2006-06-26 09:25:05 +02:00
|
|
|
/*
|
|
|
|
* linux/kernel/time/clocksource.c
|
|
|
|
*
|
|
|
|
* This file contains the functions which manage clocksource drivers.
|
|
|
|
*
|
|
|
|
* Copyright (C) 2004, 2005 IBM, John Stultz (johnstul@us.ibm.com)
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|
|
|
*
|
|
|
|
* TODO WishList:
|
|
|
|
* o Allow clocksource drivers to be unregistered
|
|
|
|
*/
|
|
|
|
|
2015-05-25 20:49:55 +02:00
|
|
|
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
|
|
|
|
|
2011-12-15 00:28:51 +01:00
|
|
|
#include <linux/device.h>
|
2006-06-26 09:25:05 +02:00
|
|
|
#include <linux/clocksource.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/module.h>
|
[PATCH] kernel/time/clocksource.c needs struct task_struct on m68k
kernel/time/clocksource.c needs struct task_struct on m68k.
Because it uses spin_unlock_irq(), which, on m68k, uses hardirq_count(), which
uses preempt_count(), which needs to dereference struct task_struct, we
have to include sched.h. Because it would cause a loop inclusion, we
cannot include sched.h in any other of asm-m68k/system.h,
linux/thread_info.h, linux/hardirq.h, which leaves this ugly include in
a C file as the only simple solution.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-10 10:43:43 +01:00
|
|
|
#include <linux/sched.h> /* for spin_unlock_irq() using preempt_count() m68k */
|
2007-02-16 10:28:03 +01:00
|
|
|
#include <linux/tick.h>
|
2009-08-18 17:09:42 +02:00
|
|
|
#include <linux/kthread.h>
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2015-03-25 13:07:37 +01:00
|
|
|
#include "tick-internal.h"
|
2014-07-16 23:05:10 +02:00
|
|
|
#include "timekeeping_internal.h"
|
2013-04-25 22:31:50 +02:00
|
|
|
|
2009-11-11 15:05:29 +01:00
|
|
|
/**
|
|
|
|
* clocks_calc_mult_shift - calculate mult/shift factors for scaled math of clocks
|
|
|
|
* @mult: pointer to mult variable
|
|
|
|
* @shift: pointer to shift variable
|
|
|
|
* @from: frequency to convert from
|
|
|
|
* @to: frequency to convert to
|
2011-01-11 18:18:12 +01:00
|
|
|
* @maxsec: guaranteed runtime conversion range in seconds
|
2009-11-11 15:05:29 +01:00
|
|
|
*
|
|
|
|
* The function evaluates the shift/mult pair for the scaled math
|
|
|
|
* operations of clocksources and clockevents.
|
|
|
|
*
|
|
|
|
* @to and @from are frequency values in HZ. For clock sources @to is
|
|
|
|
* NSEC_PER_SEC == 1GHz and @from is the counter frequency. For clock
|
|
|
|
* event @to is the counter frequency and @from is NSEC_PER_SEC.
|
|
|
|
*
|
2011-01-11 18:18:12 +01:00
|
|
|
* The @maxsec conversion range argument controls the time frame in
|
2009-11-11 15:05:29 +01:00
|
|
|
* seconds which must be covered by the runtime conversion with the
|
|
|
|
* calculated mult and shift factors. This guarantees that no 64bit
|
|
|
|
* overflow happens when the input value of the conversion is
|
|
|
|
* multiplied with the calculated mult factor. Larger ranges may
|
|
|
|
* reduce the conversion accuracy by chosing smaller mult and shift
|
|
|
|
* factors.
|
|
|
|
*/
|
|
|
|
void
|
2011-01-11 18:18:12 +01:00
|
|
|
clocks_calc_mult_shift(u32 *mult, u32 *shift, u32 from, u32 to, u32 maxsec)
|
2009-11-11 15:05:29 +01:00
|
|
|
{
|
|
|
|
u64 tmp;
|
|
|
|
u32 sft, sftacc= 32;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculate the shift factor which is limiting the conversion
|
|
|
|
* range:
|
|
|
|
*/
|
2011-01-11 18:18:12 +01:00
|
|
|
tmp = ((u64)maxsec * from) >> 32;
|
2009-11-11 15:05:29 +01:00
|
|
|
while (tmp) {
|
|
|
|
tmp >>=1;
|
|
|
|
sftacc--;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Find the conversion shift/mult pair which has the best
|
|
|
|
* accuracy and fits the maxsec conversion range:
|
|
|
|
*/
|
|
|
|
for (sft = 32; sft > 0; sft--) {
|
|
|
|
tmp = (u64) to << sft;
|
2010-12-16 20:03:27 +01:00
|
|
|
tmp += from / 2;
|
2009-11-11 15:05:29 +01:00
|
|
|
do_div(tmp, from);
|
|
|
|
if ((tmp >> sftacc) == 0)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
*mult = tmp;
|
|
|
|
*shift = sft;
|
|
|
|
}
|
|
|
|
|
2006-06-26 09:25:05 +02:00
|
|
|
/*[Clocksource internal variables]---------
|
|
|
|
* curr_clocksource:
|
2009-08-14 15:47:21 +02:00
|
|
|
* currently selected clocksource.
|
2006-06-26 09:25:05 +02:00
|
|
|
* clocksource_list:
|
|
|
|
* linked list with the registered clocksources
|
2009-08-14 15:47:30 +02:00
|
|
|
* clocksource_mutex:
|
|
|
|
* protects manipulations to curr_clocksource and the clocksource_list
|
2006-06-26 09:25:05 +02:00
|
|
|
* override_name:
|
|
|
|
* Name of the user-specified clocksource.
|
|
|
|
*/
|
2009-08-14 15:47:21 +02:00
|
|
|
static struct clocksource *curr_clocksource;
|
2006-06-26 09:25:05 +02:00
|
|
|
static LIST_HEAD(clocksource_list);
|
2009-08-14 15:47:30 +02:00
|
|
|
static DEFINE_MUTEX(clocksource_mutex);
|
2013-04-25 22:31:45 +02:00
|
|
|
static char override_name[CS_NAME_LEN];
|
2009-09-14 19:49:02 +02:00
|
|
|
static int finished_booting;
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2007-02-16 10:27:43 +01:00
|
|
|
#ifdef CONFIG_CLOCKSOURCE_WATCHDOG
|
2009-09-11 15:33:05 +02:00
|
|
|
static void clocksource_watchdog_work(struct work_struct *work);
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
static void clocksource_select(void);
|
2009-09-11 15:33:05 +02:00
|
|
|
|
2007-02-16 10:27:43 +01:00
|
|
|
static LIST_HEAD(watchdog_list);
|
|
|
|
static struct clocksource *watchdog;
|
|
|
|
static struct timer_list watchdog_timer;
|
2009-09-11 15:33:05 +02:00
|
|
|
static DECLARE_WORK(watchdog_work, clocksource_watchdog_work);
|
2007-02-16 10:27:43 +01:00
|
|
|
static DEFINE_SPINLOCK(watchdog_lock);
|
2009-08-14 15:47:24 +02:00
|
|
|
static int watchdog_running;
|
2011-09-12 13:32:23 +02:00
|
|
|
static atomic_t watchdog_reset_pending;
|
2007-05-09 11:35:15 +02:00
|
|
|
|
2009-08-18 17:09:42 +02:00
|
|
|
static int clocksource_watchdog_kthread(void *data);
|
2009-08-19 11:26:09 +02:00
|
|
|
static void __clocksource_change_rating(struct clocksource *cs, int rating);
|
2009-08-14 15:47:25 +02:00
|
|
|
|
2007-02-16 10:27:43 +01:00
|
|
|
/*
|
2007-05-09 11:33:40 +02:00
|
|
|
* Interval: 0.5sec Threshold: 0.0625s
|
2007-02-16 10:27:43 +01:00
|
|
|
*/
|
|
|
|
#define WATCHDOG_INTERVAL (HZ >> 1)
|
2007-05-09 11:33:40 +02:00
|
|
|
#define WATCHDOG_THRESHOLD (NSEC_PER_SEC >> 4)
|
2007-02-16 10:27:43 +01:00
|
|
|
|
2009-08-18 17:09:42 +02:00
|
|
|
static void clocksource_watchdog_work(struct work_struct *work)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* If kthread_run fails the next watchdog scan over the
|
|
|
|
* watchdog_list will find the unstable clock again.
