percpu_counter: FBC_BATCH might be too big

PATCH In review Eric Dumazet 4 participants 9 messages Dec 5, 2008 – Dec 8, 2008

Subsystems (3)

Library code ext4 file system Per-CPU memory allocator

lore

Share & tools

lore

Eric Dumazet

author

2008-12-05 16:05 UTC #0 lore raw

PATCH 1/1 percpu_counter: FBC_BATCH might be too big

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS

Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.

A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())

We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

Signed-off-by Eric Dumazet <[email protected]>

---
 include/linux/percpu_counter.h |    6 +-----
 lib/percpu_counter.c           |   14 +++++++++++++-
 2 files changed, 14 insertions(+), 6 deletions(-)

include/linux/percpu_counter.h +1 -5

index 9007ccd..c42a184 100644
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int FBC_BATCH;
 
 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

lib/percpu_counter.c +13 -1

index a866389..2fcf943 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int FBC_BATCH __read_mostly = 32;
+EXPORT_SYMBOL(FBC_BATCH);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	FBC_BATCH = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,12 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#endif
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

Peter Zijlstra

Acked-by on v1 1/1

2008-12-05 16:40 UTC #1 lore raw

Quoted 9 lines · Eric Dumazet

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS
Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.
A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())
We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

Agreed, that is one of the reasons the proportion code doesn't use the FBC_BATCH stuff.

Acked-by Peter Zijlstra <[email protected]>

Quoted 66 lines

Signed-off-by Eric Dumazet <[email protected]>

---
 include/linux/percpu_counter.h |    6 +-----
 lib/percpu_counter.c           |   14 +++++++++++++-
 2 files changed, 14 insertions(+), 6 deletions(-)

include/linux/percpu_counter.h +1 -5

index 9007ccd..c42a184 100644
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int FBC_BATCH;
 
 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

lib/percpu_counter.c +13 -1

index a866389..2fcf943 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int FBC_BATCH __read_mostly = 32;
+EXPORT_SYMBOL(FBC_BATCH);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	FBC_BATCH = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,12 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#endif
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

David Miller

Acked-by on v1 1/1

2008-12-05 20:32 UTC #2 lore raw

In reply to Eric Dumazet

From: Eric Dumazet <[email protected]>
Date: Fri, 05 Dec 2008 17:05:16 +0100

Quoted 11 lines

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS
Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.
A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())
We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.
Signed-off-by Eric Dumazet <[email protected]>

The downside is now we must load this value in these common routines. But I think the gain outweights the loss so:

Acked-by David S. Miller <[email protected]>

Eric Dumazet

author

2008-12-07 09:36 UTC #3 lore raw

PATCH 1/1 percpu_counter: FBC_BATCH might be too big

In reply to Eric Dumazet

In this second version I guarded hotcpu_notifier() call by a #ifdef CONFIG_HOTPLUG_CPU

I wonder why hotcpu_notifier() is not a null op if !CONFIG_HOTPLUG_CPU

Thank you

[PATCH] percpu_counter: FBC_BATCH might be too big

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS

Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.

A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())

We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

Signed-off-by Eric Dumazet <[email protected]>

Acked-by David S. Miller <[email protected]>

Acked-by Peter Zijlstra <[email protected]>

---
 include/linux/percpu_counter.h |    6 +-----
 lib/percpu_counter.c           |   16 +++++++++++++++-
 2 files changed, 16 insertions(+), 6 deletions(-)

include/linux/percpu_counter.h +1 -5

index 9007ccd..c42a184 100644
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int FBC_BATCH;
 
 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

lib/percpu_counter.c +15 -1

index a866389..e21ce7c 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int FBC_BATCH __read_mostly = 32;
+EXPORT_SYMBOL(FBC_BATCH);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	FBC_BATCH = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#endif
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
+#ifdef CONFIG_HOTPLUG_CPU
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
+#endif
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

Andrew Morton

maintainer

2008-12-07 17:10 UTC #4 lore raw

Quoted 11 lines · On Sun, 07 Dec 2008 10:25:45 +0100 Eric Dumazet <[email protected]> wrote:

In this second version I guarded hotcpu_notifier() call by a #ifdef CONFIG_HOTPLUG_CPU
I wonder why hotcpu_notifier() is not a null op if !CONFIG_HOTPLUG_CPU
Thank you
[PATCH] percpu_counter: FBC_BATCH might be too big
For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS

Quoted 7 lines

Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.
A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())
We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

Yup, anything using NR_CPUS is probably wrong.

