thr3ads.net - Ocfs2 devel - [PATCH V3 3/8] Cleancache: core ops functions and configuration [Jun 2010]

If this information is useful, please help other people find it:
Share via:

Dan Magenheimer

2010-Jun-21 23:19 UTC

[Ocfs2-devel] [PATCH V3 3/8] Cleancache: core ops functions and configuration

[PATCH V3 3/8] Cleancache: core ops functions and configuration

Cleancache core ops functions and configuration

Credits: Cleancache_ops design derived from Jeremy Fitzhardinge
design for tmem; sysfs code modelled after mm/ksm.c

Note that CONFIG_CLEANCACHE defaults to on; all hooks devolve
to a compare-function-pointer-to-NULL so performance impact should
be negligible, but can be reduced to zero impact if config'ed off.

Signed-off-by: Dan Magenheimer <dan.magenheimer at oracle.com>

Diffstat:
 include/linux/cleancache.h               |   88 ++++++++++
 mm/Kconfig                               |   22 ++
 mm/Makefile                              |    1 
 mm/cleancache.c                          |  169 +++++++++++++++++++++
 4 files changed, 280 insertions(+)

--- linux-2.6.35-rc2/include/linux/cleancache.h	1969-12-31 17:00:00.000000000
-0700
+++ linux-2.6.35-rc2-cleancache/include/linux/cleancache.h	2010-06-21
14:45:18.000000000 -0600
@@ -0,0 +1,88 @@
+#ifndef _LINUX_CLEANCACHE_H
+#define _LINUX_CLEANCACHE_H
+
+#include <linux/fs.h>
+#include <linux/mm.h>
+
+struct cleancache_ops {
+	int (*init_fs)(size_t);
+	int (*init_shared_fs)(char *uuid, size_t);
+	int (*get_page)(int, ino_t, pgoff_t, struct page *);
+	void (*put_page)(int, ino_t, pgoff_t, struct page *);
+	void (*flush_page)(int, ino_t, pgoff_t);
+	void (*flush_inode)(int, ino_t);
+	void (*flush_fs)(int);
+};
+
+extern struct cleancache_ops cleancache_ops;
+extern int __cleancache_get_page(struct page *);
+extern void __cleancache_put_page(struct page *);
+extern void __cleancache_flush_page(struct address_space *, struct page *);
+extern void __cleancache_flush_inode(struct address_space *);
+
+#ifdef CONFIG_CLEANCACHE
+#define cleancache_enabled (cleancache_ops.init_fs)
+#else
+#define cleancache_enabled (0)
+#endif
+
+/* called by a cleancache-enabled filesystem at time of mount */
+static inline int cleancache_init_fs(size_t pagesize)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = (*cleancache_ops.init_fs)(pagesize);
+	return ret;
+}
+
+/* called by a cleancache-enabled clustered filesystem at time of mount */
+static inline int cleancache_init_shared_fs(char *uuid, size_t pagesize)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = (*cleancache_ops.init_shared_fs)(uuid, pagesize);
+	return ret;
+}
+
+static inline int cleancache_get_page(struct page *page)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = __cleancache_get_page(page);
+	return ret;
+}
+
+static inline void cleancache_put_page(struct page *page)
+{
+	if (cleancache_enabled)
+		__cleancache_put_page(page);
+}
+
+static inline void cleancache_flush_page(struct address_space *mapping,
+					struct page *page)
+{
+	if (cleancache_enabled)
+		__cleancache_flush_page(mapping, page);
+}
+
+static inline void cleancache_flush_inode(struct address_space *mapping)
+{
+	if (cleancache_enabled)
+		__cleancache_flush_inode(mapping);
+}
+
+/*
+ * called by any cleancache-enabled filesystem at time of unmount;
+ * note that pool_id is surrendered and may be returned by a subsequent
+ * cleancache_init_fs or cleancache_init_shared_fs
+ */
+static inline void cleancache_flush_fs(int pool_id)
+{
+	if (cleancache_enabled && pool_id >= 0)
+		(*cleancache_ops.flush_fs)(pool_id);
+}
+
+#endif /* _LINUX_CLEANCACHE_H */
--- linux-2.6.35-rc2/mm/cleancache.c	1969-12-31 17:00:00.000000000 -0700
+++ linux-2.6.35-rc2-cleancache/mm/cleancache.c	2010-06-11 10:32:35.000000000
-0600
@@ -0,0 +1,169 @@
+/*
+ * Cleancache frontend
+ *
+ * This code provides the generic "frontend" layer to call a matching
+ * "backend" driver implementation of cleancache.  See
+ * Documentation/vm/cleancache.txt for more information.
+ *
+ * Copyright (C) 2009-2010 Oracle Corp. All rights reserved.
+ * Author: Dan Magenheimer
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.
+ */
+