|
|
|
|
*/
|
|
|
|
kthread_run(clocksource_watchdog_kthread, NULL, "kwatchdog");
|
|
|
|
}
|
|
|
|
|
2009-08-28 20:25:24 +02:00
|
|
|
static void __clocksource_unstable(struct clocksource *cs)
|
2007-02-16 10:27:43 +01:00
|
|
|
{
|
|
|
|
cs->flags &= ~(CLOCK_SOURCE_VALID_FOR_HRES | CLOCK_SOURCE_WATCHDOG);
|
2009-08-14 15:47:25 +02:00
|
|
|
cs->flags |= CLOCK_SOURCE_UNSTABLE;
|
2009-09-14 19:49:02 +02:00
|
|
|
if (finished_booting)
|
|
|
|
schedule_work(&watchdog_work);
|
2007-02-16 10:27:43 +01:00
|
|
|
}
|
|
|
|
|
2009-08-28 20:25:24 +02:00
|
|
|
/**
|
|
|
|
* clocksource_mark_unstable - mark clocksource unstable via watchdog
|
|
|
|
* @cs: clocksource to be marked unstable
|
|
|
|
*
|
|
|
|
* This function is called instead of clocksource_change_rating from
|
|
|
|
* cpu hotplug code to avoid a deadlock between the clocksource mutex
|
|
|
|
* and the cpu hotplug mutex. It defers the update of the clocksource
|
|
|
|
* to the watchdog thread.
|
|
|
|
*/
|
|
|
|
void clocksource_mark_unstable(struct clocksource *cs)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&watchdog_lock, flags);
|
|
|
|
if (!(cs->flags & CLOCK_SOURCE_UNSTABLE)) {
|
|
|
|
if (list_empty(&cs->wd_list))
|
|
|
|
list_add(&cs->wd_list, &watchdog_list);
|
|
|
|
__clocksource_unstable(cs);
|
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&watchdog_lock, flags);
|
|
|
|
}
|
|
|
|
|
2007-02-16 10:27:43 +01:00
|
|
|
static void clocksource_watchdog(unsigned long data)
|
|
|
|
{
|
2009-08-14 15:47:25 +02:00
|
|
|
struct clocksource *cs;
|
2015-03-12 05:16:36 +01:00
|
|
|
cycle_t csnow, wdnow, cslast, wdlast, delta;
|
2007-02-16 10:27:43 +01:00
|
|
|
int64_t wd_nsec, cs_nsec;
|
2011-09-12 13:32:23 +02:00
|
|
|
int next_cpu, reset_pending;
|
2007-02-16 10:27:43 +01:00
|
|
|
|
|
|
|
spin_lock(&watchdog_lock);
|
2009-08-14 15:47:24 +02:00
|
|
|
if (!watchdog_running)
|
|
|
|
goto out;
|
2007-02-16 10:27:43 +01:00
|
|
|
|
2011-09-12 13:32:23 +02:00
|
|
|
reset_pending = atomic_read(&watchdog_reset_pending);
|
|
|
|
|
2009-08-14 15:47:25 +02:00
|
|
|
list_for_each_entry(cs, &watchdog_list, wd_list) {
|
|
|
|
|
|
|
|
/* Clocksource already marked unstable? */
|
2009-08-18 17:09:42 +02:00
|
|
|
if (cs->flags & CLOCK_SOURCE_UNSTABLE) {
|
2009-09-14 19:49:02 +02:00
|
|
|
if (finished_booting)
|
|
|
|
schedule_work(&watchdog_work);
|
2009-08-14 15:47:25 +02:00
|
|
|
continue;
|
2009-08-18 17:09:42 +02:00
|
|
|
}
|
2009-08-14 15:47:25 +02:00
|
|
|
|
2011-06-16 16:22:08 +02:00
|
|
|
local_irq_disable();
|
2009-04-21 21:24:00 +02:00
|
|
|
csnow = cs->read(cs);
|
2011-06-16 16:22:08 +02:00
|
|
|
wdnow = watchdog->read(watchdog);
|
|
|
|
local_irq_enable();
|
2007-05-09 11:35:15 +02:00
|
|
|
|
2009-08-14 15:47:22 +02:00
|
|
|
/* Clocksource initialized ? */
|
2011-09-12 13:32:23 +02:00
|
|
|
if (!(cs->flags & CLOCK_SOURCE_WATCHDOG) ||
|
|
|
|
atomic_read(&watchdog_reset_pending)) {
|
2009-08-14 15:47:22 +02:00
|
|
|
cs->flags |= CLOCK_SOURCE_WATCHDOG;
|
2011-06-16 16:22:08 +02:00
|
|
|
cs->wd_last = wdnow;
|
|
|
|
cs->cs_last = csnow;
|
2007-05-09 11:35:15 +02:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2014-07-16 23:05:10 +02:00
|
|
|
delta = clocksource_delta(wdnow, cs->wd_last, watchdog->mask);
|
|
|
|
wd_nsec = clocksource_cyc2ns(delta, watchdog->mult,
|
|
|
|
watchdog->shift);
|
2011-06-16 16:22:08 +02:00
|
|
|
|
2014-07-16 23:05:10 +02:00
|
|
|
delta = clocksource_delta(csnow, cs->cs_last, cs->mask);
|
|
|
|
cs_nsec = clocksource_cyc2ns(delta, cs->mult, cs->shift);
|
2015-03-12 05:16:36 +01:00
|
|
|
wdlast = cs->wd_last; /* save these in case we print them */
|
|
|
|
cslast = cs->cs_last;
|
2011-06-16 16:22:08 +02:00
|
|
|
cs->cs_last = csnow;
|
|
|
|
cs->wd_last = wdnow;
|
|
|
|
|
2011-09-12 13:32:23 +02:00
|
|
|
if (atomic_read(&watchdog_reset_pending))
|
|
|
|
continue;
|
|
|
|
|
2011-06-16 16:22:08 +02:00
|
|
|
/* Check the deviation from the watchdog clocksource. */
|
2015-11-09 23:58:13 +01:00
|
|
|
if (abs(cs_nsec - wd_nsec) > WATCHDOG_THRESHOLD) {
|
2015-09-10 11:01:56 +02:00
|
|
|
pr_warn("timekeeping watchdog on CPU%d: Marking clocksource '%s' as unstable because the skew is too large:\n",
|
|
|
|
smp_processor_id(), cs->name);
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_warn(" '%s' wd_now: %llx wd_last: %llx mask: %llx\n",
|
2015-03-12 05:16:36 +01:00
|
|
|
watchdog->name, wdnow, wdlast, watchdog->mask);
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_warn(" '%s' cs_now: %llx cs_last: %llx mask: %llx\n",
|
2015-03-12 05:16:36 +01:00
|
|
|
cs->name, csnow, cslast, cs->mask);
|
|
|
|
__clocksource_unstable(cs);
|
2009-08-14 15:47:22 +02:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) &&
|
|
|
|
(cs->flags & CLOCK_SOURCE_IS_CONTINUOUS) &&
|
|
|
|
(watchdog->flags & CLOCK_SOURCE_IS_CONTINUOUS)) {
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
/* Mark it valid for high-res. */
|
2009-08-14 15:47:22 +02:00
|
|
|
cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES;
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* clocksource_done_booting() will sort it if
|
|
|
|
* finished_booting is not set yet.
|
|
|
|
*/
|
|
|
|
if (!finished_booting)
|
|
|
|
continue;
|
|
|
|
|
2009-08-14 15:47:22 +02:00
|
|
|
/*
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
* If this is not the current clocksource let
|
|
|
|
* the watchdog thread reselect it. Due to the
|
|
|
|
* change to high res this clocksource might
|
|
|
|
* be preferred now. If it is the current
|
|
|
|
* clocksource let the tick code know about
|
|
|
|
* that change.
