lkml.org 
[lkml]   [2007]   [Jan]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH] high_res_timers: precisely update_process_times; Re: [patch 36/46] tick-management: dyntick / highres functionality
Date
Am Dienstag, 23. Januar 2007 23:01 schrieb Thomas Gleixner:
>
> Add functions to provide dynamic ticks and high resolution timers.
> The code which keeps track of jiffies and handles the long idle
> periods is shared between tick based and high resolution timer based
> dynticks. The dyntick functionality can be disabled on the kernel
> commandline. Provide also the infrastructure to support high resolution
> timers.

cpufreq_ondemand didn't work here on kernels 2.6.20-rc4-rt1 and above.
Following patch against 20-rc6-rt2 helps.

Karsten
-------------------------------------------------------------------------------
From: Karsten Wiese <fzu@wemgehoertderstaat.de>

high_res_timers: precisely update_process_times

Sometimes tick_sched_timer() calls tick_do_update_jiffies64() and
jiffies is updated by !=1.
Cope with these situations by splitting update_process_times() into
__update_process_times() and tick() and
exchanging call to update_process_times() from tick_sched_timer()
by calls to the splits.
Fixes cpufreq_ondemand going nuts here.

Signed-off-by: Karsten Wiese <fzu@wemgehoertderstaat.de>


--- rc6-rt2/include/linux/sched.h 2007-01-26 14:42:55.000000000 +0100
+++ rc6-rt2-kw/include/linux/sched.h 2007-01-28 02:02:29.000000000 +0100
@@ -264,6 +264,13 @@ long io_schedule_timeout(long timeout);
extern void cpu_init (void);
extern void trap_init(void);
extern void update_process_times(int user);
+#ifdef CONFIG_HIGH_RES_TIMERS
+extern void __update_process_times(int user_tick, unsigned long ticks);
+extern void tick(int user_tick);
+#define NO_HIGH_RES_TIMERS_STATIC
+#else
+#define NO_HIGH_RES_TIMERS_STATIC static
+#endif
extern void scheduler_tick(void);

#ifdef CONFIG_GENERIC_HARDIRQS
diff -pur rc6-rt2/kernel/time/tick-sched.c rc6-rt2-kw/kernel/time/tick-sched.c
--- rc6-rt2/kernel/time/tick-sched.c 2007-01-26 14:42:55.000000000 +0100
+++ rc6-rt2-kw/kernel/time/tick-sched.c 2007-01-28 01:45:14.000000000 +0100
@@ -41,9 +41,9 @@ struct tick_sched *tick_get_tick_sched(i
/*
* Must be called with interrupts disabled !
*/
-static void tick_do_update_jiffies64(ktime_t now)
+static unsigned long tick_do_update_jiffies64(ktime_t now)
{
- unsigned long ticks = 1;
+ unsigned long ticks = 0;
ktime_t delta;

/* Reevalute with xtime_lock held */
@@ -51,7 +51,7 @@ static void tick_do_update_jiffies64(kti

delta = ktime_sub(now, last_jiffies_update);
if (delta.tv64 >= tick_period.tv64) {
-
+ ticks = 1;
delta = ktime_sub(delta, tick_period);
last_jiffies_update = ktime_add(last_jiffies_update,
tick_period);
@@ -68,6 +68,7 @@ static void tick_do_update_jiffies64(kti
do_timer(ticks);
}
write_sequnlock(&xtime_lock);
+ return ticks;
}

/*
@@ -423,32 +424,36 @@ static enum hrtimer_restart tick_sched_t
ktime_t now = ktime_get();

/* Check, if the jiffies need an update */
- tick_do_update_jiffies64(now);
+ unsigned long ticks = tick_do_update_jiffies64(now);

/*
* Do not call, when we are not in irq context and have
* no valid regs pointer
*/
if (regs) {
- /*
- * When we are idle and the tick is stopped, we have to touch
- * the watchdog as we might not schedule for a really long
- * time. This happens on complete idle SMP systems while
- * waiting on the login prompt. We also increment the "start of
- * idle" jiffy stamp so the idle accounting adjustment we do
- * when we go busy again does not account too much ticks.
- */
- if (ts->tick_stopped) {
- touch_softlockup_watchdog();
- ts->idle_jiffies++;
+ if (ticks) {
+ /*
+ * When we are idle and the tick is stopped, we have to touch
+ * the watchdog as we might not schedule for a really long
+ * time. This happens on complete idle SMP systems while
+ * waiting on the login prompt. We also increment the "start of
+ * idle" jiffy stamp so the idle accounting adjustment we do
+ * when we go busy again does not account too much ticks.
+ */
+ if (ts->tick_stopped) {
+ touch_softlockup_watchdog();
+ ts->idle_jiffies++;
+ __update_process_times(user_mode(regs), 1);
+ } else
+ __update_process_times(user_mode(regs), ticks);
}
/*
- * update_process_times() might take tasklist_lock, hence
+ * tick() might take tasklist_lock, hence
* drop the base lock. sched-tick hrtimers are per-CPU and
* never accessible by userspace APIs, so this is safe to do.
*/
spin_unlock(&base->lock);
- update_process_times(user_mode(regs));
+ tick(user_mode(regs));
// profile_tick(CPU_PROFILING);
spin_lock(&base->lock);
}
diff -pur rc6-rt2/kernel/timer.c rc6-rt2-kw/kernel/timer.c
--- rc6-rt2/kernel/timer.c 2007-01-26 14:42:55.000000000 +0100
+++ rc6-rt2-kw/kernel/timer.c 2007-01-28 02:03:01.000000000 +0100
@@ -1312,20 +1312,24 @@ static void update_wall_time(void)
update_vsyscall(&xtime, clock);
}

-/*
- * 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)
+NO_HIGH_RES_TIMERS_STATIC
+void __update_process_times(int user_tick, unsigned long ticks)
{
- int cpu = smp_processor_id();
struct task_struct *p = current;

/* Note: this timer irq context must be accounted for as well. */
if (user_tick)
- account_user_time(p, jiffies_to_cputime(1));
+ account_user_time(p, jiffies_to_cputime(ticks));
else
- account_system_time(p, HARDIRQ_OFFSET, jiffies_to_cputime(1));
+ account_system_time(p, HARDIRQ_OFFSET, jiffies_to_cputime(ticks));
+}
+
+NO_HIGH_RES_TIMERS_STATIC
+void tick(int user_tick)
+{
+ int cpu = smp_processor_id();
+ struct task_struct *p = current;
+
scheduler_tick();
run_local_timers();
if (rcu_pending(cpu))
@@ -1334,6 +1338,16 @@ void update_process_times(int user_tick)
}

/*
+ * 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)
+{
+ __update_process_times(user_tick, 1);
+ tick(user_tick);
+}
+
+/*
* Nr of active tasks - counted in fixed-point numbers
*/
static unsigned long count_active_tasks(void)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
\
 
 \ /
  Last update: 2007-01-28 03:07    [W:0.126 / U:41.536 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site