issignal(): ignore signals when process is single-threading for exit

Places that will wait for curproc->p_singlethr to become zero (in the
next commit, the counter of number of external single-threading is
to be introduced), must wait for it interruptible, otherwise we
deadlock.  On the other hand, a signal delivered during this window,
if directed to the waiting thread, would cause the wait loop to become
a busy loop.

Since we are exiting, it is safe to ignore the signals.

Reviewed by:	markj
Tested by:	pho
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
Differential revision:	https://reviews.freebsd.org/D35310

1 files changed, 11 insertions, 1 deletions

diff --git a/sys/kern/kern_sig.c b/sys/kern/kern_sig.c
index afb2fd78c231..e3bbbd23ae6c 100644
--- a/sys/kern/kern_sig.c
+++ b/sys/kern/kern_sig.c
@@ -2928,7 +2928,7 @@ sigprocess(struct thread *td, int sig)
 
 	/*
 	 * We should allow pending but ignored signals below
-	 * only if there is sigwait() active, or P_TRACED was
+	 * if there is sigwait() active, or P_TRACED was
 	 * on when they were posted.
 	 */
 	if (SIGISMEMBER(ps->ps_sigignore, sig) &&
@@ -2937,6 +2937,16 @@ sigprocess(struct thread *td, int sig)
 		return (SIGSTATUS_IGNORE);
 	}
 
+	/*
+	 * If the process is going to single-thread mode to prepare
+	 * for exit, there is no sense in delivering any signal
+	 * to usermode.  Another important consequence is that
+	 * msleep(..., PCATCH, ...) now is only interruptible by a
+	 * suspend request.
+	 */
+	if ((p->p_flag2 & P2_WEXIT) != 0)
+		return (SIGSTATUS_IGNORE);
+
 	if ((p->p_flag & (P_TRACED | P_PPTRACE)) == P_TRACED) {
 		/*
 		 * If traced, always stop.