History

fs/namespace.c: WARN if mnt_count has become negative

Missing calls to mntget() (or equivalently, too many calls to mntput()) are hard to detect because mntput() delays freeing mounts using task_work_add(), then again using call_rcu(). As a result, mnt_count can often be decremented to -1 without getting a KASAN use-after-free report. Such cases are still bugs though, and they point to real use-after-frees being possible. For an example of this, see the bug fixed by commit 1b0b9cc8d379 ("vfs: fsmount: add missing mntget()"), discussed at https://lkml.kernel.org/linux-fsdevel/20190605135401.GB30925@xxxxxxxxxxxxxxxxxxxxxxxxx/T/#u. This bug *should* have been trivial to find. But actually, it wasn't found until syzkaller happened to use fchdir() to manipulate the reference count just right for the bug to be noticeable. Address this by making mntput_no_expire() issue a WARN if mnt_count has become negative.

Eric Biggers committed 6 years ago

Commit edf7ddb1 parent b6505452 files · +7 −4Browse files

Unified · Split

2 files
  1. fs/namespace.c +6 −3
  2. fs/pnode.h +1 −1
fs/namespace.c+6−3
@@ -156,10 +156,10 @@ static inline void mnt_add_count(struct mount *mnt, int n) · 155 unmodified lines
156156 /*
157157 * vfsmount lock must be held for write
158158 */
159−unsigned int mnt_get_count(struct mount *mnt)
159+int mnt_get_count(struct mount *mnt)
160160 {
161161 #ifdef CONFIG_SMP
162− unsigned int count = 0;
162+ int count = 0;
163163 int cpu;
164164
165165 for_each_possible_cpu(cpu) {
@@ -1139,6 +1139,7 @@ static DECLARE_DELAYED_WORK(delayed_mntput_work, delayed_mntput); · 973 unmodified lines
11391139 static void mntput_no_expire(struct mount *mnt)
11401140 {
11411141 LIST_HEAD(list);
1142+ int count;
11421143
11431144 rcu_read_lock();
11441145 if (likely(READ_ONCE(mnt->mnt_ns))) {
@@ -1162,7 +1163,9 @@ static void mntput_no_expire(struct mount *mnt) · 17 unmodified lines
11621163 */
11631164 smp_mb();
11641165 mnt_add_count(mnt, -1);
1165− if (mnt_get_count(mnt)) {
1166+ count = mnt_get_count(mnt);
1167+ if (count != 0) {
1168+ WARN_ON(count < 0);
11661169 rcu_read_unlock();
11671170 unlock_mount_hash();
11681171 return;
fs/pnode.h+1−1
@@ -44,7 +44,7 @@ int propagate_mount_busy(struct mount *, int); · 43 unmodified lines
4444 void propagate_mount_unlock(struct mount *);
4545 void mnt_release_group_id(struct mount *);
4646 int get_dominating_id(struct mount *mnt, const struct path *root);
47−unsigned int mnt_get_count(struct mount *mnt);
47+int mnt_get_count(struct mount *mnt);
4848 void mnt_set_mountpoint(struct mount *, struct mountpoint *,
4949 struct mount *);
5050 void mnt_change_mountpoint(struct mount *parent, struct mountpoint *mp,

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t