irq: next irq tracking

Hi Thomas,

this pull request contains the next irq prediction as posted for V11
[1] which takes into account all the commments. It is based on top of irq/core.

As previously discussed the code is not enabled by default, hence compiled out.

The first patch adds the IRQF_TIMER flag to the timers which are percpu in order to discard any timing measurement when the interrupt is coming from a timer. All the timers changes have been discarded from V10.

The second patch adds a circular buffer and the small routine to store the local time when an interrupt occurs. The t-uple <irq,timestamp> are encoded together into a single 64bits as suggested by Nicolas Pitre.

The third patch provides the mathematic to compute the regular intervals. The rational is in changelog and the math is commented in details in the code.

Thanks.

-- Daniel

The following changes since commit b50fb7c99217922ea36d6e38bae34d84c0587cad:

Merge branch 'linus' into irq/core (2017-06-20 22:08:32 +0200)

are available in the git repository at:

http://git.linaro.org/people/daniel.lezcano/linux.git next-irq/4.13-v11

for you to fetch changes up to eacb5f29a1ec71ee3e43e2729b8a1b5ec51320d5:

irq: Compute the periodic interval for interrupts (2017-06-20 23:14:18 +0200)

----------------------------------------------------------------
Daniel Lezcano (3):
irq: Allow to pass the IRQF_TIMER flag with percpu irq request
irq: Track the interrupt timings
irq: Compute the periodic interval for interrupts

 include/linux/interrupt.h |  17 +++++++-
 kernel/irq/Kconfig        |   3 ++
 kernel/irq/Makefile       |   1 +
 kernel/irq/handle.c       |   2 +
 kernel/irq/internals.h    | 103 +++++++++++++++++++++++++++++++++++++++++++++
 kernel/irq/manage.c       |  18 +++++---
 kernel/irq/timings.c      | 366 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 7 files changed, 504 insertions(+), 6 deletions(-)
 create mode 100644 kernel/irq/timings.c

 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

In the next changes, we track when the interrupts occur in order to statistically compute when is supposed to happen the next interrupt.

In all the interruptions, it does not make sense to store the timer interrupt occurences and try to predict the next interrupt as we know the expiration time.

The request_irq() has a irq flags parameter and the timer drivers use it to pass the IRQF_TIMER flag, letting us know the interrupt is coming from a timer. Based on this flag, we can discard these interrupts when tracking them.

But, the API request_percpu_irq does not allow to pass a flag, hence specifying if the interrupt type is a timer.

Add a function __request_percpu_irq() where we can specify the flags. The request_percpu_irq() function is changed to be a wrapper to __request_percpu_irq() passing a zero flag parameter.

For now, in order to prevent a misusage of this parameter, only the IRQF_TIMER flag (or zero) is a valid parameter to be passed to the __request_percpu_irq() function.

---
Changelog:

V11:
- Drop the changes in the timer drivers and rename the function request_percpu_irq_flags to __request_percpu_irq.
V10:

	- Drop the change for arch arm virtual timer
	  https://lkml.org/lkml/2017/4/25/184

V9:
- Clarified the patch description
- Fixed EXPORT_SYMBOL_GPL(request_percpu_irq_flags)

---
 include/linux/interrupt.h | 11 ++++++++++-
 kernel/irq/manage.c       | 15 ++++++++++-----
 2 files changed, 20 insertions(+), 6 deletions(-)

index a6fba48..d4ae50e 100644
@@ -152,8 +152,17 @@ request_any_context_irq(unsigned int irq, irq_handler_t handler,
 			unsigned long flags, const char *name, void *dev_id);
 
 extern int __must_check
+__request_percpu_irq(unsigned int irq, irq_handler_t handler,
+		     unsigned long flags, const char *devname,
+		     void __percpu *percpu_dev_id);
+
+static inline int __must_check
 request_percpu_irq(unsigned int irq, irq_handler_t handler,
-		   const char *devname, void __percpu *percpu_dev_id);
+		   const char *devname, void __percpu *percpu_dev_id)
+{
+	return __request_percpu_irq(irq, handler, 0,
+				    devname, percpu_dev_id);
+}
 
 extern const void *free_irq(unsigned int, void *);
 extern void free_percpu_irq(unsigned int, void __percpu *);

index 4c34696..5424cfb 100644
@@ -1974,9 +1974,10 @@ int setup_percpu_irq(unsigned int irq, struct irqaction *act)
 }
 