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/cleancache.h>
+
+/*
+ * cleancache_ops contains the pointers to the cleancache "backend"
+ * implementation functions
+ */
+struct cleancache_ops cleancache_ops;
+EXPORT_SYMBOL(cleancache_ops);
+
+/* useful stats available in /sys/kernel/mm/cleancache */
+static unsigned long succ_gets;
+static unsigned long failed_gets;
+static unsigned long puts;
+static unsigned long flushes;
+
+/*
+ * "Get" data from cleancache associated with the poolid/inode/index
+ * that were specified when the data was put to cleanache and, if
+ * successful, use it to fill the specified page with data and return 0.
+ * The pageframe is unchanged and returns -1 if the get fails.
+ * Page must be locked by caller.
+ */
+int __cleancache_get_page(struct page *page)
+{
+	int ret = -1;
+	int pool_id;
+
+	VM_BUG_ON(!PageLocked(page));
+	pool_id = page->mapping->host->i_sb->cleancache_poolid;
+	if (pool_id >= 0) {
+		ret = (*cleancache_ops.get_page)(pool_id,
+						 page->mapping->host->i_ino,
+						 page->index,
+						 page);
+		if (ret == 0)
+			succ_gets++;
+		else
+			failed_gets++;
+	}
+	return ret;
+}
+EXPORT_SYMBOL(__cleancache_get_page);
+
+/*
+ * "Put" data from a page to cleancache and associate it with the
+ * (previously-obtained per-filesystem) poolid and the page's,
+ * inode and page index.  Page must be locked.  Note that a put_page
+ * always "succeeds", though a subsequent get_page may succeed or
fail.
+ */
+void __cleancache_put_page(struct page *page)
+{
+	int pool_id;
+
+	VM_BUG_ON(!PageLocked(page));
+	pool_id = page->mapping->host->i_sb->cleancache_poolid;
+	if (pool_id >= 0) {
+		(*cleancache_ops.put_page)(pool_id, page->mapping->host->i_ino,
+					   page->index, page);
+		puts++;
+	}
+}
+
+/*
+ * Flush any data from cleancache associated with the poolid and the
+ * page's inode and page index so that a subsequent "get" will
fail.
+ */
+void __cleancache_flush_page(struct address_space *mapping, struct page *page)
+{
+	int pool_id = mapping->host->i_sb->cleancache_poolid;
+
+	if (pool_id >= 0) {
+		VM_BUG_ON(!PageLocked(page));
+		(*cleancache_ops.flush_page)(pool_id, mapping->host->i_ino,
+					     page->index);
+		flushes++;
+	}
+}
+EXPORT_SYMBOL(__cleancache_flush_page);
+
+/*
+ * Flush all data from cleancache associated with the poolid and the
+ * mappings's inode so that all subsequent gets to this poolid/inode
+ * will fail.
+ */
+void __cleancache_flush_inode(struct address_space *mapping)
+{
+	int pool_id = mapping->host->i_sb->cleancache_poolid;
+
+	if (pool_id >= 0)
+		(*cleancache_ops.flush_inode)(pool_id, mapping->host->i_ino);
+}
+EXPORT_SYMBOL(__cleancache_flush_inode);
+
+#ifdef CONFIG_SYSFS
+
+/* see Documentation/ABI/xxx/sysfs-kernel-mm-cleancache */
+
+#define CLEANCACHE_ATTR_RO(_name) \
+	static struct kobj_attribute _name##_attr = __ATTR_RO(_name)
+
+static ssize_t succ_gets_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", succ_gets);
+}
+CLEANCACHE_ATTR_RO(succ_gets);
+
+static ssize_t failed_gets_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", failed_gets);
+}
+CLEANCACHE_ATTR_RO(failed_gets);
+
+static ssize_t puts_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", puts);
+}
+CLEANCACHE_ATTR_RO(puts);
+
+static ssize_t flushes_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", flushes);
+}
+CLEANCACHE_ATTR_RO(flushes);
+
+static struct attribute *cleancache_attrs[] = {
+	&succ_gets_attr.attr,
+	&failed_gets_attr.attr,
+	&puts_attr.attr,
+	&flushes_attr.attr,
+	NULL,
+};
+
+static struct attribute_group cleancache_attr_group = {
+	.attrs = cleancache_attrs,
+	.name = "cleancache",
+};
+
+#endif /* CONFIG_SYSFS */
+
+static int __init init_cleancache(void)
+{
+#ifdef CONFIG_SYSFS
+	int err;
+
+	err = sysfs_create_group(mm_kobj, &cleancache_attr_group);
+#endif /* CONFIG_SYSFS */
+	return 0;
+}
+module_init(init_cleancache)
--- linux-2.6.35-rc2/mm/Kconfig	2010-06-05 21:43:24.000000000 -0600
+++ linux-2.6.35-rc2-cleancache/mm/Kconfig	2010-06-11 09:01:37.000000000 -0600
@@ -298,3 +298,25 @@ config NOMMU_INITIAL_TRIM_EXCESS
 	  of 1 says that all excess pages should be trimmed.
 
