All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	akpm@linux-foundation.org, neilb@suse.de, dgc@sgi.com,
	tomoki.sekiyama.qu@hitachi.com
Subject: Re: [PATCH 6/6] mm: per device dirty threshold
Date: Wed, 04 Apr 2007 14:43:37 +0200	[thread overview]
Message-ID: <1175690617.6483.85.camel@twins> (raw)
In-Reply-To: <E1HZ4ep-00069u-00@dorka.pomaz.szeredi.hu>

On Wed, 2007-04-04 at 14:32 +0200, Miklos Szeredi wrote:
> > Preferably you'd want to be able to 'flush' the per cpu diffs or
> > something like that in cases where thresh ~< NR_CPUS * stat_diff.
> > 
> > How about something like this:
> 
> Yes, maybe underscores and EXPORT_SYMBOLs are a bit excessive.

probably, here is one that actually compiles and handles cpu hotplug;
albeit a bit racy - which lock does exclude cpu hotplug these days?

---
 include/linux/backing-dev.h |   13 +++++++
 mm/backing-dev.c            |   79 ++++++++++++++++++++++++++++++++++++++++++++
 mm/page-writeback.c         |    4 ++
 3 files changed, 96 insertions(+)

Index: linux-2.6/include/linux/backing-dev.h
===================================================================
--- linux-2.6.orig/include/linux/backing-dev.h
+++ linux-2.6/include/linux/backing-dev.h
@@ -8,6 +8,7 @@
 #ifndef _LINUX_BACKING_DEV_H
 #define _LINUX_BACKING_DEV_H
 
+#include <linux/cpumask.h>
 #include <linux/spinlock.h>
 #include <asm/atomic.h>
 
@@ -117,6 +118,13 @@ void mod_bdi_stat(struct backing_dev_inf
 void inc_bdi_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
 void dec_bdi_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
 
+void bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
+void bdi_flush_all(struct backing_dev_info *bdi);
+
+static inline unsigned long bdi_stat_delta(void)
+{
+	return 8UL * num_online_cpus() * ilog2(num_online_cpus());
+}
 #else /* CONFIG_SMP */
 
 static inline void __mod_bdi_stat(struct backing_dev_info *bdi,
@@ -142,6 +150,11 @@ static inline void __dec_bdi_stat(struct
 #define mod_bdi_stat __mod_bdi_stat
 #define inc_bdi_stat __inc_bdi_stat
 #define dec_bdi_stat __dec_bdi_stat
+
+#define bdi_flush_stat(bdi, item) do { } while (0)
+#define bdi_flush_all(bdi) do { } while (0)
+
+#define bdi_stat_delta() 1UL
 #endif
 
 void bdi_stat_init(struct backing_dev_info *bdi);
Index: linux-2.6/mm/backing-dev.c
===================================================================
--- linux-2.6.orig/mm/backing-dev.c
+++ linux-2.6/mm/backing-dev.c
@@ -188,4 +188,83 @@ void dec_bdi_stat(struct backing_dev_inf
 	local_irq_restore(flags);
 }
 EXPORT_SYMBOL(dec_bdi_stat);
+
+void ___bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item,
+		int cpu)
+{
+	struct bdi_per_cpu_data *pcd = &bdi->pcd[cpu];
+	s8 *p = pcd->bdi_stat_diff + item;
+
+	bdi_stat_add(*p, bdi, item);
+	*p = 0;
+}
+
+struct bdi_flush_struct {
+	struct backing_dev_info *bdi;
+	enum bdi_stat_item item;
+};
+
+void __bdi_flush_stat(struct bdi_flush_struct *flush)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	___bdi_flush_stat(flush->bdi, flush->item, smp_processor_id());
+	local_irq_restore(flags);
+}
+
+void __bdi_flush_all(struct backing_dev_info *bdi)
+{
+	unsigned long flags;
+	int i, cpu;
+
+	local_irq_save(flags);
+	cpu = smp_processor_id();
+	for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
+		___bdi_flush_stat(bdi, i, cpu);
+	local_irq_restore(flags);
+}
+
+void bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item)
+{
+	struct bdi_flush_struct flush = {
+		bdi,
+		item
+	};
+
+#ifdef CONFIG_HOTPLUG_CPU
+	cpumask_t mask;
+	int cpu;
+
+	cpus_complement(mask, cpu_online_map);
+	for_each_cpu_mask(cpu, mask) {
+		unsigned long flags;
+
+		local_irq_save(flags);
+		___bdi_flush_stat(bdi, item, cpu);
+		local_irq_restore(flags);
+	}
+#endif
+	on_each_cpu((void (*)(void *))__bdi_flush_stat, &flush, 0, 1);
+}
+
+void bdi_flush_all(struct backing_dev_info *bdi)
+{
+#ifdef CONFIG_HOTPLUG_CPU
+	cpumask_t mask;
+	int cpu;
+
+	cpus_complement(mask, cpu_online_map);
+	for_each_cpu_mask(cpu, mask) {
+		int i;
+		unsigned long flags;
+
+		local_irq_save(flags);
+		for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
+			___bdi_flush_stat(bdi, i, cpu);
+		local_irq_restore(flags);
+	}
+#endif
+	on_each_cpu((void (*)(void *))__bdi_flush_all, bdi, 0, 1);
+}
 #endif
Index: linux-2.6/mm/page-writeback.c
===================================================================
--- linux-2.6.orig/mm/page-writeback.c
+++ linux-2.6/mm/page-writeback.c
@@ -345,6 +345,10 @@ static void balance_dirty_pages(struct a
 
