[net-next-2.6.git] / fs / sync.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 sync_inodes_sb(sb, 0)
 * 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);
	else
		sync_quota_sb(sb, -1);
	sync_inodes_sb(sb, wait);
	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_pdflush(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);
Commit	Line	Data
f79e2abb AM	1	/*
	2	* High-level sync()-related operations
	3	*/
	4
	5	#include <linux/kernel.h>
	6	#include <linux/file.h>
	7	#include <linux/fs.h>
	8	#include <linux/module.h>
914e2637	9	#include <linux/sched.h>
f79e2abb AM	10	#include <linux/writeback.h>
	11	#include <linux/syscalls.h>
	12	#include <linux/linkage.h>
	13	#include <linux/pagemap.h>
cf9a2ae8 DH	14	#include <linux/quotaops.h>
cf9a2ae8 DH	15	#include <linux/buffer_head.h>
5a3e5cb8	16	#include "internal.h"
f79e2abb AM	17
	18	#define VALID_FLAGS (SYNC_FILE_RANGE_WAIT_BEFORE\|SYNC_FILE_RANGE_WRITE\| \
	19	SYNC_FILE_RANGE_WAIT_AFTER)
	20
c15c54f5 JK	21	/*
	22	* Do the filesystem syncing work. For simple filesystems sync_inodes_sb(sb, 0)
	23	* just dirties buffers with inodes so we have to submit IO for these buffers
	24	* via __sync_blockdev(). This also speeds up the wait == 1 case since in that
	25	* case write_inode() functions do sync_dirty_buffer() and thus effectively
	26	* write one block at a time.
	27	*/
60b0680f	28	static int __sync_filesystem(struct super_block *sb, int wait)
c15c54f5	29	{
c3f8a40c JK	30	/* Avoid doing twice syncing and cache pruning for quota sync */
	31	if (!wait)
	32	writeout_quota_sb(sb, -1);
	33	else
	34	sync_quota_sb(sb, -1);
c15c54f5	35	sync_inodes_sb(sb, wait);
c15c54f5 JK	36	if (sb->s_op->sync_fs)
	37	sb->s_op->sync_fs(sb, wait);
	38	return __sync_blockdev(sb->s_bdev, wait);
	39	}
	40
	41	/*
	42	* Write out and wait upon all dirty data associated with this
	43	* superblock. Filesystem data as well as the underlying block
	44	* device. Takes the superblock lock.
	45	*/
60b0680f	46	int sync_filesystem(struct super_block *sb)
c15c54f5 JK	47	{
	48	int ret;
	49
5af7926f CH	50	/*
	51	* We need to be protected against the filesystem going from
	52	* r/o to r/w or vice versa.
	53	*/
	54	WARN_ON(!rwsem_is_locked(&sb->s_umount));
	55
	56	/*
	57	* No point in syncing out anything if the filesystem is read-only.
	58	*/
	59	if (sb->s_flags & MS_RDONLY)
	60	return 0;
	61
60b0680f	62	ret = __sync_filesystem(sb, 0);
c15c54f5 JK	63	if (ret < 0)
c15c54f5 JK	64	return ret;
60b0680f	65	return __sync_filesystem(sb, 1);
c15c54f5	66	}
60b0680f	67	EXPORT_SYMBOL_GPL(sync_filesystem);
c15c54f5 JK	68
	69	/*
	70	* Sync all the data for all the filesystems (called by sys_sync() and
	71	* emergency sync)
	72	*
	73	* This operation is careful to avoid the livelock which could easily happen
	74	* if two or more filesystems are being continuously dirtied. s_need_sync
	75	* is used only here. We set it against all filesystems and then clear it as
	76	* we sync them. So redirtied filesystems are skipped.
	77	*
	78	* But if process A is currently running sync_filesystems and then process B
	79	* calls sync_filesystems as well, process B will set all the s_need_sync
	80	* flags again, which will cause process A to resync everything. Fix that with
	81	* a local mutex.