 	  See Documentation/nommu-mmap.txt for more information.
+
+config CLEANCACHE
+	bool "Enable cleancache pseudo-RAM driver to cache clean pages"
+	default y
+	help
+ 	  Cleancache can be thought of as a page-granularity victim cache
+	  for clean pages that the kernel's pageframe replacement algorithm
+	  (PFRA) would like to keep around, but can't since there isn't enough
+	  memory.  So when the PFRA "evicts" a page, it first attempts to
put
+	  it into a synchronous concurrency-safe page-oriented pseudo-RAM
+	  device (such as Xen's Transcendent Memory, aka "tmem") which
is not
+	  directly accessible or addressable by the kernel and is of unknown
+	  (and possibly time-varying) size.  And when a cleancache-enabled
+	  filesystem wishes to access a page in a file on disk, it first
+	  checks cleancache to see if it already contains it; if it does,
+ 	  the page is copied into the kernel and a disk access is avoided.
+	  When a pseudo-RAM device is available, a significant I/O reduction
+	  may be achieved.  When none is available, all cleancache calls
+	  are reduced to a single pointer-compare-against-NULL resulting
+	  in a negligible performance hit.
+
+	  If unsure, say Y to enable cleancache
--- linux-2.6.35-rc2/mm/Makefile	2010-06-05 21:43:24.000000000 -0600
+++ linux-2.6.35-rc2-cleancache/mm/Makefile	2010-06-11 09:01:37.000000000 -0600
@@ -45,3 +45,4 @@ obj-$(CONFIG_MEMORY_FAILURE) += memory-f
 obj-$(CONFIG_HWPOISON_INJECT) += hwpoison-inject.o
 obj-$(CONFIG_DEBUG_KMEMLEAK) += kmemleak.o
 obj-$(CONFIG_DEBUG_KMEMLEAK_TEST) += kmemleak-test.o
+obj-$(CONFIG_CLEANCACHE) += cleancache.o

