2009-06-23 04:43:39

by Paul E. McKenney

[permalink] [raw]
Subject: [PATCH] v5 RCU: the bloatwatch edition

This patch is a version of RCU designed for (!SMP && EMBEDDED)
provided as a proof of concept of a small-footprint RCU implementation.
In particular, the implementation of synchronize_rcu() is extremely
lightweight and high performance. It passes rcutorture testing in each
of the four relevant configurations (combinations of NO_HZ and PREEMPT)
on x86. This saves 1263 bytes compared to Classic RCU, and more than
three kilobytes compared to Hierarchical RCU (updated to 2.6.30):

CONFIG_CLASSIC_RCU:

text data bss dec filename
544 32 16 592 kernel/rcupdate.o
1495 200 0 1695 kernel/rcuclassic.o
2287 Total (vs 2112 for v4)

CONFIG_TREE_RCU:

text data bss dec filename
544 32 16 592 kernel/rcupdate.o
3005 448 0 3453 kernel/rcutree.o
4045 Total (vs 3622 for v4)

CONFIG_TINY_RCU:

text data bss dec filename
506 32 16 554 kernel/rcupdate.o
442 28 0 470 kernel/rcutiny.o
1024 Total (vs 1041 for v4)

The above is for x86. Your mileage may vary on other platforms.

Changes from v4 (http://lkml.org/lkml/2009/5/2/91) include:

o Squeeze the size down a bit further by removing the
->completed field from struct rcu_ctrlblk.

o This permits synchronize_rcu() to become the empty function.
Previous concerns about rcutorture were unfounded, as
rcutorture correctly handles a constant value from
rcu_batches_completed() and rcu_batches_completed_bh().

Changes from v3 (http://lkml.org/lkml/2009/3/29/221) include:

o Changed rcu_batches_completed(), rcu_batches_completed_bh()
rcu_enter_nohz(), rcu_exit_nohz(), rcu_nmi_enter(), and
rcu_nmi_exit(), to be static inlines, as suggested by David
Howells. Doing this saves about 100 bytes from rcutiny.o.
(The numbers between v3 and this v4 of the patch are not directly
comparable, since they are against different versions of Linux.)

Changes from v2 (http://lkml.org/lkml/2009/2/3/333) include:

o Fix whitespace issues.

o Change short-circuit "||" operator to instead be "+" in order to
fix performance bug noted by "kraai" on LWN.

(http://lwn.net/Articles/324348/)

Changes from v1 (http://lkml.org/lkml/2009/1/13/440) include:

o This version depends on EMBEDDED as well as !SMP, as suggested
by Ingo.

o Updated rcu_needs_cpu() to unconditionally return zero,
permitting the CPU to enter dynticks-idle mode at any time.
This works because callbacks can be invoked upon entry to
dynticks-idle mode.

o I am now OK with this being included, based on a poll at the
Kernel Miniconf at linux.conf.au, where about ten people said
that they cared about saving 900 bytes on single-CPU systems.

o Applies to both mainline and tip/core/rcu.

Signed-off-by: Paul E. McKenney <[email protected]>
---

include/linux/hardirq.h | 52 +++++++++-
include/linux/rcupdate.h | 2
include/linux/rcutiny.h | 102 +++++++++++++++++++++
init/Kconfig | 7 +
kernel/Makefile | 1
kernel/rcupdate.c | 4
kernel/rcutiny.c | 227 +++++++++++++++++++++++++++++++++++++++++++++++
7 files changed, 388 insertions(+), 7 deletions(-)

diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h
index 4525747..b0d1e83 100644
--- a/include/linux/hardirq.h
+++ b/include/linux/hardirq.h
@@ -130,17 +130,55 @@ static inline void account_system_vtime(struct task_struct *tsk)
}
#endif

-#if defined(CONFIG_NO_HZ) && !defined(CONFIG_CLASSIC_RCU)
+#if !defined(CONFIG_NO_HZ) || defined(CONFIG_CLASSIC_RCU)
+
+static inline void rcu_irq_enter(void)
+{
+}
+
+static inline void rcu_irq_exit(void)
+{
+}
+
+static inline void rcu_nmi_enter(void)
+{
+}
+
+static inline void rcu_nmi_exit(void)
+{
+}
+
+#elif defined(CONFIG_TINY_RCU)
+
+extern void rcu_enter_nohz(void);
+extern void rcu_exit_nohz(void);
+
+static inline void rcu_irq_enter(void)
+{
+ rcu_exit_nohz();
+}
+
+static inline void rcu_irq_exit(void)
+{
+ rcu_enter_nohz();
+}
+
+static inline void rcu_nmi_enter(void)
+{
+}
+
+static inline void rcu_nmi_exit(void)
+{
+}
+
+#else
+
extern void rcu_irq_enter(void);
extern void rcu_irq_exit(void);
extern void rcu_nmi_enter(void);
extern void rcu_nmi_exit(void);
-#else
-# define rcu_irq_enter() do { } while (0)
-# define rcu_irq_exit() do { } while (0)
-# define rcu_nmi_enter() do { } while (0)
-# define rcu_nmi_exit() do { } while (0)
-#endif /* #if defined(CONFIG_NO_HZ) && !defined(CONFIG_CLASSIC_RCU) */
+
+#endif

