A while ago vtime_init_idle() used to invoke sched_clock_cpu() and had
write_seqlock_irqsave() because sched_clock_cpu() required to be called
with interrupts off.
This requirement is written before the body of the function and was
introduced in 2010 via c676329abb2b ("sched_clock: Add local_clock() API
and improve documentation"). This requirement has been dropped in 2013
via ef08f0fff876 ("sched/clock: Remove local_irq_disable() from the clocks")
but the body of the function still has the comment.
Now, vtime_init_idle() was converted from seqlock to seqcount via
b7ce2277f087 ("sched/cputime: Convert vtime_seqlock to seqcount") and
while doing so the IRQ-off region around sched_clock_cpu() was preserved
while not strictly required (given the first part of this commit).
A little later, sched_clock_cpu() was replaced with jiffies via
ff9a9b4c4334 ("sched, time: Switch VIRT_CPU_ACCOUNTING_GEN to jiffy
granularity").
Based on this events I assume it is safe to drop the local_irq_safe()
section.
Cc: Peter Zijlstra <a.p.zijls...@chello.nl>
Cc: Frederic Weisbecker <fweis...@gmail.com>
Cc: Rik van Riel <r...@redhat.com>
Cc: Thomas Gleixner <t...@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bige...@linutronix.de>
---
kernel/sched/cputime.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c
index 75f98c5498d5..816cd4ef5873 100644
--- a/kernel/sched/cputime.c
+++ b/kernel/sched/cputime.c
@@ -785,14 +785,10 @@ void arch_vtime_task_switch(struct task_struct *prev)
void vtime_init_idle(struct task_struct *t, int cpu)
{
- unsigned long flags;
-
- local_irq_save(flags);
write_seqcount_begin(&t->vtime_seqcount);
t->vtime_snap_whence = VTIME_SYS;
t->vtime_snap = jiffies;
write_seqcount_end(&t->vtime_seqcount);
- local_irq_restore(flags);
}
cputime_t task_gtime(struct task_struct *t)
--
2.8.0.rc3
