aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2012-08-04 11:32:46 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2012-08-04 11:32:46 -0400
commite7882d6c40874a5b5033ca85f7508a602a60b662 (patch)
treed477954bbc6ad932cda97fca1def0f5b48a2a378
parentd8579fd834de224d2833ce2e82e0f691cc67e8f7 (diff)
parentfe7c80518e34d1786f4a940ce673a0bfcbe53298 (diff)
Merge branch 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull VFS fixes and cleanups from Al Viro. Most of this is the series to remove sync_supers() and the ->write_supers VFS callback from Artem Bityutskiy. One commit to do the actual removal work, a whole series of commits to fix up stale comments etc all over the tree. There's also a regression fix for an incorrect use of mnt_drop_write() in do_dentry_open(). * 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: missed mnt_drop_write() in do_dentry_open() UBIFS: nuke pdflush from comments gfs2: nuke pdflush from comments drbd: nuke pdflush from comments nilfs2: nuke write_super from comments hfs: nuke write_super from comments vfs: nuke pdflush from comments jbd/jbd2: nuke write_super from comments btrfs: nuke pdflush from comments btrfs: nuke write_super from comments ext4: nuke pdflush from comments ext4: nuke write_super from comments ext3: nuke write_super from comments Documentation: fix the VM knobs descritpion WRT pdflush Documentation: get rid of write_super vfs: kill write_super and sync_supers
-rw-r--r--Documentation/DocBook/filesystems.tmpl4
-rw-r--r--Documentation/filesystems/Locking2
-rw-r--r--Documentation/filesystems/porting5
-rw-r--r--Documentation/filesystems/vfs.txt4
-rw-r--r--Documentation/laptops/laptop-mode.txt12
-rw-r--r--Documentation/sysctl/vm.txt14
-rw-r--r--drivers/block/drbd/drbd_main.c4
-rw-r--r--fs/bio.c2
-rw-r--r--fs/btrfs/inode.c3
-rw-r--r--fs/btrfs/ordered-data.c2
-rw-r--r--fs/btrfs/super.c4
-rw-r--r--fs/btrfs/volumes.c4
-rw-r--r--fs/ext3/inode.c8
-rw-r--r--fs/ext3/super.c11
-rw-r--r--fs/ext4/inode.c10
-rw-r--r--fs/ext4/super.c11
-rw-r--r--fs/gfs2/meta_io.c2
-rw-r--r--fs/hfs/mdb.c4
-rw-r--r--fs/jbd/journal.c4
-rw-r--r--fs/jbd2/journal.c4
-rw-r--r--fs/nilfs2/super.c4
-rw-r--r--fs/nilfs2/the_nilfs.h2
-rw-r--r--fs/open.c2
-rw-r--r--fs/super.c40
-rw-r--r--fs/ubifs/file.c10
-rw-r--r--fs/ubifs/super.c2
-rw-r--r--include/linux/backing-dev.h1
-rw-r--r--include/linux/fs.h3
-rw-r--r--include/linux/writeback.h1
-rw-r--r--mm/backing-dev.c52
-rw-r--r--mm/page-writeback.c1
31 files changed, 38 insertions, 194 deletions
diff --git a/Documentation/DocBook/filesystems.tmpl b/Documentation/DocBook/filesystems.tmpl
index 3fca32c41927..25b58efd955d 100644
--- a/Documentation/DocBook/filesystems.tmpl
+++ b/Documentation/DocBook/filesystems.tmpl
@@ -224,8 +224,8 @@ all your transactions.
224</para> 224</para>
225 225
226<para> 226<para>
227Then at umount time , in your put_super() (2.4) or write_super() (2.5) 227Then at umount time , in your put_super() you can then call journal_destroy()
228you can then call journal_destroy() to clean up your in-core journal object. 228to clean up your in-core journal object.
229</para> 229</para>
230 230
231<para> 231<para>
diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking
index 0f103e39b4f6..e540a24e5d06 100644
--- a/Documentation/filesystems/Locking
+++ b/Documentation/filesystems/Locking
@@ -114,7 +114,6 @@ prototypes:
114 int (*drop_inode) (struct inode *); 114 int (*drop_inode) (struct inode *);
115 void (*evict_inode) (struct inode *); 115 void (*evict_inode) (struct inode *);
116 void (*put_super) (struct super_block *); 116 void (*put_super) (struct super_block *);
117 void (*write_super) (struct super_block *);
118 int (*sync_fs)(struct super_block *sb, int wait); 117 int (*sync_fs)(struct super_block *sb, int wait);
119 int (*freeze_fs) (struct super_block *); 118 int (*freeze_fs) (struct super_block *);
120 int (*unfreeze_fs) (struct super_block *); 119 int (*unfreeze_fs) (struct super_block *);
@@ -136,7 +135,6 @@ write_inode:
136drop_inode: !!!inode->i_lock!!! 135drop_inode: !!!inode->i_lock!!!
