#include <linux/pagevec.h>
#include <linux/writeback.h>
-/*
- * Types of I/O for bmap clustering and I/O completion tracking.
- */
-enum {
- IO_READ, /* mapping for a read */
- IO_DELAY, /* mapping covers delalloc region */
- IO_UNWRITTEN, /* mapping covers allocated but uninitialized data */
- IO_NEW /* just allocated */
-};
/*
* Prime number of hash buckets since address is used as the key.
xfs_inode_t *ip = XFS_I(ioend->io_inode);
xfs_fsize_t isize;
- ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
- ASSERT(ioend->io_type != IO_READ);
-
if (unlikely(ioend->io_error))
return 0;
* We might have to update the on-disk file size after extending
* writes.
*/
- if (ioend->io_type != IO_READ) {
- error = xfs_setfilesize(ioend);
- ASSERT(!error || error == EAGAIN);
- }
+ error = xfs_setfilesize(ioend);
+ ASSERT(!error || error == EAGAIN);
/*
* If we didn't complete processing of the ioend, requeue it to the
loff_t offset,
ssize_t count,
struct xfs_bmbt_irec *imap,
- int flags)
+ int type,
+ int nonblocking)
{
- int nmaps = 1;
- int new = 0;
+ struct xfs_inode *ip = XFS_I(inode);
+ struct xfs_mount *mp = ip->i_mount;
+ xfs_fileoff_t offset_fsb, end_fsb;
+ int error = 0;
+ int lockmode = 0;
+ int bmapi_flags = XFS_BMAPI_ENTIRE;
+ int nimaps = 1;
+
+ if (XFS_FORCED_SHUTDOWN(mp))
+ return -XFS_ERROR(EIO);
+
+ switch (type) {
+ case IO_OVERWRITE:
+ lockmode = xfs_ilock_map_shared(ip);
+ break;
+ case IO_UNWRITTEN:
+ lockmode = XFS_ILOCK_EXCL;
+ bmapi_flags |= XFS_BMAPI_IGSTATE;
+ xfs_ilock(ip, lockmode);
+ break;
+ case IO_DELALLOC:
+ lockmode = XFS_ILOCK_SHARED;
+
+ if (!xfs_ilock_nowait(ip, lockmode)) {
+ if (nonblocking)
+ return -XFS_ERROR(EAGAIN);
+ xfs_ilock(ip, lockmode);
+ }
+ break;
+ }
+
+ ASSERT(offset <= mp->m_maxioffset);
+ if (offset + count > mp->m_maxioffset)
+ count = mp->m_maxioffset - offset;
+ end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
+ offset_fsb = XFS_B_TO_FSBT(mp, offset);
+
+ error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+ bmapi_flags, NULL, 0, imap, &nimaps, NULL);
+ if (error)
+ goto out;
+
+ switch (type) {
+ case IO_UNWRITTEN:
+ /* If we found an extent, return it */
+ if (nimaps &&
+ (imap->br_startblock != HOLESTARTBLOCK) &&
+ (imap->br_startblock != DELAYSTARTBLOCK)) {
+ trace_xfs_map_blocks_found(ip, offset, count, type, imap);
+ break;
+ }
+
+ error = xfs_iomap_write_delay(ip, offset, count, imap);
+ if (!error)
+ trace_xfs_map_blocks_alloc(ip, offset, count, type, imap);
+ break;
+ case IO_DELALLOC:
+ /* If we found an extent, return it */
+ xfs_iunlock(ip, lockmode);
+ lockmode = 0;
+
+ if (nimaps && !isnullstartblock(imap->br_startblock)) {
+ trace_xfs_map_blocks_found(ip, offset, count, type, imap);
+ break;
+ }
+
+ error = xfs_iomap_write_allocate(ip, offset, count, imap);
+ if (!error)
+ trace_xfs_map_blocks_alloc(ip, offset, count, type, imap);
+ break;
+ default:
+ if (nimaps)
+ trace_xfs_map_blocks_found(ip, offset, count, type, imap);