/*
* It is safe to do non-atomic ops on ->hardirq_context,
diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 15fbb3c..bd91ffa 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -60,6 +60,8 @@ extern int rcu_scheduler_active;
#include <linux/rcutree.h>
#elif defined(CONFIG_PREEMPT_RCU)
#include <linux/rcupreempt.h>
+#elif CONFIG_TINY_RCU
+#include <linux/rcutiny.h>
#else
#error "Unknown RCU implementation specified to kernel configuration"
#endif /* #else #if defined(CONFIG_CLASSIC_RCU) */
diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h
new file mode 100644
index 0000000..02e9ef4
--- /dev/null
+++ b/include/linux/rcutiny.h
@@ -0,0 +1,102 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion, the Bloatwatch edition.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2008
+ *
+ * Author: Paul E. McKenney <[email protected]>
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * Documentation/RCU
+ */
+
+#ifndef __LINUX_TINY_H
+#define __LINUX_TINY_H
+
+#include <linux/cache.h>
+
+/* Global control variables for rcupdate callback mechanism. */
+struct rcu_ctrlblk {
+ struct rcu_head *rcucblist; /* List of pending callbacks (CBs). */
+ struct rcu_head **donetail; /* ->next pointer of last "done" CB. */
+ struct rcu_head **curtail; /* ->next pointer of last CB. */
+};
+
+extern struct rcu_ctrlblk rcu_ctrlblk;
+extern struct rcu_ctrlblk rcu_bh_ctrlblk;
+
+/*
+ * Return non-zero if there is RCU work remaining to be done.
+ */
+static inline int rcu_needs_cpu(int cpu)
+{
+ return 0;
+}
+
+void rcu_qsctr_inc(int cpu);
+void rcu_bh_qsctr_inc(int cpu);
+
+#define __rcu_read_lock() preempt_disable()
+#define __rcu_read_unlock() preempt_enable()
+#define __rcu_read_lock_bh() local_bh_disable()
+#define __rcu_read_unlock_bh() local_bh_enable()
+#define __synchronize_sched synchronize_rcu
+#define call_rcu_sched call_rcu
+
+#define rcu_init_sched() do { } while (0)
+extern void rcu_check_callbacks(int cpu, int user);
+extern void __rcu_init(void);
+/* extern void rcu_restart_cpu(int cpu); */
+
+/*
+ * Return the number of grace periods.
+ */
+static inline long rcu_batches_completed(void)
+{
+ return 0;
+}
+
+/*
+ * Return the number of bottom-half grace periods.
+ */
+static inline long rcu_batches_completed_bh(void)
+{
+ return 0;
+}
+
+static inline int rcu_pending(int cpu)
+{
+ return 1;
+}
+
+#ifdef CONFIG_NO_HZ
+
+extern void rcu_enter_nohz(void);
+extern void rcu_exit_nohz(void);
+
+#else /* #ifdef CONFIG_NO_HZ */
+
+static inline void rcu_enter_nohz(void)
+{
+}
+
+static inline void rcu_exit_nohz(void)
+{
+}
+
+#endif /* #else #ifdef CONFIG_NO_HZ */
+
+#endif /* __LINUX_RCUTINY_H */
diff --git a/init/Kconfig b/init/Kconfig
index 7be4d38..f088d51 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -337,6 +337,13 @@ config PREEMPT_RCU
now-naive assumptions about each RCU read-side critical section
remaining on a given CPU through its execution.

+config TINY_RCU
+ bool "Tiny-Memory RCU"
+ depends on !SMP && EMBEDDED
+ help
+ This option greatly reduces the memory footprint of RCU,
+ but is usable only on UP systems.
+
endchoice

config RCU_TRACE
diff --git a/kernel/Makefile b/kernel/Makefile
index 4242366..c76518c 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -81,6 +81,7 @@ obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
obj-$(CONFIG_CLASSIC_RCU) += rcuclassic.o
obj-$(CONFIG_TREE_RCU) += rcutree.o
obj-$(CONFIG_PREEMPT_RCU) += rcupreempt.o
+obj-$(CONFIG_TINY_RCU) += rcutiny.o
obj-$(CONFIG_TREE_RCU_TRACE) += rcutree_trace.o
obj-$(CONFIG_PREEMPT_RCU_TRACE) += rcupreempt_trace.o
obj-$(CONFIG_RELAY) += relay.o
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index a967c9f..a866940 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -74,6 +74,8 @@ void wakeme_after_rcu(struct rcu_head *head)
complete(&rcu->completion);
}

+#ifndef CONFIG_TINY_RCU
+
/**
* synchronize_rcu - wait until a grace period has elapsed.
*
@@ -98,6 +100,8 @@ void synchronize_rcu(void)
}
EXPORT_SYMBOL_GPL(synchronize_rcu);