Dan Magenheimer

2010-Jun-21 23:19 UTC

head link

[PATCH V3 3/8] Cleancache: core ops functions and configuration

[PATCH V3 3/8] Cleancache: core ops functions and configuration

Cleancache core ops functions and configuration

Credits: Cleancache_ops design derived from Jeremy Fitzhardinge
design for tmem; sysfs code modelled after mm/ksm.c

Note that CONFIG_CLEANCACHE defaults to on; all hooks devolve
to a compare-function-pointer-to-NULL so performance impact should
be negligible, but can be reduced to zero impact if config''ed off.

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>

Diffstat:
 include/linux/cleancache.h               |   88 ++++++++++
 mm/Kconfig                               |   22 ++
 mm/Makefile                              |    1 
 mm/cleancache.c                          |  169 +++++++++++++++++++++
 4 files changed, 280 insertions(+)

--- linux-2.6.35-rc2/include/linux/cleancache.h	1969-12-31 17:00:00.000000000
-0700
+++ linux-2.6.35-rc2-cleancache/include/linux/cleancache.h	2010-06-21
14:45:18.000000000 -0600
@@ -0,0 +1,88 @@
+#ifndef _LINUX_CLEANCACHE_H
+#define _LINUX_CLEANCACHE_H
+
+#include <linux/fs.h>
+#include <linux/mm.h>
+
+struct cleancache_ops {
+	int (*init_fs)(size_t);
+	int (*init_shared_fs)(char *uuid, size_t);
+	int (*get_page)(int, ino_t, pgoff_t, struct page *);
+	void (*put_page)(int, ino_t, pgoff_t, struct page *);
+	void (*flush_page)(int, ino_t, pgoff_t);
+	void (*flush_inode)(int, ino_t);
+	void (*flush_fs)(int);
+};
+
+extern struct cleancache_ops cleancache_ops;
+extern int __cleancache_get_page(struct page *);
+extern void __cleancache_put_page(struct page *);
+extern void __cleancache_flush_page(struct address_space *, struct page *);
+extern void __cleancache_flush_inode(struct address_space *);
+
+#ifdef CONFIG_CLEANCACHE
+#define cleancache_enabled (cleancache_ops.init_fs)
+#else
+#define cleancache_enabled (0)
+#endif
+
+/* called by a cleancache-enabled filesystem at time of mount */
+static inline int cleancache_init_fs(size_t pagesize)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = (*cleancache_ops.init_fs)(pagesize);
+	return ret;
+}
+
+/* called by a cleancache-enabled clustered filesystem at time of mount */
+static inline int cleancache_init_shared_fs(char *uuid, size_t pagesize)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = (*cleancache_ops.init_shared_fs)(uuid, pagesize);
+	return ret;
+}
+
+static inline int cleancache_get_page(struct page *page)
+{
+	int ret = -1;
+
+	if (cleancache_enabled)
+		ret = __cleancache_get_page(page);
+	return ret;
+}
+
+static inline void cleancache_put_page(struct page *page)
+{
+	if (cleancache_enabled)
+		__cleancache_put_page(page);
+}
+
+static inline void cleancache_flush_page(struct address_space *mapping,
+					struct page *page)
+{
+	if (cleancache_enabled)
+		__cleancache_flush_page(mapping, page);
+}
+
+static inline void cleancache_flush_inode(struct address_space *mapping)
+{
+	if (cleancache_enabled)
+		__cleancache_flush_inode(mapping);
+}
+
+/*
+ * called by any cleancache-enabled filesystem at time of unmount;
+ * note that pool_id is surrendered and may be returned by a subsequent
+ * cleancache_init_fs or cleancache_init_shared_fs
+ */
+static inline void cleancache_flush_fs(int pool_id)
+{
+	if (cleancache_enabled && pool_id >= 0)
+		(*cleancache_ops.flush_fs)(pool_id);
+}
+
+#endif /* _LINUX_CLEANCACHE_H */
--- linux-2.6.35-rc2/mm/cleancache.c	1969-12-31 17:00:00.000000000 -0700
+++ linux-2.6.35-rc2-cleancache/mm/cleancache.c	2010-06-11 10:32:35.000000000
-0600
@@ -0,0 +1,169 @@
+/*
+ * Cleancache frontend
+ *
+ * This code provides the generic "frontend" layer to call a matching
+ * "backend" driver implementation of cleancache.  See
+ * Documentation/vm/cleancache.txt for more information.
+ *
+ * Copyright (C) 2009-2010 Oracle Corp. All rights reserved.
+ * Author: Dan Magenheimer
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.
+ */
+
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/cleancache.h>
+
+/*
+ * cleancache_ops contains the pointers to the cleancache "backend"