+ }
- return -xfs_iomap(XFS_I(inode), offset, count, flags, imap, &nmaps, &new);
+out:
+ if (lockmode)
+ xfs_iunlock(ip, lockmode);
+ return -XFS_ERROR(error);
}
STATIC int
if (buffer_unwritten(bh))
acceptable = (type == IO_UNWRITTEN);
else if (buffer_delay(bh))
- acceptable = (type == IO_DELAY);
+ acceptable = (type == IO_DELALLOC);
else if (buffer_dirty(bh) && buffer_mapped(bh))
- acceptable = (type == IO_NEW);
+ acceptable = (type == IO_OVERWRITE);
else
break;
} while ((bh = bh->b_this_page) != head);
if (buffer_unwritten(bh))
type = IO_UNWRITTEN;
else
- type = IO_DELAY;
+ type = IO_DELALLOC;
if (!xfs_imap_valid(inode, imap, offset)) {
done = 1;
page_dirty--;
count++;
} else {
- type = IO_NEW;
+ type = IO_OVERWRITE;
if (buffer_mapped(bh) && all_bh) {
lock_buffer(bh);
xfs_add_to_ioend(inode, bh, offset,
struct buffer_head *bh, *head;
loff_t offset = page_offset(page);
- if (!xfs_is_delayed_page(page, IO_DELAY))
+ if (!xfs_is_delayed_page(page, IO_DELALLOC))
goto out_invalidate;
if (XFS_FORCED_SHUTDOWN(ip->i_mount))
__uint64_t end_offset;
pgoff_t end_index, last_index;
ssize_t size, len;
- int flags, err, imap_valid = 0, uptodate = 1;
+ int err, imap_valid = 0, uptodate = 1;
int count = 0;
int all_bh = 0;
+ int nonblocking = 0;
trace_xfs_writepage(inode, page, 0);
bh = head = page_buffers(page);
offset = page_offset(page);
- flags = BMAPI_READ;
- type = IO_NEW;
+ type = IO_OVERWRITE;
+
+ if (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking)
+ nonblocking = 1;
do {
int new_ioend = 0;
type = IO_UNWRITTEN;
imap_valid = 0;
}
- flags = BMAPI_WRITE | BMAPI_IGNSTATE;
} else if (buffer_delay(bh)) {
- if (type != IO_DELAY) {
- type = IO_DELAY;
+ if (type != IO_DELALLOC) {
+ type = IO_DELALLOC;
imap_valid = 0;
}
- flags = BMAPI_ALLOCATE;
-
- if (wbc->sync_mode == WB_SYNC_NONE)
- flags |= BMAPI_TRYLOCK;
}
if (!imap_valid) {
* for unwritten extent conversion.
*/
new_ioend = 1;
- err = xfs_map_blocks(inode, offset, len,
- &imap, flags);
+ err = xfs_map_blocks(inode, offset, len, &imap,
+ type, nonblocking);
if (err)
goto error;
imap_valid = xfs_imap_valid(inode, &imap,
* That means it must already have extents allocated
* underneath it. Map the extent by reading it.
*/
- if (flags != BMAPI_READ) {
- flags = BMAPI_READ;
+ if (type != IO_OVERWRITE) {
+ type = IO_OVERWRITE;
imap_valid = 0;
}
if (!imap_valid) {
new_ioend = 1;
size = xfs_probe_cluster(inode, page, bh, head);
err = xfs_map_blocks(inode, offset, size,
- &imap, flags);
+ &imap, type, nonblocking);
if (err)
goto error;
imap_valid = xfs_imap_valid(inode, &imap,
offset);
}
- /*
- * We set the type to IO_NEW in case we are doing a
- * small write at EOF that is extending the file but
- * without needing an allocation. We need to update the
- * file size on I/O completion in this case so it is
- * the same case as having just allocated a new extent
- * that we are writing into for the first time.
