kernel-ark/fs/sync.c
Jens Axboe 03ba3782e8 writeback: switch to per-bdi threads for flushing data
This gets rid of pdflush for bdi writeout and kupdated style cleaning.
pdflush writeout suffers from lack of locality and also requires more
threads to handle the same workload, since it has to work in a
non-blocking fashion against each queue. This also introduces lumpy
behaviour and potential request starvation, since pdflush can be starved
for queue access if others are accessing it. A sample ffsb workload that
does random writes to files is about 8% faster here on a simple SATA drive
during the benchmark phase. File layout also seems a LOT more smooth in
vmstat:

 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  1      0 608848   2652 375372    0    0     0 71024  604    24  1 10 48 42
 0  1      0 549644   2712 433736    0    0     0 60692  505    27  1  8 48 44
 1  0      0 476928   2784 505192    0    0     4 29540  553    24  0  9 53 37
 0  1      0 457972   2808 524008    0    0     0 54876  331    16  0  4 38 58
 0  1      0 366128   2928 614284    0    0     4 92168  710    58  0 13 53 34
 0  1      0 295092   3000 684140    0    0     0 62924  572    23  0  9 53 37
 0  1      0 236592   3064 741704    0    0     4 58256  523    17  0  8 48 44
 0  1      0 165608   3132 811464    0    0     0 57460  560    21  0  8 54 38
 0  1      0 102952   3200 873164    0    0     4 74748  540    29  1 10 48 41
 0  1      0  48604   3252 926472    0    0     0 53248  469    29  0  7 47 45

where vanilla tends to fluctuate a lot in the creation phase:

 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 1  1      0 678716   5792 303380    0    0     0 74064  565    50  1 11 52 36
 1  0      0 662488   5864 319396    0    0     4   352  302   329  0  2 47 51
 0  1      0 599312   5924 381468    0    0     0 78164  516    55  0  9 51 40
 0  1      0 519952   6008 459516    0    0     4 78156  622    56  1 11 52 37
 1  1      0 436640   6092 541632    0    0     0 82244  622    54  0 11 48 41
 0  1      0 436640   6092 541660    0    0     0     8  152    39  0  0 51 49
 0  1      0 332224   6200 644252    0    0     4 102800  728    46  1 13 49 36
 1  0      0 274492   6260 701056    0    0     4 12328  459    49  0  7 50 43
 0  1      0 211220   6324 763356    0    0     0 106940  515    37  1 10 51 39
 1  0      0 160412   6376 813468    0    0     0  8224  415    43  0  6 49 45
 1  1      0  85980   6452 886556    0    0     4 113516  575    39  1 11 54 34
 0  2      0  85968   6452 886620    0    0     0  1640  158   211  0  0 46 54

A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A
SSD based writeback test on XFS performs over 20% better as well, with
the throughput being very stable around 1GB/sec, where pdflush only
manages 750MB/sec and fluctuates wildly while doing so. Random buffered
writes to many files behave a lot better as well, as does random mmap'ed
writes.

A separate thread is added to sync the super blocks. In the long term,
adding sync_supers_bdi() functionality could get rid of this thread again.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-11 09:20:25 +02:00