Quoted 5 lines

include/linux/percpu_counter.h +0 -0
index 9007ccd..c42a184 100644

include/linux/percpu_counter.h struct percpu_counter { · line 27

Quoted 9 lines

include/linux/percpu_counter.h +1 -5

@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int FBC_BATCH;

y:/usr/src/linux-2.6.28-rc7> grep -r FBC_BATCH . | wc -l 7

Can we fix this properly please? It should now become lower case, and it was a pretty dopey name anyway - now would be a good time to improve it. `percpu_counter_batch'?

Quoted 37 lines

include/linux/percpu_counter.h +0 -0

 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

lib/percpu_counter.c +12 -0

index a866389..e21ce7c 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int FBC_BATCH __read_mostly = 32;
+EXPORT_SYMBOL(FBC_BATCH);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	FBC_BATCH = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#endif

lib/percpu_counter.c static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb, · line 149

Quoted 10 lines

lib/percpu_counter.c +3 -1

 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
+#ifdef CONFIG_HOTPLUG_CPU
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
+#endif
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

hm, now what's going on in there? We should be able to drop the #ifdef CONFIG_HOTPLUG_CPU from lib/percpu_counter.c altogether. hotcpu_notifier() will do the right thing, and the compiler should generate no code for percpu_counter_hotcpu_callback() if CONFIG_HOTPLUG_CPU=n.

	$EDITOR  $(grep -l hotcpu_notifier */*.c)

and you'll see lots of code gets it right, and lots of code gets it wrong.

Eric Dumazet

author Acked-by David S. Miller Acked-by Peter Zijlstra

2008-12-07 18:40 UTC #5 lore raw

Quoted 41 lines · Andrew Morton

Quoted 11 lines · On Sun, 07 Dec 2008 10:25:45 +0100 Eric Dumazet <[email protected]> wrote:
In this second version I guarded hotcpu_notifier() call by a #ifdef CONFIG_HOTPLUG_CPU
I wonder why hotcpu_notifier() is not a null op if !CONFIG_HOTPLUG_CPU
Thank you
[PATCH] percpu_counter: FBC_BATCH might be too big
For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS
Quoted 7 lines
Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.
A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())
We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.
Yup, anything using NR_CPUS is probably wrong.
Quoted 5 lines
include/linux/percpu_counter.h +0 -0
index 9007ccd..c42a184 100644
include/linux/percpu_counter.h struct percpu_counter { · line 27
Quoted 9 lines
include/linux/percpu_counter.h +1 -5
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int FBC_BATCH;
y:/usr/src/linux-2.6.28-rc7> grep -r FBC_BATCH . | wc -l 7

Quoted 3 lines

Can we fix this properly please? It should now become lower case, and it was a pretty dopey name anyway - now would be a good time to improve it. `percpu_counter_batch'?

Yes

Quoted 55 lines

Quoted 38 lines

include/linux/percpu_counter.h +0 -0

 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

lib/percpu_counter.c +12 -0

index a866389..e21ce7c 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int FBC_BATCH __read_mostly = 32;
+EXPORT_SYMBOL(FBC_BATCH);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	FBC_BATCH = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#endif

lib/percpu_counter.c static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb, · line 149

Quoted 10 lines

lib/percpu_counter.c +3 -1

 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
+#ifdef CONFIG_HOTPLUG_CPU
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
+#endif
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

Quoted 5 lines

Do
	$EDITOR  $(grep -l hotcpu_notifier */*.c)
and you'll see lots of code gets it right, and lots of code gets it wrong.

