2005-04-17 02:20:36 +04:00
|
|
|
/*
|
2005-05-02 22:25:02 +04:00
|
|
|
* Copyright (C) International Business Machines Corp., 2000-2005
|
2005-04-17 02:20:36 +04:00
|
|
|
* Portions Copyright (C) Christoph Hellwig, 2001-2002
|
|
|
|
*
|
|
|
|
* 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
|
2006-10-02 18:55:27 +04:00
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
2005-04-17 02:20:36 +04:00
|
|
|
* (at your option) any later version.
|
2006-10-02 18:55:27 +04:00
|
|
|
*
|
2005-04-17 02:20:36 +04:00
|
|
|
* 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
|
2006-10-02 18:55:27 +04:00
|
|
|
* along with this program; if not, write to the Free Software
|
2005-04-17 02:20:36 +04:00
|
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/fs.h>
|
2005-05-02 22:25:02 +04:00
|
|
|
#include <linux/mm.h>
|
2008-05-13 17:22:10 +04:00
|
|
|
#include <linux/module.h>
|
2005-05-02 22:25:02 +04:00
|
|
|
#include <linux/bio.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
|
|
|
#include <linux/slab.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
#include <linux/mempool.h>
|
2008-05-13 17:22:10 +04:00
|
|
|
#include <linux/seq_file.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include "jfs_incore.h"
|
|
|
|
#include "jfs_superblock.h"
|
|
|
|
#include "jfs_filsys.h"
|
|
|
|
#include "jfs_metapage.h"
|
|
|
|
#include "jfs_txnmgr.h"
|
|
|
|
#include "jfs_debug.h"
|
|
|
|
|
|
|
|
#ifdef CONFIG_JFS_STATISTICS
|
|
|
|
static struct {
|
|
|
|
uint pagealloc; /* # of page allocations */
|
|
|
|
uint pagefree; /* # of page frees */
|
|
|
|
uint lockwait; /* # of sleeping lock_metapage() calls */
|
|
|
|
} mpStat;
|
|
|
|
#endif
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
#define metapage_locked(mp) test_bit(META_locked, &(mp)->flag)
|
2007-11-26 23:58:10 +03:00
|
|
|
#define trylock_metapage(mp) test_and_set_bit_lock(META_locked, &(mp)->flag)
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
static inline void unlock_metapage(struct metapage *mp)
|
|
|
|
{
|
2007-11-26 23:58:10 +03:00
|
|
|
clear_bit_unlock(META_locked, &mp->flag);
|
2005-04-17 02:20:36 +04:00
|
|
|
wake_up(&mp->wait);
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
static inline void __lock_metapage(struct metapage *mp)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
|
|
|
DECLARE_WAITQUEUE(wait, current);
|
|
|
|
INCREMENT(mpStat.lockwait);
|
|
|
|
add_wait_queue_exclusive(&mp->wait, &wait);
|
|
|
|
do {
|
|
|
|
set_current_state(TASK_UNINTERRUPTIBLE);
|
|
|
|
if (metapage_locked(mp)) {
|
2005-05-02 22:25:02 +04:00
|
|
|
unlock_page(mp->page);
|
2007-01-18 06:18:35 +03:00
|
|
|
io_schedule();
|
2005-05-02 22:25:02 +04:00
|
|
|
lock_page(mp->page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
} while (trylock_metapage(mp));
|
|
|
|
__set_current_state(TASK_RUNNING);
|
|
|
|
remove_wait_queue(&mp->wait, &wait);
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
/*
|
|
|
|
* Must have mp->page locked
|
|
|
|
*/
|
2005-04-17 02:20:36 +04:00
|
|
|
static inline void lock_metapage(struct metapage *mp)
|
|
|
|
{
|
|
|
|
if (trylock_metapage(mp))
|
|
|
|
__lock_metapage(mp);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define METAPOOL_MIN_PAGES 32
|
2006-12-07 07:33:20 +03:00
|
|
|
static struct kmem_cache *metapage_cache;
|
2005-04-17 02:20:36 +04:00
|
|
|
static mempool_t *metapage_mempool;
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
#define MPS_PER_PAGE (PAGE_CACHE_SIZE >> L2PSIZE)
|
|
|
|
|
|
|
|
#if MPS_PER_PAGE > 1
|
|
|
|
|
|
|
|
struct meta_anchor {
|
|
|
|
int mp_count;
|
|
|
|
atomic_t io_count;
|
|
|
|
struct metapage *mp[MPS_PER_PAGE];
|
|
|
|
};
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
#define mp_anchor(page) ((struct meta_anchor *)page_private(page))
|
2005-05-02 22:25:02 +04:00
|
|
|
|
2008-01-11 01:04:25 +03:00
|
|
|
static inline struct metapage *page_to_mp(struct page *page, int offset)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
|
|
|
if (!PagePrivate(page))
|
|
|
|
return NULL;
|
|
|
|
return mp_anchor(page)->mp[offset >> L2PSIZE];
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int insert_metapage(struct page *page, struct metapage *mp)
|
|
|
|
{
|
|
|
|
struct meta_anchor *a;
|
|
|
|
int index;
|
|
|
|
int l2mp_blocks; /* log2 blocks per metapage */
|
|
|
|
|
|
|
|
if (PagePrivate(page))
|
|
|
|
a = mp_anchor(page);
|
|
|
|
else {
|
2006-02-23 18:47:13 +03:00
|
|
|
a = kzalloc(sizeof(struct meta_anchor), GFP_NOFS);
|
2005-05-02 22:25:02 +04:00
|
|
|
if (!a)
|
|
|
|
return -ENOMEM;
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
set_page_private(page, (unsigned long)a);
|
2005-05-02 22:25:02 +04:00
|
|
|
SetPagePrivate(page);
|
|
|
|
kmap(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (mp) {
|
|
|
|
