Skip to contentSkip to conversation

pid: annotate data-races around pid_ns->pid_allocated

PATCH In review Jiayuan Chen 3 participants 4 messages Apr 25, 2025 – Apr 28, 2025

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

PATCH v2 1/1 pid: annotate data-races around pid_ns->pid_allocated

Suppress syzbot reports by annotating these accesses using data_race().

---
 kernel/fork.c          | 2 +-
 kernel/pid.c           | 8 ++++----
 kernel/pid_namespace.c | 2 +-
 3 files changed, 6 insertions(+), 6 deletions(-)
kernel/fork.c +1 -1
index c4b26cd8998b..5aa050418fda 100644
@@ -2584,7 +2584,7 @@ __latent_entropy struct task_struct *copy_process(
 	rseq_fork(p, clone_flags);
 
 	/* Don't start children in a dying pid namespace */
-	if (unlikely(!(ns_of_pid(pid)->pid_allocated & PIDNS_ADDING))) {
+	if (unlikely(!(data_race(ns_of_pid(pid)->pid_allocated & PIDNS_ADDING)))) {
 		retval = -ENOMEM;
 		goto bad_fork_core_free;
 	}
kernel/pid.c +4 -4
index 4ac2ce46817f..df59228dd27e 100644
@@ -122,7 +122,7 @@ void free_pid(struct pid *pid)
 	for (i = 0; i <= pid->level; i++) {
 		struct upid *upid = pid->numbers + i;
 		struct pid_namespace *ns = upid->ns;
-		switch (--ns->pid_allocated) {
+		switch (data_race(--ns->pid_allocated)) {
 		case 2:
 		case 1:
 			/* When all that is left in the pid namespace
@@ -134,7 +134,7 @@ void free_pid(struct pid *pid)
 		case PIDNS_ADDING:
 			/* Handle a fork failure of the first process */
 			WARN_ON(ns->child_reaper);
-			ns->pid_allocated = 0;
+			data_race(ns->pid_allocated = 0);
 			break;
 		}
 
@@ -271,13 +271,13 @@ struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid,
 	upid = pid->numbers + ns->level;
 	idr_preload(GFP_KERNEL);
 	spin_lock(&pidmap_lock);
-	if (!(ns->pid_allocated & PIDNS_ADDING))
+	if (!(data_race(ns->pid_allocated & PIDNS_ADDING)))
 		goto out_unlock;
 	pidfs_add_pid(pid);
 	for ( ; upid >= pid->numbers; --upid) {
 		/* Make the PID visible to find_pid_ns. */
 		idr_replace(&upid->ns->idr, pid, upid->nr);
-		upid->ns->pid_allocated++;
+		data_race(upid->ns->pid_allocated++);
 	}
 	spin_unlock(&pidmap_lock);
 	idr_preload_end();
kernel/pid_namespace.c +1 -1
index 7098ed44e717..6e2a32641904 100644
@@ -268,7 +268,7 @@ void zap_pid_ns_processes(struct pid_namespace *pid_ns)
 	 */
 	for (;;) {
 		set_current_state(TASK_INTERRUPTIBLE);
-		if (pid_ns->pid_allocated == init_pids)
+		if (data_race(pid_ns->pid_allocated == init_pids))
 			break;
 		schedule();
 	}
Signature · 1 line
2.47.1
#1 lore raw
Quoted 1 line · Jiayuan Chen

Suppress syzbot reports by annotating these accesses using data_race().

Thanks for trying this approach.
scripts/checkpatch.pl has quite some remarks about the current form :-) I mean, the data_race annotation should document why the race is harmless. I only glanced over this so I can't tell myself whether it's a bug or OK.

Quoted 3 lines
Reported-by [email protected]
Reported-by [email protected]
Reported-by [email protected]

How can I get to see full syzbot reports? (Stacktrace would be helpful, I cannot resolve those as message-ids on (LK)ML.)

Thanks,
Michal

#3 lore raw
In reply to Jiayuan Chen

kernel/fork.c __latent_entropy struct task_struct *copy_process( · line 2587

Quoted 7 lines · Jiayuan Chen
kernel/fork.c +1 -1
@@ -2584,7 +2584,7 @@ __latent_entropy struct task_struct *copy_process(
 	rseq_fork(p, clone_flags);
 
 	/* Don't start children in a dying pid namespace */
-	if (unlikely(!(ns_of_pid(pid)->pid_allocated & PIDNS_ADDING))) {
+	if (unlikely(!(data_race(ns_of_pid(pid)->pid_allocated & PIDNS_ADDING)))) {

Well. data_race() just hides the potential problem. READ_ONCE() makes more sense imo, even if I think there are no real problems with the current code.

Either way,

kernel/pid.c struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid, · line 274

Quoted 6 lines
kernel/pid.c +1 -1
@@ -271,13 +271,13 @@ struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid,
 	upid = pid->numbers + ns->level;
 	idr_preload(GFP_KERNEL);
 	spin_lock(&pidmap_lock);
-	if (!(ns->pid_allocated & PIDNS_ADDING))
+	if (!(data_race(ns->pid_allocated & PIDNS_ADDING)))

again, you do not need data_race() or READ_ONCE() if you read the data protected by pidmap_lock. But you still need WRITE_ONCE() when ->pid_allocated is modified.

Oleg.

#2 lore raw
In reply to Michal Koutný
Quoted 25 lines · Michal Koutný
Quoted 3 lines · Jiayuan Chen

Suppress syzbot reports by annotating these accesses using data_race().

Thanks for trying this approach.

scripts/checkpatch.pl has quite some remarks about the current form :-)

I mean, the data_race annotation should document why the race is

harmless. I only glanced over this so I can't tell myself whether it's a

bug or OK.

Quoted 7 lines
Reported-by [email protected]
Reported-by [email protected]
Reported-by [email protected]
Quoted 8 lines

How can I get to see full syzbot reports? (Stacktrace would be helpful,

I cannot resolve those as message-ids on (LK)ML.)

Thanks,

Michal

Hi Michal,

We can find such reports at https://syzkaller.appspot.com/upstream/s/kernel