+#endif /* #ifndef CONFIG_TINY_RCU */
+
static void rcu_barrier_callback(struct rcu_head *notused)
{
if (atomic_dec_and_test(&rcu_barrier_cpu_count))
diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c
new file mode 100644
index 0000000..94bef81
--- /dev/null
+++ b/kernel/rcutiny.c
@@ -0,0 +1,227 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion, the Bloatwatch edition.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2008
+ *
+ * Author: Paul E. McKenney <[email protected]>
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * Documentation/RCU
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/module.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/notifier.h>
+#include <linux/cpu.h>
+#include <linux/mutex.h>
+#include <linux/time.h>
+
+/* Definition for rcupdate control block. */
+struct rcu_ctrlblk rcu_ctrlblk = {
+ .rcucblist = NULL,
+ .donetail = &rcu_ctrlblk.rcucblist,
+ .curtail = &rcu_ctrlblk.rcucblist,
+};
+EXPORT_SYMBOL_GPL(rcu_ctrlblk);
+struct rcu_ctrlblk rcu_bh_ctrlblk = {
+ .rcucblist = NULL,
+ .donetail = &rcu_bh_ctrlblk.rcucblist,
+ .curtail = &rcu_bh_ctrlblk.rcucblist,
+};
+EXPORT_SYMBOL_GPL(rcu_bh_ctrlblk);
+
+#ifdef CONFIG_NO_HZ
+
+static long rcu_dynticks_nesting = 1;
+
+/*
+ * Enter dynticks-idle mode, which is an extended quiescent state
+ * if we have fully entered that mode (i.e., if the new value of
+ * dynticks_nesting is zero).
+ */
+void rcu_enter_nohz(void)
+{
+ if (--rcu_dynticks_nesting == 0)
+ rcu_qsctr_inc(0); /* implies rcu_bh_qsctr_inc(0) */
+}
+
+/*
+ * Exit dynticks-idle mode, so that we are no longer in an extended
+ * quiescent state.
+ */
+void rcu_exit_nohz(void)
+{
+ rcu_dynticks_nesting++;
+}
+
+#endif /* #ifdef CONFIG_NO_HZ */
+
+/*
+ * Helper function for rcu_qsctr_inc() and rcu_bh_qsctr_inc().
+ */
+static int rcu_qsctr_help(struct rcu_ctrlblk *rcp)
+{
+ if (rcp->rcucblist != NULL &&
+ rcp->donetail != rcp->curtail) {
+ rcp->donetail = rcp->curtail;
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * Record an rcu quiescent state. And an rcu_bh quiescent state while we
+ * are at it, given that any rcu quiescent state is also an rcu_bh
+ * quiescent state. Use "+" instead of "||" to defeat short circuiting.
+ */
+void rcu_qsctr_inc(int cpu)
+{
+ if (rcu_qsctr_help(&rcu_ctrlblk) + rcu_qsctr_help(&rcu_bh_ctrlblk))
+ raise_softirq(RCU_SOFTIRQ);
+}
+
+/*
+ * Record an rcu_bh quiescent state.
+ */
+void rcu_bh_qsctr_inc(int cpu)
+{
+ if (rcu_qsctr_help(&rcu_bh_ctrlblk))
+ raise_softirq(RCU_SOFTIRQ);
+}
+
+/*
+ * Check to see if the scheduling-clock interrupt came from an extended
+ * quiescent state, and, if so, tell RCU about it.
+ */
+void rcu_check_callbacks(int cpu, int user)
+{
+ if (!rcu_needs_cpu(0))
+ return; /* RCU doesn't need anything to be done. */
+ if (user ||
+ (idle_cpu(cpu) &&
+ !in_softirq() &&
+ hardirq_count() <= (1 << HARDIRQ_SHIFT)))
+ rcu_qsctr_inc(cpu);
+ else if (!in_softirq())
+ rcu_bh_qsctr_inc(cpu);
+}
+
+/*
+ * Helper function for rcu_process_callbacks() that operates on the
+ * specified rcu_ctrlkblk structure.
+ */
+static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp)
+{
+ unsigned long flags;
+ struct rcu_head *next, *list;
+
+ /* If no RCU callbacks ready to invoke, just return. */
+ if (&rcp->rcucblist == rcp->donetail)
+ return;
+
+ /* Move the ready-to-invoke callbacks to a local list. */
+ local_irq_save(flags);
+ list = rcp->rcucblist;
+ rcp->rcucblist = *rcp->donetail;
+ *rcp->donetail = NULL;
+ if (rcp->curtail == rcp->donetail)
+ rcp->curtail = &rcp->rcucblist;
+ rcp->donetail = &rcp->rcucblist;
+ local_irq_restore(flags);
+
+ /* Invoke the callbacks on the local list. */
+ while (list) {
+ next = list->next;
+ prefetch(next);
+ list->func(list);
+ list = next;
+ }
+}
+
+/*
+ * Invoke any callbacks whose grace period has completed.
+ */
+static void rcu_process_callbacks(struct softirq_action *unused)
+{
+ __rcu_process_callbacks(&rcu_ctrlblk);
+ __rcu_process_callbacks(&rcu_bh_ctrlblk);
+}
+
+/*
+ * Wait for a grace period to elapse. But it is illegal to invoke
+ * synchronize_rcu() from within an RCU read-side critical section.
+ * Therefore, any legal call to synchronize_rcu() is a quiescent
+ * state, and so on a UP system, synchronize_rcu() need do nothing.
+ *
+ * Cool, huh? (Due to Josh Triplett.)
+ */
+void synchronize_rcu(void)
+{
+}
+EXPORT_SYMBOL_GPL(synchronize_rcu);
+
+/*
+ * Helper function for call_rcu() and call_rcu_bh().
+ */
+static void __call_rcu(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu),
+ struct rcu_ctrlblk *rcp)
+{
+ unsigned long flags;
+
+ head->func = func;
+ head->next = NULL;
+ local_irq_save(flags);
+ *rcp->curtail = head;
+ rcp->curtail = &head->next;
+ local_irq_restore(flags);
+}
+
+/*
+ * Post an RCU callback to be invoked after the end of an RCU grace
+ * period. But since we have but one CPU, that would be after any
+ * quiescent state.
+ */
+void call_rcu(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu))
+{
+ __call_rcu(head, func, &rcu_ctrlblk);
+}
+EXPORT_SYMBOL_GPL(call_rcu);
+
+/*
+ * Post an RCU bottom-half callback to be invoked after any subsequent
+ * quiescent state.
+ */
+void call_rcu_bh(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu))
+{
+ __call_rcu(head, func, &rcu_bh_ctrlblk);
+}
+EXPORT_SYMBOL_GPL(call_rcu_bh);
+
+void __rcu_init(void)
+{
+ open_softirq(RCU_SOFTIRQ, rcu_process_callbacks);
+}


2009-06-23 09:13:31

by David Howells

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

Paul E. McKenney <[email protected]> wrote:

> This patch is a version of RCU designed for (!SMP && EMBEDDED)
> provided as a proof of concept of a small-footprint RCU implementation.
> In particular, the implementation of synchronize_rcu() is extremely
> lightweight and high performance. It passes rcutorture testing in each
> of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> on x86. This saves 1263 bytes compared to Classic RCU, and more than
> three kilobytes compared to Hierarchical RCU (updated to 2.6.30):

On FRV:

CONFIG_CLASSIC_RCU=y

text data bss dec hex filename
884 32 20 936 3a8 kernel/rcupdate.o
2616 184 0 2800 af0 kernel/rcuclassic.o

CONFIG_TREE_RCU=y

text data bss dec hex filename
884 32 20 936 3a8 kernel/rcupdate.o
4068 384 0 4452 1164 kernel/rcutree.o

CONFIG_TINY_RCU=y

text data bss dec hex filename
836 32 20 888 378 kernel/rcupdate.o
816 24 0 840 348 kernel/rcutiny.o

On MN10300:

CONFIG_CLASSIC_RCU=y

text data bss dec hex filename
900 28 48 976 3d0 kernel/rcupdate.o
1777 184 0 1961 7a9 kernel/rcuclassic.o

CONFIG_TREE_RCU=y

text data bss dec hex filename
900 28 48 976 3d0 kernel/rcupdate.o
2733 384 0 3117 c2d kernel/rcutree.o

CONFIG_PREEMPT_RCU=y

text data bss dec hex filename
961 28 48 1037 40d kernel/rcupdate.o
3314 128 8 3450 d7a kernel/rcupreempt.o

CONFIG_TINY_RCU=y

text data bss dec hex filename
865 28 48 941 3ad kernel/rcupdate.o
500 24 0 524 20c kernel/rcutiny.o

> Signed-off-by: Paul E. McKenney <[email protected]>

Acked-by: David Howells <[email protected]

2009-06-23 09:32:27

by Ingo Molnar

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition


* David Howells <[email protected]> wrote:

> Paul E. McKenney <[email protected]> wrote:
>
> > This patch is a version of RCU designed for (!SMP && EMBEDDED)
> > provided as a proof of concept of a small-footprint RCU implementation.
> > In particular, the implementation of synchronize_rcu() is extremely
> > lightweight and high performance. It passes rcutorture testing in each
> > of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> > on x86. This saves 1263 bytes compared to Classic RCU, and more than
> > three kilobytes compared to Hierarchical RCU (updated to 2.6.30):
>
> On FRV:
>
> CONFIG_CLASSIC_RCU=y
>
> text data bss dec hex filename
> 884 32 20 936 3a8 kernel/rcupdate.o
> 2616 184 0 2800 af0 kernel/rcuclassic.o
>
> CONFIG_TREE_RCU=y
>
> text data bss dec hex filename
> 884 32 20 936 3a8 kernel/rcupdate.o
> 4068 384 0 4452 1164 kernel/rcutree.o
>
> CONFIG_TINY_RCU=y
>
> text data bss dec hex filename
> 836 32 20 888 378 kernel/rcupdate.o
> 816 24 0 840 348 kernel/rcutiny.o
>
> On MN10300:
>
> CONFIG_CLASSIC_RCU=y
>
> text data bss dec hex filename
> 900 28 48 976 3d0 kernel/rcupdate.o
> 1777 184 0 1961 7a9 kernel/rcuclassic.o
>
> CONFIG_TREE_RCU=y
>
> text data bss dec hex filename
> 900 28 48 976 3d0 kernel/rcupdate.o
> 2733 384 0 3117 c2d kernel/rcutree.o
>
> CONFIG_PREEMPT_RCU=y
>
> text data bss dec hex filename
> 961 28 48 1037 40d kernel/rcupdate.o
> 3314 128 8 3450 d7a kernel/rcupreempt.o
>
> CONFIG_TINY_RCU=y
>
> text data bss dec hex filename
> 865 28 48 941 3ad kernel/rcupdate.o
> 500 24 0 524 20c kernel/rcutiny.o
>
> > Signed-off-by: Paul E. McKenney <[email protected]>
>
> Acked-by: David Howells <[email protected]

Ok, that's more convincing than 900 bytes on x86.

Paul, i guess this is .32 material anyway, right?

Since there's plenty of time, could we perhaps also do the
classic-rcu removal and integrate-rcupreempt-into-tree-rcu steps, to
reduce the (significant) config complexity perhaps?

Ingo

2009-06-23 12:58:39

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

On Tue, Jun 23, 2009 at 11:32:08AM +0200, Ingo Molnar wrote:
>
> * David Howells <[email protected]> wrote:
>
> > Paul E. McKenney <[email protected]> wrote:
> >
> > > This patch is a version of RCU designed for (!SMP && EMBEDDED)
> > > provided as a proof of concept of a small-footprint RCU implementation.
> > > In particular, the implementation of synchronize_rcu() is extremely
> > > lightweight and high performance. It passes rcutorture testing in each
> > > of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> > > on x86. This saves 1263 bytes compared to Classic RCU, and more than
> > > three kilobytes compared to Hierarchical RCU (updated to 2.6.30):
> >
> > On FRV:
> >
> > CONFIG_CLASSIC_RCU=y
> >
> > text data bss dec hex filename
> > 884 32 20 936 3a8 kernel/rcupdate.o
> > 2616 184 0 2800 af0 kernel/rcuclassic.o
> >
> > CONFIG_TREE_RCU=y
> >
> > text data bss dec hex filename
> > 884 32 20 936 3a8 kernel/rcupdate.o
> > 4068 384 0 4452 1164 kernel/rcutree.o
> >
> > CONFIG_TINY_RCU=y
> >
> > text data bss dec hex filename
> > 836 32 20 888 378 kernel/rcupdate.o
> > 816 24 0 840 348 kernel/rcutiny.o
> >
> > On MN10300:
> >
> > CONFIG_CLASSIC_RCU=y
> >
> > text data bss dec hex filename
> > 900 28 48 976 3d0 kernel/rcupdate.o
> > 1777 184 0 1961 7a9 kernel/rcuclassic.o
> >
> > CONFIG_TREE_RCU=y
> >
> > text data bss dec hex filename
> > 900 28 48 976 3d0 kernel/rcupdate.o
> > 2733 384 0 3117 c2d kernel/rcutree.o
> >
> > CONFIG_PREEMPT_RCU=y
> >
> > text data bss dec hex filename
> > 961 28 48 1037 40d kernel/rcupdate.o
> > 3314 128 8 3450 d7a kernel/rcupreempt.o
> >
> > CONFIG_TINY_RCU=y
> >
> > text data bss dec hex filename
> > 865 28 48 941 3ad kernel/rcupdate.o
> > 500 24 0 524 20c kernel/rcutiny.o
> >
> > > Signed-off-by: Paul E. McKenney <[email protected]>
> >
> > Acked-by: David Howells <[email protected]
>
> Ok, that's more convincing than 900 bytes on x86.

It used to be 900 bytes. In 2.6.30, it is 1263 bytes. ;-)

That said, I agree that FRV and MN10300 savings are more impressive.

> Paul, i guess this is .32 material anyway, right?

I am personally OK with it being .32, as long as David Howells,
Lennert Buytenhek, and Paul Mundt are OK with that schedule.

> Since there's plenty of time, could we perhaps also do the
> classic-rcu removal and integrate-rcupreempt-into-tree-rcu steps, to
> reduce the (significant) config complexity perhaps?

Classic-RCU removal most certainly. I am hoping to get the
integrate-rcupreempt-into-tree-rcu piece done in .32 timeframe, but
there is a chance that it may be a bit later.

Thanx, Paul

2009-06-23 13:47:34

by Paul Mundt

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

On Tue, Jun 23, 2009 at 05:58:32AM -0700, Paul E. McKenney wrote:
> On Tue, Jun 23, 2009 at 11:32:08AM +0200, Ingo Molnar wrote:
> > Paul, i guess this is .32 material anyway, right?
>
> I am personally OK with it being .32, as long as David Howells,
> Lennert Buytenhek, and Paul Mundt are OK with that schedule.
>
It's hard to see how this is going to cause any regressions, so I don't
see any problem with this going in to 2.6.31. OTOH, it's not too
pressing, so 2.6.32 is ok for me too.

2009-06-23 14:07:59

by David Howells

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

Paul E. McKenney <[email protected]> wrote:

> I am personally OK with it being .32, as long as David Howells,
> Lennert Buytenhek, and Paul Mundt are OK with that schedule.

Whilst it might be nice to get it in .31, I don't mind it being lodged in
linux-next till .32.