137evict_inode: 136evict_inode:
138put_super: write 137put_super: write
139write_super: read
140sync_fs: read 138sync_fs: read
141freeze_fs: write 139freeze_fs: write
142unfreeze_fs: write 140unfreeze_fs: write
diff --git a/Documentation/filesystems/porting b/Documentation/filesystems/porting
index 2bef2b3843d1..0742feebc6e2 100644
--- a/Documentation/filesystems/porting
+++ b/Documentation/filesystems/porting
@@ -94,9 +94,8 @@ protected.
94--- 94---
95[mandatory] 95[mandatory]
96 96
97BKL is also moved from around sb operations. ->write_super() Is now called 97BKL is also moved from around sb operations. BKL should have been shifted into
98without BKL held. BKL should have been shifted into individual fs sb_op 98individual fs sb_op functions. If you don't need it, remove it.
99functions. If you don't need it, remove it.
100 99
101--- 100---
102[informational] 101[informational]
diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index 065aa2dc0835..2ee133e030c3 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -216,7 +216,6 @@ struct super_operations {
216 void (*drop_inode) (struct inode *); 216 void (*drop_inode) (struct inode *);
217 void (*delete_inode) (struct inode *); 217 void (*delete_inode) (struct inode *);
218 void (*put_super) (struct super_block *); 218 void (*put_super) (struct super_block *);
219 void (*write_super) (struct super_block *);
220 int (*sync_fs)(struct super_block *sb, int wait); 219 int (*sync_fs)(struct super_block *sb, int wait);
221 int (*freeze_fs) (struct super_block *); 220 int (*freeze_fs) (struct super_block *);
222 int (*unfreeze_fs) (struct super_block *); 221 int (*unfreeze_fs) (struct super_block *);
@@ -273,9 +272,6 @@ or bottom half).
273 put_super: called when the VFS wishes to free the superblock 272 put_super: called when the VFS wishes to free the superblock
274 (i.e. unmount). This is called with the superblock lock held 273 (i.e. unmount). This is called with the superblock lock held
275 274
276 write_super: called when the VFS superblock needs to be written to
277 disc. This method is optional
278
279 sync_fs: called when VFS is writing out all dirty data associated with 275 sync_fs: called when VFS is writing out all dirty data associated with
280 a superblock. The second parameter indicates whether the method 276 a superblock. The second parameter indicates whether the method
281 should wait until the write out has been completed. Optional. 277 should wait until the write out has been completed. Optional.
diff --git a/Documentation/laptops/laptop-mode.txt b/Documentation/laptops/laptop-mode.txt
index 0bf25eebce94..4ebbfc3f1c6e 100644
--- a/Documentation/laptops/laptop-mode.txt
+++ b/Documentation/laptops/laptop-mode.txt
@@ -262,9 +262,9 @@ MINIMUM_BATTERY_MINUTES=10
262 262
263# 263#
264# Allowed dirty background ratio, in percent. Once DIRTY_RATIO has been 264# Allowed dirty background ratio, in percent. Once DIRTY_RATIO has been
265# exceeded, the kernel will wake pdflush which will then reduce the amount 265# exceeded, the kernel will wake flusher threads which will then reduce the
266# of dirty memory to dirty_background_ratio. Set this nice and low, so once 266# amount of dirty memory to dirty_background_ratio. Set this nice and low,
267# some writeout has commenced, we do a lot of it. 267# so once some writeout has commenced, we do a lot of it.
268# 268#
269#DIRTY_BACKGROUND_RATIO=5 269#DIRTY_BACKGROUND_RATIO=5
270 270
@@ -384,9 +384,9 @@ CPU_MAXFREQ=${CPU_MAXFREQ:-'slowest'}
384 384
385# 385#
386# Allowed dirty background ratio, in percent. Once DIRTY_RATIO has been 386# Allowed dirty background ratio, in percent. Once DIRTY_RATIO has been
387# exceeded, the kernel will wake pdflush which will then reduce the amount 387# exceeded, the kernel will wake flusher threads which will then reduce the
388# of dirty memory to dirty_background_ratio. Set this nice and low, so once 388# amount of dirty memory to dirty_background_ratio. Set this nice and low,
389# some writeout has commenced, we do a lot of it. 389# so once some writeout has commenced, we do a lot of it.
390# 390#
391DIRTY_BACKGROUND_RATIO=${DIRTY_BACKGROUND_RATIO:-'5'} 391DIRTY_BACKGROUND_RATIO=${DIRTY_BACKGROUND_RATIO:-'5'}
392 392
diff --git a/Documentation/sysctl/vm.txt b/Documentation/sysctl/vm.txt
index dcc2a94ae34e..078701fdbd4d 100644
--- a/Documentation/sysctl/vm.txt
+++ b/Documentation/sysctl/vm.txt
@@ -76,8 +76,8 @@ huge pages although processes will also directly compact memory as required.
76 76
77dirty_background_bytes 77dirty_background_bytes
78 78
79Contains the amount of dirty memory at which the pdflush background writeback 79Contains the amount of dirty memory at which the background kernel
80daemon will start writeback. 80flusher threads will start writeback.
