/*
* This routine is called to handle zeroing any space in the last
* block of the file that is beyond the EOF. We do this since the
* size is being increased without writing anything to that block
* and we don't want anyone to read the garbage on the disk.
*/
STATIC int /* error (positive) */
xfs_zero_last_block(
xfs_vnode_t *vp,
xfs_iocore_t *io,
xfs_fsize_t isize,
xfs_fsize_t end_size)
{
xfs_fileoff_t last_fsb;
xfs_mount_t *mp;
int nimaps;
int zero_offset;
int zero_len;
int error = 0;
xfs_bmbt_irec_t imap;
xfs_off_t loff;
ASSERT(ismrlocked(io->io_lock, MR_UPDATE) != 0);
mp = io->io_mount;
zero_offset = XFS_B_FSB_OFFSET(mp, isize);
if (zero_offset == 0) {
/*
* There are no extra bytes in the last block on disk to
* zero, so return.
*/
return 0;
}
last_fsb = XFS_B_TO_FSBT(mp, isize);
nimaps = 1;
error = XFS_BMAPI(mp, NULL, io, last_fsb, 1, 0, NULL, 0, &imap,
&nimaps, NULL, NULL);
if (error) {
return error;
}
ASSERT(nimaps > 0);
/*
* If the block underlying isize is just a hole, then there
* is nothing to zero.
*/
if (imap.br_startblock == HOLESTARTBLOCK) {
return 0;
}
/*
* Zero the part of the last block beyond the EOF, and write it
* out sync. We need to drop the ilock while we do this so we
* don't deadlock when the buffer cache calls back to us.
*/
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL| XFS_EXTSIZE_RD);
loff = XFS_FSB_TO_B(mp, last_fsb);
zero_len = mp->m_sb.sb_blocksize - zero_offset;
error = xfs_iozero(vp, loff + zero_offset, zero_len, end_size);
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
ASSERT(error >= 0);
return error;
}
/*
* xfs_inval_cached_pages
*
* This routine is responsible for keeping direct I/O and buffered I/O
* somewhat coherent. From here we make sure that we're at least
* temporarily holding the inode I/O lock exclusively and then call
* the page cache to flush and invalidate any cached pages. If there
* are no cached pages this routine will be very quick.
*/
void
xfs_inval_cached_pages(
vnode_t *vp,
xfs_iocore_t *io,
xfs_off_t offset,
int write,
int relock)
{
xfs_mount_t *mp;
if (!VN_CACHED(vp)) {
return;
}
mp = io->io_mount;
/*
* We need to get the I/O lock exclusively in order
* to safely invalidate pages and mappings.
*/
if (relock) {
XFS_IUNLOCK(mp, io, XFS_IOLOCK_SHARED);
XFS_ILOCK(mp, io, XFS_IOLOCK_EXCL);
}
/* Writing beyond EOF creates a hole that must be zeroed */
if (write && (offset > XFS_SIZE(mp, io))) {
xfs_fsize_t isize;
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
isize = XFS_SIZE(mp, io);
if (offset > isize) {
xfs_zero_eof(vp, io, offset, isize, offset);
}
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
}
xfs_inval_cached_trace(io, offset, -1, ctooff(offtoct(offset)), -1);
VOP_FLUSHINVAL_PAGES(vp, ctooff(offtoct(offset)), -1, FI_REMAPF_LOCKED);
if (relock) {
XFS_ILOCK_DEMOTE(mp, io, XFS_IOLOCK_EXCL);
}
}
int /* error (positive) */
xfs_zero_eof(
vnode_t *vp,
xfs_iocore_t *io,
xfs_off_t offset, /* starting I/O offset */
xfs_fsize_t isize, /* current inode size */
xfs_fsize_t end_size) /* terminal inode size */
{
struct inode *ip = LINVFS_GET_IP(vp);
xfs_fileoff_t start_zero_fsb;
xfs_fileoff_t end_zero_fsb;
xfs_fileoff_t prev_zero_fsb;
xfs_fileoff_t zero_count_fsb;
xfs_fileoff_t last_fsb;
xfs_extlen_t buf_len_fsb;
xfs_extlen_t prev_zero_count;
xfs_mount_t *mp;
int nimaps;
int error = 0;
xfs_bmbt_irec_t imap;
loff_t loff;
size_t lsize;
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
mp = io->io_mount;
/*
* First handle zeroing the block on which isize resides.
* We only zero a part of that block so it is handled specially.
*/
error = xfs_zero_last_block(ip, io, offset, isize, end_size);
if (error) {
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
return error;
}
/*
* Calculate the range between the new size and the old
* where blocks needing to be zeroed may exist. To get the
* block where the last byte in the file currently resides,
* we need to subtract one from the size and truncate back
* to a block boundary. We subtract 1 in case the size is
* exactly on a block boundary.
*/
last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1;
start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
end_zero_fsb = XFS_B_TO_FSBT(mp, offset - 1);
ASSERT((xfs_sfiloff_t)last_fsb < (xfs_sfiloff_t)start_zero_fsb);
if (last_fsb == end_zero_fsb) {
/*
* The size was only incremented on its last block.