David

2009-06-23 15:44:47

by Lennert Buytenhek

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

On Tue, Jun 23, 2009 at 05:58:32AM -0700, Paul E. McKenney wrote:

> > > > This patch is a version of RCU designed for (!SMP && EMBEDDED)
> > > > provided as a proof of concept of a small-footprint RCU implementation.
> > > > In particular, the implementation of synchronize_rcu() is extremely
> > > > lightweight and high performance. It passes rcutorture testing in each
> > > > of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> > > > on x86. This saves 1263 bytes compared to Classic RCU, and more than
> > > > three kilobytes compared to Hierarchical RCU (updated to 2.6.30):
> > >
> > > On FRV:
> > >
> > > CONFIG_CLASSIC_RCU=y
> > >
> > > text data bss dec hex filename
> > > 884 32 20 936 3a8 kernel/rcupdate.o
> > > 2616 184 0 2800 af0 kernel/rcuclassic.o
> > >
> > > CONFIG_TREE_RCU=y
> > >
> > > text data bss dec hex filename
> > > 884 32 20 936 3a8 kernel/rcupdate.o
> > > 4068 384 0 4452 1164 kernel/rcutree.o
> > >
> > > CONFIG_TINY_RCU=y
> > >
> > > text data bss dec hex filename
> > > 836 32 20 888 378 kernel/rcupdate.o
> > > 816 24 0 840 348 kernel/rcutiny.o
> > >
> > > On MN10300:
> > >
> > > CONFIG_CLASSIC_RCU=y
> > >
> > > text data bss dec hex filename
> > > 900 28 48 976 3d0 kernel/rcupdate.o
> > > 1777 184 0 1961 7a9 kernel/rcuclassic.o
> > >
> > > CONFIG_TREE_RCU=y
> > >
> > > text data bss dec hex filename
> > > 900 28 48 976 3d0 kernel/rcupdate.o
> > > 2733 384 0 3117 c2d kernel/rcutree.o
> > >
> > > CONFIG_PREEMPT_RCU=y
> > >
> > > text data bss dec hex filename
> > > 961 28 48 1037 40d kernel/rcupdate.o
> > > 3314 128 8 3450 d7a kernel/rcupreempt.o
> > >
> > > CONFIG_TINY_RCU=y
> > >
> > > text data bss dec hex filename
> > > 865 28 48 941 3ad kernel/rcupdate.o
> > > 500 24 0 524 20c kernel/rcutiny.o
> > >
> > > > Signed-off-by: Paul E. McKenney <[email protected]>
> > >
> > > Acked-by: David Howells <[email protected]
> >
> > Ok, that's more convincing than 900 bytes on x86.
>
> It used to be 900 bytes. In 2.6.30, it is 1263 bytes. ;-)
>
> That said, I agree that FRV and MN10300 savings are more impressive.
>
> > Paul, i guess this is .32 material anyway, right?
>
> I am personally OK with it being .32, as long as David Howells,
> Lennert Buytenhek, and Paul Mundt are OK with that schedule.

I like the patch, as nearly all ARM systems in the field right now are
non-SMP. Russell King (CCd) is the ARM maintainer, so he should comment
on the proposed timeline.

2009-06-23 16:11:01

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH] v5 RCU: the bloatwatch edition

On Tue, Jun 23, 2009 at 05:47:01PM +0200, Lennert Buytenhek wrote:
> On Tue, Jun 23, 2009 at 05:58:32AM -0700, Paul E. McKenney wrote:
>
> > > > > This patch is a version of RCU designed for (!SMP && EMBEDDED)
> > > > > provided as a proof of concept of a small-footprint RCU implementation.
> > > > > In particular, the implementation of synchronize_rcu() is extremely
> > > > > lightweight and high performance. It passes rcutorture testing in each
> > > > > of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> > > > > on x86. This saves 1263 bytes compared to Classic RCU, and more than
> > > > > three kilobytes compared to Hierarchical RCU (updated to 2.6.30):
> > > >
> > > > On FRV:
> > > >
> > > > CONFIG_CLASSIC_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 884 32 20 936 3a8 kernel/rcupdate.o
> > > > 2616 184 0 2800 af0 kernel/rcuclassic.o
> > > >
> > > > CONFIG_TREE_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 884 32 20 936 3a8 kernel/rcupdate.o
> > > > 4068 384 0 4452 1164 kernel/rcutree.o
> > > >
> > > > CONFIG_TINY_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 836 32 20 888 378 kernel/rcupdate.o
> > > > 816 24 0 840 348 kernel/rcutiny.o
> > > >
> > > > On MN10300:
> > > >
> > > > CONFIG_CLASSIC_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 900 28 48 976 3d0 kernel/rcupdate.o
> > > > 1777 184 0 1961 7a9 kernel/rcuclassic.o
> > > >
> > > > CONFIG_TREE_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 900 28 48 976 3d0 kernel/rcupdate.o
> > > > 2733 384 0 3117 c2d kernel/rcutree.o
> > > >
> > > > CONFIG_PREEMPT_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 961 28 48 1037 40d kernel/rcupdate.o
> > > > 3314 128 8 3450 d7a kernel/rcupreempt.o
> > > >
> > > > CONFIG_TINY_RCU=y
> > > >
> > > > text data bss dec hex filename
> > > > 865 28 48 941 3ad kernel/rcupdate.o
> > > > 500 24 0 524 20c kernel/rcutiny.o
> > > >
> > > > > Signed-off-by: Paul E. McKenney <[email protected]>
> > > >
> > > > Acked-by: David Howells <[email protected]
> > >
> > > Ok, that's more convincing than 900 bytes on x86.
> >
> > It used to be 900 bytes. In 2.6.30, it is 1263 bytes. ;-)
> >
> > That said, I agree that FRV and MN10300 savings are more impressive.
> >
> > > Paul, i guess this is .32 material anyway, right?
> >
> > I am personally OK with it being .32, as long as David Howells,
> > Lennert Buytenhek, and Paul Mundt are OK with that schedule.
>
> I like the patch, as nearly all ARM systems in the field right now are
> non-SMP. Russell King (CCd) is the ARM maintainer, so he should comment
> on the proposed timeline.

OK, I will submit a patch against -tip. If Russell needs Bloatwatch
RCU more quickly, we still have my earlier v5 patch against 2.6.30.