l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
|
|
|
|
index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);
|
|
|
|
a->mp_count++;
|
|
|
|
a->mp[index] = mp;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void remove_metapage(struct page *page, struct metapage *mp)
|
|
|
|
{
|
|
|
|
struct meta_anchor *a = mp_anchor(page);
|
|
|
|
int l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
|
|
|
|
int index;
|
|
|
|
|
|
|
|
index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);
|
|
|
|
|
|
|
|
BUG_ON(a->mp[index] != mp);
|
|
|
|
|
|
|
|
a->mp[index] = NULL;
|
|
|
|
if (--a->mp_count == 0) {
|
|
|
|
kfree(a);
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
set_page_private(page, 0);
|
2005-05-02 22:25:02 +04:00
|
|
|
ClearPagePrivate(page);
|
|
|
|
kunmap(page);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void inc_io(struct page *page)
|
|
|
|
{
|
|
|
|
atomic_inc(&mp_anchor(page)->io_count);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dec_io(struct page *page, void (*handler) (struct page *))
|
|
|
|
{
|
|
|
|
if (atomic_dec_and_test(&mp_anchor(page)->io_count))
|
|
|
|
handler(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
#else
|
2008-01-11 01:04:25 +03:00
|
|
|
static inline struct metapage *page_to_mp(struct page *page, int offset)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
return PagePrivate(page) ? (struct metapage *)page_private(page) : NULL;
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline int insert_metapage(struct page *page, struct metapage *mp)
|
|
|
|
{
|
|
|
|
if (mp) {
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
set_page_private(page, (unsigned long)mp);
|
2005-05-02 22:25:02 +04:00
|
|
|
SetPagePrivate(page);
|
|
|
|
kmap(page);
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void remove_metapage(struct page *page, struct metapage *mp)
|
|
|
|
{
|
[PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.
This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock. (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)
In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.
Splitting the lock is not quite for free: another cacheline access. Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS. But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.
There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 04:16:40 +03:00
|
|
|
set_page_private(page, 0);
|
2005-05-02 22:25:02 +04:00
|
|
|
ClearPagePrivate(page);
|
|
|
|
kunmap(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define inc_io(page) do {} while(0)
|
|
|
|
#define dec_io(page, handler) handler(page)
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
2005-10-21 11:20:48 +04:00
|
|
|
static inline struct metapage *alloc_metapage(gfp_t gfp_mask)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2015-04-16 02:14:11 +03:00
|
|
|
struct metapage *mp = mempool_alloc(metapage_mempool, gfp_mask);
|
|
|
|
|
|
|
|
if (mp) {
|
|
|
|
mp->lid = 0;
|
|
|
|
mp->lsn = 0;
|
|
|
|
mp->data = NULL;
|
|
|
|
mp->clsn = 0;
|
|
|
|
mp->log = NULL;
|
|
|
|
init_waitqueue_head(&mp->wait);
|
|
|
|
}
|
|
|
|
return mp;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void free_metapage(struct metapage *mp)
|
|
|
|
{
|
|
|
|
mempool_free(mp, metapage_mempool);
|
|
|
|
}
|
|
|
|
|
|
|
|
int __init metapage_init(void)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Allocate the metapage structures
|
|
|
|
*/
|
|
|
|
metapage_cache = kmem_cache_create("jfs_mp", sizeof(struct metapage),
|
2015-04-16 02:14:11 +03:00
|
|
|
0, 0, NULL);
|
2005-04-17 02:20:36 +04:00
|
|
|
if (metapage_cache == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2006-03-26 13:37:50 +04:00
|
|
|
metapage_mempool = mempool_create_slab_pool(METAPOOL_MIN_PAGES,
|
|
|
|
metapage_cache);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
if (metapage_mempool == NULL) {
|
|
|
|
kmem_cache_destroy(metapage_cache);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void metapage_exit(void)
|
|
|
|
{
|
|
|
|
mempool_destroy(metapage_mempool);
|
|
|
|
kmem_cache_destroy(metapage_cache);
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
static inline void drop_metapage(struct page *page, struct metapage *mp)
|
|
|
|
{
|
|
|
|
if (mp->count || mp->nohomeok || test_bit(META_dirty, &mp->flag) ||
|
|
|
|
test_bit(META_io, &mp->flag))
|
|
|
|
return;
|
|
|
|
remove_metapage(page, mp);
|
|
|
|
INCREMENT(mpStat.pagefree);
|
|
|
|
free_metapage(mp);
|
|
|
|
}
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
2005-05-02 22:25:02 +04:00
|
|
|
* Metapage address space operations
|
2005-04-17 02:20:36 +04:00
|
|
|
*/
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
static sector_t metapage_get_blocks(struct inode *inode, sector_t lblock,
|
2008-01-11 01:04:25 +03:00
|
|
|