432 lines
11 KiB
C

/*
* High-level sync()-related operations
*/
#include <linux/kernel.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/writeback.h>
#include <linux/syscalls.h>
#include <linux/linkage.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/buffer_head.h>
#include "internal.h"
#define VALID_FLAGS (SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE| \
SYNC_FILE_RANGE_WAIT_AFTER)
/*
* Do the filesystem syncing work. For simple filesystems
* writeback_inodes_sb(sb) just dirties buffers with inodes so we have to
* submit IO for these buffers via __sync_blockdev(). This also speeds up the
* wait == 1 case since in that case write_inode() functions do
* sync_dirty_buffer() and thus effectively write one block at a time.
*/
static int __sync_filesystem(struct super_block *sb, int wait)
{
/* Avoid doing twice syncing and cache pruning for quota sync */
if (!wait) {
writeout_quota_sb(sb, -1);
writeback_inodes_sb(sb);
} else {
sync_quota_sb(sb, -1);
sync_inodes_sb(sb);
}
if (sb->s_op->sync_fs)
sb->s_op->sync_fs(sb, wait);
return __sync_blockdev(sb->s_bdev, wait);
}
/*
* Write out and wait upon all dirty data associated with this
* superblock. Filesystem data as well as the underlying block
* device. Takes the superblock lock.
*/
int sync_filesystem(struct super_block *sb)
{
int ret;
/*
* We need to be protected against the filesystem going from
* r/o to r/w or vice versa.
*/
WARN_ON(!rwsem_is_locked(&sb->s_umount));
/*
* No point in syncing out anything if the filesystem is read-only.
*/
if (sb->s_flags & MS_RDONLY)
return 0;
ret = __sync_filesystem(sb, 0);
if (ret < 0)
return ret;
return __sync_filesystem(sb, 1);
}
EXPORT_SYMBOL_GPL(sync_filesystem);
/*
* Sync all the data for all the filesystems (called by sys_sync() and
* emergency sync)
*
* This operation is careful to avoid the livelock which could easily happen
* if two or more filesystems are being continuously dirtied. s_need_sync
* is used only here. We set it against all filesystems and then clear it as
* we sync them. So redirtied filesystems are skipped.
*
* But if process A is currently running sync_filesystems and then process B
* calls sync_filesystems as well, process B will set all the s_need_sync
* flags again, which will cause process A to resync everything. Fix that with
* a local mutex.
*/
static void sync_filesystems(int wait)
{
struct super_block *sb;
static DEFINE_MUTEX(mutex);
mutex_lock(&mutex); /* Could be down_interruptible */
spin_lock(&sb_lock);
list_for_each_entry(sb, &super_blocks, s_list)
sb->s_need_sync = 1;
restart:
list_for_each_entry(sb, &super_blocks, s_list) {
if (!sb->s_need_sync)
continue;
sb->s_need_sync = 0;
sb->s_count++;
spin_unlock(&sb_lock);
down_read(&sb->s_umount);
if (!(sb->s_flags & MS_RDONLY) && sb->s_root)
__sync_filesystem(sb, wait);
up_read(&sb->s_umount);
/* restart only when sb is no longer on the list */
spin_lock(&sb_lock);
if (__put_super_and_need_restart(sb))
goto restart;
}
spin_unlock(&sb_lock);
mutex_unlock(&mutex);
}
/*
* sync everything. Start out by waking pdflush, because that writes back
* all queues in parallel.
*/
SYSCALL_DEFINE0(sync)
{
wakeup_flusher_threads(0);
sync_filesystems(0);
sync_filesystems(1);
if (unlikely(laptop_mode))
laptop_sync_completion();
return 0;
}
static void do_sync_work(struct work_struct *work)
{
/*
* Sync twice to reduce the possibility we skipped some inodes / pages
* because they were temporarily locked
*/
sync_filesystems(0);
sync_filesystems(0);
printk("Emergency Sync complete\n");
kfree(work);
}
void emergency_sync(void)
{
struct work_struct *work;
work = kmalloc(sizeof(*work), GFP_ATOMIC);
if (work) {
INIT_WORK(work, do_sync_work);
schedule_work(work);
}
}
/*
* Generic function to fsync a file.
*
* filp may be NULL if called via the msync of a vma.
*/
int file_fsync(struct file *filp, struct dentry *dentry, int datasync)
{
struct inode * inode = dentry->d_inode;
struct super_block * sb;
int ret, err;
/* sync the inode to buffers */
ret = write_inode_now(inode, 0);
/* sync the superblock to buffers */
sb = inode->i_sb;
if (sb->s_dirt && sb->s_op->write_super)
sb->s_op->write_super(sb);
/* .. finally sync the buffers to disk */
err = sync_blockdev(sb->s_bdev);
if (!ret)
ret = err;
return ret;
}
/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @dentry: dentry of @file
* @data: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*
* In case this function is called from nfsd @file may be %NULL and
* only @dentry is set. This can only happen when the filesystem
* implements the export_operations API.
*/
int vfs_fsync(struct file *file, struct dentry *dentry, int datasync)
{
const struct file_operations *fop;
struct address_space *mapping;
int err, ret;
/*
* Get mapping and operations from the file in case we have
* as file, or get the default values for them in case we
* don't have a struct file available. Damn nfsd..
*/
if (file) {
mapping = file->f_mapping;
fop = file->f_op;
} else {
mapping = dentry->d_inode->i_mapping;
fop = dentry->d_inode->i_fop;
}
if (!fop || !fop->fsync) {
ret = -EINVAL;
goto out;
}
ret = filemap_fdatawrite(mapping);
/*
* We need to protect against concurrent writers, which could cause
* livelocks in fsync_buffers_list().
*/
mutex_lock(&mapping->host->i_mutex);
err = fop->fsync(file, dentry, datasync);
if (!ret)
ret = err;