 			get_dirty_limits(&background_thresh, &dirty_thresh,
 				       &bdi_thresh, bdi);
+
+			if (bdi_thresh < bdi_stat_delta())
+				bdi_flush_all(bdi);
+
 			bdi_nr_reclaimable = bdi_stat(bdi, BDI_DIRTY) +
 						bdi_stat(bdi, BDI_UNSTABLE);
 			if (bdi_nr_reclaimable + bdi_stat(bdi, BDI_WRITEBACK) <=



WARNING: multiple messages have this Message-ID (diff)
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	akpm@linux-foundation.org, neilb@suse.de, dgc@sgi.com,
	tomoki.sekiyama.qu@hitachi.com
Subject: Re: [PATCH 6/6] mm: per device dirty threshold
Date: Wed, 04 Apr 2007 14:43:37 +0200	[thread overview]
Message-ID: <1175690617.6483.85.camel@twins> (raw)
In-Reply-To: <E1HZ4ep-00069u-00@dorka.pomaz.szeredi.hu>

On Wed, 2007-04-04 at 14:32 +0200, Miklos Szeredi wrote:
> > Preferably you'd want to be able to 'flush' the per cpu diffs or
> > something like that in cases where thresh ~< NR_CPUS * stat_diff.
> > 
> > How about something like this:
> 
> Yes, maybe underscores and EXPORT_SYMBOLs are a bit excessive.

probably, here is one that actually compiles and handles cpu hotplug;
albeit a bit racy - which lock does exclude cpu hotplug these days?

---
 include/linux/backing-dev.h |   13 +++++++
 mm/backing-dev.c            |   79 ++++++++++++++++++++++++++++++++++++++++++++
 mm/page-writeback.c         |    4 ++
 3 files changed, 96 insertions(+)

Index: linux-2.6/include/linux/backing-dev.h
===================================================================
--- linux-2.6.orig/include/linux/backing-dev.h
+++ linux-2.6/include/linux/backing-dev.h
@@ -8,6 +8,7 @@
 #ifndef _LINUX_BACKING_DEV_H
 #define _LINUX_BACKING_DEV_H
 
+#include <linux/cpumask.h>
 #include <linux/spinlock.h>
 #include <asm/atomic.h>
 
@@ -117,6 +118,13 @@ void mod_bdi_stat(struct backing_dev_inf
 void inc_bdi_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
 void dec_bdi_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
 
+void bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item);
+void bdi_flush_all(struct backing_dev_info *bdi);
+
+static inline unsigned long bdi_stat_delta(void)
+{
+	return 8UL * num_online_cpus() * ilog2(num_online_cpus());
+}
 #else /* CONFIG_SMP */
 
 static inline void __mod_bdi_stat(struct backing_dev_info *bdi,
@@ -142,6 +150,11 @@ static inline void __dec_bdi_stat(struct
 #define mod_bdi_stat __mod_bdi_stat
 #define inc_bdi_stat __inc_bdi_stat
 #define dec_bdi_stat __dec_bdi_stat
+
+#define bdi_flush_stat(bdi, item) do { } while (0)
+#define bdi_flush_all(bdi) do { } while (0)
+
+#define bdi_stat_delta() 1UL
 #endif
 