81 81
82Note: dirty_background_bytes is the counterpart of dirty_background_ratio. Only 82Note: dirty_background_bytes is the counterpart of dirty_background_ratio. Only
83one of them may be specified at a time. When one sysctl is written it is 83one of them may be specified at a time. When one sysctl is written it is
@@ -89,7 +89,7 @@ other appears as 0 when read.
89dirty_background_ratio 89dirty_background_ratio
90 90
91Contains, as a percentage of total system memory, the number of pages at which 91Contains, as a percentage of total system memory, the number of pages at which
92the pdflush background writeback daemon will start writing out dirty data. 92the background kernel flusher threads will start writing out dirty data.
93 93
94============================================================== 94==============================================================
95 95
@@ -112,9 +112,9 @@ retained.
112dirty_expire_centisecs 112dirty_expire_centisecs
113 113
114This tunable is used to define when dirty data is old enough to be eligible 114This tunable is used to define when dirty data is old enough to be eligible
115for writeout by the pdflush daemons. It is expressed in 100'ths of a second. 115for writeout by the kernel flusher threads. It is expressed in 100'ths
116Data which has been dirty in-memory for longer than this interval will be 116of a second. Data which has been dirty in-memory for longer than this
117written out next time a pdflush daemon wakes up. 117interval will be written out next time a flusher thread wakes up.
118 118
119============================================================== 119==============================================================
120 120
@@ -128,7 +128,7 @@ data.
128 128
129dirty_writeback_centisecs 129dirty_writeback_centisecs
130 130
131The pdflush writeback daemons will periodically wake up and write `old' data 131The kernel flusher threads will periodically wake up and write `old' data
132out to disk. This tunable expresses the interval between those wakeups, in 132out to disk. This tunable expresses the interval between those wakeups, in
133100'ths of a second. 133100'ths of a second.
134 134
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 2e0e7fc1dbba..dbe6135a2abe 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -3537,9 +3537,9 @@ static void drbd_cleanup(void)
3537} 3537}
3538 3538
3539/** 3539/**
3540 * drbd_congested() - Callback for pdflush 3540 * drbd_congested() - Callback for the flusher thread
3541 * @congested_data: User data 3541 * @congested_data: User data
3542 * @bdi_bits: Bits pdflush is currently interested in 3542 * @bdi_bits: Bits the BDI flusher thread is currently interested in
3543 * 3543 *
3544 * Returns 1<<BDI_async_congested and/or 1<<BDI_sync_congested if we are congested. 3544 * Returns 1<<BDI_async_congested and/or 1<<BDI_sync_congested if we are congested.
3545 */ 3545 */
diff --git a/fs/bio.c b/fs/bio.c
index 73922abba832..5eaa70c9d96e 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -1312,7 +1312,7 @@ EXPORT_SYMBOL(bio_copy_kern);
1312 * Note that this code is very hard to test under normal circumstances because 1312 * Note that this code is very hard to test under normal circumstances because
1313 * direct-io pins the pages with get_user_pages(). This makes 1313 * direct-io pins the pages with get_user_pages(). This makes
1314 * is_page_cache_freeable return false, and the VM will not clean the pages. 1314 * is_page_cache_freeable return false, and the VM will not clean the pages.
1315 * But other code (eg, pdflush) could clean the pages if they are mapped 1315 * But other code (eg, flusher threads) could clean the pages if they are mapped
1316 * pagecache. 1316 * pagecache.
1317 * 1317 *
1318 * Simply disabling the call to bio_set_pages_dirty() is a good way to test the 1318 * Simply disabling the call to bio_set_pages_dirty() is a good way to test the
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 83baec24946d..6e8f416773d4 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -324,7 +324,8 @@ static noinline int add_async_extent(struct async_cow *cow,
324 * If this code finds it can't get good compression, it puts an 324 * If this code finds it can't get good compression, it puts an
325 * entry onto the work queue to write the uncompressed bytes. This 325 * entry onto the work queue to write the uncompressed bytes. This
326 * makes sure that both compressed inodes and uncompressed inodes 326 * makes sure that both compressed inodes and uncompressed inodes
327 * are written in the same order that pdflush sent them down. 327 * are written in the same order that the flusher thread sent them
328 * down.
328 */ 329 */
329static noinline int compress_file_range(struct inode *inode, 330static noinline int compress_file_range(struct inode *inode,
330 struct page *locked_page, 331 struct page *locked_page,
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 643335a4fe3c..051c7fe551dd 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -596,7 +596,7 @@ void btrfs_start_ordered_extent(struct inode *inode,
596 /* 596 /*
597 * pages in the range can be dirty, clean or writeback. We 597 * pages in the range can be dirty, clean or writeback. We
598 * start IO on any dirty ones so the wait doesn't stall waiting 598 * start IO on any dirty ones so the wait doesn't stall waiting
599 * for pdflush to find them 599 * for the flusher thread to find them
600 */ 600 */
601 if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags)) 601 if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
602 filemap_fdatawrite_range(inode->i_mapping, start, end); 602 filemap_fdatawrite_range(inode->i_mapping, start, end);
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 8c6e61d6eed5..f2eb24c477a3 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -100,10 +100,6 @@ static void __save_error_info(struct btrfs_fs_info *fs_info)
100 fs_info->fs_state = BTRFS_SUPER_FLAG_ERROR; 100 fs_info->fs_state = BTRFS_SUPER_FLAG_ERROR;
101} 101}
102 102
103/* NOTE:
104 * We move write_super stuff at umount in order to avoid deadlock
105 * for umount hold all lock.