 /**
- *	request_percpu_irq - allocate a percpu interrupt line
+ *	__request_percpu_irq - allocate a percpu interrupt line
  *	@irq: Interrupt line to allocate
  *	@handler: Function to be called when the IRQ occurs.
+ *	@flags: Interrupt type flags (IRQF_TIMER only)
  *	@devname: An ascii name for the claiming device
  *	@dev_id: A percpu cookie passed back to the handler function
  *

@@ -1989,8 +1990,9 @@ int setup_percpu_irq(unsigned int irq, struct irqaction *act)
  *	the handler gets called with the interrupted CPU's instance of
  *	that variable.
  */
-int request_percpu_irq(unsigned int irq, irq_handler_t handler,
-		       const char *devname, void __percpu *dev_id)
+int __request_percpu_irq(unsigned int irq, irq_handler_t handler,
+			 unsigned long flags, const char *devname,
+			 void __percpu *dev_id)
 {
 	struct irqaction *action;
 	struct irq_desc *desc;

@@ -2004,12 +2006,15 @@ int request_percpu_irq(unsigned int irq, irq_handler_t handler,
 	    !irq_settings_is_per_cpu_devid(desc))
 		return -EINVAL;
 
+	if (flags && flags != IRQF_TIMER)
+		return -EINVAL;
+
 	action = kzalloc(sizeof(struct irqaction), GFP_KERNEL);
 	if (!action)
 		return -ENOMEM;
 
 	action->handler = handler;
-	action->flags = IRQF_PERCPU | IRQF_NO_SUSPEND;
+	action->flags = flags | IRQF_PERCPU | IRQF_NO_SUSPEND;
 	action->name = devname;
 	action->percpu_dev_id = dev_id;
 

@@ -2030,7 +2035,7 @@ int request_percpu_irq(unsigned int irq, irq_handler_t handler,
 
 	return retval;
 }
-EXPORT_SYMBOL_GPL(request_percpu_irq);
+EXPORT_SYMBOL_GPL(__request_percpu_irq);
 
 /**
  *	irq_get_irqchip_state - returns the irqchip state of a interrupt.

2.7.4

The interrupt framework gives a lot of information about each interrupt.

It does not keep track of when those interrupts occur though.

This patch provides a mean to record the timestamp for each interrupt occurrences in a per-CPU circular buffer to help with the prediction of the next occurrence using a statistical model.

Each CPU can store IRQ_TIMINGS_SIZE events <irq, timestamp>, the current value of IRQ_TIMINGS_SIZE is 32.

Each event is encoded into a single u64, where the 48 bits are used for the timestamp and the next 16 bits are for the irq number.

A static key is introduced so when the irq prediction is switched off at runtime, we can reduce the overhead near to zero.

It results in most of the code in internals.h for inline reason and a very few in the new file timings.c. The latter will contain more in the next patch which will provide the statistical model for the next event prediction.

Note this code is by default *not* compiled in the kernel.

---

Changelog:

V11:
- No changes
V10:
- No changes
V9:
- Changed indentation level by inverting the static key condition
- Encoded interrupt and timestamp into a u64 variable
- Boolean enable instead of refcount for the static key
V8:
- Replaced percpu field in the irqdesc by a percpu array containing the timings and the associated irq. The function irq_timings_get_next()is no longer needed, so it is removed
- Removed all unused code resulting from the conversion irqdesc->percpu timings storage
V7:
- Mentionned in the irq_timings_get_next() function description, the function must be called inside a rcu read locked section
V6:
- Renamed handle_irq_timings to record_irq_time
- Stored the event time instead of the interval time
- Removed the 'timestamp' field from the timings structure
- Moved _handle_irq_timings content inside record_irq_time
V5:
- Changed comment about 'deterministic' as the comment is confusing
- Added license comment in the header
- Replaced irq_timings_get/put by irq_timings_enable/disable
- Moved IRQS_TIMINGS check in the handle_timings inline function
- Dropped 'if !prev' as it is pointless
- Stored time interval in nsec basis with u64 instead of u32
- Removed redundant store
- Removed the math
V4:
- Added a static key
- Added more comments for irq_timings_get_next()
- Unified some function names to be prefixed by 'irq_timings_...'
- Fixed a rebase error
V3:
- Replaced ktime_get() by local_clock()
- Shared irq are not handled
- Simplified code by adding the timing in the irqdesc struct
- Added a function to browse the irq timings
V2:
- Fixed kerneldoc comment
- Removed data field from the struct irq timing
- Changed the lock section comment
- Removed semi-colon style with empty stub
- Replaced macro by static inline
- Fixed static functions declaration
RFC:
- initial posting