|
2009-08-14 15:47:22 +02:00
|
|
|
*/
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
if (cs != curr_clocksource) {
|
|
|
|
cs->flags |= CLOCK_SOURCE_RESELECT;
|
|
|
|
schedule_work(&watchdog_work);
|
|
|
|
} else {
|
|
|
|
tick_clock_notify();
|
|
|
|
}
|
2007-02-16 10:27:43 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-09-12 13:32:23 +02:00
|
|
|
/*
|
|
|
|
* We only clear the watchdog_reset_pending, when we did a
|
|
|
|
* full cycle through all clocksources.
|
|
|
|
*/
|
|
|
|
if (reset_pending)
|
|
|
|
atomic_dec(&watchdog_reset_pending);
|
|
|
|
|
2009-08-14 15:47:25 +02:00
|
|
|
/*
|
|
|
|
* Cycle through CPUs to check if the CPUs stay synchronized
|
|
|
|
* to each other.
|
|
|
|
*/
|
|
|
|
next_cpu = cpumask_next(raw_smp_processor_id(), cpu_online_mask);
|
|
|
|
if (next_cpu >= nr_cpu_ids)
|
|
|
|
next_cpu = cpumask_first(cpu_online_mask);
|
|
|
|
watchdog_timer.expires += WATCHDOG_INTERVAL;
|
|
|
|
add_timer_on(&watchdog_timer, next_cpu);
|
2009-08-14 15:47:24 +02:00
|
|
|
out:
|
2007-02-16 10:27:43 +01:00
|
|
|
spin_unlock(&watchdog_lock);
|
|
|
|
}
|
2009-08-14 15:47:23 +02:00
|
|
|
|
2009-08-14 15:47:24 +02:00
|
|
|
static inline void clocksource_start_watchdog(void)
|
|
|
|
{
|
|
|
|
if (watchdog_running || !watchdog || list_empty(&watchdog_list))
|
|
|
|
return;
|
|
|
|
init_timer(&watchdog_timer);
|
|
|
|
watchdog_timer.function = clocksource_watchdog;
|
|
|
|
watchdog_timer.expires = jiffies + WATCHDOG_INTERVAL;
|
|
|
|
add_timer_on(&watchdog_timer, cpumask_first(cpu_online_mask));
|
|
|
|
watchdog_running = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void clocksource_stop_watchdog(void)
|
|
|
|
{
|
|
|
|
if (!watchdog_running || (watchdog && !list_empty(&watchdog_list)))
|
|
|
|
return;
|
|
|
|
del_timer(&watchdog_timer);
|
|
|
|
watchdog_running = 0;
|
|
|
|
}
|
|
|
|
|
2009-08-14 15:47:23 +02:00
|
|
|
static inline void clocksource_reset_watchdog(void)
|
|
|
|
{
|
|
|
|
struct clocksource *cs;
|
|
|
|
|
|
|
|
list_for_each_entry(cs, &watchdog_list, wd_list)
|
|
|
|
cs->flags &= ~CLOCK_SOURCE_WATCHDOG;
|
|
|
|
}
|
|
|
|
|
2007-05-09 11:35:15 +02:00
|
|
|
static void clocksource_resume_watchdog(void)
|
|
|
|
{
|
2011-09-12 13:32:23 +02:00
|
|
|
atomic_inc(&watchdog_reset_pending);
|
2007-05-09 11:35:15 +02:00
|
|
|
}
|
|
|
|
|
2009-08-14 15:47:24 +02:00
|
|
|
static void clocksource_enqueue_watchdog(struct clocksource *cs)
|
2007-02-16 10:27:43 +01:00
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&watchdog_lock, flags);
|
|
|
|
if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) {
|
2009-08-14 15:47:24 +02:00
|
|
|
/* cs is a clocksource to be watched. */
|
2007-02-16 10:27:43 +01:00
|
|
|
list_add(&cs->wd_list, &watchdog_list);
|
2009-08-14 15:47:24 +02:00
|
|
|
cs->flags &= ~CLOCK_SOURCE_WATCHDOG;
|
2007-03-25 14:42:51 +02:00
|
|
|
} else {
|
2009-08-14 15:47:24 +02:00
|
|
|
/* cs is a watchdog. */
|
2007-03-25 14:42:51 +02:00
|
|
|
if (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS)
|
2007-02-16 10:27:43 +01:00
|
|
|
cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES;
|
2016-01-22 18:31:53 +01:00
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&watchdog_lock, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void clocksource_select_watchdog(bool fallback)
|
|
|
|
{
|
|
|
|
struct clocksource *cs, *old_wd;
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&watchdog_lock, flags);
|
|
|
|
/* save current watchdog */
|
|
|
|
old_wd = watchdog;
|
|
|
|
if (fallback)
|
|
|
|
watchdog = NULL;
|
|
|
|
|
|
|
|
list_for_each_entry(cs, &clocksource_list, list) {
|
|
|
|
/* cs is a clocksource to be watched. */
|
|
|
|
if (cs->flags & CLOCK_SOURCE_MUST_VERIFY)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/* Skip current if we were requested for a fallback. */
|
|
|
|
if (fallback && cs == old_wd)
|
|
|
|
continue;
|
|
|
|
|
2009-08-14 15:47:24 +02:00
|
|
|
/* Pick the best watchdog. */
|
2016-01-22 18:31:53 +01:00
|
|
|
if (!watchdog || cs->rating > watchdog->rating)
|
2007-02-16 10:27:43 +01:00
|
|
|
watchdog = cs;
|
|
|
|
}
|
2016-01-22 18:31:53 +01:00
|
|
|
/* If we failed to find a fallback restore the old one. */
|
|
|
|
if (!watchdog)
|
|
|
|
watchdog = old_wd;
|
|
|
|
|
|
|
|
/* If we changed the watchdog we need to reset cycles. */
|
|
|
|
if (watchdog != old_wd)
|
|
|
|
clocksource_reset_watchdog();
|
|
|
|
|
2009-08-14 15:47:24 +02:00
|
|
|
/* Check if the watchdog timer needs to be started. */
|
|
|
|
clocksource_start_watchdog();
|
2007-02-16 10:27:43 +01:00
|
|
|
spin_unlock_irqrestore(&watchdog_lock, flags);
|
|
|
|
}
|
2009-08-14 15:47:24 +02:00
|
|
|
|
|
|
|
static void clocksource_dequeue_watchdog(struct clocksource *cs)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&watchdog_lock, flags);
|
2013-04-25 22:31:46 +02:00
|
|
|
if (cs != watchdog) {
|
|
|
|
if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) {
|
|
|
|
/* cs is a watched clocksource. */
|
|
|
|
list_del_init(&cs->wd_list);
|
|
|
|
/* Check if the watchdog timer needs to be stopped. */
|
|
|
|
clocksource_stop_watchdog();
|
2009-08-14 15:47:24 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&watchdog_lock, flags);
|
|
|
|
}
|
|
|
|
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
static int __clocksource_watchdog_kthread(void)
|
2009-08-14 15:47:25 +02:00
|
|
|
{
|
|
|
|
struct clocksource *cs, *tmp;
|
|
|
|
unsigned long flags;
|
2009-08-15 13:20:42 +02:00
|
|
|
LIST_HEAD(unstable);
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
int select = 0;
|
2009-08-14 15:47:25 +02:00
|
|
|
|
|
|
|
spin_lock_irqsave(&watchdog_lock, flags);
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
list_for_each_entry_safe(cs, tmp, &watchdog_list, wd_list) {
|
2009-08-14 15:47:25 +02:00
|
|
|
if (cs->flags & CLOCK_SOURCE_UNSTABLE) {
|
|
|
|
list_del_init(&cs->wd_list);
|
2009-08-15 13:20:42 +02:00
|
|
|
list_add(&cs->wd_list, &unstable);
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
select = 1;
|
|
|
|
}
|
|
|
|
if (cs->flags & CLOCK_SOURCE_RESELECT) {
|
|
|
|
cs->flags &= ~CLOCK_SOURCE_RESELECT;
|
|
|
|
select = 1;
|
2009-08-14 15:47:25 +02:00
|
|
|
}
