Re: [PATCH] remove i386 dynamic ticks ifdefs
On Tue, 1 Jan 2002 21:33, Pavel Machek wrote: > Hi! > > > I assume you're maintaining the dyn tick patches for i386 posted on the > > muru website as your email is listed there. I thought you might be > > interested in this patch for dyn-ticks which removes most of the #ifdefs > > out of common code paths as per linux kernel style and moves more code > > into dyn-tick.c. Most of it is straight forward code reorganisation, but > > to keep do_timer_interrupt inlined I'd have to move it's code around > > somewhat. That may be a better option but I've tried to fiddle with the > > mainline code as little as possible. > > > > Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 > > applied > > > > cc'ed lkml just for public record of the patch. > > Please inline patches... Not everyone uses console email clients :| It was an inlined attachment rather than an ordinary attachment but clearly that doesn't suit those with console clients. I'm sorry. > > You broke indentation in one of first hunks, and you probably > want empty functions to be static inline, so that we do not eat > function call overhead. Will examine. Thanks very much for your code comments! Cheers, Con - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] remove i386 dynamic ticks ifdefs
Hi! > I assume you're maintaining the dyn tick patches for i386 posted on the muru > website as your email is listed there. I thought you might be interested in > this patch for dyn-ticks which removes most of the #ifdefs out of common code > paths as per linux kernel style and moves more code into dyn-tick.c. Most of > it is straight forward code reorganisation, but to keep do_timer_interrupt > inlined I'd have to move it's code around somewhat. That may be a better > option but I've tried to fiddle with the mainline code as little as possible. > > Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 applied > > cc'ed lkml just for public record of the patch. Please inline patches... You broke indentation in one of first hunks, and you probably want empty functions to be static inline, so that we do not eat function call overhead. Pavel -- 64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] remove i386 dynamic ticks ifdefs
Hi! I assume you're maintaining the dyn tick patches for i386 posted on the muru website as your email is listed there. I thought you might be interested in this patch for dyn-ticks which removes most of the #ifdefs out of common code paths as per linux kernel style and moves more code into dyn-tick.c. Most of it is straight forward code reorganisation, but to keep do_timer_interrupt inlined I'd have to move it's code around somewhat. That may be a better option but I've tried to fiddle with the mainline code as little as possible. Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 applied cc'ed lkml just for public record of the patch. Please inline patches... You broke indentation in one of first hunks, and you probably want empty functions to be static inline, so that we do not eat function call overhead. Pavel -- 64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH] remove i386 dynamic ticks ifdefs
On Tue, 1 Jan 2002 21:33, Pavel Machek wrote: Hi! I assume you're maintaining the dyn tick patches for i386 posted on the muru website as your email is listed there. I thought you might be interested in this patch for dyn-ticks which removes most of the #ifdefs out of common code paths as per linux kernel style and moves more code into dyn-tick.c. Most of it is straight forward code reorganisation, but to keep do_timer_interrupt inlined I'd have to move it's code around somewhat. That may be a better option but I've tried to fiddle with the mainline code as little as possible. Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 applied cc'ed lkml just for public record of the patch. Please inline patches... Not everyone uses console email clients :| It was an inlined attachment rather than an ordinary attachment but clearly that doesn't suit those with console clients. I'm sorry. You broke indentation in one of first hunks, and you probably want empty functions to be static inline, so that we do not eat function call overhead. Will examine. Thanks very much for your code comments! Cheers, Con - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[PATCH] remove i386 dynamic ticks ifdefs
Hi Tony
I assume you're maintaining the dyn tick patches for i386 posted on the muru
website as your email is listed there. I thought you might be interested in
this patch for dyn-ticks which removes most of the #ifdefs out of common code
paths as per linux kernel style and moves more code into dyn-tick.c. Most of
it is straight forward code reorganisation, but to keep do_timer_interrupt
inlined I'd have to move it's code around somewhat. That may be a better
option but I've tried to fiddle with the mainline code as little as possible.
Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 applied
cc'ed lkml just for public record of the patch.