---
 include/linux/interrupt.h |  5 +++
 kernel/irq/Kconfig        |  3 ++
 kernel/irq/Makefile       |  1 +
 kernel/irq/handle.c       |  2 ++
 kernel/irq/internals.h    | 84 +++++++++++++++++++++++++++++++++++++++++++++++
 kernel/irq/manage.c       |  3 ++
 kernel/irq/timings.c      | 30 +++++++++++++++++
 7 files changed, 128 insertions(+)
 create mode 100644 kernel/irq/timings.c

index d4ae50e..5f98aa7 100644
@@ -712,6 +712,11 @@ static inline void init_irq_proc(void)
 }
 #endif
 
+#ifdef CONFIG_IRQ_TIMINGS
+void irq_timings_enable(void);
+void irq_timings_disable(void);
+#endif
+
 struct seq_file;
 int show_interrupts(struct seq_file *p, void *v);
 int arch_show_interrupts(struct seq_file *p, int prec);

index 3bbfd6a..38e551d 100644
@@ -81,6 +81,9 @@ config GENERIC_MSI_IRQ_DOMAIN
 config HANDLE_DOMAIN_IRQ
 	bool
 
+config IRQ_TIMINGS
+	bool
+
 config IRQ_DOMAIN_DEBUG
 	bool "Expose hardware/virtual IRQ mapping via debugfs"
 	depends on IRQ_DOMAIN && DEBUG_FS

index 1d3ee31..efb5f14 100644
@@ -10,3 +10,4 @@ obj-$(CONFIG_PM_SLEEP) += pm.o
 obj-$(CONFIG_GENERIC_MSI_IRQ) += msi.o
 obj-$(CONFIG_GENERIC_IRQ_IPI) += ipi.o
 obj-$(CONFIG_SMP) += affinity.o
+obj-$(CONFIG_IRQ_TIMINGS) += timings.o

index d3f2490..eb4d3e8 100644
@@ -138,6 +138,8 @@ irqreturn_t __handle_irq_event_percpu(struct irq_desc *desc, unsigned int *flags
 	unsigned int irq = desc->irq_data.irq;
 	struct irqaction *action;
 
+	record_irq_time(desc);
+
 	for_each_action_of_desc(desc, action) {
 		irqreturn_t res;
 

index 5424cfb..d3f63f1 100644
@@ -1388,6 +1388,8 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new)
 
 	raw_spin_unlock_irqrestore(&desc->lock, flags);
 
+	setup_timings(desc, new);
+
 	/*
 	 * Strictly no need to wake it up, but hung_task complains
 	 * when no hard interrupt wakes the thread up.

@@ -1516,6 +1518,7 @@ static struct irqaction *__free_irq(unsigned int irq, void *dev_id)
 		irq_settings_clr_disable_unlazy(desc);
 		irq_shutdown(desc);
 		irq_release_resources(desc);
+		remove_timings(desc);
 	}
 
 #ifdef CONFIG_SMP

new file mode 100644
index 0000000..56cf687
@@ -0,0 +1,30 @@
+/*
+ * linux/kernel/irq/timings.c
+ *
+ * Copyright (C) 2016, Linaro Ltd - Daniel Lezcano <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/static_key.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+
+#include "internals.h"
+
+DEFINE_STATIC_KEY_FALSE(irq_timing_enabled);
+
+DEFINE_PER_CPU(struct irq_timings, irq_timings);
+
+void irq_timings_enable(void)
+{
+	static_branch_enable(&irq_timing_enabled);
+}
+
+void irq_timings_disable(void)
+{
+	static_branch_disable(&irq_timing_enabled);
+}

2.7.4

...

Hi Thomas,

thanks for the review.

[ ... ]

Ok.

Nothing wrong, as we are expecting two values I don't like the idea to have one returned and the other one passed as a pointer. It is a matter of taste. I can return the irq if you prefer.