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
}
|
2009-08-14 15:47:25 +02:00
|
|
|
/* Check if the watchdog timer needs to be stopped. */
|
|
|
|
clocksource_stop_watchdog();
|
2009-08-15 13:20:42 +02:00
|
|
|
spin_unlock_irqrestore(&watchdog_lock, flags);
|
|
|
|
|
|
|
|
/* Needs to be done outside of watchdog lock */
|
|
|
|
list_for_each_entry_safe(cs, tmp, &unstable, wd_list) {
|
|
|
|
list_del_init(&cs->wd_list);
|
2009-08-19 11:26:09 +02:00
|
|
|
__clocksource_change_rating(cs, 0);
|
2009-08-15 13:20:42 +02:00
|
|
|
}
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
return select;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int clocksource_watchdog_kthread(void *data)
|
|
|
|
{
|
|
|
|
mutex_lock(&clocksource_mutex);
|
|
|
|
if (__clocksource_watchdog_kthread())
|
|
|
|
clocksource_select();
|
2009-08-19 11:26:09 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2009-08-18 17:09:42 +02:00
|
|
|
return 0;
|
2009-08-14 15:47:25 +02:00
|
|
|
}
|
|
|
|
|
2013-04-25 22:31:46 +02:00
|
|
|
static bool clocksource_is_watchdog(struct clocksource *cs)
|
|
|
|
{
|
|
|
|
return cs == watchdog;
|
|
|
|
}
|
|
|
|
|
2009-08-14 15:47:24 +02:00
|
|
|
#else /* CONFIG_CLOCKSOURCE_WATCHDOG */
|
|
|
|
|
|
|
|
static void clocksource_enqueue_watchdog(struct clocksource *cs)
|
2007-02-16 10:27:43 +01:00
|
|
|
{
|
|
|
|
if (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS)
|
|
|
|
cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES;
|
|
|
|
}
|
2007-05-09 11:35:15 +02:00
|
|
|
|
2016-01-22 18:31:53 +01:00
|
|
|
static void clocksource_select_watchdog(bool fallback) { }
|
2009-08-14 15:47:24 +02:00
|
|
|
static inline void clocksource_dequeue_watchdog(struct clocksource *cs) { }
|
2007-05-09 11:35:15 +02:00
|
|
|
static inline void clocksource_resume_watchdog(void) { }
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
static inline int __clocksource_watchdog_kthread(void) { return 0; }
|
2013-04-25 22:31:46 +02:00
|
|
|
static bool clocksource_is_watchdog(struct clocksource *cs) { return false; }
|
2013-02-22 21:08:56 +01:00
|
|
|
void clocksource_mark_unstable(struct clocksource *cs) { }
|
2009-08-14 15:47:24 +02:00
|
|
|
|
|
|
|
#endif /* CONFIG_CLOCKSOURCE_WATCHDOG */
|
2007-02-16 10:27:43 +01:00
|
|
|
|
2010-02-02 23:41:41 +01:00
|
|
|
/**
|
|
|
|
* clocksource_suspend - suspend the clocksource(s)
|
|
|
|
*/
|
|
|
|
void clocksource_suspend(void)
|
|
|
|
{
|
|
|
|
struct clocksource *cs;
|
|
|
|
|
|
|
|
list_for_each_entry_reverse(cs, &clocksource_list, list)
|
|
|
|
if (cs->suspend)
|
|
|
|
cs->suspend(cs);
|
|
|
|
}
|
|
|
|
|
2007-05-09 11:35:15 +02:00
|
|
|
/**
|
|
|
|
* clocksource_resume - resume the clocksource(s)
|
|
|
|
*/
|
|
|
|
void clocksource_resume(void)
|
|
|
|
{
|
2007-10-19 08:39:58 +02:00
|
|
|
struct clocksource *cs;
|
2007-05-09 11:35:15 +02:00
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
list_for_each_entry(cs, &clocksource_list, list)
|
2007-05-09 11:35:15 +02:00
|
|
|
if (cs->resume)
|
2010-02-02 23:41:39 +01:00
|
|
|
cs->resume(cs);
|
2007-05-09 11:35:15 +02:00
|
|
|
|
|
|
|
clocksource_resume_watchdog();
|
|
|
|
}
|
|
|
|
|
2008-02-15 21:55:54 +01:00
|
|
|
/**
|
|
|
|
* clocksource_touch_watchdog - Update watchdog
|
|
|
|
*
|
|
|
|
* Update the watchdog after exception contexts such as kgdb so as not
|
2010-01-26 12:51:10 +01:00
|
|
|
* to incorrectly trip the watchdog. This might fail when the kernel
|
|
|
|
* was stopped in code which holds watchdog_lock.
|
2008-02-15 21:55:54 +01:00
|
|
|
*/
|
|
|
|
void clocksource_touch_watchdog(void)
|
|
|
|
{
|
|
|
|
clocksource_resume_watchdog();
|
|
|
|
}
|
|
|
|
|
2011-10-31 22:06:35 +01:00
|
|
|
/**
|
|
|
|
* clocksource_max_adjustment- Returns max adjustment amount
|
|
|
|
* @cs: Pointer to clocksource
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
static u32 clocksource_max_adjustment(struct clocksource *cs)
|
|
|
|
{
|
|
|
|
u64 ret;
|
|
|
|
/*
|
2012-03-15 04:28:56 +01:00
|
|
|
* We won't try to correct for more than 11% adjustments (110,000 ppm),
|
2011-10-31 22:06:35 +01:00
|
|
|
*/
|
|
|
|
ret = (u64)cs->mult * 11;
|
|
|
|
do_div(ret,100);
|
|
|
|
return (u32)ret;
|
|
|
|
}
|
|
|
|
|
2009-08-18 19:45:10 +02:00
|
|
|
/**
|
2013-07-19 01:21:14 +02:00
|
|
|
* clocks_calc_max_nsecs - Returns maximum nanoseconds that can be converted
|
|
|
|
* @mult: cycle to nanosecond multiplier
|
|
|
|
* @shift: cycle to nanosecond divisor (power of two)
|
|
|
|
* @maxadj: maximum adjustment value to mult (~11%)
|
|
|
|
* @mask: bitmask for two's complement subtraction of non 64 bit counters
|
2015-03-12 05:16:31 +01:00
|
|
|
* @max_cyc: maximum cycle value before potential overflow (does not include
|
|
|
|
* any safety margin)
|
2015-03-12 05:16:30 +01:00
|
|
|
*
|
2015-04-02 05:34:39 +02:00
|
|
|
* NOTE: This function includes a safety margin of 50%, in other words, we
|
|
|
|
* return half the number of nanoseconds the hardware counter can technically
|
|
|
|
* cover. This is done so that we can potentially detect problems caused by
|
|
|
|
* delayed timers or bad hardware, which might result in time intervals that
|
2015-08-25 08:42:53 +02:00
|
|
|
* are larger than what the math used can handle without overflows.
|
2009-08-18 19:45:10 +02:00
|
|
|
*/
|
2015-03-12 05:16:31 +01:00
|
|
|
u64 clocks_calc_max_nsecs(u32 mult, u32 shift, u32 maxadj, u64 mask, u64 *max_cyc)
|
2009-08-18 19:45:10 +02:00
|
|
|
{
|
|
|
|
u64 max_nsecs, max_cycles;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculate the maximum number of cycles that we can pass to the
|
2015-03-12 05:16:29 +01:00
|
|
|
* cyc2ns() function without overflowing a 64-bit result.
|
2009-08-18 19:45:10 +02:00
|
|
|
*/
|
2015-03-12 05:16:29 +01:00
|
|
|
max_cycles = ULLONG_MAX;
|
|
|
|
do_div(max_cycles, mult+maxadj);
|
2009-08-18 19:45:10 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The actual maximum number of cycles we can defer the clocksource is
|
2013-07-19 01:21:14 +02:00
|
|
|
* determined by the minimum of max_cycles and mask.
|
2011-10-31 22:06:35 +01:00
|
|
|
* Note: Here we subtract the maxadj to make sure we don't sleep for
|
|
|
|
* too long if there's a large negative adjustment.