* We took care of that above, so just return.
*/
return 0;
}
ASSERT(start_zero_fsb <= end_zero_fsb);
prev_zero_fsb = NULLFILEOFF;
prev_zero_count = 0;
while (start_zero_fsb <= end_zero_fsb) {
nimaps = 1;
zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
error = XFS_BMAPI(mp, NULL, io, start_zero_fsb, zero_count_fsb,
0, NULL, 0, &imap, &nimaps, NULL);
if (error) {
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
return error;
}
ASSERT(nimaps > 0);
if (imap.br_state == XFS_EXT_UNWRITTEN ||
imap.br_startblock == HOLESTARTBLOCK) {
/*
* This loop handles initializing pages that were
* partially initialized by the code below this
* loop. It basically zeroes the part of the page
* that sits on a hole and sets the page as P_HOLE
* and calls remapf if it is a mapped file.
*/
prev_zero_fsb = NULLFILEOFF;
prev_zero_count = 0;
start_zero_fsb = imap.br_startoff +
imap.br_blockcount;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
continue;
}
/*
* There are blocks in the range requested.
* Zero them a single write at a time. We actually
* don't zero the entire range returned if it is
* too big and simply loop around to get the rest.
* That is not the most efficient thing to do, but it
* is simple and this path should not be exercised often.
*/
buf_len_fsb = XFS_FILBLKS_MIN(imap.br_blockcount,
mp->m_writeio_blocks << 8);
/*
* Drop the inode lock while we're doing the I/O.
* We'll still have the iolock to protect us.
*/
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
loff = XFS_FSB_TO_B(mp, start_zero_fsb);
lsize = XFS_FSB_TO_B(mp, buf_len_fsb);
error = xfs_iozero(ip, loff, lsize, end_size);
if (error) {
goto out_lock;
}
prev_zero_fsb = start_zero_fsb;
prev_zero_count = buf_len_fsb;
start_zero_fsb = imap.br_startoff + buf_len_fsb;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
}
return 0;
out_lock:
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
ASSERT(error >= 0);
return error;
}
int
xfs_iomap(
xfs_iocore_t *io,
xfs_off_t offset,
ssize_t count,
int flags,
xfs_iomap_t *iomapp,
int *niomaps)
{
xfs_mount_t *mp = io->io_mount;
xfs_fileoff_t offset_fsb, end_fsb;
int error = 0;
int lockmode = 0;
xfs_bmbt_irec_t imap;
int nimaps = 1;
int bmapi_flags = 0;
int iomap_flags = 0;
if (XFS_FORCED_SHUTDOWN(mp))
return XFS_ERROR(EIO);
switch (flags &
(BMAPI_READ | BMAPI_WRITE | BMAPI_ALLOCATE |
BMAPI_UNWRITTEN | BMAPI_DEVICE)) {
case BMAPI_READ:
xfs_iomap_enter_trace(XFS_IOMAP_READ_ENTER, io, offset, count);
lockmode = XFS_LCK_MAP_SHARED(mp, io);
bmapi_flags = XFS_BMAPI_ENTIRE;
break;
case BMAPI_WRITE:
xfs_iomap_enter_trace(XFS_IOMAP_WRITE_ENTER, io, offset, count);
lockmode = XFS_ILOCK_EXCL|XFS_EXTSIZE_WR;
if (flags & BMAPI_IGNSTATE)
bmapi_flags |= XFS_BMAPI_IGSTATE|XFS_BMAPI_ENTIRE;
XFS_ILOCK(mp, io, lockmode);
break;
case BMAPI_ALLOCATE:
xfs_iomap_enter_trace(XFS_IOMAP_ALLOC_ENTER, io, offset, count);
lockmode = XFS_ILOCK_SHARED|XFS_EXTSIZE_RD;
bmapi_flags = XFS_BMAPI_ENTIRE;
/* Attempt non-blocking lock */
if (flags & BMAPI_TRYLOCK) {
if (!XFS_ILOCK_NOWAIT(mp, io, lockmode))
return XFS_ERROR(EAGAIN);
} else {
XFS_ILOCK(mp, io, lockmode);
}
break;
case BMAPI_UNWRITTEN:
goto phase2;
case BMAPI_DEVICE:
lockmode = XFS_LCK_MAP_SHARED(mp, io);
iomapp->iomap_target = io->io_flags & XFS_IOCORE_RT ?