	82	*/
	83	static void sync_filesystems(int wait)
	84	{
	85	struct super_block *sb;
	86	static DEFINE_MUTEX(mutex);
	87
	88	mutex_lock(&mutex); /* Could be down_interruptible */
	89	spin_lock(&sb_lock);
5af7926f	90	list_for_each_entry(sb, &super_blocks, s_list)
c15c54f5	91	sb->s_need_sync = 1;
c15c54f5 JK	92
	93	restart:
	94	list_for_each_entry(sb, &super_blocks, s_list) {
	95	if (!sb->s_need_sync)
	96	continue;
	97	sb->s_need_sync = 0;
c15c54f5 JK	98	sb->s_count++;
c15c54f5 JK	99	spin_unlock(&sb_lock);
5af7926f	100
c15c54f5	101	down_read(&sb->s_umount);
5af7926f	102	if (!(sb->s_flags & MS_RDONLY) && sb->s_root)
60b0680f	103	__sync_filesystem(sb, wait);
c15c54f5	104	up_read(&sb->s_umount);
5af7926f	105
c15c54f5 JK	106	/* restart only when sb is no longer on the list */
	107	spin_lock(&sb_lock);
	108	if (__put_super_and_need_restart(sb))
	109	goto restart;
	110	}
	111	spin_unlock(&sb_lock);
	112	mutex_unlock(&mutex);
	113	}
	114
3beab0b4 ZY	115	/*
	116	* sync everything. Start out by waking pdflush, because that writes back
	117	* all queues in parallel.
	118	*/
5cee5815	119	SYSCALL_DEFINE0(sync)
cf9a2ae8	120	{
3beab0b4	121	wakeup_pdflush(0);
5cee5815 JK	122	sync_filesystems(0);
5cee5815 JK	123	sync_filesystems(1);
cf9a2ae8 DH	124	if (unlikely(laptop_mode))
cf9a2ae8 DH	125	laptop_sync_completion();
cf9a2ae8 DH	126	return 0;
	127	}
	128
a2a9537a JA	129	static void do_sync_work(struct work_struct *work)
a2a9537a JA	130	{
5cee5815 JK	131	/*
	132	* Sync twice to reduce the possibility we skipped some inodes / pages
	133	* because they were temporarily locked
	134	*/
	135	sync_filesystems(0);
	136	sync_filesystems(0);
	137	printk("Emergency Sync complete\n");
a2a9537a JA	138	kfree(work);
	139	}
	140
cf9a2ae8 DH	141	void emergency_sync(void)
cf9a2ae8 DH	142	{
a2a9537a JA	143	struct work_struct *work;
	144
	145	work = kmalloc(sizeof(*work), GFP_ATOMIC);
	146	if (work) {
	147	INIT_WORK(work, do_sync_work);
	148	schedule_work(work);
	149	}
cf9a2ae8 DH	150	}
	151
	152	/*
	153	* Generic function to fsync a file.
	154	*
	155	* filp may be NULL if called via the msync of a vma.
	156	*/
	157	int file_fsync(struct file filp, struct dentry dentry, int datasync)
	158	{
	159	struct inode * inode = dentry->d_inode;
	160	struct super_block * sb;
	161	int ret, err;
	162
	163	/* sync the inode to buffers */
	164	ret = write_inode_now(inode, 0);
	165
	166	/* sync the superblock to buffers */
	167	sb = inode->i_sb;
762873c2	168	if (sb->s_dirt && sb->s_op->write_super)
cf9a2ae8	169	sb->s_op->write_super(sb);
cf9a2ae8 DH	170
	171	/* .. finally sync the buffers to disk */
	172	err = sync_blockdev(sb->s_bdev);
	173	if (!ret)
	174	ret = err;
	175	return ret;
	176	}
	177
4c728ef5 CH	178	/**
	179	* vfs_fsync - perform a fsync or fdatasync on a file
	180	* @file: file to sync
	181	* @dentry: dentry of @file
	182	* @data: only perform a fdatasync operation
	183	*
	184	* Write back data and metadata for @file to disk. If @datasync is
	185	* set only metadata needed to access modified file data is written.
	186	*
	187	* In case this function is called from nfsd @file may be %NULL and
	188	* only @dentry is set. This can only happen when the filesystem
	189	* implements the export_operations API.
	190	*/
	191	int vfs_fsync(struct file file, struct dentry dentry, int datasync)
cf9a2ae8	192	{
4c728ef5 CH	193	const struct file_operations *fop;
	194	struct address_space *mapping;
	195	int err, ret;
	196
	197	/*
	198	* Get mapping and operations from the file in case we have
	199	* as file, or get the default values for them in case we
	200	* don't have a struct file available. Damn nfsd..
	201	*/
	202	if (file) {
	203	mapping = file->f_mapping;
	204	fop = file->f_op;
	205	} else {
	206	mapping = dentry->d_inode->i_mapping;
	207	fop = dentry->d_inode->i_fop;
	208	}
cf9a2ae8	209
4c728ef5	210	if (!fop \|\| !fop->fsync) {
cf9a2ae8 DH	211	ret = -EINVAL;
	212	goto out;
	213	}
	214
	215	ret = filemap_fdatawrite(mapping);
	216
	217	/*
	218	* We need to protect against concurrent writers, which could cause
	219	* livelocks in fsync_buffers_list().