|
2009-08-18 19:45:10 +02:00
|
|
|
*/
|
2013-07-19 01:21:14 +02:00
|
|
|
max_cycles = min(max_cycles, mask);
|
|
|
|
max_nsecs = clocksource_cyc2ns(max_cycles, mult - maxadj, shift);
|
|
|
|
|
2015-03-12 05:16:31 +01:00
|
|
|
/* return the max_cycles value as well if requested */
|
|
|
|
if (max_cyc)
|
|
|
|
*max_cyc = max_cycles;
|
|
|
|
|
2015-03-12 05:16:30 +01:00
|
|
|
/* Return 50% of the actual maximum, so we can detect bad values */
|
|
|
|
max_nsecs >>= 1;
|
|
|
|
|
2013-07-19 01:21:14 +02:00
|
|
|
return max_nsecs;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2015-03-12 05:16:31 +01:00
|
|
|
* clocksource_update_max_deferment - Updates the clocksource max_idle_ns & max_cycles
|
|
|
|
* @cs: Pointer to clocksource to be updated
|
2013-07-19 01:21:14 +02:00
|
|
|
*
|
|
|
|
*/
|
2015-03-12 05:16:31 +01:00
|
|
|
static inline void clocksource_update_max_deferment(struct clocksource *cs)
|
2013-07-19 01:21:14 +02:00
|
|
|
{
|
2015-03-12 05:16:31 +01:00
|
|
|
cs->max_idle_ns = clocks_calc_max_nsecs(cs->mult, cs->shift,
|
|
|
|
cs->maxadj, cs->mask,
|
|
|
|
&cs->max_cycles);
|
2009-08-18 19:45:10 +02:00
|
|
|
}
|
|
|
|
|
2010-07-14 02:56:20 +02:00
|
|
|
#ifndef CONFIG_ARCH_USES_GETTIMEOFFSET
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2013-04-25 22:31:45 +02:00
|
|
|
static struct clocksource *clocksource_find_best(bool oneshot, bool skipcur)
|
2013-04-25 22:31:43 +02:00
|
|
|
{
|
|
|
|
struct clocksource *cs;
|
|
|
|
|
|
|
|
if (!finished_booting || list_empty(&clocksource_list))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We pick the clocksource with the highest rating. If oneshot
|
|
|
|
* mode is active, we pick the highres valid clocksource with
|
|
|
|
* the best rating.
|
|
|
|
*/
|
|
|
|
list_for_each_entry(cs, &clocksource_list, list) {
|
2013-04-25 22:31:45 +02:00
|
|
|
if (skipcur && cs == curr_clocksource)
|
|
|
|
continue;
|
2013-04-25 22:31:43 +02:00
|
|
|
if (oneshot && !(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES))
|
|
|
|
continue;
|
|
|
|
return cs;
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2013-04-25 22:31:45 +02:00
|
|
|
static void __clocksource_select(bool skipcur)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2013-04-25 22:31:43 +02:00
|
|
|
bool oneshot = tick_oneshot_mode_active();
|
2009-08-14 15:47:21 +02:00
|
|
|
struct clocksource *best, *cs;
|
2007-02-16 10:27:43 +01:00
|
|
|
|
2013-04-25 22:31:43 +02:00
|
|
|
/* Find the best suitable clocksource */
|
2013-04-25 22:31:45 +02:00
|
|
|
best = clocksource_find_best(oneshot, skipcur);
|
2013-04-25 22:31:43 +02:00
|
|
|
if (!best)
|
2009-08-14 15:47:21 +02:00
|
|
|
return;
|
2013-04-25 22:31:43 +02:00
|
|
|
|
2009-08-14 15:47:21 +02:00
|
|
|
/* Check for the override clocksource. */
|
|
|
|
list_for_each_entry(cs, &clocksource_list, list) {
|
2013-04-25 22:31:45 +02:00
|
|
|
if (skipcur && cs == curr_clocksource)
|
|
|
|
continue;
|
2009-08-14 15:47:21 +02:00
|
|
|
if (strcmp(cs->name, override_name) != 0)
|
|
|
|
continue;
|
|
|
|
/*
|
|
|
|
* Check to make sure we don't switch to a non-highres
|
|
|
|
* capable clocksource if the tick code is in oneshot
|
|
|
|
* mode (highres or nohz)
|
|
|
|
*/
|
2013-04-25 22:31:43 +02:00
|
|
|
if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && oneshot) {
|
2009-08-14 15:47:21 +02:00
|
|
|
/* Override clocksource cannot be used. */
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_warn("Override clocksource %s is not HRT compatible - cannot switch while in HRT/NOHZ mode\n",
|
|
|
|
cs->name);
|
2009-08-14 15:47:21 +02:00
|
|
|
override_name[0] = 0;
|
|
|
|
} else
|
|
|
|
/* Override clocksource can be used. */
|
|
|
|
best = cs;
|
|
|
|
break;
|
|
|
|
}
|
2013-04-25 22:31:44 +02:00
|
|
|
|
|
|
|
if (curr_clocksource != best && !timekeeping_notify(best)) {
|
|
|
|
pr_info("Switched to clocksource %s\n", best->name);
|
2009-08-14 15:47:30 +02:00
|
|
|
curr_clocksource = best;
|
|
|
|
}
|
2009-08-14 15:47:21 +02:00
|
|
|
}
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2013-04-25 22:31:45 +02:00
|
|
|
/**
|
|
|
|
* clocksource_select - Select the best clocksource available
|
|
|
|
*
|
|
|
|
* Private function. Must hold clocksource_mutex when called.
|
|
|
|
*
|
|
|
|
* Select the clocksource with the best rating, or the clocksource,
|
|
|
|
* which is selected by userspace override.
|
|
|
|
*/
|
|
|
|
static void clocksource_select(void)
|
|
|
|
{
|
2015-09-23 13:19:19 +02:00
|
|
|
__clocksource_select(false);
|
2013-04-25 22:31:45 +02:00
|
|
|
}
|
|
|
|
|
2013-04-25 22:31:46 +02:00
|
|
|
static void clocksource_select_fallback(void)
|
|
|
|
{
|
2015-09-23 13:19:19 +02:00
|
|
|
__clocksource_select(true);
|
2013-04-25 22:31:46 +02:00
|
|
|
}
|
|
|
|
|
2010-07-14 02:56:20 +02:00
|
|
|
#else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */
|
2009-09-14 19:49:02 +02:00
|
|
|
static inline void clocksource_select(void) { }
|
2013-05-28 09:48:46 +02:00
|
|
|
static inline void clocksource_select_fallback(void) { }
|
2009-09-14 19:49:02 +02:00
|
|
|
|
|
|
|
#endif
|
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
/*
|
|
|
|
* clocksource_done_booting - Called near the end of core bootup
|
|
|
|
*
|
|
|
|
* Hack to avoid lots of clocksource churn at boot time.
|
|
|
|
* We use fs_initcall because we want this to start before
|
|
|
|
* device_initcall but after subsys_initcall.
|
|
|
|
*/
|
|
|
|
static int __init clocksource_done_booting(void)
|
|
|
|
{
|
2010-03-01 21:34:43 +01:00
|
|
|
mutex_lock(&clocksource_mutex);
|
|
|
|
curr_clocksource = clocksource_default_clock();
|
2009-08-14 15:47:30 +02:00
|
|
|
finished_booting = 1;
|
2009-09-14 19:49:02 +02:00
|
|
|
/*
|
|
|
|
* Run the watchdog first to eliminate unstable clock sources
|
|
|
|
*/
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
__clocksource_watchdog_kthread();
|
2009-08-14 15:47:30 +02:00
|
|
|
clocksource_select();
|
2009-09-14 19:51:11 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2009-08-14 15:47:30 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
fs_initcall(clocksource_done_booting);
|
|
|
|
|
2007-02-16 10:27:33 +01:00
|
|
|
/*
|
|
|
|
* Enqueue the clocksource sorted by rating
|
2006-06-26 09:25:05 +02:00
|
|
|
*/
|
2009-08-14 15:47:21 +02:00
|
|
|
static void clocksource_enqueue(struct clocksource *cs)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2009-08-14 15:47:21 +02:00
|
|
|
struct list_head *entry = &clocksource_list;
|
|
|
|
struct clocksource *tmp;
|
2007-02-16 10:27:33 +01:00
|
|
|
|
2009-08-14 15:47:21 +02:00
|
|
|
list_for_each_entry(tmp, &clocksource_list, list)
|
2007-02-16 10:27:33 +01:00
|
|
|
/* Keep track of the place, where to insert */
|
2009-08-14 15:47:21 +02:00
|
|
|
if (tmp->rating >= cs->rating)
|
|
|
|
entry = &tmp->list;
|
|
|
|
list_add(&cs->list, entry);
|
2006-06-26 09:25:05 +02:00
|
|
|
}
|
|
|
|
|
2010-05-08 03:07:38 +02:00
|
|
|
/**
|
2015-03-12 05:16:40 +01:00
|
|
|
* __clocksource_update_freq_scale - Used update clocksource with new freq
|
2011-12-19 10:13:19 +01:00
|
|
|
* @cs: clocksource to be registered
|
2010-05-08 03:07:38 +02:00
|
|
|
* @scale: Scale factor multiplied against freq to get clocksource hz
|
|
|
|
* @freq: clocksource frequency (cycles per second) divided by scale
|
|
|
|
*
|
2010-07-14 02:56:28 +02:00
|
|
|
* This should only be called from the clocksource->enable() method.