int *len)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
int rc = 0;
|
|
|
|
int xflag;
|
|
|
|
s64 xaddr;
|
2006-09-27 12:50:49 +04:00
|
|
|
sector_t file_blocks = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
|
2005-05-02 22:25:02 +04:00
|
|
|
inode->i_blkbits;
|
|
|
|
|
|
|
|
if (lblock >= file_blocks)
|
|
|
|
return 0;
|
|
|
|
if (lblock + *len > file_blocks)
|
|
|
|
*len = file_blocks - lblock;
|
|
|
|
|
|
|
|
if (inode->i_ino) {
|
|
|
|
rc = xtLookup(inode, (s64)lblock, *len, &xflag, &xaddr, len, 0);
|
|
|
|
if ((rc == 0) && *len)
|
|
|
|
lblock = (sector_t)xaddr;
|
|
|
|
else
|
|
|
|
lblock = 0;
|
|
|
|
} /* else no mapping */
|
|
|
|
|
|
|
|
return lblock;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
static void last_read_complete(struct page *page)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
if (!PageError(page))
|
|
|
|
SetPageUptodate(page);
|
|
|
|
unlock_page(page);
|
|
|
|
}
|
|
|
|
|
2015-07-20 16:29:37 +03:00
|
|
|
static void metapage_read_end_io(struct bio *bio)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
|
|
|
struct page *page = bio->bi_private;
|
|
|
|
|
2015-07-20 16:29:37 +03:00
|
|
|
if (bio->bi_error) {
|
2005-05-02 22:25:02 +04:00
|
|
|
printk(KERN_ERR "metapage_read_end_io: I/O error\n");
|
|
|
|
SetPageError(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
dec_io(page, last_read_complete);
|
|
|
|
bio_put(bio);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
static void remove_from_logsync(struct metapage *mp)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
struct jfs_log *log = mp->log;
|
|
|
|
unsigned long flags;
|
|
|
|
/*
|
|
|
|
* This can race. Recheck that log hasn't been set to null, and after
|
|
|
|
* acquiring logsync lock, recheck lsn
|
|
|
|
*/
|
|
|
|
if (!log)
|
|
|
|
return;
|
|
|
|
|
|
|
|
LOGSYNC_LOCK(log, flags);
|
|
|
|
if (mp->lsn) {
|
|
|
|
mp->log = NULL;
|
|
|
|
mp->lsn = 0;
|
|
|
|
mp->clsn = 0;
|
|
|
|
log->count--;
|
|
|
|
list_del(&mp->synclist);
|
|
|
|
}
|
|
|
|
LOGSYNC_UNLOCK(log, flags);
|
|
|
|
}
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
static void last_write_complete(struct page *page)
|
|
|
|
{
|
|
|
|
struct metapage *mp;
|
|
|
|
unsigned int offset;
|
|
|
|
|
|
|
|
for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
|
|
|
|
mp = page_to_mp(page, offset);
|
|
|
|
if (mp && test_bit(META_io, &mp->flag)) {
|
|
|
|
if (mp->lsn)
|
|
|
|
remove_from_logsync(mp);
|
|
|
|
clear_bit(META_io, &mp->flag);
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* I'd like to call drop_metapage here, but I don't think it's
|
|
|
|
* safe unless I have the page locked
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
end_page_writeback(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2015-07-20 16:29:37 +03:00
|
|
|
static void metapage_write_end_io(struct bio *bio)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
struct page *page = bio->bi_private;
|
|
|
|
|
|
|
|
BUG_ON(!PagePrivate(page));
|
|
|
|
|
2015-07-20 16:29:37 +03:00
|
|
|
if (bio->bi_error) {
|
2005-05-02 22:25:02 +04:00
|
|
|
printk(KERN_ERR "metapage_write_end_io: I/O error\n");
|
|
|
|
SetPageError(page);
|
|
|
|
}
|
|
|
|
dec_io(page, last_write_complete);
|
|
|
|
bio_put(bio);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int metapage_writepage(struct page *page, struct writeback_control *wbc)
|
|
|
|
{
|
|
|
|
struct bio *bio = NULL;
|
2008-01-11 01:04:25 +03:00
|
|
|
int block_offset; /* block offset of mp within page */
|
2005-05-02 22:25:02 +04:00
|
|
|
struct inode *inode = page->mapping->host;
|
2008-01-11 01:04:25 +03:00
|
|
|
int blocks_per_mp = JFS_SBI(inode->i_sb)->nbperpage;
|
|
|
|
int len;
|
|
|
|
int xlen;
|
2005-05-02 22:25:02 +04:00
|
|
|
struct metapage *mp;
|
|
|
|
int redirty = 0;
|
|
|
|
sector_t lblock;
|
2008-01-03 22:09:33 +03:00
|
|
|
int nr_underway = 0;
|
2005-05-02 22:25:02 +04:00
|
|
|
sector_t pblock;
|
|
|
|
sector_t next_block = 0;
|
|
|
|
sector_t page_start;
|
|
|
|
unsigned long bio_bytes = 0;
|
|
|
|
unsigned long bio_offset = 0;
|
2008-01-11 01:04:25 +03:00
|
|
|
int offset;
|
2009-01-30 23:09:06 +03:00
|
|
|
int bad_blocks = 0;
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
page_start = (sector_t)page->index <<
|
|
|
|
(PAGE_CACHE_SHIFT - inode->i_blkbits);
|
|
|
|
BUG_ON(!PageLocked(page));
|
|
|
|
BUG_ON(PageWriteback(page));
|
2008-01-03 22:09:33 +03:00
|
|
|
set_page_writeback(page);
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
|
|
|
|
mp = page_to_mp(page, offset);
|
|
|
|
|
|
|
|
if (!mp || !test_bit(META_dirty, &mp->flag))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (mp->nohomeok && !test_bit(META_forcewrite, &mp->flag)) {
|
|
|
|
redirty = 1;
|
2005-10-04 00:32:11 +04:00
|
|
|
/*
|
|
|
|
* Make sure this page isn't blocked indefinitely.