Thanx, Paul

2009-06-23 16:12:35

by Paul E. McKenney

[permalink] [raw]
Subject: [PATCH -tip] v5 RCU: the bloatwatch edition

This patch is a -tip version of RCU designed for (!SMP && EMBEDDED)
provided as a proof of concept of a small-footprint RCU implementation.
In particular, the implementation of synchronize_rcu() is extremely
lightweight and high performance. It passes rcutorture testing in each
of the four relevant configurations (combinations of NO_HZ and PREEMPT)
on x86. This saves 1263 bytes compared to Classic RCU, and more than
three kilobytes compared to Hierarchical RCU (updated to 2.6.30):

CONFIG_CLASSIC_RCU:

text data bss dec filename
544 32 16 592 kernel/rcupdate.o
1495 200 0 1695 kernel/rcuclassic.o
2287 Total (vs 2112 for v4)

CONFIG_TREE_RCU:

text data bss dec filename
544 32 16 592 kernel/rcupdate.o
3005 448 0 3453 kernel/rcutree.o
4045 Total (vs 3622 for v4)

CONFIG_TINY_RCU:

text data bss dec filename
506 32 16 554 kernel/rcupdate.o
442 28 0 470 kernel/rcutiny.o
1024 Total (vs 1041 for v4)
1263 savings vs Classic

The above is for x86. Your mileage may vary on other platforms.

On FRV (courtesy of David Howells):

CONFIG_CLASSIC_RCU=y

text data bss dec filename
884 32 20 936 kernel/rcupdate.o
2616 184 0 2800 kernel/rcuclassic.o
3736 Total

CONFIG_TREE_RCU=y

text data bss dec filename
884 32 20 936 kernel/rcupdate.o
4068 384 0 4452 kernel/rcutree.o
5388 Total

CONFIG_TINY_RCU=y

text data bss dec filename
836 32 20 888 kernel/rcupdate.o
816 24 0 840 kernel/rcutiny.o
1728 Total, 2008 bytes saved vs Classic

On MN10300 (also courtesy of David Howells):

CONFIG_CLASSIC_RCU=y

text data bss dec filename
900 28 48 976 kernel/rcupdate.o
1777 184 0 1961 kernel/rcuclassic.o
2937 Total

CONFIG_TREE_RCU=y

text data bss dec filename
900 28 48 976 kernel/rcupdate.o
2733 384 0 3117 kernel/rcutree.o
4093 Total

CONFIG_PREEMPT_RCU=y

text data bss dec filename
961 28 48 1037 kernel/rcupdate.o
3314 128 8 3450 kernel/rcupreempt.o
4487 Total

CONFIG_TINY_RCU=y

text data bss dec filename
865 28 48 941 kernel/rcupdate.o
500 24 0 524 kernel/rcutiny.o
1465 Total, 1472 savings vs Classic

Ingo, please include in -tip slating for 2.6.32.

Changes from v4 (http://lkml.org/lkml/2009/5/2/91) include:

o Squeeze the size down a bit further by removing the
->completed field from struct rcu_ctrlblk.

o This permits synchronize_rcu() to become the empty function.
Previous concerns about rcutorture were unfounded, as
rcutorture correctly handles a constant value from
rcu_batches_completed() and rcu_batches_completed_bh().

Changes from v3 (http://lkml.org/lkml/2009/3/29/221) include:

o Changed rcu_batches_completed(), rcu_batches_completed_bh()
rcu_enter_nohz(), rcu_exit_nohz(), rcu_nmi_enter(), and
rcu_nmi_exit(), to be static inlines, as suggested by David
Howells. Doing this saves about 100 bytes from rcutiny.o.
(The numbers between v3 and this v4 of the patch are not directly
comparable, since they are against different versions of Linux.)

Changes from v2 (http://lkml.org/lkml/2009/2/3/333) include:

o Fix whitespace issues.

o Change short-circuit "||" operator to instead be "+" in order to
fix performance bug noted by "kraai" on LWN.

(http://lwn.net/Articles/324348/)

Changes from v1 (http://lkml.org/lkml/2009/1/13/440) include:

o This version depends on EMBEDDED as well as !SMP, as suggested
by Ingo.

o Updated rcu_needs_cpu() to unconditionally return zero,
permitting the CPU to enter dynticks-idle mode at any time.
This works because callbacks can be invoked upon entry to
dynticks-idle mode.

o I am now OK with this being included, based on a poll at the
Kernel Miniconf at linux.conf.au, where about ten people said
that they cared about saving 900 bytes on single-CPU systems.

o Applies to both mainline and tip/core/rcu.

[ Impact: reduce kernel size for UP builds ]

Signed-off-by: Paul E. McKenney <[email protected]>
Acked-by: David Howells <[email protected]>
---

include/linux/hardirq.h | 52 +++++++++-
include/linux/rcupdate.h | 2
include/linux/rcutiny.h | 102 +++++++++++++++++++++
init/Kconfig | 7 +
kernel/Makefile | 1
kernel/rcupdate.c | 4
kernel/rcutiny.c | 227 +++++++++++++++++++++++++++++++++++++++++++++++
7 files changed, 388 insertions(+), 7 deletions(-)

diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h
index 4525747..b0d1e83 100644
--- a/include/linux/hardirq.h
+++ b/include/linux/hardirq.h
@@ -130,17 +130,55 @@ static inline void account_system_vtime(struct task_struct *tsk)
}
#endif

-#if defined(CONFIG_NO_HZ) && !defined(CONFIG_CLASSIC_RCU)
+#if !defined(CONFIG_NO_HZ) || defined(CONFIG_CLASSIC_RCU)
+
+static inline void rcu_irq_enter(void)
+{
+}
+
+static inline void rcu_irq_exit(void)
+{
+}
+
+static inline void rcu_nmi_enter(void)
+{
+}
+
+static inline void rcu_nmi_exit(void)
+{
+}
+
+#elif defined(CONFIG_TINY_RCU)
+
+extern void rcu_enter_nohz(void);
+extern void rcu_exit_nohz(void);
+
+static inline void rcu_irq_enter(void)
+{
+ rcu_exit_nohz();
+}
+
+static inline void rcu_irq_exit(void)
+{
+ rcu_enter_nohz();
+}
+
+static inline void rcu_nmi_enter(void)
+{
+}
+
+static inline void rcu_nmi_exit(void)
+{
+}
+
+#else
+
extern void rcu_irq_enter(void);
extern void rcu_irq_exit(void);
extern void rcu_nmi_enter(void);
extern void rcu_nmi_exit(void);
-#else
-# define rcu_irq_enter() do { } while (0)
-# define rcu_irq_exit() do { } while (0)
-# define rcu_nmi_enter() do { } while (0)
-# define rcu_nmi_exit() do { } while (0)
-#endif /* #if defined(CONFIG_NO_HZ) && !defined(CONFIG_CLASSIC_RCU) */
+
+#endif