	220	*/
	221	mutex_lock(&mapping->host->i_mutex);
4c728ef5	222	err = fop->fsync(file, dentry, datasync);
cf9a2ae8 DH	223	if (!ret)
	224	ret = err;
	225	mutex_unlock(&mapping->host->i_mutex);
	226	err = filemap_fdatawait(mapping);
	227	if (!ret)
	228	ret = err;
	229	out:
	230	return ret;
	231	}
4c728ef5	232	EXPORT_SYMBOL(vfs_fsync);
cf9a2ae8	233
4c728ef5	234	static int do_fsync(unsigned int fd, int datasync)
cf9a2ae8 DH	235	{
	236	struct file *file;
	237	int ret = -EBADF;
	238
	239	file = fget(fd);
	240	if (file) {
4c728ef5	241	ret = vfs_fsync(file, file->f_path.dentry, datasync);
cf9a2ae8 DH	242	fput(file);
	243	}
	244	return ret;
	245	}
	246
a5f8fa9e	247	SYSCALL_DEFINE1(fsync, unsigned int, fd)
cf9a2ae8	248	{
4c728ef5	249	return do_fsync(fd, 0);
cf9a2ae8 DH	250	}
cf9a2ae8 DH	251
a5f8fa9e	252	SYSCALL_DEFINE1(fdatasync, unsigned int, fd)
cf9a2ae8	253	{
4c728ef5	254	return do_fsync(fd, 1);
cf9a2ae8 DH	255	}
cf9a2ae8 DH	256
f79e2abb AM	257	/*
	258	* sys_sync_file_range() permits finely controlled syncing over a segment of
	259	* a file in the range offset .. (offset+nbytes-1) inclusive. If nbytes is
	260	* zero then sys_sync_file_range() will operate from offset out to EOF.
	261	*
	262	* The flag bits are:
	263	*
	264	* SYNC_FILE_RANGE_WAIT_BEFORE: wait upon writeout of all pages in the range
	265	* before performing the write.
	266	*
	267	* SYNC_FILE_RANGE_WRITE: initiate writeout of all those dirty pages in the
cce77081 PM	268	* range which are not presently under writeback. Note that this may block for
cce77081 PM	269	* significant periods due to exhaustion of disk request structures.
f79e2abb AM	270	*
	271	* SYNC_FILE_RANGE_WAIT_AFTER: wait upon writeout of all pages in the range
	272	* after performing the write.
	273	*
	274	* Useful combinations of the flag bits are:
	275	*
	276	* SYNC_FILE_RANGE_WAIT_BEFORE\|SYNC_FILE_RANGE_WRITE: ensures that all pages
	277	* in the range which were dirty on entry to sys_sync_file_range() are placed
	278	* under writeout. This is a start-write-for-data-integrity operation.
	279	*
	280	* SYNC_FILE_RANGE_WRITE: start writeout of all dirty pages in the range which
	281	* are not presently under writeout. This is an asynchronous flush-to-disk
	282	* operation. Not suitable for data integrity operations.
	283	*
	284	* SYNC_FILE_RANGE_WAIT_BEFORE (or SYNC_FILE_RANGE_WAIT_AFTER): wait for
	285	* completion of writeout of all pages in the range. This will be used after an
	286	* earlier SYNC_FILE_RANGE_WAIT_BEFORE\|SYNC_FILE_RANGE_WRITE operation to wait
	287	* for that operation to complete and to return the result.
	288	*
	289	* SYNC_FILE_RANGE_WAIT_BEFORE\|SYNC_FILE_RANGE_WRITE\|SYNC_FILE_RANGE_WAIT_AFTER:
	290	* a traditional sync() operation. This is a write-for-data-integrity operation
	291	* which will ensure that all pages in the range which were dirty on entry to
	292	* sys_sync_file_range() are committed to disk.
	293	*
	294	*
	295	* SYNC_FILE_RANGE_WAIT_BEFORE and SYNC_FILE_RANGE_WAIT_AFTER will detect any
	296	* I/O errors or ENOSPC conditions and will return those to the caller, after
	297	* clearing the EIO and ENOSPC flags in the address_space.
	298	*
	299	* It should be noted that none of these operations write out the file's
	300	* metadata. So unless the application is strictly performing overwrites of
	301	* already-instantiated disk blocks, there are no guarantees here that the data
	302	* will be available after a crash.