+ * implementation functions
+ */
+struct cleancache_ops cleancache_ops;
+EXPORT_SYMBOL(cleancache_ops);
+
+/* useful stats available in /sys/kernel/mm/cleancache */
+static unsigned long succ_gets;
+static unsigned long failed_gets;
+static unsigned long puts;
+static unsigned long flushes;
+
+/*
+ * "Get" data from cleancache associated with the poolid/inode/index
+ * that were specified when the data was put to cleanache and, if
+ * successful, use it to fill the specified page with data and return 0.
+ * The pageframe is unchanged and returns -1 if the get fails.
+ * Page must be locked by caller.
+ */
+int __cleancache_get_page(struct page *page)
+{
+	int ret = -1;
+	int pool_id;
+
+	VM_BUG_ON(!PageLocked(page));
+	pool_id = page->mapping->host->i_sb->cleancache_poolid;
+	if (pool_id >= 0) {
+		ret = (*cleancache_ops.get_page)(pool_id,
+						 page->mapping->host->i_ino,
+						 page->index,
+						 page);
+		if (ret == 0)
+			succ_gets++;
+		else
+			failed_gets++;
+	}
+	return ret;
+}
+EXPORT_SYMBOL(__cleancache_get_page);
+
+/*
+ * "Put" data from a page to cleancache and associate it with the
+ * (previously-obtained per-filesystem) poolid and the page''s,
+ * inode and page index.  Page must be locked.  Note that a put_page
+ * always "succeeds", though a subsequent get_page may succeed or
fail.
+ */
+void __cleancache_put_page(struct page *page)
+{
+	int pool_id;
+
+	VM_BUG_ON(!PageLocked(page));
+	pool_id = page->mapping->host->i_sb->cleancache_poolid;
+	if (pool_id >= 0) {
+		(*cleancache_ops.put_page)(pool_id, page->mapping->host->i_ino,
+					   page->index, page);
+		puts++;
+	}
+}
+
+/*
+ * Flush any data from cleancache associated with the poolid and the
+ * page''s inode and page index so that a subsequent "get"
will fail.
+ */
+void __cleancache_flush_page(struct address_space *mapping, struct page *page)
+{
+	int pool_id = mapping->host->i_sb->cleancache_poolid;
+
+	if (pool_id >= 0) {
+		VM_BUG_ON(!PageLocked(page));
+		(*cleancache_ops.flush_page)(pool_id, mapping->host->i_ino,
+					     page->index);
+		flushes++;
+	}
+}
+EXPORT_SYMBOL(__cleancache_flush_page);
+
+/*
+ * Flush all data from cleancache associated with the poolid and the
+ * mappings''s inode so that all subsequent gets to this poolid/inode
+ * will fail.
+ */
+void __cleancache_flush_inode(struct address_space *mapping)
+{
+	int pool_id = mapping->host->i_sb->cleancache_poolid;
+
+	if (pool_id >= 0)
+		(*cleancache_ops.flush_inode)(pool_id, mapping->host->i_ino);
+}
+EXPORT_SYMBOL(__cleancache_flush_inode);
+
+#ifdef CONFIG_SYSFS
+
+/* see Documentation/ABI/xxx/sysfs-kernel-mm-cleancache */
+
+#define CLEANCACHE_ATTR_RO(_name) \
+	static struct kobj_attribute _name##_attr = __ATTR_RO(_name)
+
+static ssize_t succ_gets_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", succ_gets);
+}
+CLEANCACHE_ATTR_RO(succ_gets);
+
+static ssize_t failed_gets_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", failed_gets);
+}
+CLEANCACHE_ATTR_RO(failed_gets);
+
+static ssize_t puts_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", puts);
+}
+CLEANCACHE_ATTR_RO(puts);
+
+static ssize_t flushes_show(struct kobject *kobj,
+			       struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%lu\n", flushes);
+}
+CLEANCACHE_ATTR_RO(flushes);
+
+static struct attribute *cleancache_attrs[] = {
+	&succ_gets_attr.attr,
+	&failed_gets_attr.attr,
+	&puts_attr.attr,
+	&flushes_attr.attr,
+	NULL,
+};
+
+static struct attribute_group cleancache_attr_group = {
+	.attrs = cleancache_attrs,
+	.name = "cleancache",
+};
+
+#endif /* CONFIG_SYSFS */
+
+static int __init init_cleancache(void)
+{
+#ifdef CONFIG_SYSFS
+	int err;
+
+	err = sysfs_create_group(mm_kobj, &cleancache_attr_group);
+#endif /* CONFIG_SYSFS */
+	return 0;
+}
+module_init(init_cleancache)
--- linux-2.6.35-rc2/mm/Kconfig	2010-06-05 21:43:24.000000000 -0600
+++ linux-2.6.35-rc2-cleancache/mm/Kconfig	2010-06-11 09:01:37.000000000 -0600
@@ -298,3 +298,25 @@ config NOMMU_INITIAL_TRIM_EXCESS
 	  of 1 says that all excess pages should be trimmed.
 