- */
- type = IO_NEW;
if (imap_valid) {
all_bh = 1;
lock_buffer(bh);
int create,
int direct)
{
- int flags = create ? BMAPI_WRITE : BMAPI_READ;
+ struct xfs_inode *ip = XFS_I(inode);
+ struct xfs_mount *mp = ip->i_mount;
+ xfs_fileoff_t offset_fsb, end_fsb;
+ int error = 0;
+ int lockmode = 0;
struct xfs_bmbt_irec imap;
+ int nimaps = 1;
xfs_off_t offset;
ssize_t size;
- int nimap = 1;
int new = 0;
- int error;
+
+ if (XFS_FORCED_SHUTDOWN(mp))
+ return -XFS_ERROR(EIO);
offset = (xfs_off_t)iblock << inode->i_blkbits;
ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
if (!create && direct && offset >= i_size_read(inode))
return 0;
- if (direct && create)
- flags |= BMAPI_DIRECT;
+ if (create) {
+ lockmode = XFS_ILOCK_EXCL;
+ xfs_ilock(ip, lockmode);
+ } else {
+ lockmode = xfs_ilock_map_shared(ip);
+ }
+
+ ASSERT(offset <= mp->m_maxioffset);
+ if (offset + size > mp->m_maxioffset)
+ size = mp->m_maxioffset - offset;
+ end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + size);
+ offset_fsb = XFS_B_TO_FSBT(mp, offset);
- error = xfs_iomap(XFS_I(inode), offset, size, flags, &imap, &nimap,
- &new);
+ error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+ XFS_BMAPI_ENTIRE, NULL, 0, &imap, &nimaps, NULL);
if (error)
- return -error;
- if (nimap == 0)
- return 0;
+ goto out_unlock;
+
+ if (create &&
+ (!nimaps ||
+ (imap.br_startblock == HOLESTARTBLOCK ||
+ imap.br_startblock == DELAYSTARTBLOCK))) {
+ if (direct) {
+ error = xfs_iomap_write_direct(ip, offset, size,
+ &imap, nimaps);
+ } else {
+ error = xfs_iomap_write_delay(ip, offset, size, &imap);
+ }
+ if (error)
+ goto out_unlock;
+
+ trace_xfs_get_blocks_alloc(ip, offset, size, 0, &imap);
+ } else if (nimaps) {
+ trace_xfs_get_blocks_found(ip, offset, size, 0, &imap);
+ } else {
+ trace_xfs_get_blocks_notfound(ip, offset, size);
+ goto out_unlock;
+ }
+ xfs_iunlock(ip, lockmode);
if (imap.br_startblock != HOLESTARTBLOCK &&
imap.br_startblock != DELAYSTARTBLOCK) {
}
return 0;
+
+out_unlock:
+ xfs_iunlock(ip, lockmode);
+ return -error;
}
int
ssize_t ret;
if (rw & WRITE) {
- iocb->private = xfs_alloc_ioend(inode, IO_NEW);
+ iocb->private = xfs_alloc_ioend(inode, IO_DIRECT);
ret = __blockdev_direct_IO(rw, iocb, inode, bdev, iov,
offset, nr_segs,
extern struct workqueue_struct *xfsconvertd_workqueue;
extern mempool_t *xfs_ioend_pool;
+/*
+ * Types of I/O for bmap clustering and I/O completion tracking.
+ */
+enum {
+ IO_DIRECT = 0, /* special case for direct I/O ioends */
+ IO_DELALLOC, /* mapping covers delalloc region */
+ IO_UNWRITTEN, /* mapping covers allocated but uninitialized data */
+ IO_OVERWRITE, /* mapping covers already allocated extent */
+};
+
+#define XFS_IO_TYPES \
+ { 0, "" }, \
+ { IO_DELALLOC, "delalloc" }, \
+ { IO_UNWRITTEN, "unwritten" }, \
+ { IO_OVERWRITE, "overwrite" }
+
/*
* xfs_ioend struct manages large extent writes for XFS.
* It can manage several multi-page bio's at once.
