Fix data races to pwq->triggered by using READ_ONCE. These accesses are expected to race per comment and already provided memory barriers surrounding the accesses. Original KCSAN report: write to 0xffffc90001043c30 of 4 bytes by task 17695 on cpu 0: __pollwake fs/select.c:197 [inline] pollwake+0xe3/0x140 fs/select.c:217 __wake_up_common+0x7b/0x180 kernel/sched/wait.c:93 __wake_up_common_lock+0x77/0xb0 kernel/sched/wait.c:123 __wake_up+0xe/0x10 kernel/sched/wait.c:142 signalfd_notify include/linux/signalfd.h:22 [inline] signalfd_notify include/linux/signalfd.h:19 [inline] __send_signal+0x70e/0x870 kernel/signal.c:1158 send_signal+0x224/0x2b0 kernel/signal.c:1236 __group_send_sig_info kernel/signal.c:1275 [inline] do_notify_parent+0x55b/0x5e0 kernel/signal.c:1992 exit_notify kernel/exit.c:670 [inline] do_exit+0x16ef/0x18c0 kernel/exit.c:818 do_group_exit+0xb4/0x1c0 kernel/exit.c:895 __do_sys_exit_group kernel/exit.c:906 [inline] __se_sys_exit_group kernel/exit.c:904 [inline] __x64_sys_exit_group+0x2e/0x30 kernel/exit.c:904 do_syscall_64+0xcc/0x3a0 arch/x86/entry/common.c:294 entry_SYSCALL_64_after_hwframe+0x44/0xa9 read to 0xffffc90001043c30 of 4 bytes by task 15995 on cpu 1: poll_schedule_timeout.constprop.0+0x50/0xc0 fs/select.c:242 do_poll fs/select.c:951 [inline] do_sys_poll+0x66d/0x990 fs/select.c:1001 __do_sys_poll fs/select.c:1059 [inline] __se_sys_poll fs/select.c:1047 [inline] __x64_sys_poll+0x10f/0x250 fs/select.c:1047 do_syscall_64+0xcc/0x3a0 arch/x86/entry/common.c:294 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Reported-by: syzbot+bd0076c6e839503096b5@xxxxxxxxxxxxxxxxxxxxxxxxx Signed-off-by: Marco Elver <elver@xxxxxxxxxx> --- fs/select.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/select.c b/fs/select.c index 11d0285d46b7..152987ddf40e 100644 --- a/fs/select.c +++ b/fs/select.c @@ -194,7 +194,7 @@ static int __pollwake(wait_queue_entry_t *wait, unsigned mode, int sync, void *k * and is paired with smp_store_mb() in poll_schedule_timeout. */ smp_wmb(); - pwq->triggered = 1; + WRITE_ONCE(pwq->triggered, 1); /* * Perform the default wake up operation using a dummy @@ -239,7 +239,7 @@ static int poll_schedule_timeout(struct poll_wqueues *pwq, int state, int rc = -EINTR; set_current_state(state); - if (!pwq->triggered) + if (!READ_ONCE(pwq->triggered)) rc = schedule_hrtimeout_range(expires, slack, HRTIMER_MODE_ABS); __set_current_state(TASK_RUNNING); -- 2.24.1.735.g03f4e72817-goog