|
|
|
|
* If the journal isn't undergoing I/O, push it
|
|
|
|
*/
|
|
|
|
if (mp->log && !(mp->log->cflag & logGC_PAGEOUT))
|
|
|
|
jfs_flush_journal(mp->log, 0);
|
2005-05-02 22:25:02 +04:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
clear_bit(META_dirty, &mp->flag);
|
2009-01-30 23:09:06 +03:00
|
|
|
set_bit(META_io, &mp->flag);
|
2005-05-02 22:25:02 +04:00
|
|
|
block_offset = offset >> inode->i_blkbits;
|
|
|
|
lblock = page_start + block_offset;
|
|
|
|
if (bio) {
|
|
|
|
if (xlen && lblock == next_block) {
|
|
|
|
/* Contiguous, in memory & on disk */
|
|
|
|
len = min(xlen, blocks_per_mp);
|
|
|
|
xlen -= len;
|
|
|
|
bio_bytes += len << inode->i_blkbits;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/* Not contiguous */
|
|
|
|
if (bio_add_page(bio, page, bio_bytes, bio_offset) <
|
|
|
|
bio_bytes)
|
|
|
|
goto add_failed;
|
|
|
|
/*
|
|
|
|
* Increment counter before submitting i/o to keep
|
|
|
|
* count from hitting zero before we're through
|
|
|
|
*/
|
|
|
|
inc_io(page);
|
2013-10-12 02:44:27 +04:00
|
|
|
if (!bio->bi_iter.bi_size)
|
2005-05-02 22:25:02 +04:00
|
|
|
goto dump_bio;
|
|
|
|
submit_bio(WRITE, bio);
|
2008-01-03 22:09:33 +03:00
|
|
|
nr_underway++;
|
2005-05-02 22:25:02 +04:00
|
|
|
bio = NULL;
|
2008-01-03 22:09:33 +03:00
|
|
|
} else
|
2005-05-02 22:25:02 +04:00
|
|
|
inc_io(page);
|
|
|
|
xlen = (PAGE_CACHE_SIZE - offset) >> inode->i_blkbits;
|
|
|
|
pblock = metapage_get_blocks(inode, lblock, &xlen);
|
|
|
|
if (!pblock) {
|
|
|
|
printk(KERN_ERR "JFS: metapage_get_blocks failed\n");
|
2009-01-30 23:09:06 +03:00
|
|
|
/*
|
|
|
|
* We already called inc_io(), but can't cancel it
|
|
|
|
* with dec_io() until we're done with the page
|
|
|
|
*/
|
|
|
|
bad_blocks++;
|
2005-05-02 22:25:02 +04:00
|
|
|
continue;
|
|
|
|
}
|
2008-01-11 01:04:25 +03:00
|
|
|
len = min(xlen, (int)JFS_SBI(inode->i_sb)->nbperpage);
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
bio = bio_alloc(GFP_NOFS, 1);
|
|
|
|
bio->bi_bdev = inode->i_sb->s_bdev;
|
2013-10-12 02:44:27 +04:00
|
|
|
bio->bi_iter.bi_sector = pblock << (inode->i_blkbits - 9);
|
2005-05-02 22:25:02 +04:00
|
|
|
bio->bi_end_io = metapage_write_end_io;
|
|
|
|
bio->bi_private = page;
|
|
|
|
|
|
|
|
/* Don't call bio_add_page yet, we may add to this vec */
|
|
|
|
bio_offset = offset;
|
|
|
|
bio_bytes = len << inode->i_blkbits;
|
|
|
|
|
|
|
|
xlen -= len;
|
|
|
|
next_block = lblock + len;
|
|
|
|
}
|
|
|
|
if (bio) {
|
|
|
|
if (bio_add_page(bio, page, bio_bytes, bio_offset) < bio_bytes)
|
|
|
|
goto add_failed;
|
2013-10-12 02:44:27 +04:00
|
|
|
if (!bio->bi_iter.bi_size)
|
2005-05-02 22:25:02 +04:00
|
|
|
goto dump_bio;
|
2006-10-02 18:55:27 +04:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
submit_bio(WRITE, bio);
|
2008-01-03 22:09:33 +03:00
|
|
|
nr_underway++;
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
|
|
|
if (redirty)
|
|
|
|
redirty_page_for_writepage(wbc, page);
|
|
|
|
|
|
|
|
unlock_page(page);
|
|
|
|
|
2009-01-30 23:09:06 +03:00
|
|
|
if (bad_blocks)
|
|
|
|
goto err_out;
|
|
|
|
|
2008-01-03 22:09:33 +03:00
|
|
|
if (nr_underway == 0)
|
|
|
|
end_page_writeback(page);
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
return 0;
|
|
|
|
add_failed:
|
|
|
|
/* We should never reach here, since we're only adding one vec */
|
|
|
|
printk(KERN_ERR "JFS: bio_add_page failed unexpectedly\n");
|
|
|
|
goto skip;
|
|
|
|
dump_bio:
|
2007-06-13 19:17:50 +04:00
|
|
|
print_hex_dump(KERN_ERR, "JFS: dump of bio: ", DUMP_PREFIX_ADDRESS, 16,
|
|
|
|
4, bio, sizeof(*bio), 0);
|
2005-05-02 22:25:02 +04:00
|
|
|
skip:
|
|
|
|
bio_put(bio);
|
|
|
|
unlock_page(page);