106 */
107static void save_error_info(struct btrfs_fs_info *fs_info) 103static void save_error_info(struct btrfs_fs_info *fs_info)
108{ 104{
109 __save_error_info(fs_info); 105 __save_error_info(fs_info);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index b8708f994e67..e86ae04abe6a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1744,10 +1744,6 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
1744 1744
1745 device->fs_devices = root->fs_info->fs_devices; 1745 device->fs_devices = root->fs_info->fs_devices;
1746 1746
1747 /*
1748 * we don't want write_supers to jump in here with our device
1749 * half setup
1750 */
1751 mutex_lock(&root->fs_info->fs_devices->device_list_mutex); 1747 mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1752 list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices); 1748 list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
1753 list_add(&device->dev_alloc_list, 1749 list_add(&device->dev_alloc_list,
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index 9a4a5c48b1c9..a07597307fd1 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -3459,14 +3459,6 @@ ext3_reserve_inode_write(handle_t *handle, struct inode *inode,
3459 * inode out, but prune_icache isn't a user-visible syncing function. 3459 * inode out, but prune_icache isn't a user-visible syncing function.
3460 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync) 3460 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync)
3461 * we start and wait on commits. 3461 * we start and wait on commits.
3462 *
3463 * Is this efficient/effective? Well, we're being nice to the system
3464 * by cleaning up our inodes proactively so they can be reaped
3465 * without I/O. But we are potentially leaving up to five seconds'
3466 * worth of inodes floating about which prune_icache wants us to
3467 * write out. One way to fix that would be to get prune_icache()
3468 * to do a write_super() to free up some memory. It has the desired
3469 * effect.
3470 */ 3462 */
3471int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode) 3463int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode)
3472{ 3464{
diff --git a/fs/ext3/super.c b/fs/ext3/super.c
index ff9bcdc5b0d5..8c892e93d8e7 100644
--- a/fs/ext3/super.c
+++ b/fs/ext3/super.c
@@ -64,11 +64,6 @@ static int ext3_freeze(struct super_block *sb);
64 64
65/* 65/*
66 * Wrappers for journal_start/end. 66 * Wrappers for journal_start/end.
67 *
68 * The only special thing we need to do here is to make sure that all
69 * journal_end calls result in the superblock being marked dirty, so
70 * that sync() will call the filesystem's write_super callback if
71 * appropriate.
72 */ 67 */
73handle_t *ext3_journal_start_sb(struct super_block *sb, int nblocks) 68handle_t *ext3_journal_start_sb(struct super_block *sb, int nblocks)
74{ 69{
@@ -90,12 +85,6 @@ handle_t *ext3_journal_start_sb(struct super_block *sb, int nblocks)
90 return journal_start(journal, nblocks); 85 return journal_start(journal, nblocks);
91} 86}
92 87
93/*
94 * The only special thing we need to do here is to make sure that all
95 * journal_stop calls result in the superblock being marked dirty, so
96 * that sync() will call the filesystem's write_super callback if
97 * appropriate.
98 */
99int __ext3_journal_stop(const char *where, handle_t *handle) 88int __ext3_journal_stop(const char *where, handle_t *handle)
100{ 89{
101 struct super_block *sb; 90 struct super_block *sb;
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 6324f74e0342..dff171c3a123 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1970,7 +1970,7 @@ static void ext4_end_io_buffer_write(struct buffer_head *bh, int uptodate);
1970 * This function can get called via... 1970 * This function can get called via...
1971 * - ext4_da_writepages after taking page lock (have journal handle) 1971 * - ext4_da_writepages after taking page lock (have journal handle)
1972 * - journal_submit_inode_data_buffers (no journal handle) 1972 * - journal_submit_inode_data_buffers (no journal handle)
1973 * - shrink_page_list via pdflush (no journal handle) 1973 * - shrink_page_list via the kswapd/direct reclaim (no journal handle)
1974 * - grab_page_cache when doing write_begin (have journal handle) 1974 * - grab_page_cache when doing write_begin (have journal handle)
1975 * 1975 *
1976 * We don't do any block allocation in this function. If we have page with 1976 * We don't do any block allocation in this function. If we have page with
@@ -4589,14 +4589,6 @@ static int ext4_expand_extra_isize(struct inode *inode,
4589 * inode out, but prune_icache isn't a user-visible syncing function. 4589 * inode out, but prune_icache isn't a user-visible syncing function.
4590 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync) 4590 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync)
4591 * we start and wait on commits. 4591 * we start and wait on commits.
