] [gid=] [egid=]
+				[fowner=] [fgroup=]]
 			lsm:	[[subj_user=] [subj_role=] [subj_type=]
 				 [obj_user=] [obj_role=] [obj_type=]]
 			option:	[[appraise_type=]] [template=] [permit_directio]
@@ -40,7 +41,10 @@ Description:
 			fsuuid:= file system UUID (e.g 8bcbe394-4f13-4144-be8e-5aa9ea2ce2f6)
 			uid:= decimal value
 			euid:= decimal value
+			gid:= decimal value
+			egid:= decimal value
 			fowner:= decimal value
+			fgroup:= decimal value
 		  lsm:  are LSM specific
 		  option:
 			appraise_type:= [imasig] [imasig|modsig]
security/integrity/ima/ima_policy.c +174 -27

354 lines not on this page · raw message

Signature · 1 line
2.33.0
maintainer
#1 lore raw

Hi Alex,

Quoted 7 lines · Alex Henrie

From: Curtis Veit <[email protected]>

IMA currently supports the concept of rules based on uid where the rule is based on the uid of the file owner or the uid of the user accessing the file. It is useful to have similar rules based on gid. This patch provides that ability.

Quoted 5 lines

---
v2: Trivial changes that Mimi requested

Sorry, scripts/check-patch.pl reported some warnings. Two more trivial changes.

Quoted 10 lines
---
 Documentation/ABI/testing/ima_policy |   8 +-
 security/integrity/ima/ima_policy.c  | 201 +++++++++++++++++++++++----
 2 files changed, 180 insertions(+), 29 deletions(-)
Documentation/ABI/testing/ima_policy +0 -1
index 5c2798534950..e1a04bd3b9e5 100644
--- a/Documentation/ABI/testing/ima_policy

Documentation/ABI/testing/ima_policy struct ima_rule_entry { · line 90

Quoted 10 lines
Documentation/ABI/testing/ima_policy +4 -0
@@ -78,9 +81,13 @@ struct ima_rule_entry {
 	unsigned long fsmagic;
 	uuid_t fsuuid;
 	kuid_t uid;
+	kgid_t gid;
 	kuid_t fowner;
+	kgid_t fgroup;
 	bool (*uid_op)(kuid_t, kuid_t);    /* Handlers for operators       */
+	bool (*gid_op)(kgid_t, kgid_t);
 	bool (*fowner_op)(kuid_t, kuid_t); /* uid_eq(), uid_gt(), uid_lt() */
+	bool (*fgroup_op)(kgid_t, kgid_t); /* gid_eq(), gid_gt(), gid_lt() */

scripts/checkpatch.pl complains about missing variables.

Documentation/ABI/testing/ima_policy static bool ima_match_rules(struct ima_rule_entry *rule, · line 593

Quoted 10 lines
Documentation/ABI/testing/ima_policy +1 -1
 	int pcr;
 	unsigned int allowed_algos; /* bitfield of allowed hash algorithms */
 	struct {
@@ -582,10 +590,23 @@ static bool ima_match_rules(struct ima_rule_entry *rule,
 		} else if (!rule->uid_op(cred->euid, rule->uid))
 			return false;
 	}
-
+	if ((rule->flags & IMA_GID) && !rule->gid_op(rule->gid, cred->gid))

All of uid_op/gid_op calls in ima_match_rules() pass the "cred->xxxx,
rule->xxx" except here, where it is rule->gid, cred->rule. Reversing
the parameters here will help with addressing the checkpatch.pl warning.

thanks,

Mimi

Reviewed-by
#2 lore raw

Hi Alex, Mimi,

Quoted 27 lines

Hi Alex,

Quoted 7 lines · Alex Henrie

From: Curtis Veit <[email protected]>

IMA currently supports the concept of rules based on uid where the rule is based on the uid of the file owner or the uid of the user accessing the file. It is useful to have similar rules based on gid. This patch provides that ability.

Quoted 5 lines

---
v2: Trivial changes that Mimi requested

Sorry, scripts/check-patch.pl reported some warnings. Two more trivial changes.