|
2010-05-08 03:07:38 +02:00
|
|
|
*
|
|
|
|
* This *SHOULD NOT* be called directly! Please use the
|
2015-03-12 05:16:40 +01:00
|
|
|
* __clocksource_update_freq_hz() or __clocksource_update_freq_khz() helper
|
|
|
|
* functions.
|
2010-05-08 03:07:38 +02:00
|
|
|
*/
|
2015-03-12 05:16:40 +01:00
|
|
|
void __clocksource_update_freq_scale(struct clocksource *cs, u32 scale, u32 freq)
|
2010-05-08 03:07:38 +02:00
|
|
|
{
|
2011-05-20 10:50:52 +02:00
|
|
|
u64 sec;
|
2015-03-12 05:16:37 +01:00
|
|
|
|
2010-05-08 03:07:38 +02:00
|
|
|
/*
|
2015-03-12 05:16:37 +01:00
|
|
|
* Default clocksources are *special* and self-define their mult/shift.
|
|
|
|
* But, you're not special, so you should specify a freq value.
|
2010-05-08 03:07:38 +02:00
|
|
|
*/
|
2015-03-12 05:16:37 +01:00
|
|
|
if (freq) {
|
|
|
|
/*
|
|
|
|
* Calc the maximum number of seconds which we can run before
|
|
|
|
* wrapping around. For clocksources which have a mask > 32-bit
|
|
|
|
* we need to limit the max sleep time to have a good
|
|
|
|
* conversion precision. 10 minutes is still a reasonable
|
|
|
|
* amount. That results in a shift value of 24 for a
|
|
|
|
* clocksource with mask >= 40-bit and f >= 4GHz. That maps to
|
|
|
|
* ~ 0.06ppm granularity for NTP.
|
|
|
|
*/
|
|
|
|
sec = cs->mask;
|
|
|
|
do_div(sec, freq);
|
|
|
|
do_div(sec, scale);
|
|
|
|
if (!sec)
|
|
|
|
sec = 1;
|
|
|
|
else if (sec > 600 && cs->mask > UINT_MAX)
|
|
|
|
sec = 600;
|
|
|
|
|
|
|
|
clocks_calc_mult_shift(&cs->mult, &cs->shift, freq,
|
|
|
|
NSEC_PER_SEC / scale, sec * scale);
|
|
|
|
}
|
2011-10-31 22:06:35 +01:00
|
|
|
/*
|
2015-03-12 05:16:30 +01:00
|
|
|
* Ensure clocksources that have large 'mult' values don't overflow
|
|
|
|
* when adjusted.
|
2011-10-31 22:06:35 +01:00
|
|
|
*/
|
|
|
|
cs->maxadj = clocksource_max_adjustment(cs);
|
2015-03-12 05:16:37 +01:00
|
|
|
while (freq && ((cs->mult + cs->maxadj < cs->mult)
|
|
|
|
|| (cs->mult - cs->maxadj > cs->mult))) {
|
2011-10-31 22:06:35 +01:00
|
|
|
cs->mult >>= 1;
|
|
|
|
cs->shift--;
|
|
|
|
cs->maxadj = clocksource_max_adjustment(cs);
|
|
|
|
}
|
|
|
|
|
2015-03-12 05:16:37 +01:00
|
|
|
/*
|
|
|
|
* Only warn for *special* clocksources that self-define
|
|
|
|
* their mult/shift values and don't specify a freq.
|
|
|
|
*/
|
|
|
|
WARN_ONCE(cs->mult + cs->maxadj < cs->mult,
|
|
|
|
"timekeeping: Clocksource %s might overflow on 11%% adjustment\n",
|
|
|
|
cs->name);
|
|
|
|
|
2015-03-12 05:16:31 +01:00
|
|
|
clocksource_update_max_deferment(cs);
|
2015-03-12 05:16:39 +01:00
|
|
|
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_info("%s: mask: 0x%llx max_cycles: 0x%llx, max_idle_ns: %lld ns\n",
|
|
|
|
cs->name, cs->mask, cs->max_cycles, cs->max_idle_ns);
|
2010-07-14 02:56:28 +02:00
|
|
|
}
|
2015-03-12 05:16:40 +01:00
|
|
|
EXPORT_SYMBOL_GPL(__clocksource_update_freq_scale);
|
2010-07-14 02:56:28 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* __clocksource_register_scale - Used to install new clocksources
|
2011-12-19 10:13:19 +01:00
|
|
|
* @cs: clocksource to be registered
|
2010-07-14 02:56:28 +02:00
|
|
|
* @scale: Scale factor multiplied against freq to get clocksource hz
|
|
|
|
* @freq: clocksource frequency (cycles per second) divided by scale
|
|
|
|
*
|
|
|
|
* Returns -EBUSY if registration fails, zero otherwise.
|
|
|
|
*
|
|
|
|
* This *SHOULD NOT* be called directly! Please use the
|
|
|
|
* clocksource_register_hz() or clocksource_register_khz helper functions.
|
|
|
|
*/
|
|
|
|
int __clocksource_register_scale(struct clocksource *cs, u32 scale, u32 freq)
|
|
|
|
{
|
|
|
|
|
tree-wide: fix comment/printk typos
"gadget", "through", "command", "maintain", "maintain", "controller", "address",
"between", "initiali[zs]e", "instead", "function", "select", "already",
"equal", "access", "management", "hierarchy", "registration", "interest",
"relative", "memory", "offset", "already",
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2010-11-01 20:38:34 +01:00
|
|
|
/* Initialize mult/shift and max_idle_ns */
|
2015-03-12 05:16:40 +01:00
|
|
|
__clocksource_update_freq_scale(cs, scale, freq);
|
2010-05-08 03:07:38 +02:00
|
|
|
|
2014-09-18 16:59:07 +02:00
|
|
|
/* Add clocksource to the clocksource list */
|
2010-05-08 03:07:38 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
|
|
|
clocksource_enqueue(cs);
|
|
|
|
clocksource_enqueue_watchdog(cs);
|
2011-05-05 03:16:50 +02:00
|
|
|
clocksource_select();
|
2016-01-22 18:31:53 +01:00
|
|
|
clocksource_select_watchdog(false);
|
2010-05-08 03:07:38 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(__clocksource_register_scale);
|
|
|
|
|
2009-08-19 11:26:09 +02:00
|
|
|
static void __clocksource_change_rating(struct clocksource *cs, int rating)
|
|
|
|
{
|
|
|
|
list_del(&cs->list);
|
|
|
|
cs->rating = rating;
|
|
|
|
clocksource_enqueue(cs);
|
|
|
|
}
|
|
|
|
|
2006-06-26 09:25:05 +02:00
|
|
|
/**
|
2007-02-16 10:27:33 +01:00
|
|
|
* clocksource_change_rating - Change the rating of a registered clocksource
|
2011-12-19 10:13:19 +01:00
|
|
|
* @cs: clocksource to be changed
|
|
|
|
* @rating: new rating
|
2006-06-26 09:25:05 +02:00
|
|
|
*/
|
2007-02-16 10:27:33 +01:00
|
|
|
void clocksource_change_rating(struct clocksource *cs, int rating)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2009-08-19 11:26:09 +02:00
|
|
|
__clocksource_change_rating(cs, rating);
|
clocksource: Reselect clocksource when watchdog validated high-res capability
Up to commit 5d33b883a (clocksource: Always verify highres capability)
we had no sanity check when selecting a clocksource, which prevented
that a non highres capable clocksource is used when the system already
switched to highres/nohz mode.
