scheduler_tick() has the potential of running for some time if f.e.
sched_domains for a system with 1024 processors have to be balanced.
We currently do all of that with interrupts disabled. So we may be unable
to service interrupts for some time.
I wonder if it would be possible to put the sched_tick() into a tasklet and
allow interrupts to be enabled? Preemption is still disabled and so we
are stuck on a cpu.
This can only work if we are sure that no scheduler activity is performed
from interrupt code or from other tasklets that may potentially run.
sched_tick() takes the lock on a request queue without disabling interrupts.
So any other attempt to invoke scheduler functions from interrupts or
tasklets will cause a deadlock.
I tested this patch with AIM7 and things seem to be fine.
Signed-off-by: Christoph Lameter <[email protected]>
Index: linux-2.6.19-rc2-mm1/kernel/timer.c
===================================================================
--- linux-2.6.19-rc2-mm1.orig/kernel/timer.c 2006-10-18 11:52:53.160984800 -0500
+++ linux-2.6.19-rc2-mm1/kernel/timer.c 2006-10-18 11:53:10.039399811 -0500
@@ -1210,8 +1210,14 @@ static void update_wall_time(void)
}
}
+static void sched_tick_action(unsigned long dummy)
+{
+ scheduler_tick();
+}
+
+static DECLARE_TASKLET(scheduler_tick_tasklet, sched_tick_action, 0);
/*
- * Called from the timer interrupt handler to charge one tick to the current
+ * Called from the timer interrupt handler to charge one tick to the current
* process. user_tick is 1 if the tick is user time, 0 for system.
*/
void update_process_times(int user_tick)
@@ -1227,7 +1233,7 @@ void update_process_times(int user_tick)
run_local_timers();
if (rcu_pending(cpu))
rcu_check_callbacks(cpu, user_tick);
- scheduler_tick();
+ tasklet_schedule(&scheduler_tick_tasklet);
run_posix_cpu_timers(p);
}
Index: linux-2.6.19-rc2-mm1/kernel/sched.c
===================================================================
--- linux-2.6.19-rc2-mm1.orig/kernel/sched.c 2006-10-18 11:52:51.768286047 -0500
+++ linux-2.6.19-rc2-mm1/kernel/sched.c 2006-10-18 11:53:10.003263868 -0500
@@ -1608,9 +1608,10 @@ void fastcall sched_fork(struct task_str
* runqueue lock is not a problem.
*/
current->time_slice = 1;
+ local_irq_enable();
scheduler_tick();
- }
- local_irq_enable();
+ } else
+ local_irq_enable();
put_cpu();
}
@@ -3040,7 +3041,8 @@ void account_steal_time(struct task_stru
/*
* This function gets called by the timer code, with HZ frequency.
- * We call it with interrupts disabled.
+ * We call it with interrupts enabled. However, the thread is
+ * pinned to a specific cpu.
*
* It also gets called by the fork code, when changing the parent's
* timeslices.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