Quoted 8 lines
---
 Documentation/ABI/testing/ima_policy |   8 +-
 security/integrity/ima/ima_policy.c  | 201 +++++++++++++++++++++++----
 2 files changed, 180 insertions(+), 29 deletions(-)
Documentation/ABI/testing/ima_policy +0 -1
index 5c2798534950..e1a04bd3b9e5 100644
--- a/Documentation/ABI/testing/ima_policy
Quoted 13 lines

Documentation/ABI/testing/ima_policy struct ima_rule_entry { · line 90

Quoted 11 lines
Documentation/ABI/testing/ima_policy +4 -0
@@ -78,9 +81,13 @@ struct ima_rule_entry {
 	unsigned long fsmagic;
 	uuid_t fsuuid;
 	kuid_t uid;
+	kgid_t gid;
 	kuid_t fowner;
+	kgid_t fgroup;
 	bool (*uid_op)(kuid_t, kuid_t);    /* Handlers for operators       */
+	bool (*gid_op)(kgid_t, kgid_t);
 	bool (*fowner_op)(kuid_t, kuid_t); /* uid_eq(), uid_gt(), uid_lt() */
+	bool (*fgroup_op)(kgid_t, kgid_t); /* gid_eq(), gid_gt(), gid_lt() */

scripts/checkpatch.pl complains about missing variables.

+1

Quoted 4 lines
Quoted 3 lines
Documentation/ABI/testing/ima_policy +0 -0
 	int pcr;
 	unsigned int allowed_algos; /* bitfield of allowed hash algorithms */
 	struct {
Quoted 11 lines

Documentation/ABI/testing/ima_policy static bool ima_match_rules(struct ima_rule_entry *rule, · line 593

Quoted 6 lines
Documentation/ABI/testing/ima_policy +1 -1
@@ -582,10 +590,23 @@ static bool ima_match_rules(struct ima_rule_entry *rule,
 		} else if (!rule->uid_op(cred->euid, rule->uid))
 			return false;
 	}
-
+	if ((rule->flags & IMA_GID) && !rule->gid_op(rule->gid, cred->gid))

All of uid_op/gid_op calls in ima_match_rules() pass the "cred->xxxx,
rule->xxx" except here, where it is rule->gid, cred->rule. Reversing
the parameters here will help with addressing the checkpatch.pl warning.

+1

Apart from those checkpatch issues patch LGTM.

Kind regards,
Petr

Quoted 3 lines

thanks,

Mimi

#3 lore raw
In reply to Mimi Zohar
Quoted 14 lines · On Wed, 06 Oct 2021 15:49:58 -0400 Mimi Zohar <[email protected]> wrote:

security/integrity/ima/ima_policy.c struct ima_rule_entry { · line 90

Quoted 11 lines · Alex Henrie
security/integrity/ima/ima_policy.c +4 -0
@@ -78,9 +81,13 @@ struct ima_rule_entry {
 	unsigned long fsmagic;
 	uuid_t fsuuid;
 	kuid_t uid;
+	kgid_t gid;
 	kuid_t fowner;
+	kgid_t fgroup;
 	bool (*uid_op)(kuid_t, kuid_t);    /* Handlers for operators       */
+	bool (*gid_op)(kgid_t, kgid_t);
 	bool (*fowner_op)(kuid_t, kuid_t); /* uid_eq(), uid_gt(), uid_lt() */
+	bool (*fgroup_op)(kgid_t, kgid_t); /* gid_eq(), gid_gt(), gid_lt() */

scripts/checkpatch.pl complains about missing variables.

I'll resend with a new patch that fixes the existing style problems with the UID code before adding GID support.

Quoted 11 lines

security/integrity/ima/ima_policy.c static bool ima_match_rules(struct ima_rule_entry *rule, · line 593

Quoted 6 lines
security/integrity/ima/ima_policy.c +1 -1
@@ -582,10 +590,23 @@ static bool ima_match_rules(struct ima_rule_entry *rule,
 		} else if (!rule->uid_op(cred->euid, rule->uid))
 			return false;
 	}
-
+	if ((rule->flags & IMA_GID) && !rule->gid_op(rule->gid, cred->gid))

All of uid_op/gid_op calls in ima_match_rules() pass the "cred->xxxx,
rule->xxx" except here, where it is rule->gid, cred->rule. Reversing
the parameters here will help with addressing the checkpatch.pl warning.

Good catch, thanks.

-Alex