diff options
| author | fabient <fabient@FreeBSD.org> | 2011-03-31 13:59:47 +0000 |
|---|---|---|
| committer | fabient <fabient@FreeBSD.org> | 2011-03-31 13:59:47 +0000 |
| commit | e0588db8d22fa2514698e48727b291a5bca65dbf (patch) | |
| tree | ff69eb812c81c5fd3f17576f198670edff277f82 /sys/kern/sched_ule.c | |
| parent | 94327c73822769aa17f717f6edbf6e54159cc288 (diff) | |
| download | FreeBSD-src-e0588db8d22fa2514698e48727b291a5bca65dbf.zip FreeBSD-src-e0588db8d22fa2514698e48727b291a5bca65dbf.tar.gz | |
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
Diffstat (limited to 'sys/kern/sched_ule.c')
| -rw-r--r-- | sys/kern/sched_ule.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/sys/kern/sched_ule.c b/sys/kern/sched_ule.c index c1bb98f..ac18e77 100644 --- a/sys/kern/sched_ule.c +++ b/sys/kern/sched_ule.c @@ -1783,7 +1783,8 @@ sched_switch(struct thread *td, struct thread *newtd, int flags) ts->ts_rltick = ticks; td->td_lastcpu = td->td_oncpu; td->td_oncpu = NOCPU; - td->td_flags &= ~TDF_NEEDRESCHED; + if (!(flags & SW_PREEMPT)) + td->td_flags &= ~TDF_NEEDRESCHED; td->td_owepreempt = 0; tdq->tdq_switchcnt++; /* |