I see nothing interesting, I must be blind.

lib/percpu_counter.c: In function 'percpu_counter_startup':
lib/percpu_counter.c:158: error: 'percpu_counter_hotcpu_callback' undeclared (first use in this function) lib/percpu_counter.c:158: error: (Each undeclared identifier is reported only once lib/percpu_counter.c:158: error: for each function it appears in.)
make[1]: *** [lib/percpu_counter.o] Error 1

static int __init percpu_counter_startup(void)

{
        compute_batch_value();
        hotcpu_notifier(percpu_counter_hotcpu_callback, 0); << ERROR >>
        return 0;
}

module_init(percpu_counter_startup);

# grep hotcpu_notifier include/linux/cpu.h
#define hotcpu_notifier(fn, pri) {                          \
#define hotcpu_notifier(fn, pri)    do { (void)(fn); } while (0)  << ERROR >>

If changed to :

static struct notifier_block __cpuinitdata percpu_counter_cpu_notifier = {
.notifier_call = percpu_counter_hotcpu_callback,
};
static int __init percpu_counter_startup(void)

{
        compute_batch_value();
        register_hotcpu_notifier(&percpu_counter_cpu_notifier);
        return 0;
}

module_init(percpu_counter_startup);

Then error is :
lib/percpu_counter.c:156: error: 'percpu_counter_hotcpu_callback' undeclared here (not in a function) make[1]: *** [lib/percpu_counter.o] Error 1

So the only way seems to add an ugly

#define percpu_counter_hotcpu_callback NULL

Is is what you name "the right thing" ?

Thanks

[PATCH] percpu_counter: FBC_BATCH should be a variable

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS

Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH.

A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())

We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

We rename FBC_BATCH to percpu_counter_batch since its not a constant anymore.

Signed-off-by Eric Dumazet <[email protected]>

Acked-by David S. Miller <[email protected]>

Acked-by Peter Zijlstra <[email protected]>

---
 fs/ext4/ext4.h                 |    6 +++---
 fs/ext4/inode.c                |    2 +-
 include/linux/percpu_counter.h |    8 ++------
 lib/percpu_counter.c           |   16 +++++++++++++++-
 4 files changed, 21 insertions(+), 11 deletions(-)

fs/ext4/ext4.h +3 -3

index b0537c8..6c46c64 100644
@@ -1225,11 +1225,11 @@ do {								\
 } while (0)
 
 #ifdef CONFIG_SMP
-/* Each CPU can accumulate FBC_BATCH blocks in their local
+/* Each CPU can accumulate percpu_counter_batch blocks in their local
  * counters. So we need to make sure we have free blocks more
- * than FBC_BATCH  * nr_cpu_ids. Also add a window of 4 times.
+ * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
  */
-#define EXT4_FREEBLOCKS_WATERMARK (4 * (FBC_BATCH * nr_cpu_ids))
+#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
 #else
 #define EXT4_FREEBLOCKS_WATERMARK 0
 #endif

fs/ext4/inode.c +1 -1

index be21a5a..db661e7 100644
@@ -2497,7 +2497,7 @@ static int ext4_nonda_switch(struct super_block *sb)
 	/*
 	 * switch to non delalloc mode if we are running low
 	 * on free block. The free block accounting via percpu
-	 * counters can get slightly wrong with FBC_BATCH getting
+	 * counters can get slightly wrong with percpu_counter_batch getting
 	 * accumulated on each CPU without updating global counters
 	 * Delalloc need an accurate free block accounting. So switch
 	 * to non delalloc when we are near to error range.

include/linux/percpu_counter.h +2 -6

index 9007ccd..99de7a3 100644
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int percpu_counter_batch;
 