|
|
|
|
dec_io(page, last_write_complete);
|
2009-01-30 23:09:06 +03:00
|
|
|
err_out:
|
|
|
|
while (bad_blocks--)
|
|
|
|
dec_io(page, last_write_complete);
|
2005-05-02 22:25:02 +04:00
|
|
|
return -EIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int metapage_readpage(struct file *fp, struct page *page)
|
|
|
|
{
|
|
|
|
struct inode *inode = page->mapping->host;
|
|
|
|
struct bio *bio = NULL;
|
2008-01-11 01:04:25 +03:00
|
|
|
int block_offset;
|
|
|
|
int blocks_per_page = PAGE_CACHE_SIZE >> inode->i_blkbits;
|
2005-05-02 22:25:02 +04:00
|
|
|
sector_t page_start; /* address of page in fs blocks */
|
|
|
|
sector_t pblock;
|
2008-01-11 01:04:25 +03:00
|
|
|
int xlen;
|
2005-05-02 22:25:02 +04:00
|
|
|
unsigned int len;
|
2008-01-11 01:04:25 +03:00
|
|
|
int offset;
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
BUG_ON(!PageLocked(page));
|
|
|
|
page_start = (sector_t)page->index <<
|
|
|
|
(PAGE_CACHE_SHIFT - inode->i_blkbits);
|
|
|
|
|
|
|
|
block_offset = 0;
|
|
|
|
while (block_offset < blocks_per_page) {
|
|
|
|
xlen = blocks_per_page - block_offset;
|
|
|
|
pblock = metapage_get_blocks(inode, page_start + block_offset,
|
|
|
|
&xlen);
|
|
|
|
if (pblock) {
|
|
|
|
if (!PagePrivate(page))
|
|
|
|
insert_metapage(page, NULL);
|
|
|
|
inc_io(page);
|
|
|
|
if (bio)
|
|
|
|
submit_bio(READ, bio);
|
|
|
|
|
|
|
|
bio = bio_alloc(GFP_NOFS, 1);
|
|
|
|
bio->bi_bdev = inode->i_sb->s_bdev;
|
2013-10-12 02:44:27 +04:00
|
|
|
bio->bi_iter.bi_sector =
|
|
|
|
pblock << (inode->i_blkbits - 9);
|
2005-05-02 22:25:02 +04:00
|
|
|
bio->bi_end_io = metapage_read_end_io;
|
|
|
|
bio->bi_private = page;
|
|
|
|
len = xlen << inode->i_blkbits;
|
|
|
|
offset = block_offset << inode->i_blkbits;
|
|
|
|
if (bio_add_page(bio, page, len, offset) < len)
|
|
|
|
goto add_failed;
|
|
|
|
block_offset += xlen;
|
|
|
|
} else
|
|
|
|
block_offset++;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2005-05-02 22:25:02 +04:00
|
|
|
if (bio)
|
|
|
|
submit_bio(READ, bio);
|
|
|
|
else
|
|
|
|
unlock_page(page);
|
|
|
|
|
|
|
|
return 0;
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
add_failed:
|
|
|
|
printk(KERN_ERR "JFS: bio_add_page failed unexpectedly\n");
|
|
|
|
bio_put(bio);
|
|
|
|
dec_io(page, last_read_complete);
|
|
|
|
return -EIO;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-10-21 11:20:48 +04:00
|
|
|
static int metapage_releasepage(struct page *page, gfp_t gfp_mask)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
|
|
|
struct metapage *mp;
|
2006-05-24 16:43:38 +04:00
|
|
|
int ret = 1;
|
2008-01-11 01:04:25 +03:00
|
|
|
int offset;
|
2005-05-02 22:25:02 +04:00
|
|
|
|
|
|
|
for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
|
|
|
|
mp = page_to_mp(page, offset);
|
|
|
|
|
|
|
|
if (!mp)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
jfs_info("metapage_releasepage: mp = 0x%p", mp);
|
2006-05-24 16:43:38 +04:00
|
|
|
if (mp->count || mp->nohomeok ||
|
|
|
|
test_bit(META_dirty, &mp->flag)) {
|
2005-05-02 22:25:02 +04:00
|
|
|
jfs_info("count = %ld, nohomeok = %d", mp->count,
|
|
|
|
mp->nohomeok);
|
2006-05-24 16:43:38 +04:00
|
|
|
ret = 0;
|
2005-05-02 22:25:02 +04:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (mp->lsn)
|
|
|
|
remove_from_logsync(mp);
|
|
|
|
remove_metapage(page, mp);
|
|
|
|
INCREMENT(mpStat.pagefree);
|
|
|
|
free_metapage(mp);
|
|
|
|
}
|
2006-05-24 16:43:38 +04:00
|
|
|
return ret;
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
|
|
|
|
2013-05-22 07:17:23 +04:00
|
|
|
static void metapage_invalidatepage(struct page *page, unsigned int offset,
|
|
|
|
unsigned int length)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