Cheers,
Con
Move most of the dynamic ticks code that is #ifdef'd out of code paths and
put it into dyn-tick.c
Signed-off-by: Con Kolivas <[EMAIL PROTECTED]>
arch/i386/kernel/Makefile |2
arch/i386/kernel/apic.c | 36 --
arch/i386/kernel/dyn-tick.c | 146
arch/i386/kernel/irq.c |5 -
arch/i386/kernel/time.c | 72 -
include/asm/dyn-tick.h |4 -
include/linux/dyn-tick.h| 12 +++
7 files changed, 165 insertions(+), 112 deletions(-)
Index: linux-2.6.12-dt/arch/i386/kernel/apic.c
===
--- linux-2.6.12-dt.orig/arch/i386/kernel/apic.c 2005-07-29 01:43:15.0 +1000
+++ linux-2.6.12-dt/arch/i386/kernel/apic.c 2005-07-29 01:49:05.0 +1000
@@ -932,11 +932,8 @@ static void __setup_APIC_LVTT(unsigned i
apic_timer_val = clocks/APIC_DIVISOR;
-#ifdef CONFIG_NO_IDLE_HZ
- /* Local APIC timer is 24-bit */
if (apic_timer_val)
- dyn_tick->max_skip = 0xff / apic_timer_val;
-#endif
+ set_dyn_tick_max_skip(apic_timer_val);
apic_write_around(APIC_TMICT, apic_timer_val);
}
@@ -1051,12 +1048,7 @@ void __init setup_boot_APIC_clock(void)
*/
setup_APIC_timer(calibration_result);
-#ifdef CONFIG_NO_IDLE_HZ
- if (calibration_result)
- dyn_tick->state |= DYN_TICK_USE_APIC;
- else
- printk(KERN_INFO "dyn-tick: Cannot use local APIC\n");
-#endif
+ setup_dyn_tick_use_apic(calibration_result);
local_irq_enable();
}
@@ -1086,18 +1078,6 @@ void enable_APIC_timer(void)
}
}
-#if defined(CONFIG_NO_IDLE_HZ)
-void reprogram_apic_timer(unsigned int count)
-{
- unsigned long flags;
-
- count *= apic_timer_val;
- local_irq_save(flags);
- apic_write_around(APIC_TMICT, count);
- local_irq_restore(flags);
-}
-#endif
-
/*
* the frequency of the profiling timer can be changed
* by writing a multiplier value into /proc/profile.
@@ -1210,21 +1190,11 @@ fastcall void smp_apic_timer_interrupt(s
*/
irq_enter();
-#ifdef CONFIG_NO_IDLE_HZ
/*
* Check if we need to wake up PIT interrupt handler.
* Otherwise just wake up local APIC timer.
*/
- do {
- seq = read_seqbegin(_lock);
- if (dyn_tick->state & (DYN_TICK_ENABLED | DYN_TICK_SKIPPING)) {
- if (dyn_tick->skip_cpu == cpu && dyn_tick->skip > DYN_TICK_MIN_SKIP)
-dyn_tick->interrupt(99, NULL, regs);
- else
-reprogram_apic_timer(1);
- }
- } while (read_seqretry(_lock, seq));
-#endif
+ wakeup_pit_or_apic(seq, cpu, );
smp_local_timer_interrupt(regs);
irq_exit();
Index: linux-2.6.12-dt/arch/i386/kernel/dyn-tick.c
===
--- linux-2.6.12-dt.orig/arch/i386/kernel/dyn-tick.c 2005-07-29 01:43:15.0 +1000
+++ linux-2.6.12-dt/arch/i386/kernel/dyn-tick.c 2005-07-29 11:46:48.0 +1000
@@ -17,6 +17,7 @@
#include
#include
+#ifdef CONFIG_NO_IDLE_HZ
void arch_reprogram_timer(void)
{
if (cpu_has_local_apic()) {
@@ -43,3 +44,148 @@ int __init dyn_tick_init(void)
return 0;
}
arch_initcall(dyn_tick_init);
+
+static unsigned long long last_tick;
+
+/*
+ * This interrupt handler updates the time based on number of jiffies skipped
+ * It would be somewhat more optimized to have a customa handler in each timer
+ * using hardware ticks instead of nanoseconds. Note that CONFIG_NO_IDLE_HZ
+ * currently disables timer fallback on skipped jiffies.