4592 *
4593 * Is this efficient/effective? Well, we're being nice to the system
4594 * by cleaning up our inodes proactively so they can be reaped
4595 * without I/O. But we are potentially leaving up to five seconds'
4596 * worth of inodes floating about which prune_icache wants us to
4597 * write out. One way to fix that would be to get prune_icache()
4598 * to do a write_super() to free up some memory. It has the desired
4599 * effect.
4600 */ 4592 */
4601int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode) 4593int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
4602{ 4594{
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index d76ec8277d3f..3e0851e4f468 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -326,11 +326,6 @@ static void ext4_put_nojournal(handle_t *handle)
326 326
327/* 327/*
328 * Wrappers for jbd2_journal_start/end. 328 * Wrappers for jbd2_journal_start/end.
329 *
330 * The only special thing we need to do here is to make sure that all
331 * journal_end calls result in the superblock being marked dirty, so
332 * that sync() will call the filesystem's write_super callback if
333 * appropriate.
334 */ 329 */
335handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks) 330handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks)
336{ 331{
@@ -356,12 +351,6 @@ handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks)
356 return jbd2_journal_start(journal, nblocks); 351 return jbd2_journal_start(journal, nblocks);
357} 352}
358 353
359/*
360 * The only special thing we need to do here is to make sure that all
361 * jbd2_journal_stop calls result in the superblock being marked dirty, so
362 * that sync() will call the filesystem's write_super callback if
363 * appropriate.
364 */
365int __ext4_journal_stop(const char *where, unsigned int line, handle_t *handle) 354int __ext4_journal_stop(const char *where, unsigned int line, handle_t *handle)
366{ 355{
367 struct super_block *sb; 356 struct super_block *sb;
diff --git a/fs/gfs2/meta_io.c b/fs/gfs2/meta_io.c
index 3a56c8d94de0..22255d96b27e 100644
--- a/fs/gfs2/meta_io.c
+++ b/fs/gfs2/meta_io.c
@@ -52,7 +52,7 @@ static int gfs2_aspace_writepage(struct page *page, struct writeback_control *wb
52 /* 52 /*
53 * If it's a fully non-blocking write attempt and we cannot 53 * If it's a fully non-blocking write attempt and we cannot
54 * lock the buffer then redirty the page. Note that this can 54 * lock the buffer then redirty the page. Note that this can
55 * potentially cause a busy-wait loop from pdflush and kswapd 55 * potentially cause a busy-wait loop from flusher thread and kswapd
56 * activity, but those code paths have their own higher-level 56 * activity, but those code paths have their own higher-level
57 * throttling. 57 * throttling.
58 */ 58 */
diff --git a/fs/hfs/mdb.c b/fs/hfs/mdb.c
index 5fd51a5833ff..b7ec224910c5 100644
--- a/fs/hfs/mdb.c
+++ b/fs/hfs/mdb.c
@@ -236,10 +236,10 @@ out:
236 * hfs_mdb_commit() 236 * hfs_mdb_commit()
237 * 237 *
238 * Description: 238 * Description:
239 * This updates the MDB on disk (look also at hfs_write_super()). 239 * This updates the MDB on disk.
240 * It does not check, if the superblock has been modified, or 240 * It does not check, if the superblock has been modified, or
241 * if the filesystem has been mounted read-only. It is mainly 241 * if the filesystem has been mounted read-only. It is mainly
242 * called by hfs_write_super() and hfs_btree_extend(). 242 * called by hfs_sync_fs() and flush_mdb().
243 * Input Variable(s): 243 * Input Variable(s):
244 * struct hfs_mdb *mdb: Pointer to the hfs MDB 244 * struct hfs_mdb *mdb: Pointer to the hfs MDB
245 * int backup; 245 * int backup;
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 425c2f2cf170..09357508ec9a 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -534,8 +534,8 @@ int journal_start_commit(journal_t *journal, tid_t *ptid)
534 ret = 1; 534 ret = 1;
535 } else if (journal->j_committing_transaction) { 535 } else if (journal->j_committing_transaction) {
536 /* 536 /*
537 * If ext3_write_super() recently started a commit, then we 537 * If commit has been started, then we have to wait for
538 * have to wait for completion of that transaction 538 * completion of that transaction.
539 */ 539 */
540 if (ptid) 540 if (ptid)
541 *ptid = journal->j_committing_transaction->t_tid; 541 *ptid = journal->j_committing_transaction->t_tid;
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index e9a3c4c85594..8625da27eccf 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -612,8 +612,8 @@ int jbd2_journal_start_commit(journal_t *journal, tid_t *ptid)
612 ret = 1; 612 ret = 1;
613 } else if (journal->j_committing_transaction) { 613 } else if (journal->j_committing_transaction) {
614 /* 614 /*
615 * If ext3_write_super() recently started a commit, then we 615 * If commit has been started, then we have to wait for
616 * have to wait for completion of that transaction 616 * completion of that transaction.