DEFINE_PAGE_EVENT(xfs_releasepage);
DEFINE_PAGE_EVENT(xfs_invalidatepage);
-DECLARE_EVENT_CLASS(xfs_iomap_class,
+DECLARE_EVENT_CLASS(xfs_imap_class,
TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
- int flags, struct xfs_bmbt_irec *irec),
- TP_ARGS(ip, offset, count, flags, irec),
+ int type, struct xfs_bmbt_irec *irec),
+ TP_ARGS(ip, offset, count, type, irec),
TP_STRUCT__entry(
__field(dev_t, dev)
__field(xfs_ino_t, ino)
__field(loff_t, new_size)
__field(loff_t, offset)
__field(size_t, count)
- __field(int, flags)
+ __field(int, type)
__field(xfs_fileoff_t, startoff)
__field(xfs_fsblock_t, startblock)
__field(xfs_filblks_t, blockcount)
__entry->new_size = ip->i_new_size;
__entry->offset = offset;
__entry->count = count;
- __entry->flags = flags;
+ __entry->type = type;
__entry->startoff = irec ? irec->br_startoff : 0;
__entry->startblock = irec ? irec->br_startblock : 0;
__entry->blockcount = irec ? irec->br_blockcount : 0;
),
TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
- "offset 0x%llx count %zd flags %s "
+ "offset 0x%llx count %zd type %s "
"startoff 0x%llx startblock %lld blockcount 0x%llx",
MAJOR(__entry->dev), MINOR(__entry->dev),
__entry->ino,
__entry->new_size,
__entry->offset,
__entry->count,
- __print_flags(__entry->flags, "|", BMAPI_FLAGS),
+ __print_symbolic(__entry->type, XFS_IO_TYPES),
__entry->startoff,
(__int64_t)__entry->startblock,
__entry->blockcount)
)
#define DEFINE_IOMAP_EVENT(name) \
-DEFINE_EVENT(xfs_iomap_class, name, \
+DEFINE_EVENT(xfs_imap_class, name, \
TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count, \
- int flags, struct xfs_bmbt_irec *irec), \
- TP_ARGS(ip, offset, count, flags, irec))
-DEFINE_IOMAP_EVENT(xfs_iomap_enter);
-DEFINE_IOMAP_EVENT(xfs_iomap_found);
-DEFINE_IOMAP_EVENT(xfs_iomap_alloc);
+ int type, struct xfs_bmbt_irec *irec), \
+ TP_ARGS(ip, offset, count, type, irec))
+DEFINE_IOMAP_EVENT(xfs_map_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_map_blocks_alloc);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_alloc);
DECLARE_EVENT_CLASS(xfs_simple_io_class,
TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count),
TP_ARGS(ip, offset, count))
DEFINE_SIMPLE_IO_EVENT(xfs_delalloc_enospc);
DEFINE_SIMPLE_IO_EVENT(xfs_unwritten_convert);
+DEFINE_SIMPLE_IO_EVENT(xfs_get_blocks_notfound);
TRACE_EVENT(xfs_itruncate_start,
#define XFS_WRITEIO_ALIGN(mp,off) (((off) >> mp->m_writeio_log) \
<< mp->m_writeio_log)
-#define XFS_STRAT_WRITE_IMAPS 2
#define XFS_WRITE_IMAPS XFS_BMAP_MAX_NMAP
-STATIC int xfs_iomap_write_direct(struct xfs_inode *, xfs_off_t, size_t,
- struct xfs_bmbt_irec *, int);
-STATIC int xfs_iomap_write_delay(struct xfs_inode *, xfs_off_t, size_t,
- struct xfs_bmbt_irec *);
-STATIC int xfs_iomap_write_allocate(struct xfs_inode *, xfs_off_t, size_t,
- struct xfs_bmbt_irec *);
-
-int
-xfs_iomap(
- struct xfs_inode *ip,
- xfs_off_t offset,
- ssize_t count,
- int flags,
- struct xfs_bmbt_irec *imap,
- int *nimaps,
- int *new)
-{
- struct xfs_mount *mp = ip->i_mount;
- xfs_fileoff_t offset_fsb, end_fsb;
- int error = 0;
- int lockmode = 0;
- int bmapi_flags = 0;
-
- ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
-
- *new = 0;
-
- if (XFS_FORCED_SHUTDOWN(mp))
- return XFS_ERROR(EIO);
-
- trace_xfs_iomap_enter(ip, offset, count, flags, NULL);
-
- switch (flags & (BMAPI_READ | BMAPI_WRITE | BMAPI_ALLOCATE)) {
- case BMAPI_READ:
- lockmode = xfs_ilock_map_shared(ip);
- bmapi_flags = XFS_BMAPI_ENTIRE;
- break;
- case BMAPI_WRITE:
- lockmode = XFS_ILOCK_EXCL;
- if (flags & BMAPI_IGNSTATE)