+ */
+irqreturn_t dyn_tick_timer_interrupt(int irq, void *dev_id, struct pt_regs *regs)
+{
+ unsigned long flags;
+ volatile unsigned long long now;
+ unsigned int skipped = 0;
+
+ write_seqlock_irqsave(_lock, flags);
+ now = cur_timer->monotonic_clock();
+ while (now - last_tick >= NS_TICK_LEN) {
+ last_tick += NS_TICK_LEN;
+ cur_timer->mark_offset();
+ do_timer_interrupt(irq, NULL, regs);
+ skipped++;
+ }
+ if (dyn_tick->state & (DYN_TICK_ENABLED | DYN_TICK_SKIPPING)) {
+ dyn_tick->skip = 1;
+ if (cpu_has_local_apic())
+ reprogram_apic_timer(dyn_tick->skip);
+ reprogram_pit_timer(dyn_tick->skip);
+ dyn_tick->state |= DYN_TICK_ENABLED;
+ dyn_tick->state &= ~DYN_TICK_SKIPPING;
+ }
+ write_sequnlock_irqrestore(_lock, flags);
+
+ return IRQ_HANDLED;
+}
+
+int __init dyn_tick_arch_init(void)
+{
+ unsigned long flags;
+
+ write_seqlock_irqsave(_lock, flags);
+ last_tick =
[PATCH] remove i386 dynamic ticks ifdefs
Hi Tony
I assume you're maintaining the dyn tick patches for i386 posted on the muru
website as your email is listed there. I thought you might be interested in
this patch for dyn-ticks which removes most of the #ifdefs out of common code
paths as per linux kernel style and moves more code into dyn-tick.c. Most of
it is straight forward code reorganisation, but to keep do_timer_interrupt
inlined I'd have to move it's code around somewhat. That may be a better
option but I've tried to fiddle with the mainline code as little as possible.
Patch applies to 2.6.12 with patch-dynamic-tick-2.6.12-rc6-050610-1 applied
cc'ed lkml just for public record of the patch.
Cheers,
Con
Move most of the dynamic ticks code that is #ifdef'd out of code paths and
put it into dyn-tick.c
Signed-off-by: Con Kolivas [EMAIL PROTECTED]
arch/i386/kernel/Makefile |2
arch/i386/kernel/apic.c | 36 --
arch/i386/kernel/dyn-tick.c | 146
arch/i386/kernel/irq.c |5 -
arch/i386/kernel/time.c | 72 -
include/asm/dyn-tick.h |4 -
include/linux/dyn-tick.h| 12 +++
7 files changed, 165 insertions(+), 112 deletions(-)
Index: linux-2.6.12-dt/arch/i386/kernel/apic.c
===
--- linux-2.6.12-dt.orig/arch/i386/kernel/apic.c 2005-07-29 01:43:15.0 +1000
+++ linux-2.6.12-dt/arch/i386/kernel/apic.c 2005-07-29 01:49:05.0 +1000
@@ -932,11 +932,8 @@ static void __setup_APIC_LVTT(unsigned i
apic_timer_val = clocks/APIC_DIVISOR;
-#ifdef CONFIG_NO_IDLE_HZ
- /* Local APIC timer is 24-bit */
if (apic_timer_val)
- dyn_tick-max_skip = 0xff / apic_timer_val;
-#endif
+ set_dyn_tick_max_skip(apic_timer_val);
apic_write_around(APIC_TMICT, apic_timer_val);
}
@@ -1051,12 +1048,7 @@ void __init setup_boot_APIC_clock(void)
*/
setup_APIC_timer(calibration_result);
-#ifdef CONFIG_NO_IDLE_HZ
- if (calibration_result)
- dyn_tick-state |= DYN_TICK_USE_APIC;
- else
- printk(KERN_INFO dyn-tick: Cannot use local APIC\n);
-#endif
+ setup_dyn_tick_use_apic(calibration_result);
local_irq_enable();
}
@@ -1086,18 +1078,6 @@ void enable_APIC_timer(void)
}
}
-#if defined(CONFIG_NO_IDLE_HZ)
-void reprogram_apic_timer(unsigned int count)
-{
- unsigned long flags;
-
- count *= apic_timer_val;
- local_irq_save(flags);
- apic_write_around(APIC_TMICT, count);
- local_irq_restore(flags);
-}
-#endif
-
/*
* the frequency of the profiling timer can be changed
* by writing a multiplier value into /proc/profile.
