x86/tsc, sched/clock, clocksource: Use clocksource watchdog to provide stable sync...
authorPeter Zijlstra <peterz@infradead.org>
Fri, 21 Apr 2017 10:14:13 +0000 (12:14 +0200)
committerIngo Molnar <mingo@kernel.org>
Mon, 15 May 2017 08:15:18 +0000 (10:15 +0200)
commitb421b22b00b0011f6a2ce3561176c4e79e640c49
tree26fa2e2baa7f403f66292a5f1e8b55ea0ceecf1f
parentcf15ca8deda86b27b66e27848b4b0fe58098fc0b
x86/tsc, sched/clock, clocksource: Use clocksource watchdog to provide stable sync points

Currently we keep sched_clock_tick() active for stable TSC in order to
keep the per-CPU state semi up-to-date. The (obvious) problem is that
by the time we detect TSC is borked, our per-CPU state is also borked.

So hook into the clocksource watchdog and call a method after we've
found it to still be stable.

There's the obvious race where the TSC goes wonky between finding it
stable and us running the callback, but closing that is too much work
and not really worth it, since we're already detecting TSC wobbles
after the fact, so we cannot, per definition, fully avoid funny clock
values.

And since the watchdog runs less often than the tick, this is also an
optimization.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/tsc.c
include/linux/clocksource.h
include/linux/sched/clock.h
kernel/sched/clock.c
kernel/time/clocksource.c