2013-05-22 07:17:23 +04:00
|
|
|
BUG_ON(offset || length < PAGE_CACHE_SIZE);
|
2005-05-02 22:25:02 +04:00
|
|
|
|
2006-03-26 13:37:18 +04:00
|
|
|
BUG_ON(PageWriteback(page));
|
2005-05-02 22:25:02 +04:00
|
|
|
|
2006-03-26 13:37:18 +04:00
|
|
|
metapage_releasepage(page, 0);
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
|
|
|
|
2006-06-28 15:26:44 +04:00
|
|
|
const struct address_space_operations jfs_metapage_aops = {
|
2005-05-02 22:25:02 +04:00
|
|
|
.readpage = metapage_readpage,
|
|
|
|
.writepage = metapage_writepage,
|
|
|
|
.releasepage = metapage_releasepage,
|
|
|
|
.invalidatepage = metapage_invalidatepage,
|
|
|
|
.set_page_dirty = __set_page_dirty_nobuffers,
|
|
|
|
};
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
struct metapage *__get_metapage(struct inode *inode, unsigned long lblock,
|
|
|
|
unsigned int size, int absolute,
|
|
|
|
unsigned long new)
|
|
|
|
{
|
|
|
|
int l2BlocksPerPage;
|
|
|
|
int l2bsize;
|
|
|
|
struct address_space *mapping;
|
2005-05-02 22:25:02 +04:00
|
|
|
struct metapage *mp = NULL;
|
|
|
|
struct page *page;
|
2005-04-17 02:20:36 +04:00
|
|
|
unsigned long page_index;
|
|
|
|
unsigned long page_offset;
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
jfs_info("__get_metapage: ino = %ld, lblock = 0x%lx, abs=%d",
|
|
|
|
inode->i_ino, lblock, absolute);
|
|
|
|
|
|
|
|
l2bsize = inode->i_blkbits;
|
|
|
|
l2BlocksPerPage = PAGE_CACHE_SHIFT - l2bsize;
|
|
|
|
page_index = lblock >> l2BlocksPerPage;
|
|
|
|
page_offset = (lblock - (page_index << l2BlocksPerPage)) << l2bsize;
|
|
|
|
if ((page_offset + size) > PAGE_CACHE_SIZE) {
|
|
|
|
jfs_err("MetaData crosses page boundary!!");
|
|
|
|
jfs_err("lblock = %lx, size = %d", lblock, size);
|
|
|
|
dump_stack();
|
|
|
|
return NULL;
|
|
|
|
}
|
2005-04-17 02:20:36 +04:00
|
|
|
if (absolute)
|
2005-05-02 22:25:02 +04:00
|
|
|
mapping = JFS_SBI(inode->i_sb)->direct_inode->i_mapping;
|
2005-04-17 02:20:36 +04:00
|
|
|
else {
|
|
|
|
/*
|
|
|
|
* If an nfs client tries to read an inode that is larger
|
|
|
|
* than any existing inodes, we may try to read past the
|
|
|
|
* end of the inode map
|
|
|
|
*/
|
|
|
|
if ((lblock << inode->i_blkbits) >= inode->i_size)
|
|
|
|
return NULL;
|
|
|
|
mapping = inode->i_mapping;
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
if (new && (PSIZE == PAGE_CACHE_SIZE)) {
|
|
|
|
page = grab_cache_page(mapping, page_index);
|
|
|
|
if (!page) {
|
|
|
|
jfs_err("grab_cache_page failed!");
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
SetPageUptodate(page);
|
|
|
|
} else {
|
2006-06-23 13:05:08 +04:00
|
|
|
page = read_mapping_page(mapping, page_index, NULL);
|
2005-07-22 20:08:44 +04:00
|
|
|
if (IS_ERR(page) || !PageUptodate(page)) {
|
2006-06-23 13:05:08 +04:00
|
|
|
jfs_err("read_mapping_page failed!");
|
2005-05-02 22:25:02 +04:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
lock_page(page);
|
|
|
|
}
|
|
|
|
|
|
|
|
mp = page_to_mp(page, page_offset);
|
2005-04-17 02:20:36 +04:00
|
|
|
if (mp) {
|
2005-05-02 22:25:02 +04:00
|
|
|
if (mp->logical_size != size) {
|
|
|
|
jfs_error(inode->i_sb,
|
2013-06-05 03:39:15 +04:00
|
|
|
"get_mp->logical_size != size\n");
|
2005-05-02 22:25:02 +04:00
|
|
|
jfs_err("logical_size = %d, size = %d",
|
|
|
|
mp->logical_size, size);
|
|
|
|
dump_stack();
|
2006-10-02 18:55:27 +04:00
|
|
|
goto unlock;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
mp->count++;
|
|
|
|
lock_metapage(mp);
|
|
|
|
if (test_bit(META_discard, &mp->flag)) {
|
|
|
|
if (!new) {
|
|
|
|
jfs_error(inode->i_sb,
|
2013-06-05 03:39:15 +04:00
|
|
|
"using a discarded metapage\n");