 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

@@ -39,7 +35,7 @@ s64 __percpu_counter_sum(struct percpu_counter *fbc);
 
 static inline void percpu_counter_add(struct percpu_counter *fbc, s64 amount)
 {
-	__percpu_counter_add(fbc, amount, FBC_BATCH);
+	__percpu_counter_add(fbc, amount, percpu_counter_batch);
 }
 
 static inline s64 percpu_counter_sum_positive(struct percpu_counter *fbc)

lib/percpu_counter.c +15 -1

index a866389..519b877 100644
@@ -114,6 +114,16 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
+int percpu_counter_batch __read_mostly = 32;
+EXPORT_SYMBOL(percpu_counter_batch);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	percpu_counter_batch = max(32, nr*2);
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)

@@ -121,6 +131,7 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -139,11 +150,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 	mutex_unlock(&percpu_counters_lock);
 	return NOTIFY_OK;
 }
+#else
+#define percpu_counter_hotcpu_callback NULL
+#endif /* CONFIG_HOTPLUG_CPU */
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Andrew Morton

maintainer

2008-12-08 04:33 UTC #6 lore raw

Quoted 7 lines · On Sun, 07 Dec 2008 19:30:10 +0100 Eric Dumazet <[email protected]> wrote:

Quoted 6 lines
Do
	$EDITOR  $(grep -l hotcpu_notifier */*.c)
and you'll see lots of code gets it right, and lots of code gets it wrong.

Quoted 7 lines

I see nothing interesting, I must be blind.
lib/percpu_counter.c: In function 'percpu_counter_startup':
lib/percpu_counter.c:158: error: 'percpu_counter_hotcpu_callback' undeclared (first use in this function) lib/percpu_counter.c:158: error: (Each undeclared identifier is reported only once lib/percpu_counter.c:158: error: for each function it appears in.)
make[1]: *** [lib/percpu_counter.o] Error 1

Perhaps you still had percpu_counter_hotcpu_callback inside #ifdef.

That a look at kernel/workqueue.c, fs/buffer.c. No #ifdef CONFIG_HOTPLUG_CPU needed at all.

Eric Dumazet

author

2008-12-08 08:10 UTC #7 lore raw

PATCH 1/1 percpu_counter: FBC_BATCH should be a variable

Quoted 15 lines · Andrew Morton

Quoted 4 lines · On Sun, 07 Dec 2008 19:30:10 +0100 Eric Dumazet <[email protected]> wrote:
Quoted 4 lines
Do
	$EDITOR  $(grep -l hotcpu_notifier */*.c)
Quoted 8 lines
Quoted 1 line
and you'll see lots of code gets it right, and lots of code gets it wrong.
I see nothing interesting, I must be blind.
lib/percpu_counter.c: In function 'percpu_counter_startup':
lib/percpu_counter.c:158: error: 'percpu_counter_hotcpu_callback' undeclared (first use in this function) lib/percpu_counter.c:158: error: (Each undeclared identifier is reported only once lib/percpu_counter.c:158: error: for each function it appears in.)
make[1]: *** [lib/percpu_counter.o] Error 1

Quoted 4 lines

Perhaps you still had percpu_counter_hotcpu_callback inside #ifdef.
That a look at kernel/workqueue.c, fs/buffer.c. No #ifdef CONFIG_HOTPLUG_CPU needed at all.

We still need some #ifdef or we also must also delete them around "struct list_head list" in include/linux/percpu_counter.h and grow struct percpu_counter.

lib/percpu_counter.c: In function 'percpu_counter_hotcpu_callback':
lib/percpu_counter.c:137: error: 'struct percpu_counter' has no member named 'list' lib/percpu_counter.c:137: warning: type defaults to 'int' in declaration of '__mptr' lib/percpu_counter.c:137: warning: initialization from incompatible pointer type
...

Thank you Andrew

[PATCH] percpu_counter: FBC_BATCH should be a variable

For NR_CPUS >= 16 values, FBC_BATCH is 2*NR_CPUS

Considering more and more distros are using high NR_CPUS values, it makes sense to use a more sensible value for FBC_BATCH, and get rid of NR_CPUS.