mutex_unlock(&mapping->host->i_mutex);
err = filemap_fdatawait(mapping);
if (!ret)
ret = err;
out:
return ret;
}
EXPORT_SYMBOL(vfs_fsync);
static int do_fsync(unsigned int fd, int datasync)
{
struct file *file;
int ret = -EBADF;
file = fget(fd);
if (file) {
ret = vfs_fsync(file, file->f_path.dentry, datasync);
fput(file);
}
return ret;
}
SYSCALL_DEFINE1(fsync, unsigned int, fd)
{
return do_fsync(fd, 0);
}
SYSCALL_DEFINE1(fdatasync, unsigned int, fd)
{
return do_fsync(fd, 1);
}
/*
* sys_sync_file_range() permits finely controlled syncing over a segment of
* a file in the range offset .. (offset+nbytes-1) inclusive. If nbytes is
* zero then sys_sync_file_range() will operate from offset out to EOF.
*
* The flag bits are:
*
* SYNC_FILE_RANGE_WAIT_BEFORE: wait upon writeout of all pages in the range
* before performing the write.
*
* SYNC_FILE_RANGE_WRITE: initiate writeout of all those dirty pages in the
* range which are not presently under writeback. Note that this may block for
* significant periods due to exhaustion of disk request structures.
*
* SYNC_FILE_RANGE_WAIT_AFTER: wait upon writeout of all pages in the range
* after performing the write.
*
* Useful combinations of the flag bits are:
*
* SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE: ensures that all pages
* in the range which were dirty on entry to sys_sync_file_range() are placed
* under writeout. This is a start-write-for-data-integrity operation.
*
* SYNC_FILE_RANGE_WRITE: start writeout of all dirty pages in the range which
* are not presently under writeout. This is an asynchronous flush-to-disk
* operation. Not suitable for data integrity operations.
*
* SYNC_FILE_RANGE_WAIT_BEFORE (or SYNC_FILE_RANGE_WAIT_AFTER): wait for
* completion of writeout of all pages in the range. This will be used after an
* earlier SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE operation to wait
* for that operation to complete and to return the result.
*
* SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE|SYNC_FILE_RANGE_WAIT_AFTER:
* a traditional sync() operation. This is a write-for-data-integrity operation
* which will ensure that all pages in the range which were dirty on entry to
* sys_sync_file_range() are committed to disk.
*
*
* SYNC_FILE_RANGE_WAIT_BEFORE and SYNC_FILE_RANGE_WAIT_AFTER will detect any
* I/O errors or ENOSPC conditions and will return those to the caller, after
* clearing the EIO and ENOSPC flags in the address_space.
*
* It should be noted that none of these operations write out the file's
* metadata. So unless the application is strictly performing overwrites of
* already-instantiated disk blocks, there are no guarantees here that the data
* will be available after a crash.
*/
SYSCALL_DEFINE(sync_file_range)(int fd, loff_t offset, loff_t nbytes,
unsigned int flags)
{
int ret;
struct file *file;
loff_t endbyte; /* inclusive */
int fput_needed;
umode_t i_mode;
ret = -EINVAL;
if (flags & ~VALID_FLAGS)
goto out;
endbyte = offset + nbytes;
if ((s64)offset < 0)
goto out;
if ((s64)endbyte < 0)
goto out;
if (endbyte < offset)
goto out;
if (sizeof(pgoff_t) == 4) {
if (offset >= (0x100000000ULL << PAGE_CACHE_SHIFT)) {
/*
* The range starts outside a 32 bit machine's
* pagecache addressing capabilities. Let it "succeed"
*/
ret = 0;
goto out;
}
if (endbyte >= (0x100000000ULL << PAGE_CACHE_SHIFT)) {
/*
* Out to EOF
*/
nbytes = 0;
}
}
if (nbytes == 0)
endbyte = LLONG_MAX;
else
endbyte--; /* inclusive */
ret = -EBADF;
file = fget_light(fd, &fput_needed);
if (!file)
goto out;
i_mode = file->f_path.dentry->d_inode->i_mode;
ret = -ESPIPE;
if (!S_ISREG(i_mode) && !S_ISBLK(i_mode) && !S_ISDIR(i_mode) &&
!S_ISLNK(i_mode))
goto out_put;
ret = do_sync_mapping_range(file->f_mapping, offset, endbyte, flags);
out_put:
fput_light(file, fput_needed);
out:
return ret;
}
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_sync_file_range(long fd, loff_t offset, loff_t nbytes,
long flags)
{
return SYSC_sync_file_range((int) fd, offset, nbytes,
(unsigned int) flags);
}
SYSCALL_ALIAS(sys_sync_file_range, SyS_sync_file_range);
#endif
/* It would be nice if people remember that not all the world's an i386
when they introduce new system calls */
SYSCALL_DEFINE(sync_file_range2)(int fd, unsigned int flags,
loff_t offset, loff_t nbytes)
{
return sys_sync_file_range(fd, offset, nbytes, flags);
}
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_sync_file_range2(long fd, long flags,
loff_t offset, loff_t nbytes)
{
return SYSC_sync_file_range2((int) fd, (unsigned int) flags,
offset, nbytes);
}
SYSCALL_ALIAS(sys_sync_file_range2, SyS_sync_file_range2);
#endif
/*
* `endbyte' is inclusive
*/
int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
loff_t endbyte, unsigned int flags)
{
int ret;
if (!mapping) {
ret = -EINVAL;
goto out;
}
ret = 0;
if (flags & SYNC_FILE_RANGE_WAIT_BEFORE) {
ret = wait_on_page_writeback_range(mapping,
offset >> PAGE_CACHE_SHIFT,
endbyte >> PAGE_CACHE_SHIFT);
if (ret < 0)
goto out;
}
if (flags & SYNC_FILE_RANGE_WRITE) {
ret = __filemap_fdatawrite_range(mapping, offset, endbyte,
WB_SYNC_ALL);
if (ret < 0)
goto out;
}
if (flags & SYNC_FILE_RANGE_WAIT_AFTER) {
ret = wait_on_page_writeback_range(mapping,
offset >> PAGE_CACHE_SHIFT,
endbyte >> PAGE_CACHE_SHIFT);
}
out:
return ret;
}
EXPORT_SYMBOL_GPL(do_sync_mapping_range);