 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

An interrupt behaves with a burst of activity with periodic interval of time followed by one or two peaks of longer interval.

As the time intervals are periodic, statistically speaking they follow a normal distribution and each interrupts can be tracked individually.

This patch does statistics on all interrupts, except the timers which are deterministic by essence. The goal is to extract the periodicity for each interrupt, with the last timestamp and sum them, so we have the next event.

Taking the earliest prediction gives the expected wakeup on the system (assuming a timer won't expire before).

As stated in the previous patch, this code is not enabled in the kernel by default.

---
Changelog:

V11:
- No changes
V10:
- Simplified index/count computation
V9:
- Deal with 48+16 bits encoded values
- Changed irq_stat => irqt_stat to prevent name collision on s390
- Changed div64 by constant IRQ_TIMINGS_SHIFT bit shift for average
- Changed div64 by constant IRQ_TIMINGS_SHIFT bit shift for variance

---
 include/linux/interrupt.h |   1 +
 kernel/irq/internals.h    |  21 ++-
 kernel/irq/timings.c      | 336 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 357 insertions(+), 1 deletion(-)

index 5f98aa7..5ac6e23 100644
@@ -715,6 +715,7 @@ static inline void init_irq_proc(void)
 #ifdef CONFIG_IRQ_TIMINGS
 void irq_timings_enable(void);
 void irq_timings_disable(void);
+u64 irq_timings_next_event(u64 now);
 #endif
 
 struct seq_file;

index df51b5e0..bf3d827 100644
@@ -237,18 +237,26 @@ irq_pm_remove_action(struct irq_desc *desc, struct irqaction *action) { }
 
 struct irq_timings {
 	u64 values[IRQ_TIMINGS_SIZE]; /* our circular buffer */
-	unsigned int count; /* Number of interruptions since last inspection */
+	int count; /* Number of interruptions since last inspection */
 };
 
 DECLARE_PER_CPU(struct irq_timings, irq_timings);
 
+extern void irq_timings_free(int irq);
+extern int irq_timings_alloc(int irq);
+
 static inline void remove_timings(struct irq_desc *desc)
 {
 	desc->istate &= ~IRQS_TIMINGS;
+
+	irq_timings_free(irq_desc_get_irq(desc));
 }
 
 static inline void setup_timings(struct irq_desc *desc, struct irqaction *act)
 {
+	int irq = irq_desc_get_irq(desc);
+	int ret;
+
 	/*
 	 * We don't need the measurement because the idle code already
 	 * knows the next expiry event.

@@ -256,6 +264,17 @@ static inline void setup_timings(struct irq_desc *desc, struct irqaction *act)
 	if (act->flags & __IRQF_TIMER)
 		return;
 
+	/*
+	 * In case the timing allocation fails, we just want to warn,
+	 * not fail, so letting the system boot anyway.
+	 */
+	ret = irq_timings_alloc(irq);
+	if (ret) {
+		pr_warn("Failed to allocate irq timing stats for irq%d (%d)",
+			irq, ret);
+		return;
+	}
+
 	desc->istate |= IRQS_TIMINGS;
 }
 

2.7.4

Idem.

	struct irqt_stat __percpu *s;
	int id;

On Thu, Jun 22, 2017 at 05:25:56PM +0200, Thomas Gleixner wrote:

[ ... ]

No, it could be also the value returned by idr_alloc() which could be EINVAL or ENOSPC (which are values returned also by the low level libraries radix-tree functions).

Hi Daniel,

Sorry for leading you round the garden path on the naming.

FWIW:

Thanks,
Mark.

No problem ;)

 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

I merily asked for splitting them out into a separate patch, but I'm fine with dropping them completely. But with them dropped, there is no point to actually change the interface right now. Then 1/3 can be dropped as well and we do a scripted interface change right before 4.13-rc1 and queue the timer driver changes for 4.14

Thanks,

tglx

Ok, I updated the irq timings branch without the 1/3.

http://git.linaro.org/people/daniel.lezcano/linux.git next-irq/4.13-v11.1

 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

The irq timings infrastructure tracks when the interrupts occur in order to statistically compute when is supposed to happen an interrupt.

In all the interruptions, it does not make sense to store the timer interrupt occurences and try to predict the next interrupt as we know the expiration time.