	303	*/
6673e0c3 HC	304	SYSCALL_DEFINE(sync_file_range)(int fd, loff_t offset, loff_t nbytes,
6673e0c3 HC	305	unsigned int flags)
f79e2abb AM	306	{
	307	int ret;
	308	struct file *file;
	309	loff_t endbyte; /* inclusive */
	310	int fput_needed;
	311	umode_t i_mode;
	312
	313	ret = -EINVAL;
	314	if (flags & ~VALID_FLAGS)
	315	goto out;
	316
	317	endbyte = offset + nbytes;
	318
	319	if ((s64)offset < 0)
	320	goto out;
	321	if ((s64)endbyte < 0)
	322	goto out;
	323	if (endbyte < offset)
	324	goto out;
	325
	326	if (sizeof(pgoff_t) == 4) {
	327	if (offset >= (0x100000000ULL << PAGE_CACHE_SHIFT)) {
	328	/*
	329	* The range starts outside a 32 bit machine's
	330	* pagecache addressing capabilities. Let it "succeed"
	331	*/
	332	ret = 0;
	333	goto out;
	334	}
	335	if (endbyte >= (0x100000000ULL << PAGE_CACHE_SHIFT)) {
	336	/*
	337	* Out to EOF
	338	*/
	339	nbytes = 0;
	340	}
	341	}
	342
	343	if (nbytes == 0)
111ebb6e	344	endbyte = LLONG_MAX;
f79e2abb AM	345	else
	346	endbyte--; /* inclusive */
	347
	348	ret = -EBADF;
	349	file = fget_light(fd, &fput_needed);
	350	if (!file)
	351	goto out;
	352
0f7fc9e4	353	i_mode = file->f_path.dentry->d_inode->i_mode;
f79e2abb AM	354	ret = -ESPIPE;
	355	if (!S_ISREG(i_mode) && !S_ISBLK(i_mode) && !S_ISDIR(i_mode) &&
	356	!S_ISLNK(i_mode))
	357	goto out_put;
	358
ef51c976	359	ret = do_sync_mapping_range(file->f_mapping, offset, endbyte, flags);
f79e2abb AM	360	out_put:
	361	fput_light(file, fput_needed);
	362	out:
	363	return ret;
	364	}
6673e0c3 HC	365	#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
	366	asmlinkage long SyS_sync_file_range(long fd, loff_t offset, loff_t nbytes,
	367	long flags)
	368	{
	369	return SYSC_sync_file_range((int) fd, offset, nbytes,
	370	(unsigned int) flags);
	371	}
	372	SYSCALL_ALIAS(sys_sync_file_range, SyS_sync_file_range);
	373	#endif
f79e2abb	374
edd5cd4a DW	375	/* It would be nice if people remember that not all the world's an i386
edd5cd4a DW	376	when they introduce new system calls */
6673e0c3 HC	377	SYSCALL_DEFINE(sync_file_range2)(int fd, unsigned int flags,
6673e0c3 HC	378	loff_t offset, loff_t nbytes)
edd5cd4a DW	379	{
	380	return sys_sync_file_range(fd, offset, nbytes, flags);
	381	}
6673e0c3 HC	382	#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
	383	asmlinkage long SyS_sync_file_range2(long fd, long flags,
	384	loff_t offset, loff_t nbytes)
	385	{
	386	return SYSC_sync_file_range2((int) fd, (unsigned int) flags,
	387	offset, nbytes);
	388	}
	389	SYSCALL_ALIAS(sys_sync_file_range2, SyS_sync_file_range2);
	390	#endif
edd5cd4a	391
f79e2abb AM	392	/*
	393	* `endbyte' is inclusive
	394	*/
5b04aa3a MF	395	int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
5b04aa3a MF	396	loff_t endbyte, unsigned int flags)
f79e2abb AM	397	{
f79e2abb AM	398	int ret;
f79e2abb	399
f79e2abb AM	400	if (!mapping) {
	401	ret = -EINVAL;
	402	goto out;
	403	}
	404
	405	ret = 0;
	406	if (flags & SYNC_FILE_RANGE_WAIT_BEFORE) {
	407	ret = wait_on_page_writeback_range(mapping,
	408	offset >> PAGE_CACHE_SHIFT,
	409	endbyte >> PAGE_CACHE_SHIFT);
	410	if (ret < 0)
	411	goto out;
	412	}
	413
	414	if (flags & SYNC_FILE_RANGE_WRITE) {
	415	ret = __filemap_fdatawrite_range(mapping, offset, endbyte,
ee53a891	416	WB_SYNC_ALL);
f79e2abb AM	417	if (ret < 0)
	418	goto out;
	419	}
	420
	421	if (flags & SYNC_FILE_RANGE_WAIT_AFTER) {
	422	ret = wait_on_page_writeback_range(mapping,
	423	offset >> PAGE_CACHE_SHIFT,
	424	endbyte >> PAGE_CACHE_SHIFT);
	425	}
	426	out:
	427	return ret;
	428	}
5b04aa3a	429	EXPORT_SYMBOL_GPL(do_sync_mapping_range);