617 */ 617 */
618 if (ptid) 618 if (ptid)
619 *ptid = journal->j_committing_transaction->t_tid; 619 *ptid = journal->j_committing_transaction->t_tid;
diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
index 6522cac6057c..6a10812711c1 100644
--- a/fs/nilfs2/super.c
+++ b/fs/nilfs2/super.c
@@ -676,17 +676,13 @@ static const struct super_operations nilfs_sops = {
676 .alloc_inode = nilfs_alloc_inode, 676 .alloc_inode = nilfs_alloc_inode,
677 .destroy_inode = nilfs_destroy_inode, 677 .destroy_inode = nilfs_destroy_inode,
678 .dirty_inode = nilfs_dirty_inode, 678 .dirty_inode = nilfs_dirty_inode,
679 /* .write_inode = nilfs_write_inode, */
680 /* .drop_inode = nilfs_drop_inode, */
681 .evict_inode = nilfs_evict_inode, 679 .evict_inode = nilfs_evict_inode,
682 .put_super = nilfs_put_super, 680 .put_super = nilfs_put_super,
683 /* .write_super = nilfs_write_super, */
684 .sync_fs = nilfs_sync_fs, 681 .sync_fs = nilfs_sync_fs,
685 .freeze_fs = nilfs_freeze, 682 .freeze_fs = nilfs_freeze,
686 .unfreeze_fs = nilfs_unfreeze, 683 .unfreeze_fs = nilfs_unfreeze,
687 .statfs = nilfs_statfs, 684 .statfs = nilfs_statfs,
688 .remount_fs = nilfs_remount, 685 .remount_fs = nilfs_remount,
689 /* .umount_begin */
690 .show_options = nilfs_show_options 686 .show_options = nilfs_show_options
691}; 687};
692 688
diff --git a/fs/nilfs2/the_nilfs.h b/fs/nilfs2/the_nilfs.h
index 6eee4177807b..be1267a34cea 100644
--- a/fs/nilfs2/the_nilfs.h
+++ b/fs/nilfs2/the_nilfs.h
@@ -107,8 +107,6 @@ struct the_nilfs {
107 * used for 107 * used for
108 * - loading the latest checkpoint exclusively. 108 * - loading the latest checkpoint exclusively.
109 * - allocating a new full segment. 109 * - allocating a new full segment.
110 * - protecting s_dirt in the super_block struct
111 * (see nilfs_write_super) and the following fields.
112 */ 110 */
113 struct buffer_head *ns_sbh[2]; 111 struct buffer_head *ns_sbh[2];
114 struct nilfs_super_block *ns_sbp[2]; 112 struct nilfs_super_block *ns_sbp[2];
diff --git a/fs/open.c b/fs/open.c
index f3d96e7e7b19..bc132e167d2d 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -717,7 +717,7 @@ cleanup_all:
717 * here, so just reset the state. 717 * here, so just reset the state.
718 */ 718 */
719 file_reset_write(f); 719 file_reset_write(f);
720 mnt_drop_write(f->f_path.mnt); 720 __mnt_drop_write(f->f_path.mnt);
721 } 721 }
722 } 722 }
723cleanup_file: 723cleanup_file:
diff --git a/fs/super.c b/fs/super.c
index b05cf47463d0..0902cfa6a12e 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -537,46 +537,6 @@ void drop_super(struct super_block *sb)
537EXPORT_SYMBOL(drop_super); 537EXPORT_SYMBOL(drop_super);
538 538
539/** 539/**
540 * sync_supers - helper for periodic superblock writeback
541 *
542 * Call the write_super method if present on all dirty superblocks in
543 * the system. This is for the periodic writeback used by most older
544 * filesystems. For data integrity superblock writeback use
545 * sync_filesystems() instead.
546 *
547 * Note: check the dirty flag before waiting, so we don't
548 * hold up the sync while mounting a device. (The newly
549 * mounted device won't need syncing.)
550 */
551void sync_supers(void)
552{
553 struct super_block *sb, *p = NULL;
554
555 spin_lock(&sb_lock);
556 list_for_each_entry(sb, &super_blocks, s_list) {
557 if (hlist_unhashed(&sb->s_instances))
558 continue;
559 if (sb->s_op->write_super && sb->s_dirt) {
560 sb->s_count++;
561 spin_unlock(&sb_lock);
562
563 down_read(&sb->s_umount);
564 if (sb->s_root && sb->s_dirt && (sb->s_flags & MS_BORN))
565 sb->s_op->write_super(sb);
566 up_read(&sb->s_umount);
567
568 spin_lock(&sb_lock);
569 if (p)
570 __put_super(p);
571 p = sb;
572 }
573 }
574 if (p)
575 __put_super(p);
576 spin_unlock(&sb_lock);
577}
578
579/**
580 * iterate_supers - call function for all active superblocks 540 * iterate_supers - call function for all active superblocks
581 * @f: function to call 541 * @f: function to call
582 * @arg: argument to pass to it 542 * @arg: argument to pass to it
diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 35389ca2d267..7bd6e72afd11 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -37,11 +37,11 @@
37 * 37 *
38 * A thing to keep in mind: inode @i_mutex is locked in most VFS operations we 38 * A thing to keep in mind: inode @i_mutex is locked in most VFS operations we
39 * implement. However, this is not true for 'ubifs_writepage()', which may be 39 * implement. However, this is not true for 'ubifs_writepage()', which may be
40 * called with @i_mutex unlocked. For example, when pdflush is doing background 40 * called with @i_mutex unlocked. For example, when flusher thread is doing
41 * write-back, it calls 'ubifs_writepage()' with unlocked @i_mutex. At "normal" 41 * background write-back, it calls 'ubifs_writepage()' with unlocked @i_mutex.