A sensible value is 2*num_online_cpus(), with a minimum value of 32 (This minimum value helps branch prediction in __percpu_counter_add())

We already have a hotcpu notifier, so we can adjust FBC_BATCH dynamically.

We rename FBC_BATCH to percpu_counter_batch since its not a constant anymore.

Signed-off-by Eric Dumazet <[email protected]>

Acked-by David S. Miller <[email protected]>

Acked-by Peter Zijlstra <[email protected]>

---
 fs/ext4/ext4.h                 |    6 +++---
 fs/ext4/inode.c                |    2 +-
 include/linux/percpu_counter.h |    8 ++------
 lib/percpu_counter.c           |   18 ++++++++++++++----
 4 files changed, 20 insertions(+), 14 deletions(-)

fs/ext4/ext4.h +3 -3

index b0537c8..6c46c64 100644
@@ -1225,11 +1225,11 @@ do {								\
 } while (0)
 
 #ifdef CONFIG_SMP
-/* Each CPU can accumulate FBC_BATCH blocks in their local
+/* Each CPU can accumulate percpu_counter_batch blocks in their local
  * counters. So we need to make sure we have free blocks more
- * than FBC_BATCH  * nr_cpu_ids. Also add a window of 4 times.
+ * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
  */
-#define EXT4_FREEBLOCKS_WATERMARK (4 * (FBC_BATCH * nr_cpu_ids))
+#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
 #else
 #define EXT4_FREEBLOCKS_WATERMARK 0
 #endif

fs/ext4/inode.c +1 -1

index be21a5a..db661e7 100644
@@ -2497,7 +2497,7 @@ static int ext4_nonda_switch(struct super_block *sb)
 	/*
 	 * switch to non delalloc mode if we are running low
 	 * on free block. The free block accounting via percpu
-	 * counters can get slightly wrong with FBC_BATCH getting
+	 * counters can get slightly wrong with percpu_counter_batch getting
 	 * accumulated on each CPU without updating global counters
 	 * Delalloc need an accurate free block accounting. So switch
 	 * to non delalloc when we are near to error range.

include/linux/percpu_counter.h +2 -6

index 9007ccd..99de7a3 100644
@@ -24,11 +24,7 @@ struct percpu_counter {
 	s32 *counters;
 };
 
-#if NR_CPUS >= 16
-#define FBC_BATCH	(NR_CPUS*2)
-#else
-#define FBC_BATCH	(NR_CPUS*4)
-#endif
+extern int percpu_counter_batch;
 
 int percpu_counter_init(struct percpu_counter *fbc, s64 amount);
 int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount);

@@ -39,7 +35,7 @@ s64 __percpu_counter_sum(struct percpu_counter *fbc);
 
 static inline void percpu_counter_add(struct percpu_counter *fbc, s64 amount)
 {
-	__percpu_counter_add(fbc, amount, FBC_BATCH);
+	__percpu_counter_add(fbc, amount, percpu_counter_batch);
 }
 
 static inline s64 percpu_counter_sum_positive(struct percpu_counter *fbc)

lib/percpu_counter.c +14 -4

index a866389..e73eb5b 100644
@@ -9,10 +9,8 @@
 #include <linux/cpu.h>
 #include <linux/module.h>
 
-#ifdef CONFIG_HOTPLUG_CPU
 static LIST_HEAD(percpu_counters);
 static DEFINE_MUTEX(percpu_counters_lock);
-#endif
 
 void percpu_counter_set(struct percpu_counter *fbc, s64 amount)
 {

@@ -114,13 +112,24 @@ void percpu_counter_destroy(struct percpu_counter *fbc)
 }
 EXPORT_SYMBOL(percpu_counter_destroy);
 