 	  See Documentation/nommu-mmap.txt for more information.
+
+config CLEANCACHE
+	bool "Enable cleancache pseudo-RAM driver to cache clean pages"
+	default y
+	help
+ 	  Cleancache can be thought of as a page-granularity victim cache
+	  for clean pages that the kernel''s pageframe replacement algorithm
+	  (PFRA) would like to keep around, but can''t since there
isn''t enough
+	  memory.  So when the PFRA "evicts" a page, it first attempts to
put
+	  it into a synchronous concurrency-safe page-oriented pseudo-RAM
+	  device (such as Xen''s Transcendent Memory, aka "tmem")
which is not
+	  directly accessible or addressable by the kernel and is of unknown
+	  (and possibly time-varying) size.  And when a cleancache-enabled
+	  filesystem wishes to access a page in a file on disk, it first
+	  checks cleancache to see if it already contains it; if it does,
+ 	  the page is copied into the kernel and a disk access is avoided.
+	  When a pseudo-RAM device is available, a significant I/O reduction
+	  may be achieved.  When none is available, all cleancache calls
+	  are reduced to a single pointer-compare-against-NULL resulting
+	  in a negligible performance hit.
+
+	  If unsure, say Y to enable cleancache
--- linux-2.6.35-rc2/mm/Makefile	2010-06-05 21:43:24.000000000 -0600
+++ linux-2.6.35-rc2-cleancache/mm/Makefile	2010-06-11 09:01:37.000000000 -0600
@@ -45,3 +45,4 @@ obj-$(CONFIG_MEMORY_FAILURE) += memory-f
 obj-$(CONFIG_HWPOISON_INJECT) += hwpoison-inject.o
 obj-$(CONFIG_DEBUG_KMEMLEAK) += kmemleak.o
 obj-$(CONFIG_DEBUG_KMEMLEAK_TEST) += kmemleak-test.o
+obj-$(CONFIG_CLEANCACHE) += cleancache.o
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs"
in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Dan Magenheimer