mp->m_rtdev_targp : mp->m_ddev_targp;
error = 0;
*niomaps = 1;
goto out;
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(mp, NULL, io, offset_fsb,
(xfs_filblks_t)(end_fsb - offset_fsb),
bmapi_flags, NULL, 0, &imap,
&nimaps, NULL, NULL);
if (error)
goto out;
phase2:
switch (flags & (BMAPI_WRITE|BMAPI_ALLOCATE|BMAPI_UNWRITTEN)) {
case BMAPI_WRITE:
/* If we found an extent, return it */
if (nimaps &&
(imap.br_startblock != HOLESTARTBLOCK) &&
(imap.br_startblock != DELAYSTARTBLOCK)) {
xfs_iomap_map_trace(XFS_IOMAP_WRITE_MAP, io,
offset, count, iomapp, &imap, flags);
break;
}
if (flags & (BMAPI_DIRECT|BMAPI_MMAP)) {
error = XFS_IOMAP_WRITE_DIRECT(mp, io, offset,
count, flags, &imap, &nimaps, nimaps);
} else {
error = XFS_IOMAP_WRITE_DELAY(mp, io, offset, count,
flags, &imap, &nimaps);
}
if (!error) {
xfs_iomap_map_trace(XFS_IOMAP_ALLOC_MAP, io,
offset, count, iomapp, &imap, flags);
}
iomap_flags = IOMAP_NEW;
break;
case BMAPI_ALLOCATE:
/* If we found an extent, return it */
XFS_IUNLOCK(mp, io, lockmode);
lockmode = 0;
if (nimaps && !ISNULLSTARTBLOCK(imap.br_startblock)) {
xfs_iomap_map_trace(XFS_IOMAP_WRITE_MAP, io,
offset, count, iomapp, &imap, flags);
break;
}
error = XFS_IOMAP_WRITE_ALLOCATE(mp, io, offset, count,
&imap, &nimaps);
break;
case BMAPI_UNWRITTEN:
lockmode = 0;
error = XFS_IOMAP_WRITE_UNWRITTEN(mp, io, offset, count);
nimaps = 0;
break;
}
if (nimaps) {
*niomaps = xfs_imap_to_bmap(io, offset, &imap,
iomapp, nimaps, *niomaps, iomap_flags);
} else if (niomaps) {
*niomaps = 0;
}
out:
if (lockmode)
XFS_IUNLOCK(mp, io, lockmode);
return XFS_ERROR(error);
}
int /* error (positive) */
xfs_zero_eof(
bhv_vnode_t *vp,
xfs_iocore_t *io,
xfs_off_t offset, /* starting I/O offset */
xfs_fsize_t isize) /* current inode size */
{
struct inode *ip = vn_to_inode(vp);
xfs_fileoff_t start_zero_fsb;
xfs_fileoff_t end_zero_fsb;
xfs_fileoff_t zero_count_fsb;
xfs_fileoff_t last_fsb;
xfs_fileoff_t zero_off;
xfs_fsize_t zero_len;
xfs_mount_t *mp = io->io_mount;
int nimaps;
int error = 0;
xfs_bmbt_irec_t imap;
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
ASSERT(offset > isize);
/*
* First handle zeroing the block on which isize resides.
* We only zero a part of that block so it is handled specially.
*/
error = xfs_zero_last_block(ip, io, offset, isize);
if (error) {
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
return error;
}
/*
* Calculate the range between the new size and the old
* where blocks needing to be zeroed may exist. To get the
* block where the last byte in the file currently resides,
* we need to subtract one from the size and truncate back
* to a block boundary. We subtract 1 in case the size is
* exactly on a block boundary.
*/
last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1;
start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
end_zero_fsb = XFS_B_TO_FSBT(mp, offset - 1);
ASSERT((xfs_sfiloff_t)last_fsb < (xfs_sfiloff_t)start_zero_fsb);
if (last_fsb == end_zero_fsb) {
/*
* The size was only incremented on its last block.
* We took care of that above, so just return.
*/
return 0;
}
ASSERT(start_zero_fsb <= end_zero_fsb);
while (start_zero_fsb <= end_zero_fsb) {
nimaps = 1;
zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
error = XFS_BMAPI(mp, NULL, io, start_zero_fsb, zero_count_fsb,
0, NULL, 0, &imap, &nimaps, NULL, NULL);
if (error) {
ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
return error;
}
ASSERT(nimaps > 0);
if (imap.br_state == XFS_EXT_UNWRITTEN ||
imap.br_startblock == HOLESTARTBLOCK) {
/*
* This loop handles initializing pages that were
* partially initialized by the code below this
* loop. It basically zeroes the part of the page
* that sits on a hole and sets the page as P_HOLE
* and calls remapf if it is a mapped file.
*/
start_zero_fsb = imap.br_startoff + imap.br_blockcount;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
continue;
}
/*
* There are blocks we need to zero.
* Drop the inode lock while we're doing the I/O.
* We'll still have the iolock to protect us.
*/
XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
zero_off = XFS_FSB_TO_B(mp, start_zero_fsb);
zero_len = XFS_FSB_TO_B(mp, imap.br_blockcount);
if ((zero_off + zero_len) > offset)
zero_len = offset - zero_off;
error = xfs_iozero(ip, zero_off, zero_len);
if (error) {
goto out_lock;
}
start_zero_fsb = imap.br_startoff + imap.br_blockcount;
ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
}
return 0;
out_lock:
XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
ASSERT(error >= 0);
return error;
}