proc: Report SIGEV_NONE in /proc/pid/timers if target task has died
PATCH 1/1 proc: Report SIGEV_NONE in /proc/pid/timers if target task has died
When a posix timer is created targeting a specific thread (using SIGEV_SIGNAL | SIGEV_THREAD_ID), it takes a reference to the target struct pid in timer->it_pid. If the target thread subsequently terminates, its numeric tid is freed and can be recycled for an unrelated task. However, the timer holds its reference to the original struct pid.
show_timer() in /proc/[pid]/timers previously called pid_nr_ns() directly on timer->it_pid without checking whether any task remained attached to that struct pid. As a result:
1. It reported the stale tid, which could mistakenly refer to a recycled pid.
2. In the kernel, expired signals for dead target threads are dropped by posixtimer_send_sigqueue() because posixtimer_get_target() returns NULL, so the timer functionally acts as SIGEV_NONE.
3. Checkpoint/restore tools (CRIU) parsing /proc/[pid]/timers would try to restore a timer with SIGEV_SIGNAL | SIGEV_THREAD_ID targeting a non-existent or unrelated thread.
Check pid_has_task(timer->it_pid, timer->it_pid_type) in show_timer(). If the target task has died, override notify to SIGEV_NONE and report PID 0 (e.g., 'notify: none/pid.0').
--- fs/proc/base.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
fs/proc/base.c +7 -1
@@ -2519,17 +2519,23 @@ static int show_timer(struct seq_file *m, void *v) struct k_itimer *timer = hlist_entry((struct hlist_node *)v, struct k_itimer, list);struct timers_private *tp = m->private;int notify = timer->it_sigev_notify; +pid_t nr =0 ;guard (spinlock_irq)(&timer->it_lock);if (!posixtimer_valid (timer))return 0 ; +if (timer->it_pid &&pid_has_task (timer->it_pid, timer->it_pid_type)) + nr =pid_nr_ns (timer->it_pid, tp->ns); +else + notify =SIGEV_NONE ; +seq_printf (m,"ID: %d\n" , timer->it_id);seq_printf (m,"signal: %d/%px\n" , timer->sigq.info.si_signo, timer->sigq.info.si_value.sival_ptr);seq_printf (m,"notify: %s/%s.%d\n" , nstr[notify & ~SIGEV_THREAD_ID ], (notify &SIGEV_THREAD_ID ) ?"tid" :"pid" , -pid_nr_ns (timer->it_pid, tp->ns)); + nr);seq_printf (m,"ClockID: %d\n" , timer->it_clock);return 0 ;
Signature · 1 line
2.55.0.691.gc56d675ccc-goog
Quoted 19 lines · Andrei Vagin
When a posix timer is created targeting a specific thread (using
SIGEV_SIGNAL|SIGEV_THREAD_ID), it takes a reference to the target struct pid intimer->it_pid. If the target thread subsequently terminates, its numeric tid is freed and can be recycled for an unrelated task. However, the timer holds its reference to the original struct pid.
show_timer()in /proc/[pid]/timers previously calledpid_nr_ns()directly ontimer->it_pidwithout checking whether any task remained attached to that struct pid. As a result:
1. It reported the stale tid, which could mistakenly refer to a recycled pid.
2. In the kernel, expired signals for dead target threads are dropped byposixtimer_send_sigqueue()becauseposixtimer_get_target()returns NULL, so the timer functionally acts asSIGEV_NONE.
3. Checkpoint/restore tools (CRIU) parsing /proc/[pid]/timers would try to restore a timer withSIGEV_SIGNAL|SIGEV_THREAD_IDtargeting a non-existent or unrelated thread.
Quoted 6 lines
Check
pid_has_task(timer->it_pid, timer->it_pid_type)inshow_timer(). If the target task has died, override notify toSIGEV_NONEand report PID 0 (e.g., 'notify: none/pid.0').Cc Thomas Gleixner <[email protected]>Signed-off-by Andrei Vagin <[email protected]>
Hopefully a useful note:
Though the fixed interface is not full-proof by itself, e.g. there is still window that task alive at the time of check can be already dead at the time when userspace uses the reported pid.
That gap probably can be closed easily from userspace, e.g. with taking pidfd on this reported pid and re reading the info (to make sure it's still the same task), or in CRIU we just have container processes frozen/ptrace-stopped so the reported pid task should not go away under us.
Quoted 33 lines
--- fs/proc/base.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)fs/proc/base.c +7
-1@@ -2519,17 +2519,23 @@ static int show_timer(struct seq_file *m, void *v) struct k_itimer *timer = hlist_entry((struct hlist_node *)v, struct k_itimer, list);struct timers_private *tp = m->private;int notify = timer->it_sigev_notify; +pid_t nr =0 ;guard (spinlock_irq)(&timer->it_lock);if (!posixtimer_valid (timer))return 0 ; +if (timer->it_pid &&pid_has_task (timer->it_pid, timer->it_pid_type)) + nr =pid_nr_ns (timer->it_pid, tp->ns); +else + notify =SIGEV_NONE ; +seq_printf (m,"ID: %d\n" , timer->it_id);seq_printf (m,"signal: %d/%px\n" , timer->sigq.info.si_signo, timer->sigq.info.si_value.sival_ptr);seq_printf (m,"notify: %s/%s.%d\n" , nstr[notify & ~SIGEV_THREAD_ID ], (notify &SIGEV_THREAD_ID ) ?"tid" :"pid" , -pid_nr_ns (timer->it_pid, tp->ns)); + nr);seq_printf (m,"ClockID: %d\n" , timer->it_clock);return 0 ;
Signature · 2 lines
Best regards, Pavel Tikhomirov Senior Software Developer, Virtuozzo.
