Current pselect(3) is implemented in usermode and thus vulnerable to

well-known race condition, which elimination was the reason for the
function appearance in first place. If sigmask supplied as argument to
pselect() enables a signal, the signal might be delivered before thread
called select(2), causing lost wakeup. Reimplement pselect() in kernel,
making change of sigmask and sleep atomic.

Since signal shall be delivered to the usermode, but sigmask restored,
set TDP_OLDMASK and save old mask in td_oldsigmask. The TDP_OLDMASK
should be cleared by ast() in case signal was not gelivered during
syscall execution.

Reviewed by:	davidxu
Tested by:	pho
MFC after:	1 month

1 files changed, 53 insertions, 3 deletions

diff --git a/sys/kern/sys_generic.c b/sys/kern/sys_generic.c
index 6831fe8..b34af61 100644
--- a/sys/kern/sys_generic.c
+++ b/sys/kern/sys_generic.c
@@ -751,6 +751,58 @@ poll_no_poll(int events)
 	return (events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
 }
 
+int
+pselect(struct thread *td, struct pselect_args *uap)
+{
+	struct timespec ts;
+	struct timeval tv, *tvp;
+	sigset_t set, *uset;
+	int error;
+
+	if (uap->ts != NULL) {
+		error = copyin(uap->ts, &ts, sizeof(ts));
+		if (error != 0)
+		    return (error);
+		TIMESPEC_TO_TIMEVAL(&tv, &ts);
+		tvp = &tv;
+	} else
+		tvp = NULL;
+	if (uap->sm != NULL) {
+		error = copyin(uap->sm, &set, sizeof(set));
+		if (error != 0)
+			return (error);
+		uset = &set;
+	} else
+		uset = NULL;
+	return (kern_pselect(td, uap->nd, uap->in, uap->ou, uap->ex, tvp,
+	    uset, NFDBITS));
+}
+
+int
+kern_pselect(struct thread *td, int nd, fd_set *in, fd_set *ou, fd_set *ex,
+    struct timeval *tvp, sigset_t *uset, int abi_nfdbits)
+{
+	int error;
+
+	if (uset != NULL) {
+		error = kern_sigprocmask(td, SIG_SETMASK, uset,
+		    &td->td_oldsigmask, 0);
+		if (error != 0)
+			return (error);
+		td->td_pflags |= TDP_OLDMASK;
+		/*
+		 * Make sure that ast() is called on return to
+		 * usermode and TDP_OLDMASK is cleared, restoring old
+		 * sigmask.
+		 */
+		thread_lock(td);
+		td->td_flags |= TDF_ASTPENDING;
+		thread_unlock(td);
+	}
+	error = kern_select(td, nd, in, ou, ex, tvp, abi_nfdbits);
+	return (error);
+}
+
 #ifndef _SYS_SYSPROTO_H_
 struct select_args {
 	int	nd;
@@ -759,9 +811,7 @@ struct select_args {
 };
 #endif
 int
-select(td, uap)
-	register struct thread *td;
-	register struct select_args *uap;
+select(struct thread *td, struct select_args *uap)
 {
 	struct timeval tv, *tvp;
 	int error;