2010-Jun-22 16:20 UTC

head link

[Ocfs2-devel] [PATCH V3 3/8] Cleancache: core ops functions and configuration

Hi Christoph --

Thanks for the comments... replying to both in one reply.
> Subject: Re: [PATCH V3 0/8] Cleancache: overview
> 
> What all this fails to explain is that this actually is useful for?
See FAQ #1 in patch 1/8 (and repeated in patch 0/8).

But, in a few words, it's useful for maintaining a cache
of clean pages (for which the kernel has insufficient RAM)
in "other" RAM that's not directly accessible or addressable
by the kernel (such as hypervisor-owned RAM or kernel-owned
RAM that is secretly compressed).  Like the kernel's page
cache, use of cleancache avoids lots of disk reads ("refaults").
And when kernel RAM is scarce but "other" RAM is plentiful,
it avoids LOTS and LOTS of disk reads/refaults.
> Subject: Re: [PATCH V3 3/8] Cleancache: core ops functions and
> configuration
> 
> On Mon, Jun 21, 2010 at 04:19:39PM -0700, Dan Magenheimer wrote:
> > [PATCH V3 3/8] Cleancache: core ops functions and configuration
> >
> > Cleancache core ops functions and configuration
>
> NACK for code that just adds random hooks all over VFS and even
> individual FS code, does an EXPORT_SYMBOL but doesn't actually
> introduce any users.
There's a bit of a chicken and egg here.  Since cleancache
touches code owned by a number of maintainers, it made sense to
get that code reviewed first and respond to the feedback of those
maintainers.  So if this is the only remaining objection, we
will proceed next with introducing users.  See below for
a brief description.
> And even if it had users these would have to be damn good ones given
> how invasive it is.
I need to quibble with your definition of "invasive".  The patch
adds 43 lines of code (not counting comments and blank lines)
in VFS/filesystem code.  These lines have basically stayed the
same since 2.6.18 so the hooks are clearly not in code that
is rapidly changing... so maintenance should not be an issue.
The patch covers four filesystems and implements an interface
that provides both reading/writing to an "external" cache AND
coherency with that cache.

And all of these lines of code either compile into nothingness
when CONFIG_CLEANCACHE is off, or become compare function-pointer-
to-NULL if no user ("backend") claims the ops function.

I consider that very very NON-invasive.  (And should credit
Chris Mason for the hook placement and Jeremy Fitzhardinge
for the clean layering.)
> So what exactly is this going to help us?  Given your
> affiliation probably something Xen related, so some real use case would
> be interesting as well instead of just making Xen suck slightly less.
As I was typing this reply, I saw Nitin's reply talking
about zcache.  That's the non-Xen-related "real" use case...
it may even help KVM suck slightly less ;-)

Making-Xen-suck-slightly-less is another user... Transcendent
Memory ("tmem") has been in Xen for over a year now and distros
are already shipping an earlier version of cleancache that works
with Xen tmem.  Some shim code is required between cleancache and
Xen tmem, and this shim will live in the drivers/xen directory.
Excellent performance results for this "user" have been presented
at OLS'09 and LCA'10.

And the patch provides a very generic clean interface that
will likely be useful for future TBD forms of "other RAM".
While I honestly believe these additional users will eventually
appear, the first two users (zcache and Xen tmem) should be
sufficient to resolve your NACK.

Thanks,
Dan

Ocfs2 devel - Jun 2010 - [PATCH V3 3/8] Cleancache: core ops functions and configuration

[Ocfs2-devel] [PATCH V3 3/8] Cleancache: core ops functions and configuration

[PATCH V3 3/8] Cleancache: core ops functions and configuration

[Ocfs2-devel] [PATCH V3 3/8] Cleancache: core ops functions and configuration