Clearing the flag when preempting will let the preempted thread run

too much time. This can finish in a scheduler deadlock with ping-pong
between two threads.

One sample of this is:
- device lapic (to have a preemption point on critical_exit())
- options DEVICE_POLLING with HZ>1499 (to have lapic freq = hardclock freq)
- running a cpu intensive task (that does not enter the kernel)
- only one CPU on SMP or no SMP.

As requested by jhb@ 4BSD have received the same type of fix instead of
propagating the flag to the new thread.

Reviewed by:	jhb, jeff
MFC after:	1 month

1 files changed, 2 insertions, 6 deletions

diff --git a/sys/kern/sched_4bsd.c b/sys/kern/sched_4bsd.c
index f7025b0..c96d7b5 100644
--- a/sys/kern/sched_4bsd.c
+++ b/sys/kern/sched_4bsd.c
@@ -940,13 +940,9 @@ sched_switch(struct thread *td, struct thread *newtd, int flags)
 	if ((td->td_flags & TDF_NOLOAD) == 0)
 		sched_load_rem();
 
-	if (newtd) {
-		MPASS(newtd->td_lock == &sched_lock);
-		newtd->td_flags |= (td->td_flags & TDF_NEEDRESCHED);
-	}
-
 	td->td_lastcpu = td->td_oncpu;
-	td->td_flags &= ~TDF_NEEDRESCHED;
+	if (!(flags & SW_PREEMPT))
+		td->td_flags &= ~TDF_NEEDRESCHED;
 	td->td_owepreempt = 0;
 	td->td_oncpu = NOCPU;