Quoted 30 lines · Pavel Tikhomirov
Quoted 19 lines · Andrei Vagin
When a posix timer is created targeting a specific thread (using
SIGEV_SIGNAL|SIGEV_THREAD_ID), it takes a reference to the target struct pid intimer->it_pid. If the target thread subsequently terminates, its numeric tid is freed and can be recycled for an unrelated task. However, the timer holds its reference to the original struct pid.
show_timer()in /proc/[pid]/timers previously calledpid_nr_ns()directly ontimer->it_pidwithout checking whether any task remained attached to that struct pid. As a result:
1. It reported the stale tid, which could mistakenly refer to a recycled pid.
2. In the kernel, expired signals for dead target threads are dropped byposixtimer_send_sigqueue()becauseposixtimer_get_target()returns NULL, so the timer functionally acts asSIGEV_NONE.
3. Checkpoint/restore tools (CRIU) parsing /proc/[pid]/timers would try to restore a timer withSIGEV_SIGNAL|SIGEV_THREAD_IDtargeting a non-existent or unrelated thread.Quoted 6 lines
Check
pid_has_task(timer->it_pid, timer->it_pid_type)inshow_timer(). If the target task has died, override notify toSIGEV_NONEand report PID 0 (e.g., 'notify: none/pid.0').Cc Thomas Gleixner <[email protected]>Signed-off-by Andrei Vagin <[email protected]>
Quoted 7 lines
Reviewed-by Pavel Tikhomirov <[email protected]>Hopefully a useful note:
Though the fixed interface is not full-proof by itself, e.g. there is still window that task alive at the time of check can be already dead at the time when userspace uses the reported pid.
Pavel, this isn't a unique problem. We can say that about any pid reported in /proc.
Quoted 5 lines
That gap probably can be closed easily from userspace, e.g. with taking pidfd on this reported pid and re reading the info (to make sure it's still the same task), or in CRIU we just have container processes frozen/ptrace-stopped so the reported pid task should not go away under us.
I think you answered your question:)
Thanks,
Andrei
Quoted 7 lines · On Sun, 16 Aug 2026 16:12:15 +0000 Andrei Vagin <[email protected]> wrote:
When a posix timer is created targeting a specific thread (using
SIGEV_SIGNAL|SIGEV_THREAD_ID), it takes a reference to the target struct pid intimer->it_pid. If the target thread subsequently terminates, its numeric tid is freed and can be recycled for an unrelated task. However, the timer holds its reference to the original struct pid.
Quoted 11 lines
show_timer()in /proc/[pid]/timers previously calledpid_nr_ns()directly ontimer->it_pidwithout checking whether any task remained attached to that struct pid. As a result:
1. It reported the stale tid, which could mistakenly refer to a recycled pid.
2. In the kernel, expired signals for dead target threads are dropped byposixtimer_send_sigqueue()becauseposixtimer_get_target()returns NULL, so the timer functionally acts asSIGEV_NONE.
3. Checkpoint/restore tools (CRIU) parsing /proc/[pid]/timers would try to restore a timer withSIGEV_SIGNAL|SIGEV_THREAD_IDtargeting a non-existent or unrelated thread.
This sounds like a somewhat significant issue for CRIU but that's just me wildly guessing.
Quoted 3 lines
Check
pid_has_task(timer->it_pid, timer->it_pid_type)inshow_timer(). If the target task has died, override notify toSIGEV_NONEand report PID 0 (e.g., 'notify: none/pid.0').
No Fixes: and no cc:stable? Maybe my guess was wrong?
Quoted 22 lines · Andrew Morton
Quoted 7 lines · On Sun, 16 Aug 2026 16:12:15 +0000 Andrei Vagin <[email protected]> wrote:
When a posix timer is created targeting a specific thread (using
SIGEV_SIGNAL|SIGEV_THREAD_ID), it takes a reference to the target struct pid intimer->it_pid. If the target thread subsequently terminates, its numeric tid is freed and can be recycled for an unrelated task. However, the timer holds its reference to the original struct pid.Quoted 11 lines
show_timer()in /proc/[pid]/timers previously calledpid_nr_ns()directly ontimer->it_pidwithout checking whether any task remained attached to that struct pid. As a result:
1. It reported the stale tid, which could mistakenly refer to a recycled pid.
2. In the kernel, expired signals for dead target threads are dropped byposixtimer_send_sigqueue()becauseposixtimer_get_target()returns NULL, so the timer functionally acts asSIGEV_NONE.
3. Checkpoint/restore tools (CRIU) parsing /proc/[pid]/timers would try to restore a timer withSIGEV_SIGNAL|SIGEV_THREAD_IDtargeting a non-existent or unrelated thread.
Quoted 8 lines
This sounds like a somewhat significant issue for CRIU but that's just me wildly guessing.
Quoted 3 lines
Check
pid_has_task(timer->it_pid, timer->it_pid_type)inshow_timer(). If the target task has died, override notify toSIGEV_NONEand report PID 0 (e.g., 'notify: none/pid.0').No Fixes: and no cc:stable? Maybe my guess was wrong?
I was not sure this "qualifies" as a regression, as this behavior has existed since /proc/[pid]/timers was introduced more than 10 years ago.
In CRIU, we already have a workaround checking whether the process has a thread with the specified TID. The only edge case this workaround cannot handle is if the original thread dies and a new thread is later created reusing that same recycled TID. This is a reason why this kernel change is required.
The proper "fixes" tag is:
Thanks,
Andrei