- bmapi_flags |= XFS_BMAPI_IGSTATE|XFS_BMAPI_ENTIRE;
- xfs_ilock(ip, lockmode);
- break;
- case BMAPI_ALLOCATE:
- lockmode = XFS_ILOCK_SHARED;
- bmapi_flags = XFS_BMAPI_ENTIRE;
-
- /* Attempt non-blocking lock */
- if (flags & BMAPI_TRYLOCK) {
- if (!xfs_ilock_nowait(ip, lockmode))
- return XFS_ERROR(EAGAIN);
- } else {
- xfs_ilock(ip, lockmode);
- }
- break;
- default:
- BUG();
- }
-
- ASSERT(offset <= mp->m_maxioffset);
- if ((xfs_fsize_t)offset + count > mp->m_maxioffset)
- count = mp->m_maxioffset - offset;
- end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
- offset_fsb = XFS_B_TO_FSBT(mp, offset);
-
- error = xfs_bmapi(NULL, ip, offset_fsb,
- (xfs_filblks_t)(end_fsb - offset_fsb),
- bmapi_flags, NULL, 0, imap,
- nimaps, NULL);
-
- if (error)
- goto out;
-
- switch (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)) {
- case BMAPI_WRITE:
- /* If we found an extent, return it */
- if (*nimaps &&
- (imap->br_startblock != HOLESTARTBLOCK) &&
- (imap->br_startblock != DELAYSTARTBLOCK)) {
- trace_xfs_iomap_found(ip, offset, count, flags, imap);
- break;
- }
-
- if (flags & BMAPI_DIRECT) {
- error = xfs_iomap_write_direct(ip, offset, count, imap,
- *nimaps);
- } else {
- error = xfs_iomap_write_delay(ip, offset, count, imap);
- }
-
- if (!error) {
- trace_xfs_iomap_alloc(ip, offset, count, flags, imap);
- }
- *new = 1;
- break;
- case BMAPI_ALLOCATE:
- /* If we found an extent, return it */
- xfs_iunlock(ip, lockmode);
- lockmode = 0;
-
- if (*nimaps && !isnullstartblock(imap->br_startblock)) {
- trace_xfs_iomap_found(ip, offset, count, flags, imap);
- break;
- }
-
- error = xfs_iomap_write_allocate(ip, offset, count, imap);
- break;
- }
-
-out:
- if (lockmode)
- xfs_iunlock(ip, lockmode);
- return XFS_ERROR(error);
-}
-
STATIC int
xfs_iomap_eof_align_last_fsb(
xfs_mount_t *mp,
return EFSCORRUPTED;
}
-STATIC int
+int
xfs_iomap_write_direct(
xfs_inode_t *ip,
xfs_off_t offset,
return 0;
}
-STATIC int
+int
xfs_iomap_write_delay(
xfs_inode_t *ip,
xfs_off_t offset,
* We no longer bother to look at the incoming map - all we have to
* guarantee is that whatever we allocate fills the required range.
*/
-STATIC int
+int
xfs_iomap_write_allocate(
xfs_inode_t *ip,
xfs_off_t offset,
#ifndef __XFS_IOMAP_H__
#define __XFS_IOMAP_H__
-/* base extent manipulation calls */
-#define BMAPI_READ (1 << 0) /* read extents */
-#define BMAPI_WRITE (1 << 1) /* create extents */
-#define BMAPI_ALLOCATE (1 << 2) /* delayed allocate to real extents */
-
-/* modifiers */
-#define BMAPI_IGNSTATE (1 << 4) /* ignore unwritten state on read */
-#define BMAPI_DIRECT (1 << 5) /* direct instead of buffered write */
-#define BMAPI_MMA (1 << 6) /* allocate for mmap write */
-#define BMAPI_TRYLOCK (1 << 7) /* non-blocking request */
-
-#define BMAPI_FLAGS \
- { BMAPI_READ, "READ" }, \
- { BMAPI_WRITE, "WRITE" }, \
- { BMAPI_ALLOCATE, "ALLOCATE" }, \
- { BMAPI_IGNSTATE, "IGNSTATE" }, \
- { BMAPI_DIRECT, "DIRECT" }, \
- { BMAPI_TRYLOCK, "TRYLOCK" }
-
struct xfs_inode;
struct xfs_bmbt_irec;
-extern int xfs_iomap(struct xfs_inode *, xfs_off_t, ssize_t, int,
- struct xfs_bmbt_irec *, int *, int *);
+extern int xfs_iomap_write_direct(struct xfs_inode *, xfs_off_t, size_t,
+ struct xfs_bmbt_irec *, int);
+extern int xfs_iomap_write_delay(struct xfs_inode *, xfs_off_t, size_t,
+ struct xfs_bmbt_irec *);
+extern int xfs_iomap_write_allocate(struct xfs_inode *, xfs_off_t, size_t,
+ struct xfs_bmbt_irec *);
extern int xfs_iomap_write_unwritten(struct xfs_inode *, xfs_off_t, size_t);
#endif /* __XFS_IOMAP_H__*/
git.openpandora.org / pandora-kernel.git / commitdiff