42 * work-paths the @i_mutex is locked in 'ubifs_writepage()', e.g. in the 42 * At "normal" work-paths the @i_mutex is locked in 'ubifs_writepage()', e.g.
43 * "sys_write -> alloc_pages -> direct reclaim path". So, in 'ubifs_writepage()' 43 * in the "sys_write -> alloc_pages -> direct reclaim path". So, in
44 * we are only guaranteed that the page is locked. 44 * 'ubifs_writepage()' we are only guaranteed that the page is locked.
45 * 45 *
46 * Similarly, @i_mutex is not always locked in 'ubifs_readpage()', e.g., the 46 * Similarly, @i_mutex is not always locked in 'ubifs_readpage()', e.g., the
47 * read-ahead path does not lock it ("sys_read -> generic_file_aio_read -> 47 * read-ahead path does not lock it ("sys_read -> generic_file_aio_read ->
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 1c766c39c038..c3fa6c5327a3 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -303,7 +303,7 @@ static int ubifs_write_inode(struct inode *inode, struct writeback_control *wbc)
303 mutex_lock(&ui->ui_mutex); 303 mutex_lock(&ui->ui_mutex);
304 /* 304 /*
305 * Due to races between write-back forced by budgeting 305 * Due to races between write-back forced by budgeting
306 * (see 'sync_some_inodes()') and pdflush write-back, the inode may 306 * (see 'sync_some_inodes()') and background write-back, the inode may
307 * have already been synchronized, do not do this again. This might 307 * have already been synchronized, do not do this again. This might
308 * also happen if it was synchronized in an VFS operation, e.g. 308 * also happen if it was synchronized in an VFS operation, e.g.
309 * 'ubifs_link()'. 309 * 'ubifs_link()'.
diff --git a/include/linux/backing-dev.h b/include/linux/backing-dev.h
index c97c6b9cd38e..2a9a9abc9126 100644
--- a/include/linux/backing-dev.h
+++ b/include/linux/backing-dev.h
@@ -124,7 +124,6 @@ void bdi_start_writeback(struct backing_dev_info *bdi, long nr_pages,
124void bdi_start_background_writeback(struct backing_dev_info *bdi); 124void bdi_start_background_writeback(struct backing_dev_info *bdi);
125int bdi_writeback_thread(void *data); 125int bdi_writeback_thread(void *data);
126int bdi_has_dirty_io(struct backing_dev_info *bdi); 126int bdi_has_dirty_io(struct backing_dev_info *bdi);
127void bdi_arm_supers_timer(void);
128void bdi_wakeup_thread_delayed(struct backing_dev_info *bdi); 127void bdi_wakeup_thread_delayed(struct backing_dev_info *bdi);
129void bdi_lock_two(struct bdi_writeback *wb1, struct bdi_writeback *wb2); 128void bdi_lock_two(struct bdi_writeback *wb1, struct bdi_writeback *wb2);
130 129
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 38dba16c4176..aa110476a95b 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1491,7 +1491,6 @@ struct sb_writers {
1491struct super_block { 1491struct super_block {
1492 struct list_head s_list; /* Keep this first */ 1492 struct list_head s_list; /* Keep this first */
1493 dev_t s_dev; /* search index; _not_ kdev_t */ 1493 dev_t s_dev; /* search index; _not_ kdev_t */
1494 unsigned char s_dirt;
1495 unsigned char s_blocksize_bits; 1494 unsigned char s_blocksize_bits;
1496 unsigned long s_blocksize; 1495 unsigned long s_blocksize;
1497 loff_t s_maxbytes; /* Max file size */ 1496 loff_t s_maxbytes; /* Max file size */
@@ -1861,7 +1860,6 @@ struct super_operations {
1861 int (*drop_inode) (struct inode *); 1860 int (*drop_inode) (struct inode *);
1862 void (*evict_inode) (struct inode *); 1861 void (*evict_inode) (struct inode *);
1863 void (*put_super) (struct super_block *); 1862 void (*put_super) (struct super_block *);
1864 void (*write_super) (struct super_block *);
1865 int (*sync_fs)(struct super_block *sb, int wait); 1863 int (*sync_fs)(struct super_block *sb, int wait);
1866 int (*freeze_fs) (struct super_block *); 1864 int (*freeze_fs) (struct super_block *);
1867 int (*unfreeze_fs) (struct super_block *); 1865 int (*unfreeze_fs) (struct super_block *);
@@ -2397,7 +2395,6 @@ extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2397 int datasync); 2395 int datasync);
2398extern int vfs_fsync(struct file *file, int datasync); 2396extern int vfs_fsync(struct file *file, int datasync);