The request_irq() has a irq flags parameter and the timer drivers use it to pass the IRQF_TIMER flag, letting us know the interrupt is coming from a timer. Based on this flag, we can discard these interrupts when tracking them.

But, the API request_percpu_irq does not allow to pass a flag, hence specifying if the interrupt type is a timer.

Add a function __request_percpu_irq() where we can specify the flags. The request_percpu_irq() function is changed to be a wrapper to __request_percpu_irq() passing a zero flag parameter.

For now, in order to prevent a misusage of this parameter, only the IRQF_TIMER flag (or zero) is a valid parameter to be passed to the __request_percpu_irq() function.

---
Changelog:

V11:
- Drop the changes in the timer drivers and rename the function request_percpu_irq_flags to __request_percpu_irq.
V10:

	- Drop the change for arch arm virtual timer
	  https://lkml.org/lkml/2017/4/25/184

V9:
- Clarified the patch description
- Fixed EXPORT_SYMBOL_GPL(request_percpu_irq_flags)

---
 include/linux/interrupt.h | 11 ++++++++++-
 kernel/irq/manage.c       | 15 ++++++++++-----
 2 files changed, 20 insertions(+), 6 deletions(-)

index 37f8e35..5ac6e23 100644
@@ -152,8 +152,17 @@ request_any_context_irq(unsigned int irq, irq_handler_t handler,
 			unsigned long flags, const char *name, void *dev_id);
 
 extern int __must_check
+__request_percpu_irq(unsigned int irq, irq_handler_t handler,
+		     unsigned long flags, const char *devname,
+		     void __percpu *percpu_dev_id);
+
+static inline int __must_check
 request_percpu_irq(unsigned int irq, irq_handler_t handler,
-		   const char *devname, void __percpu *percpu_dev_id);
+		   const char *devname, void __percpu *percpu_dev_id)
+{
+	return __request_percpu_irq(irq, handler, 0,
+				    devname, percpu_dev_id);
+}
 
 extern const void *free_irq(unsigned int, void *);
 extern void free_percpu_irq(unsigned int, void __percpu *);

index 5c11c17..f80baf7 100644
@@ -1935,9 +1935,10 @@ int setup_percpu_irq(unsigned int irq, struct irqaction *act)
 }
 
 /**
- *	request_percpu_irq - allocate a percpu interrupt line
+ *	__request_percpu_irq - allocate a percpu interrupt line
  *	@irq: Interrupt line to allocate
  *	@handler: Function to be called when the IRQ occurs.
+ *	@flags: Interrupt type flags (IRQF_TIMER only)
  *	@devname: An ascii name for the claiming device
  *	@dev_id: A percpu cookie passed back to the handler function
  *

@@ -1950,8 +1951,9 @@ int setup_percpu_irq(unsigned int irq, struct irqaction *act)
  *	the handler gets called with the interrupted CPU's instance of
  *	that variable.
  */
-int request_percpu_irq(unsigned int irq, irq_handler_t handler,
-		       const char *devname, void __percpu *dev_id)
+int __request_percpu_irq(unsigned int irq, irq_handler_t handler,
+			 unsigned long flags, const char *devname,
+			 void __percpu *dev_id)
 {
 	struct irqaction *action;
 	struct irq_desc *desc;

@@ -1965,12 +1967,15 @@ int request_percpu_irq(unsigned int irq, irq_handler_t handler,
 	    !irq_settings_is_per_cpu_devid(desc))
 		return -EINVAL;
 
+	if (flags && flags != IRQF_TIMER)
+		return -EINVAL;
+
 	action = kzalloc(sizeof(struct irqaction), GFP_KERNEL);
 	if (!action)
 		return -ENOMEM;
 
 	action->handler = handler;
-	action->flags = IRQF_PERCPU | IRQF_NO_SUSPEND;
+	action->flags = flags | IRQF_PERCPU | IRQF_NO_SUSPEND;
 	action->name = devname;
 	action->percpu_dev_id = dev_id;
 

@@ -1991,7 +1996,7 @@ int request_percpu_irq(unsigned int irq, irq_handler_t handler,
 
 	return retval;
 }
-EXPORT_SYMBOL_GPL(request_percpu_irq);
+EXPORT_SYMBOL_GPL(__request_percpu_irq);
 
 /**
  *	irq_get_irqchip_state - returns the irqchip state of a interrupt.

2.7.4