/*
* It is safe to do non-atomic ops on ->hardirq_context,
diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 15fbb3c..bd91ffa 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -60,6 +60,8 @@ extern int rcu_scheduler_active;
#include <linux/rcutree.h>
#elif defined(CONFIG_PREEMPT_RCU)
#include <linux/rcupreempt.h>
+#elif CONFIG_TINY_RCU
+#include <linux/rcutiny.h>
#else
#error "Unknown RCU implementation specified to kernel configuration"
#endif /* #else #if defined(CONFIG_CLASSIC_RCU) */
diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h
new file mode 100644
index 0000000..02e9ef4
--- /dev/null
+++ b/include/linux/rcutiny.h
@@ -0,0 +1,102 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion, the Bloatwatch edition.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2008
+ *
+ * Author: Paul E. McKenney <[email protected]>
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * Documentation/RCU
+ */
+
+#ifndef __LINUX_TINY_H
+#define __LINUX_TINY_H
+
+#include <linux/cache.h>
+
+/* Global control variables for rcupdate callback mechanism. */
+struct rcu_ctrlblk {
+ struct rcu_head *rcucblist; /* List of pending callbacks (CBs). */
+ struct rcu_head **donetail; /* ->next pointer of last "done" CB. */
+ struct rcu_head **curtail; /* ->next pointer of last CB. */
+};
+
+extern struct rcu_ctrlblk rcu_ctrlblk;
+extern struct rcu_ctrlblk rcu_bh_ctrlblk;
+
+/*
+ * Return non-zero if there is RCU work remaining to be done.
+ */
+static inline int rcu_needs_cpu(int cpu)
+{
+ return 0;
+}
+
+void rcu_qsctr_inc(int cpu);
+void rcu_bh_qsctr_inc(int cpu);
+
+#define __rcu_read_lock() preempt_disable()
+#define __rcu_read_unlock() preempt_enable()
+#define __rcu_read_lock_bh() local_bh_disable()
+#define __rcu_read_unlock_bh() local_bh_enable()
+#define __synchronize_sched synchronize_rcu
+#define call_rcu_sched call_rcu
+
+#define rcu_init_sched() do { } while (0)
+extern void rcu_check_callbacks(int cpu, int user);
+extern void __rcu_init(void);
+/* extern void rcu_restart_cpu(int cpu); */
+
+/*
+ * Return the number of grace periods.
+ */
+static inline long rcu_batches_completed(void)
+{
+ return 0;
+}
+
+/*
+ * Return the number of bottom-half grace periods.
+ */
+static inline long rcu_batches_completed_bh(void)
+{
+ return 0;
+}
+
+static inline int rcu_pending(int cpu)
+{
+ return 1;
+}
+
+#ifdef CONFIG_NO_HZ
+
+extern void rcu_enter_nohz(void);
+extern void rcu_exit_nohz(void);
+
+#else /* #ifdef CONFIG_NO_HZ */
+
+static inline void rcu_enter_nohz(void)
+{
+}
+
+static inline void rcu_exit_nohz(void)
+{
+}
+
+#endif /* #else #ifdef CONFIG_NO_HZ */
+
+#endif /* __LINUX_RCUTINY_H */
diff --git a/init/Kconfig b/init/Kconfig
index 7be4d38..f088d51 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -337,6 +337,13 @@ config PREEMPT_RCU
now-naive assumptions about each RCU read-side critical section
remaining on a given CPU through its execution.

+config TINY_RCU
+ bool "Tiny-Memory RCU"
+ depends on !SMP && EMBEDDED
+ help
+ This option greatly reduces the memory footprint of RCU,
+ but is usable only on UP systems.
+
endchoice

config RCU_TRACE
diff --git a/kernel/Makefile b/kernel/Makefile
index 4242366..c76518c 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -81,6 +81,7 @@ obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
obj-$(CONFIG_CLASSIC_RCU) += rcuclassic.o
obj-$(CONFIG_TREE_RCU) += rcutree.o
obj-$(CONFIG_PREEMPT_RCU) += rcupreempt.o
+obj-$(CONFIG_TINY_RCU) += rcutiny.o
obj-$(CONFIG_TREE_RCU_TRACE) += rcutree_trace.o
obj-$(CONFIG_PREEMPT_RCU_TRACE) += rcupreempt_trace.o
obj-$(CONFIG_RELAY) += relay.o
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index a967c9f..a866940 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -74,6 +74,8 @@ void wakeme_after_rcu(struct rcu_head *head)
complete(&rcu->completion);
}

+#ifndef CONFIG_TINY_RCU
+
/**
* synchronize_rcu - wait until a grace period has elapsed.
*
@@ -98,6 +100,8 @@ void synchronize_rcu(void)
}
EXPORT_SYMBOL_GPL(synchronize_rcu);