2399extern int generic_write_sync(struct file *file, loff_t pos, loff_t count); 2397extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2400extern void sync_supers(void);
2401extern void emergency_sync(void); 2398extern void emergency_sync(void);
2402extern void emergency_remount(void); 2399extern void emergency_remount(void);
2403#ifdef CONFIG_BLOCK 2400#ifdef CONFIG_BLOCK
diff --git a/include/linux/writeback.h b/include/linux/writeback.h
index c66fe3332d83..50c3e8fa06a8 100644
--- a/include/linux/writeback.h
+++ b/include/linux/writeback.h
@@ -104,7 +104,6 @@ static inline void wait_on_inode(struct inode *inode)
104 wait_on_bit(&inode->i_state, __I_NEW, inode_wait, TASK_UNINTERRUPTIBLE); 104 wait_on_bit(&inode->i_state, __I_NEW, inode_wait, TASK_UNINTERRUPTIBLE);
105} 105}
106 106
107
108/* 107/*
109 * mm/page-writeback.c 108 * mm/page-writeback.c
110 */ 109 */
diff --git a/mm/backing-dev.c b/mm/backing-dev.c
index 6b4718e2ee34..b41823cc05e6 100644
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -39,12 +39,6 @@ DEFINE_SPINLOCK(bdi_lock);
39LIST_HEAD(bdi_list); 39LIST_HEAD(bdi_list);
40LIST_HEAD(bdi_pending_list); 40LIST_HEAD(bdi_pending_list);
41 41
42static struct task_struct *sync_supers_tsk;
43static struct timer_list sync_supers_timer;
44
45static int bdi_sync_supers(void *);
46static void sync_supers_timer_fn(unsigned long);
47
48void bdi_lock_two(struct bdi_writeback *wb1, struct bdi_writeback *wb2) 42void bdi_lock_two(struct bdi_writeback *wb1, struct bdi_writeback *wb2)
49{ 43{
50 if (wb1 < wb2) { 44 if (wb1 < wb2) {
@@ -250,12 +244,6 @@ static int __init default_bdi_init(void)
250{ 244{
251 int err; 245 int err;
252 246
253 sync_supers_tsk = kthread_run(bdi_sync_supers, NULL, "sync_supers");
254 BUG_ON(IS_ERR(sync_supers_tsk));
255
256 setup_timer(&sync_supers_timer, sync_supers_timer_fn, 0);
257 bdi_arm_supers_timer();
258
259 err = bdi_init(&default_backing_dev_info); 247 err = bdi_init(&default_backing_dev_info);
260 if (!err) 248 if (!err)
261 bdi_register(&default_backing_dev_info, NULL, "default"); 249 bdi_register(&default_backing_dev_info, NULL, "default");
@@ -270,46 +258,6 @@ int bdi_has_dirty_io(struct backing_dev_info *bdi)
270 return wb_has_dirty_io(&bdi->wb); 258 return wb_has_dirty_io(&bdi->wb);
271} 259}
272 260
273/*
274 * kupdated() used to do this. We cannot do it from the bdi_forker_thread()
275 * or we risk deadlocking on ->s_umount. The longer term solution would be
276 * to implement sync_supers_bdi() or similar and simply do it from the
277 * bdi writeback thread individually.
278 */
279static int bdi_sync_supers(void *unused)
280{
281 set_user_nice(current, 0);
282
283 while (!kthread_should_stop()) {
284 set_current_state(TASK_INTERRUPTIBLE);
285 schedule();
286
287 /*
288 * Do this periodically, like kupdated() did before.
289 */
290 sync_supers();
291 }
292
293 return 0;
294}
295
296void bdi_arm_supers_timer(void)
297{
298 unsigned long next;
299
300 if (!dirty_writeback_interval)
301 return;
302
303 next = msecs_to_jiffies(dirty_writeback_interval * 10) + jiffies;
304 mod_timer(&sync_supers_timer, round_jiffies_up(next));
305}
306
307static void sync_supers_timer_fn(unsigned long unused)
308{
309 wake_up_process(sync_supers_tsk);
310 bdi_arm_supers_timer();
311}
312
313static void wakeup_timer_fn(unsigned long data) 261static void wakeup_timer_fn(unsigned long data)
314{ 262{
315 struct backing_dev_info *bdi = (struct backing_dev_info *)data; 263 struct backing_dev_info *bdi = (struct backing_dev_info *)data;
diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index e5363f34e025..5ad5ce23c1e0 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -1532,7 +1532,6 @@ int dirty_writeback_centisecs_handler(ctl_table *table, int write,
1532 void __user *buffer, size_t *length, loff_t *ppos) 1532 void __user *buffer, size_t *length, loff_t *ppos)
1533{ 1533{
1534 proc_dointvec(table, write, buffer, length, ppos); 1534 proc_dointvec(table, write, buffer, length, ppos);
1535 bdi_arm_supers_timer();
1536 return 0; 1535 return 0;
1537} 1536}
1538 1537