|
2005-05-02 22:25:02 +04:00
|
|
|
discard_metapage(mp);
|
2006-10-02 18:55:27 +04:00
|
|
|
goto unlock;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
clear_bit(META_discard, &mp->flag);
|
|
|
|
}
|
|
|
|
} else {
|
2005-05-02 22:25:02 +04:00
|
|
|
INCREMENT(mpStat.pagealloc);
|
|
|
|
mp = alloc_metapage(GFP_NOFS);
|
|
|
|
mp->page = page;
|
2005-04-17 02:20:36 +04:00
|
|
|
mp->flag = 0;
|
|
|
|
mp->xflag = COMMIT_PAGE;
|
|
|
|
mp->count = 1;
|
2005-05-02 22:25:02 +04:00
|
|
|
mp->nohomeok = 0;
|
2005-04-17 02:20:36 +04:00
|
|
|
mp->logical_size = size;
|
2005-05-02 22:25:02 +04:00
|
|
|
mp->data = page_address(page) + page_offset;
|
|
|
|
mp->index = lblock;
|
|
|
|
if (unlikely(insert_metapage(page, mp))) {
|
|
|
|
free_metapage(mp);
|
|
|
|
goto unlock;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2005-05-02 22:25:02 +04:00
|
|
|
lock_metapage(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
if (new) {
|
|
|
|
jfs_info("zeroing mp = 0x%p", mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
memset(mp->data, 0, PSIZE);
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
unlock_page(page);
|
|
|
|
jfs_info("__get_metapage: returning = 0x%p data = 0x%p", mp, mp->data);
|
2005-04-17 02:20:36 +04:00
|
|
|
return mp;
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
unlock:
|
|
|
|
unlock_page(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
void grab_metapage(struct metapage * mp)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
jfs_info("grab_metapage: mp = 0x%p", mp);
|
|
|
|
page_cache_get(mp->page);
|
|
|
|
lock_page(mp->page);
|
2005-04-17 02:20:36 +04:00
|
|
|
mp->count++;
|
2005-05-02 22:25:02 +04:00
|
|
|
lock_metapage(mp);
|
|
|
|
unlock_page(mp->page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
void force_metapage(struct metapage *mp)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
struct page *page = mp->page;
|
|
|
|
jfs_info("force_metapage: mp = 0x%p", mp);
|
|
|
|
set_bit(META_forcewrite, &mp->flag);
|
|
|
|
clear_bit(META_sync, &mp->flag);
|
|
|
|
page_cache_get(page);
|
|
|
|
lock_page(page);
|
|
|
|
set_page_dirty(page);
|
|
|
|
write_one_page(page, 1);
|
|
|
|
clear_bit(META_forcewrite, &mp->flag);
|
|
|
|
page_cache_release(page);
|
|
|
|
}
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-05-05 00:29:35 +04:00
|
|
|
void hold_metapage(struct metapage *mp)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
2005-04-17 02:20:36 +04:00
|
|
|
lock_page(mp->page);
|
2005-05-02 22:25:02 +04:00
|
|
|
}
|
|
|
|
|
2005-05-05 00:29:35 +04:00
|
|
|
void put_metapage(struct metapage *mp)
|
2005-05-02 22:25:02 +04:00
|
|
|
{
|
|
|
|
if (mp->count || mp->nohomeok) {
|
|
|
|
/* Someone else will release this */
|
2005-04-17 02:20:36 +04:00
|
|
|
unlock_page(mp->page);
|
|
|
|
return;
|
|
|
|
}
|
2005-05-02 22:25:02 +04:00
|
|
|
page_cache_get(mp->page);
|
|
|
|
mp->count++;
|
|
|
|
lock_metapage(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
unlock_page(mp->page);
|
2005-05-02 22:25:02 +04:00
|
|
|
release_metapage(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
void release_metapage(struct metapage * mp)
|
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
struct page *page = mp->page;
|
2005-04-17 02:20:36 +04:00
|
|
|
jfs_info("release_metapage: mp = 0x%p, flag = 0x%lx", mp, mp->flag);
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
BUG_ON(!page);
|
|
|
|
|
|
|
|
lock_page(page);
|
|
|
|
unlock_metapage(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
assert(mp->count);
|
2005-05-02 22:25:02 +04:00
|
|
|
if (--mp->count || mp->nohomeok) {
|
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
