Skip to contentSkip to conversation

kernel/auditfilter.c: need process 'tree' when audit_add_watch() failed in audit_add_rule().

PATCH In review Chen Gang 2 participants 3 messages May 10, 2013 – May 23, 2013 Audit

lore
Share & tools
lore

Chen Gang

author
#0 lore raw

PATCH 1/1 kernel/auditfilter.c: need process 'tree' when audit_add_watch() failed in audit_add_rule().

If both 'tree' and 'watch' are valid, need call audit_put_tree(), just like the upper area has done within function audit_add_rule().

---
 kernel/auditfilter.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)
kernel/auditfilter.c +6 -0
index f9fc54b..81f63f9 100644
@@ -952,6 +952,12 @@ static inline int audit_add_rule(struct audit_entry *entry)
 		err = audit_add_watch(&entry->rule, &list);
 		if (err) {
 			mutex_unlock(&audit_filter_mutex);
+			/*
+			 * normally audit_add_tree_rule() will free it
+			 * on failure
+			 */
+			if (tree)
+				audit_put_tree(tree);
 			goto error;
 		}
 	}
Signature · 1 line
1.7.7.6
#1 lore raw
Quoted 16 lines · On Fri, 10 May 2013 18:12:26 +0800 Chen Gang <[email protected]> wrote:

If both 'tree' and 'watch' are valid, need call audit_put_tree(), just like the upper area has done within function audit_add_rule().

---
 kernel/auditfilter.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)
kernel/auditfilter.c +0 -0
index f9fc54b..81f63f9 100644
@@ -952,6 +952,12 @@ static inline int audit_add_rule(struct audit_entry *entry)
 		err = audit_add_watch(&entry->rule, &list);
 		if (err) {

kernel/auditfilter.c static inline int audit_add_rule(struct audit_entry *entry) · line 963

Quoted 10 lines
kernel/auditfilter.c +6 -0
 			mutex_unlock(&audit_filter_mutex);
+			/*
+			 * normally audit_add_tree_rule() will free it
+			 * on failure
+			 */
+			if (tree)
+				audit_put_tree(tree);
 			goto error;
 		}
 	}

Are you sure? Or is the earlier audit_put_tree(tree) wrong?

Where is the "get" which this "put" is undoing?

author
#2 lore raw
Quoted 21 lines · Andrew Morton
Quoted 19 lines · On Fri, 10 May 2013 18:12:26 +0800 Chen Gang <[email protected]> wrote:
Quoted 19 lines

If both 'tree' and 'watch' are valid, need call audit_put_tree(), just like the upper area has done within function audit_add_rule().

---
 kernel/auditfilter.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)
kernel/auditfilter.c +2 -0
index f9fc54b..81f63f9 100644
@@ -952,6 +952,12 @@ static inline int audit_add_rule(struct audit_entry *entry)
 		err = audit_add_watch(&entry->rule, &list);
 		if (err) {
 			mutex_unlock(&audit_filter_mutex);
+			/*
+			 * normally audit_add_tree_rule() will free it
Quoted 9 lines
Quoted 7 lines

kernel/auditfilter.c static inline int audit_add_rule(struct audit_entry *entry) · line 963

Quoted 7 lines
kernel/auditfilter.c +4 -0
+			 * on failure
+			 */
+			if (tree)
+				audit_put_tree(tree);
 			goto error;
 		}
 	}

Are you sure? Or is the earlier audit_put_tree(tree) wrong?

Yes I am sure.

Since audit_add_tree_rule() will really free it on failure, we have to be sure to free it in another area within audit_add_rule().

Quoted 3 lines

Where is the "get" which this "put" is undoing?

"Allocating tree" is the "get", this "put" will really free the tree.

Thanks.

Signature · 3 lines
Chen Gang

Asianux Corporation