The new sanity check works as Alex and Tim found out. It prevents the
TSC from being used. This happens because on x86 the boot process
looks like this:
tsc_start_freqency_validation(TSC);
clocksource_register(HPET);
clocksource_done_booting();
clocksource_select()
Selects HPET which is valid for high-res
switch_to_highres();
clocksource_register(TSC);
TSC is not selected, because it is not yet
flagged as VALID_HIGH_RES
clocksource_watchdog()
Validates TSC for highres, but that does not make TSC
the current clocksource.
Before the sanity check was added, we installed TSC unvalidated which
worked most of the time. If the TSC was really detected as unstable,
then the unstable logic removed it and installed HPET again.
The sanity check is correct and needed. So the watchdog needs to kick
a reselection of the clocksource, when it qualifies TSC as a valid
high res clocksource.
To solve this, we mark the clocksource which got the flag
CLOCK_SOURCE_VALID_FOR_HRES set by the watchdog with an new flag
CLOCK_SOURCE_RESELECT and trigger the watchdog thread. The watchdog
thread evaluates the flag and invokes clocksource_select() when set.
To avoid that the clocksource_done_booting() code, which is about to
install the first real clocksource anyway, needs to go through
clocksource_select and tick_oneshot_notify() pointlessly, split out
the clocksource_watchdog_kthread() list walk code and invoke the
select/notify only when called from clocksource_watchdog_kthread().
So clocksource_done_booting() can utilize the same splitout code
without the select/notify invocation and the clocksource_mutex
unlock/relock dance.
Reported-and-tested-by: Alex Shi <alex.shi@intel.com>
Cc: Hans Peter Anvin <hpa@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1307042239150.11637@ionos.tec.linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2013-07-04 22:46:45 +02:00
|
|
|
clocksource_select();
|
2016-01-22 18:31:53 +01:00
|
|
|
clocksource_select_watchdog(false);
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
}
|
2009-08-14 15:47:24 +02:00
|
|
|
EXPORT_SYMBOL(clocksource_change_rating);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2013-04-25 22:31:46 +02:00
|
|
|
/*
|
|
|
|
* Unbind clocksource @cs. Called with clocksource_mutex held
|
|
|
|
*/
|
|
|
|
static int clocksource_unbind(struct clocksource *cs)
|
|
|
|
{
|
2016-01-22 18:31:53 +01:00
|
|
|
if (clocksource_is_watchdog(cs)) {
|
|
|
|
/* Select and try to install a replacement watchdog. */
|
|
|
|
clocksource_select_watchdog(true);
|
|
|
|
if (clocksource_is_watchdog(cs))
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
2013-04-25 22:31:46 +02:00
|
|
|
|
|
|
|
if (cs == curr_clocksource) {
|
|
|
|
/* Select and try to install a replacement clock source */
|
|
|
|
clocksource_select_fallback();
|
|
|
|
if (curr_clocksource == cs)
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
clocksource_dequeue_watchdog(cs);
|
|
|
|
list_del_init(&cs->list);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-01-30 13:30:02 +01:00
|
|
|
/**
|
|
|
|
* clocksource_unregister - remove a registered clocksource
|
2011-12-19 10:13:19 +01:00
|
|
|
* @cs: clocksource to be unregistered
|
2008-01-30 13:30:02 +01:00
|
|
|
*/
|
2013-04-25 22:31:46 +02:00
|
|
|
int clocksource_unregister(struct clocksource *cs)
|
2008-01-30 13:30:02 +01:00
|
|
|
{
|
2013-04-25 22:31:46 +02:00
|
|
|
int ret = 0;
|
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2013-04-25 22:31:46 +02:00
|
|
|
if (!list_empty(&cs->list))
|
|
|
|
ret = clocksource_unbind(cs);
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2013-04-25 22:31:46 +02:00
|
|
|
return ret;
|
2008-01-30 13:30:02 +01:00
|
|
|
}
|
2009-08-14 15:47:24 +02:00
|
|
|
EXPORT_SYMBOL(clocksource_unregister);
|
2008-01-30 13:30:02 +01:00
|
|
|
|
2006-12-10 11:21:30 +01:00
|
|
|
#ifdef CONFIG_SYSFS
|
2006-06-26 09:25:05 +02:00
|
|
|
/**
|
|
|
|
* sysfs_show_current_clocksources - sysfs interface for current clocksource
|
|
|
|
* @dev: unused
|
2011-12-19 10:13:19 +01:00
|
|
|
* @attr: unused
|
2006-06-26 09:25:05 +02:00
|
|
|
* @buf: char buffer to be filled with clocksource list
|
|
|
|
*
|
|
|
|
* Provides sysfs interface for listing current clocksource.
|
|
|
|
*/
|
|
|
|
static ssize_t
|
2011-12-15 00:28:51 +01:00
|
|
|
sysfs_show_current_clocksources(struct device *dev,
|
|
|
|
struct device_attribute *attr, char *buf)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2008-02-06 10:36:53 +01:00
|
|
|
ssize_t count = 0;
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2008-02-06 10:36:53 +01:00
|
|
|
count = snprintf(buf, PAGE_SIZE, "%s\n", curr_clocksource->name);
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2008-02-06 10:36:53 +01:00
|
|
|
return count;
|
2006-06-26 09:25:05 +02:00
|
|
|
}
|
|
|
|
|
2013-10-11 19:11:55 +02:00
|
|
|
ssize_t sysfs_get_uname(const char *buf, char *dst, size_t cnt)
|
2013-04-25 22:31:45 +02:00
|
|
|
{
|
|
|
|
size_t ret = cnt;
|
|
|
|
|
|
|
|
/* strings from sysfs write are not 0 terminated! */
|
|
|
|
if (!cnt || cnt >= CS_NAME_LEN)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* strip of \n: */
|
|
|
|
if (buf[cnt-1] == '\n')
|
|
|
|
cnt--;
|
|
|
|
if (cnt > 0)
|
|
|
|
memcpy(dst, buf, cnt);
|
|
|
|
dst[cnt] = 0;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2006-06-26 09:25:05 +02:00
|
|
|
/**
|
|
|
|
* sysfs_override_clocksource - interface for manually overriding clocksource
|
|
|
|
* @dev: unused
|
2011-12-19 10:13:19 +01:00
|
|
|
* @attr: unused
|
2006-06-26 09:25:05 +02:00
|
|
|
* @buf: name of override clocksource
|
|
|
|
* @count: length of buffer
|
|
|
|
*
|
|
|
|
* Takes input from sysfs interface for manually overriding the default
|
2009-10-06 12:42:51 +02:00
|
|
|
* clocksource selection.
|
2006-06-26 09:25:05 +02:00
|
|
|
*/
|
2011-12-15 00:28:51 +01:00
|
|
|
static ssize_t sysfs_override_clocksource(struct device *dev,
|
|
|
|
struct device_attribute *attr,
|
2006-06-26 09:25:05 +02:00
|
|
|
const char *buf, size_t count)
|
|
|
|
{
|
2013-09-12 12:28:54 +02:00
|
|
|
ssize_t ret;
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2013-04-25 22:31:50 +02:00
|
|
|
ret = sysfs_get_uname(buf, override_name, count);
|
2013-04-25 22:31:45 +02:00
|
|
|
if (ret >= 0)
|
|
|
|
clocksource_select();
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-04-25 22:31:46 +02:00
|
|
|
/**
|
|
|
|
* sysfs_unbind_current_clocksource - interface for manually unbinding clocksource
|
|
|
|
* @dev: unused
|
|
|
|
* @attr: unused
|
|
|
|
* @buf: unused
|
|
|
|
* @count: length of buffer
|
|
|
|
*
|
|
|
|
* Takes input from sysfs interface for manually unbinding a clocksource.