@@ -1210,21 +1190,11 @@ fastcall void smp_apic_timer_interrupt(s
*/
irq_enter();
-#ifdef CONFIG_NO_IDLE_HZ
/*
* Check if we need to wake up PIT interrupt handler.
* Otherwise just wake up local APIC timer.
*/
- do {
- seq = read_seqbegin(xtime_lock);
- if (dyn_tick-state (DYN_TICK_ENABLED | DYN_TICK_SKIPPING)) {
- if (dyn_tick-skip_cpu == cpu dyn_tick-skip DYN_TICK_MIN_SKIP)
-dyn_tick-interrupt(99, NULL, regs);
- else
-reprogram_apic_timer(1);
- }
- } while (read_seqretry(xtime_lock, seq));
-#endif
+ wakeup_pit_or_apic(seq, cpu, regs);
smp_local_timer_interrupt(regs);
irq_exit();
Index: linux-2.6.12-dt/arch/i386/kernel/dyn-tick.c
===
--- linux-2.6.12-dt.orig/arch/i386/kernel/dyn-tick.c 2005-07-29 01:43:15.0 +1000
+++ linux-2.6.12-dt/arch/i386/kernel/dyn-tick.c 2005-07-29 11:46:48.0 +1000
@@ -17,6 +17,7 @@
#include linux/module.h
#include linux/dyn-tick.h
+#ifdef CONFIG_NO_IDLE_HZ
void arch_reprogram_timer(void)
{
if (cpu_has_local_apic()) {
@@ -43,3 +44,148 @@ int __init dyn_tick_init(void)
return 0;
}
arch_initcall(dyn_tick_init);
+
+static unsigned long long last_tick;
+
+/*
+ * This interrupt handler updates the time based on number of jiffies skipped
+ * It would be somewhat more optimized to have a customa handler in each timer
+ * using hardware ticks instead of nanoseconds. Note that CONFIG_NO_IDLE_HZ
+ * currently disables timer fallback on skipped jiffies.
+ */
+irqreturn_t dyn_tick_timer_interrupt(int irq, void *dev_id, struct pt_regs *regs)
+{
+ unsigned long flags;
+ volatile unsigned long long now;
+ unsigned int skipped = 0;
+
+ write_seqlock_irqsave(xtime_lock, flags);
+ now = cur_timer-monotonic_clock();
+ while (now - last_tick = NS_TICK_LEN) {
+ last_tick += NS_TICK_LEN;
+ cur_timer-mark_offset();
+ do_timer_interrupt(irq, NULL, regs);
+ skipped++;
+ }
+ if (dyn_tick-state (DYN_TICK_ENABLED | DYN_TICK_SKIPPING)) {
+ dyn_tick-skip = 1;
+ if (cpu_has_local_apic())
+ reprogram_apic_timer(dyn_tick-skip);
+ reprogram_pit_timer(dyn_tick-skip);
+ dyn_tick-state |= DYN_TICK_ENABLED;
+ dyn_tick-state = ~DYN_TICK_SKIPPING;
+ }
+ write_sequnlock_irqrestore(xtime_lock, flags);
+
+ return IRQ_HANDLED;
+}
+
+int __init dyn_tick_arch_init(void)
+{
+ unsigned long flags;
+
+