+#endif /* #ifndef CONFIG_TINY_RCU */
+
static void rcu_barrier_callback(struct rcu_head *notused)
{
if (atomic_dec_and_test(&rcu_barrier_cpu_count))
diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c
new file mode 100644
index 0000000..94bef81
--- /dev/null
+++ b/kernel/rcutiny.c
@@ -0,0 +1,227 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion, the Bloatwatch edition.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2008
+ *
+ * Author: Paul E. McKenney <[email protected]>
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * Documentation/RCU
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/module.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/notifier.h>
+#include <linux/cpu.h>
+#include <linux/mutex.h>
+#include <linux/time.h>
+
+/* Definition for rcupdate control block. */
+struct rcu_ctrlblk rcu_ctrlblk = {
+ .rcucblist = NULL,
+ .donetail = &rcu_ctrlblk.rcucblist,
+ .curtail = &rcu_ctrlblk.rcucblist,
+};
+EXPORT_SYMBOL_GPL(rcu_ctrlblk);
+struct rcu_ctrlblk rcu_bh_ctrlblk = {
+ .rcucblist = NULL,
+ .donetail = &rcu_bh_ctrlblk.rcucblist,
+ .curtail = &rcu_bh_ctrlblk.rcucblist,
+};
+EXPORT_SYMBOL_GPL(rcu_bh_ctrlblk);
+
+#ifdef CONFIG_NO_HZ
+
+static long rcu_dynticks_nesting = 1;
+
+/*
+ * Enter dynticks-idle mode, which is an extended quiescent state
+ * if we have fully entered that mode (i.e., if the new value of
+ * dynticks_nesting is zero).
+ */
+void rcu_enter_nohz(void)
+{
+ if (--rcu_dynticks_nesting == 0)
+ rcu_qsctr_inc(0); /* implies rcu_bh_qsctr_inc(0) */
+}
+
+/*
+ * Exit dynticks-idle mode, so that we are no longer in an extended
+ * quiescent state.
+ */
+void rcu_exit_nohz(void)
+{
+ rcu_dynticks_nesting++;
+}
+
+#endif /* #ifdef CONFIG_NO_HZ */
+
+/*
+ * Helper function for rcu_qsctr_inc() and rcu_bh_qsctr_inc().
+ */
+static int rcu_qsctr_help(struct rcu_ctrlblk *rcp)
+{
+ if (rcp->rcucblist != NULL &&
+ rcp->donetail != rcp->curtail) {
+ rcp->donetail = rcp->curtail;
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * Record an rcu quiescent state. And an rcu_bh quiescent state while we
+ * are at it, given that any rcu quiescent state is also an rcu_bh
+ * quiescent state. Use "+" instead of "||" to defeat short circuiting.
+ */
+void rcu_qsctr_inc(int cpu)
+{
+ if (rcu_qsctr_help(&rcu_ctrlblk) + rcu_qsctr_help(&rcu_bh_ctrlblk))
+ raise_softirq(RCU_SOFTIRQ);
+}
+
+/*
+ * Record an rcu_bh quiescent state.
+ */
+void rcu_bh_qsctr_inc(int cpu)
+{
+ if (rcu_qsctr_help(&rcu_bh_ctrlblk))
+ raise_softirq(RCU_SOFTIRQ);
+}
+
+/*
+ * Check to see if the scheduling-clock interrupt came from an extended
+ * quiescent state, and, if so, tell RCU about it.
+ */
+void rcu_check_callbacks(int cpu, int user)
+{
+ if (!rcu_needs_cpu(0))
+ return; /* RCU doesn't need anything to be done. */
+ if (user ||
+ (idle_cpu(cpu) &&
+ !in_softirq() &&
+ hardirq_count() <= (1 << HARDIRQ_SHIFT)))
+ rcu_qsctr_inc(cpu);
+ else if (!in_softirq())
+ rcu_bh_qsctr_inc(cpu);
+}
+
+/*
+ * Helper function for rcu_process_callbacks() that operates on the
+ * specified rcu_ctrlkblk structure.
+ */
+static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp)
+{
+ unsigned long flags;
+ struct rcu_head *next, *list;
+
+ /* If no RCU callbacks ready to invoke, just return. */
+ if (&rcp->rcucblist == rcp->donetail)
+ return;
+
+ /* Move the ready-to-invoke callbacks to a local list. */
+ local_irq_save(flags);
+ list = rcp->rcucblist;
+ rcp->rcucblist = *rcp->donetail;
+ *rcp->donetail = NULL;
+ if (rcp->curtail == rcp->donetail)
+ rcp->curtail = &rcp->rcucblist;
+ rcp->donetail = &rcp->rcucblist;
+ local_irq_restore(flags);
+
+ /* Invoke the callbacks on the local list. */
+ while (list) {
+ next = list->next;
+ prefetch(next);
+ list->func(list);
+ list = next;
+ }
+}
+
+/*
+ * Invoke any callbacks whose grace period has completed.
+ */
+static void rcu_process_callbacks(struct softirq_action *unused)
+{
+ __rcu_process_callbacks(&rcu_ctrlblk);
+ __rcu_process_callbacks(&rcu_bh_ctrlblk);
+}
+
+/*
+ * Wait for a grace period to elapse. But it is illegal to invoke
+ * synchronize_rcu() from within an RCU read-side critical section.
+ * Therefore, any legal call to synchronize_rcu() is a quiescent
+ * state, and so on a UP system, synchronize_rcu() need do nothing.
+ *
+ * Cool, huh? (Due to Josh Triplett.)
+ */
+void synchronize_rcu(void)
+{
+}
+EXPORT_SYMBOL_GPL(synchronize_rcu);
+
+/*
+ * Helper function for call_rcu() and call_rcu_bh().
+ */
+static void __call_rcu(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu),
+ struct rcu_ctrlblk *rcp)
+{
+ unsigned long flags;
+
+ head->func = func;
+ head->next = NULL;
+ local_irq_save(flags);
+ *rcp->curtail = head;
+ rcp->curtail = &head->next;
+ local_irq_restore(flags);
+}
+
+/*
+ * Post an RCU callback to be invoked after the end of an RCU grace
+ * period. But since we have but one CPU, that would be after any
+ * quiescent state.
+ */
+void call_rcu(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu))
+{
+ __call_rcu(head, func, &rcu_ctrlblk);
+}
+EXPORT_SYMBOL_GPL(call_rcu);
+
+/*
+ * Post an RCU bottom-half callback to be invoked after any subsequent
+ * quiescent state.
+ */
+void call_rcu_bh(struct rcu_head *head,
+ void (*func)(struct rcu_head *rcu))
+{
+ __call_rcu(head, func, &rcu_bh_ctrlblk);
+}
+EXPORT_SYMBOL_GPL(call_rcu_bh);
+
+void __rcu_init(void)
+{
+ open_softirq(RCU_SOFTIRQ, rcu_process_callbacks);
+}

2009-06-23 16:40:46

by Paul Mundt

[permalink] [raw]
Subject: Re: [PATCH -tip] v5 RCU: the bloatwatch edition

On Tue, Jun 23, 2009 at 09:12:17AM -0700, Paul E. McKenney wrote:
> This patch is a -tip version of RCU designed for (!SMP && EMBEDDED)
> provided as a proof of concept of a small-footprint RCU implementation.
> In particular, the implementation of synchronize_rcu() is extremely
> lightweight and high performance. It passes rcutorture testing in each
> of the four relevant configurations (combinations of NO_HZ and PREEMPT)
> on x86. This saves 1263 bytes compared to Classic RCU, and more than
> three kilobytes compared to Hierarchical RCU (updated to 2.6.30):
>
[snip]

> Signed-off-by: Paul E. McKenney <[email protected]>
> Acked-by: David Howells <[email protected]>

Acked-by: Paul Mundt <[email protected]>