-#ifdef CONFIG_HOTPLUG_CPU
+int percpu_counter_batch __read_mostly = 32;
+EXPORT_SYMBOL(percpu_counter_batch);
+
+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	percpu_counter_batch = max(32, nr*2);
+}
+
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)
 {
+#ifdef CONFIG_HOTPLUG_CPU
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -137,13 +146,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 		spin_unlock_irqrestore(&fbc->lock, flags);
 	}
 	mutex_unlock(&percpu_counters_lock);
+#endif
 	return NOTIFY_OK;
 }
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
 	return 0;
 }
 module_init(percpu_counter_startup);
-#endif

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Andrew Morton

maintainer

2008-12-08 08:19 UTC #8 lore raw

Re: [PATCH] percpu_counter: FBC_BATCH should be a variable

Quoted 21 lines · On Mon, 08 Dec 2008 09:05:04 +0100 Eric Dumazet <[email protected]> wrote:

Andrew Morton a __crit :
Quoted 15 lines · On Sun, 07 Dec 2008 19:30:10 +0100 Eric Dumazet <[email protected]> wrote:
Quoted 5 lines
Quoted 5 lines
Do
	$EDITOR  $(grep -l hotcpu_notifier */*.c)
Quoted 8 lines
Quoted 1 line
and you'll see lots of code gets it right, and lots of code gets it wrong.
I see nothing interesting, I must be blind.
lib/percpu_counter.c: In function 'percpu_counter_startup':
lib/percpu_counter.c:158: error: 'percpu_counter_hotcpu_callback' undeclared (first use in this function) lib/percpu_counter.c:158: error: (Each undeclared identifier is reported only once lib/percpu_counter.c:158: error: for each function it appears in.)
make[1]: *** [lib/percpu_counter.o] Error 1
Quoted 4 lines
Perhaps you still had percpu_counter_hotcpu_callback inside #ifdef.
That a look at kernel/workqueue.c, fs/buffer.c. No #ifdef CONFIG_HOTPLUG_CPU needed at all.

Quoted 9 lines

We still need some #ifdef or we also must also delete them around "struct list_head list" in include/linux/percpu_counter.h and grow struct percpu_counter.
lib/percpu_counter.c: In function 'percpu_counter_hotcpu_callback':
lib/percpu_counter.c:137: error: 'struct percpu_counter' has no member named 'list' lib/percpu_counter.c:137: warning: type defaults to 'int' in declaration of '__mptr' lib/percpu_counter.c:137: warning: initialization from incompatible pointer type
...

Yes, this conditionality:

struct percpu_counter {
	spinlock_t lock;
	s64 count;
#ifdef CONFIG_HOTPLUG_CPU
	struct list_head list;	/* All percpu_counters are on a list */
#endif
	s32 *counters;
};

mucked up our nice scheme. Oh well.

Quoted 23 lines

lib/percpu_counter.c +10 -0

+static void compute_batch_value(void)
+{
+	int nr = num_online_cpus();
+
+	percpu_counter_batch = max(32, nr*2);
+}
+
 static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 					unsigned long action, void *hcpu)
 {
+#ifdef CONFIG_HOTPLUG_CPU
 	unsigned int cpu;
 	struct percpu_counter *fbc;
 
+	compute_batch_value();
 	if (action != CPU_DEAD)
 		return NOTIFY_OK;

@@ -137,13 +146,14 @@ static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb,
 		spin_unlock_irqrestore(&fbc->lock, flags);
 	}
 	mutex_unlock(&percpu_counters_lock);
+#endif

lib/percpu_counter.c static int __cpuinit percpu_counter_hotcpu_callback(struct notifier_block *nb, · line 159

Quoted 10 lines

lib/percpu_counter.c +1 -0

 	return NOTIFY_OK;
 }
 
 static int __init percpu_counter_startup(void)
 {
+	compute_batch_value();
 	hotcpu_notifier(percpu_counter_hotcpu_callback, 0);
 	return 0;
 }
 module_init(percpu_counter_startup);

compute_batch_value() can be __cpuinit, but I think we're close enough here ;)