Commit a14e0110 authored Nov 30, 2014 by Steve Muckle

sched: refresh sched_clock() after acquiring rq lock in irq path



The wallclock time passed to sched_account_irqtime() may be stale
after we wait to acquire the runqueue lock. This could cause problems
in update_task_ravg because a different CPU may have advanced
this CPU's window_start based on a more up-to-date wallclock value,
triggering a BUG_ON(window_start > wallclock).

Change-Id: I316af62d1716e9b59c4a2898a2d9b44d6c7a75d8
Signed-off-by: Steve Muckle <smuckle@codeaurora.org>

parent 5fdc1d3a

kernel/sched/core.c

+4 −2

Original line number	Diff line number	Diff line
		@@ -1832,14 +1832,16 @@ void sched_account_irqtime(int cpu, struct task_struct *curr,
		{
		struct rq *rq = cpu_rq(cpu);
		unsigned long flags, nr_windows;
		u64 cur_jiffies_ts;
		u64 cur_jiffies_ts, now;

		raw_spin_lock_irqsave(&rq->lock, flags);

		now = sched_clock();
		delta += (now - wallclock);
		cur_jiffies_ts = get_jiffies_64();

		if (is_idle_task(curr))
		update_task_ravg(curr, rq, IRQ_UPDATE, wallclock, delta);
		update_task_ravg(curr, rq, IRQ_UPDATE, now, delta);

		nr_windows = cur_jiffies_ts - rq->irqload_ts;