|
|
|
|
*/
|
|
|
|
static ssize_t sysfs_unbind_clocksource(struct device *dev,
|
|
|
|
struct device_attribute *attr,
|
|
|
|
const char *buf, size_t count)
|
|
|
|
{
|
|
|
|
struct clocksource *cs;
|
|
|
|
char name[CS_NAME_LEN];
|
2013-09-12 12:28:54 +02:00
|
|
|
ssize_t ret;
|
2013-04-25 22:31:46 +02:00
|
|
|
|
2013-04-25 22:31:50 +02:00
|
|
|
ret = sysfs_get_uname(buf, name, count);
|
2013-04-25 22:31:46 +02:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ret = -ENODEV;
|
|
|
|
mutex_lock(&clocksource_mutex);
|
|
|
|
list_for_each_entry(cs, &clocksource_list, list) {
|
|
|
|
if (strcmp(cs->name, name))
|
|
|
|
continue;
|
|
|
|
ret = clocksource_unbind(cs);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
mutex_unlock(&clocksource_mutex);
|
|
|
|
|
|
|
|
return ret ? ret : count;
|
|
|
|
}
|
|
|
|
|
2006-06-26 09:25:05 +02:00
|
|
|
/**
|
|
|
|
* sysfs_show_available_clocksources - sysfs interface for listing clocksource
|
|
|
|
* @dev: unused
|
2011-12-19 10:13:19 +01:00
|
|
|
* @attr: unused
|
2006-06-26 09:25:05 +02:00
|
|
|
* @buf: char buffer to be filled with clocksource list
|
|
|
|
*
|
|
|
|
* Provides sysfs interface for listing registered clocksources
|
|
|
|
*/
|
|
|
|
static ssize_t
|
2011-12-15 00:28:51 +01:00
|
|
|
sysfs_show_available_clocksources(struct device *dev,
|
|
|
|
struct device_attribute *attr,
|
2008-07-01 18:48:41 +02:00
|
|
|
char *buf)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2007-10-19 08:39:58 +02:00
|
|
|
struct clocksource *src;
|
2008-02-06 10:36:53 +01:00
|
|
|
ssize_t count = 0;
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2007-10-19 08:39:58 +02:00
|
|
|
list_for_each_entry(src, &clocksource_list, list) {
|
2009-06-12 11:29:27 +02:00
|
|
|
/*
|
|
|
|
* Don't show non-HRES clocksource if the tick code is
|
|
|
|
* in one shot mode (highres=on or nohz=on)
|
|
|
|
*/
|
|
|
|
if (!tick_oneshot_mode_active() ||
|
|
|
|
(src->flags & CLOCK_SOURCE_VALID_FOR_HRES))
|
2009-01-22 06:53:22 +01:00
|
|
|
count += snprintf(buf + count,
|
2008-02-06 10:36:53 +01:00
|
|
|
max((ssize_t)PAGE_SIZE - count, (ssize_t)0),
|
|
|
|
"%s ", src->name);
|
2006-06-26 09:25:05 +02:00
|
|
|
}
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2008-02-06 10:36:53 +01:00
|
|
|
count += snprintf(buf + count,
|
|
|
|
max((ssize_t)PAGE_SIZE - count, (ssize_t)0), "\n");
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2008-02-06 10:36:53 +01:00
|
|
|
return count;
|
2006-06-26 09:25:05 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Sysfs setup bits:
|
|
|
|
*/
|
2011-12-15 00:28:51 +01:00
|
|
|
static DEVICE_ATTR(current_clocksource, 0644, sysfs_show_current_clocksources,
|
2006-12-10 11:21:33 +01:00
|
|
|
sysfs_override_clocksource);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2013-04-25 22:31:46 +02:00
|
|
|
static DEVICE_ATTR(unbind_clocksource, 0200, NULL, sysfs_unbind_clocksource);
|
|
|
|
|
2011-12-15 00:28:51 +01:00
|
|
|
static DEVICE_ATTR(available_clocksource, 0444,
|
2006-12-10 11:21:33 +01:00
|
|
|
sysfs_show_available_clocksources, NULL);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
2011-12-15 00:28:51 +01:00
|
|
|
static struct bus_type clocksource_subsys = {
|
2007-12-20 02:09:39 +01:00
|
|
|
.name = "clocksource",
|
2011-12-15 00:28:51 +01:00
|
|
|
.dev_name = "clocksource",
|
2006-06-26 09:25:05 +02:00
|
|
|
};
|
|
|
|
|
2011-12-15 00:28:51 +01:00
|
|
|
static struct device device_clocksource = {
|
2006-06-26 09:25:05 +02:00
|
|
|
.id = 0,
|
2011-12-15 00:28:51 +01:00
|
|
|
.bus = &clocksource_subsys,
|
2006-06-26 09:25:05 +02:00
|
|
|
};
|
|
|
|
|
2006-06-26 09:25:06 +02:00
|
|
|
static int __init init_clocksource_sysfs(void)
|
2006-06-26 09:25:05 +02:00
|
|
|
{
|
2011-12-15 00:28:51 +01:00
|
|
|
int error = subsys_system_register(&clocksource_subsys, NULL);
|
2006-06-26 09:25:05 +02:00
|
|
|
|
|
|
|
if (!error)
|
2011-12-15 00:28:51 +01:00
|
|
|
error = device_register(&device_clocksource);
|
2006-06-26 09:25:05 +02:00
|
|
|
if (!error)
|
2011-12-15 00:28:51 +01:00
|
|
|
error = device_create_file(
|
2006-06-26 09:25:05 +02:00
|
|
|
&device_clocksource,
|
2011-12-15 00:28:51 +01:00
|
|
|
&dev_attr_current_clocksource);
|
2013-04-25 22:31:46 +02:00
|
|
|
if (!error)
|
|
|
|
error = device_create_file(&device_clocksource,
|
|
|
|
&dev_attr_unbind_clocksource);
|
2006-06-26 09:25:05 +02:00
|
|
|
if (!error)
|
2011-12-15 00:28:51 +01:00
|
|
|
error = device_create_file(
|
2006-06-26 09:25:05 +02:00
|
|
|
&device_clocksource,
|
2011-12-15 00:28:51 +01:00
|
|
|
&dev_attr_available_clocksource);
|
2006-06-26 09:25:05 +02:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
device_initcall(init_clocksource_sysfs);
|
2006-12-10 11:21:30 +01:00
|
|
|
#endif /* CONFIG_SYSFS */
|
2006-06-26 09:25:05 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* boot_override_clocksource - boot clock override
|
|
|
|
* @str: override name
|
|
|
|
*
|
|
|
|
* Takes a clocksource= boot argument and uses it
|
|
|
|
* as the clocksource override name.
|
|
|
|
*/
|
|
|
|
static int __init boot_override_clocksource(char* str)
|
|
|
|
{
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_lock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
if (str)
|
|
|
|
strlcpy(override_name, str, sizeof(override_name));
|
2009-08-14 15:47:30 +02:00
|
|
|
mutex_unlock(&clocksource_mutex);
|
2006-06-26 09:25:05 +02:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
__setup("clocksource=", boot_override_clocksource);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* boot_override_clock - Compatibility layer for deprecated boot option
|
|
|
|
* @str: override name
|
|
|
|
*
|
|
|
|
* DEPRECATED! Takes a clock= boot argument and uses it
|
|
|
|
* as the clocksource override name
|
|
|
|
*/
|
|
|
|
static int __init boot_override_clock(char* str)
|
|
|
|
{
|
2006-06-26 09:25:12 +02:00
|
|
|
if (!strcmp(str, "pmtmr")) {
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_warn("clock=pmtmr is deprecated - use clocksource=acpi_pm\n");
|
2006-06-26 09:25:12 +02:00
|
|
|
return boot_override_clocksource("acpi_pm");
|
|
|
|
}
|
2015-05-25 20:49:55 +02:00
|
|
|
pr_warn("clock= boot option is deprecated - use clocksource=xyz\n");
|
2006-06-26 09:25:05 +02:00
|
|
|
return boot_override_clocksource(str);
|
|
|
|
}
|
|
|
|
|
|
|
|
__setup("clock=", boot_override_clock);
|