 void bdi_stat_init(struct backing_dev_info *bdi);
Index: linux-2.6/mm/backing-dev.c
===================================================================
--- linux-2.6.orig/mm/backing-dev.c
+++ linux-2.6/mm/backing-dev.c
@@ -188,4 +188,83 @@ void dec_bdi_stat(struct backing_dev_inf
 	local_irq_restore(flags);
 }
 EXPORT_SYMBOL(dec_bdi_stat);
+
+void ___bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item,
+		int cpu)
+{
+	struct bdi_per_cpu_data *pcd = &bdi->pcd[cpu];
+	s8 *p = pcd->bdi_stat_diff + item;
+
+	bdi_stat_add(*p, bdi, item);
+	*p = 0;
+}
+
+struct bdi_flush_struct {
+	struct backing_dev_info *bdi;
+	enum bdi_stat_item item;
+};
+
+void __bdi_flush_stat(struct bdi_flush_struct *flush)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	___bdi_flush_stat(flush->bdi, flush->item, smp_processor_id());
+	local_irq_restore(flags);
+}
+
+void __bdi_flush_all(struct backing_dev_info *bdi)
+{
+	unsigned long flags;
+	int i, cpu;
+
+	local_irq_save(flags);
+	cpu = smp_processor_id();
+	for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
+		___bdi_flush_stat(bdi, i, cpu);
+	local_irq_restore(flags);
+}
+
+void bdi_flush_stat(struct backing_dev_info *bdi, enum bdi_stat_item item)
+{
+	struct bdi_flush_struct flush = {
+		bdi,
+		item
+	};
+
+#ifdef CONFIG_HOTPLUG_CPU
+	cpumask_t mask;
+	int cpu;
+
+	cpus_complement(mask, cpu_online_map);
+	for_each_cpu_mask(cpu, mask) {
+		unsigned long flags;
+
+		local_irq_save(flags);
+		___bdi_flush_stat(bdi, item, cpu);
+		local_irq_restore(flags);
+	}
+#endif
+	on_each_cpu((void (*)(void *))__bdi_flush_stat, &flush, 0, 1);
+}
+
+void bdi_flush_all(struct backing_dev_info *bdi)
+{
+#ifdef CONFIG_HOTPLUG_CPU
+	cpumask_t mask;
+	int cpu;
+
+	cpus_complement(mask, cpu_online_map);
+	for_each_cpu_mask(cpu, mask) {
+		int i;
+		unsigned long flags;
+
+		local_irq_save(flags);
+		for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
+			___bdi_flush_stat(bdi, i, cpu);
+		local_irq_restore(flags);
+	}
+#endif
+	on_each_cpu((void (*)(void *))__bdi_flush_all, bdi, 0, 1);
+}
 #endif
Index: linux-2.6/mm/page-writeback.c
===================================================================
--- linux-2.6.orig/mm/page-writeback.c
+++ linux-2.6/mm/page-writeback.c
@@ -345,6 +345,10 @@ static void balance_dirty_pages(struct a
 
 			get_dirty_limits(&background_thresh, &dirty_thresh,
 				       &bdi_thresh, bdi);
+
+			if (bdi_thresh < bdi_stat_delta())
+				bdi_flush_all(bdi);
+
 			bdi_nr_reclaimable = bdi_stat(bdi, BDI_DIRTY) +
 						bdi_stat(bdi, BDI_UNSTABLE);
 			if (bdi_nr_reclaimable + bdi_stat(bdi, BDI_WRITEBACK) <=


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  reply	other threads:[~2007-04-04 12:43 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-04-03 14:40 [PATCH 0/6] per device dirty throttling -V2 Peter Zijlstra
2007-04-03 14:40 ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 1/6] mm: scalable bdi statistics counters Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-04  9:20   ` Miklos Szeredi
2007-04-04  9:20     ` Miklos Szeredi
2007-04-04  9:25     ` Peter Zijlstra
2007-04-04  9:25       ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 2/6] mm: count dirty pages per BDI Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 3/6] mm: count writeback " Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 4/6] mm: count unstable " Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 5/6] mm: expose BDI statistics in sysfs Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-03 14:40 ` [PATCH 6/6] mm: per device dirty threshold Peter Zijlstra
2007-04-03 14:40   ` Peter Zijlstra
2007-04-04  9:34   ` Miklos Szeredi
2007-04-04  9:34     ` Miklos Szeredi
2007-04-04 10:16     ` Peter Zijlstra
2007-04-04 10:16       ` Peter Zijlstra
2007-04-04 10:29       ` Miklos Szeredi
2007-04-04 10:29         ` Miklos Szeredi
2007-04-04 11:01         ` Peter Zijlstra
2007-04-04 11:01           ` Peter Zijlstra
2007-04-04 11:12           ` Miklos Szeredi
2007-04-04 11:12             ` Miklos Szeredi
2007-04-04 12:05             ` Peter Zijlstra
2007-04-04 12:05               ` Peter Zijlstra
2007-04-04 12:32               ` Miklos Szeredi
2007-04-04 12:32                 ` Miklos Szeredi
2007-04-04 12:43                 ` Peter Zijlstra [this message]
2007-04-04 12:43                   ` Peter Zijlstra
2007-04-04 20:03                   ` Peter Zijlstra
2007-04-04 20:03                     ` Peter Zijlstra

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1175690617.6483.85.camel@twins \
    --to=a.p.zijlstra@chello.nl \
    --cc=akpm@linux-foundation.org \
    --cc=dgc@sgi.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=miklos@szeredi.hu \
    --cc=neilb@suse.de \
    --cc=tomoki.sekiyama.qu@hitachi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.