if (test_bit(META_dirty, &mp->flag)) {
|
|
|
|
set_page_dirty(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
if (test_bit(META_sync, &mp->flag)) {
|
|
|
|
clear_bit(META_sync, &mp->flag);
|
2005-05-02 22:25:02 +04:00
|
|
|
write_one_page(page, 1);
|
|
|
|
lock_page(page); /* write_one_page unlocks the page */
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2005-05-02 22:25:02 +04:00
|
|
|
} else if (mp->lsn) /* discard_metapage doesn't remove it */
|
|
|
|
remove_from_logsync(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
/* Try to keep metapages from using up too much memory */
|
|
|
|
drop_metapage(page, mp);
|
2006-12-21 00:55:35 +03:00
|
|
|
|
2005-05-02 22:25:02 +04:00
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
void __invalidate_metapages(struct inode *ip, s64 addr, int len)
|
|
|
|
{
|
2005-05-02 22:25:02 +04:00
|
|
|
sector_t lblock;
|
2005-04-17 02:20:36 +04:00
|
|
|
int l2BlocksPerPage = PAGE_CACHE_SHIFT - ip->i_blkbits;
|
2005-05-02 22:25:02 +04:00
|
|
|
int BlocksPerPage = 1 << l2BlocksPerPage;
|
2005-04-17 02:20:36 +04:00
|
|
|
/* All callers are interested in block device's mapping */
|
2005-05-02 22:25:02 +04:00
|
|
|
struct address_space *mapping =
|
|
|
|
JFS_SBI(ip->i_sb)->direct_inode->i_mapping;
|
2005-04-17 02:20:36 +04:00
|
|
|
struct metapage *mp;
|
|
|
|
struct page *page;
|
2005-05-02 22:25:02 +04:00
|
|
|
unsigned int offset;
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
/*
|
2005-05-02 22:25:02 +04:00
|
|
|
* Mark metapages to discard. They will eventually be
|
2005-04-17 02:20:36 +04:00
|
|
|
* released, but should not be written.
|
|
|
|
*/
|
2005-05-02 22:25:02 +04:00
|
|
|
for (lblock = addr & ~(BlocksPerPage - 1); lblock < addr + len;
|
|
|
|
lblock += BlocksPerPage) {
|
|
|
|
page = find_lock_page(mapping, lblock >> l2BlocksPerPage);
|
|
|
|
if (!page)
|
|
|
|
continue;
|
|
|
|
for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
|
|
|
|
mp = page_to_mp(page, offset);
|
|
|
|
if (!mp)
|
|
|
|
continue;
|
|
|
|
if (mp->index < addr)
|
|
|
|
continue;
|
|
|
|
if (mp->index >= addr + len)
|
|
|
|
break;
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
clear_bit(META_dirty, &mp->flag);
|
|
|
|
set_bit(META_discard, &mp->flag);
|
2005-05-02 22:25:02 +04:00
|
|
|
if (mp->lsn)
|
|
|
|
remove_from_logsync(mp);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2005-05-02 22:25:02 +04:00
|
|
|
unlock_page(page);
|
|
|
|
page_cache_release(page);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_JFS_STATISTICS
|
2008-05-13 17:22:10 +04:00
|
|
|
static int jfs_mpstat_proc_show(struct seq_file *m, void *v)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2008-05-13 17:22:10 +04:00
|
|
|
seq_printf(m,
|
2005-04-17 02:20:36 +04:00
|
|
|
"JFS Metapage statistics\n"
|
|
|
|
"=======================\n"
|
|
|
|
"page allocations = %d\n"
|
|
|
|
"page frees = %d\n"
|
|
|
|
"lock waits = %d\n",
|
|
|
|
mpStat.pagealloc,
|
|
|
|
mpStat.pagefree,
|
|
|
|
mpStat.lockwait);
|
2008-05-13 17:22:10 +04:00
|
|
|
return 0;
|
|
|
|
}
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2008-05-13 17:22:10 +04:00
|
|
|
static int jfs_mpstat_proc_open(struct inode *inode, struct file *file)
|
|
|
|
{
|
|
|
|
return single_open(file, jfs_mpstat_proc_show, NULL);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2008-05-13 17:22:10 +04:00
|
|
|
|
|
|
|
const struct file_operations jfs_mpstat_proc_fops = {
|
|
|
|
.owner = THIS_MODULE,
|
|
|
|
.open = jfs_mpstat_proc_open,
|
|
|
|
.read = seq_read,
|
|
|
|
.llseek = seq_lseek,
|
|
|
|
.release = single_release,
|
|
|
|
};
|
2005-04-17 02:20:36 +04:00
|
|
|
#endif
|