xfs_iomap.c

linux 内核源代码

/*
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_dir2.h"
#include "xfs_alloc.h"
#include "xfs_dmapi.h"
#include "xfs_quota.h"
#include "xfs_mount.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_ialloc.h"
#include "xfs_btree.h"
#include "xfs_bmap.h"
#include "xfs_rtalloc.h"
#include "xfs_error.h"
#include "xfs_itable.h"
#include "xfs_rw.h"
#include "xfs_acl.h"
#include "xfs_attr.h"
#include "xfs_buf_item.h"
#include "xfs_trans_space.h"
#include "xfs_utils.h"
#include "xfs_iomap.h"

#if defined(XFS_RW_TRACE)
void
xfs_iomap_enter_trace(
	int		tag,
	xfs_iocore_t	*io,
	xfs_off_t	offset,
	ssize_t		count)
{
	xfs_inode_t	*ip = XFS_IO_INODE(io);

	if (!ip->i_rwtrace)
		return;

	ktrace_enter(ip->i_rwtrace,
		(void *)((unsigned long)tag),
		(void *)ip,
		(void *)((unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff)),
		(void *)((unsigned long)(ip->i_d.di_size & 0xffffffff)),
		(void *)((unsigned long)((offset >> 32) & 0xffffffff)),
		(void *)((unsigned long)(offset & 0xffffffff)),
		(void *)((unsigned long)count),
		(void *)((unsigned long)((io->io_new_size >> 32) & 0xffffffff)),
		(void *)((unsigned long)(io->io_new_size & 0xffffffff)),
		(void *)((unsigned long)current_pid()),
		(void *)NULL,
		(void *)NULL,
		(void *)NULL,
		(void *)NULL,
		(void *)NULL,
		(void *)NULL);
}

void
xfs_iomap_map_trace(
	int		tag,
	xfs_iocore_t	*io,
	xfs_off_t	offset,
	ssize_t		count,
	xfs_iomap_t	*iomapp,
	xfs_bmbt_irec_t	*imapp,
	int		flags)
{
	xfs_inode_t	*ip = XFS_IO_INODE(io);

	if (!ip->i_rwtrace)
		return;

	ktrace_enter(ip->i_rwtrace,
		(void *)((unsigned long)tag),
		(void *)ip,
		(void *)((unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff)),
		(void *)((unsigned long)(ip->i_d.di_size & 0xffffffff)),
		(void *)((unsigned long)((offset >> 32) & 0xffffffff)),
		(void *)((unsigned long)(offset & 0xffffffff)),
		(void *)((unsigned long)count),
		(void *)((unsigned long)flags),
		(void *)((unsigned long)((iomapp->iomap_offset >> 32) & 0xffffffff)),
		(void *)((unsigned long)(iomapp->iomap_offset & 0xffffffff)),
		(void *)((unsigned long)(iomapp->iomap_delta)),
		(void *)((unsigned long)(iomapp->iomap_bsize)),
		(void *)((unsigned long)(iomapp->iomap_bn)),
		(void *)(__psint_t)(imapp->br_startoff),
		(void *)((unsigned long)(imapp->br_blockcount)),
		(void *)(__psint_t)(imapp->br_startblock));
}
#else
#define xfs_iomap_enter_trace(tag, io, offset, count)
#define xfs_iomap_map_trace(tag, io, offset, count, iomapp, imapp, flags)
#endif

#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_imap_to_bmap(
	xfs_iocore_t	*io,
	xfs_off_t	offset,
	xfs_bmbt_irec_t *imap,
	xfs_iomap_t	*iomapp,
	int		imaps,			/* Number of imap entries */
	int		iomaps,			/* Number of iomap entries */
	int		flags)
{
	xfs_mount_t	*mp;
	int		pbm;
	xfs_fsblock_t	start_block;

	mp = io->io_mount;

	for (pbm = 0; imaps && pbm < iomaps; imaps--, iomapp++, imap++, pbm++) {
		iomapp->iomap_offset = XFS_FSB_TO_B(mp, imap->br_startoff);
		iomapp->iomap_delta = offset - iomapp->iomap_offset;
		iomapp->iomap_bsize = XFS_FSB_TO_B(mp, imap->br_blockcount);
		iomapp->iomap_flags = flags;

		if (io->io_flags & XFS_IOCORE_RT) {
			iomapp->iomap_flags |= IOMAP_REALTIME;
			iomapp->iomap_target = mp->m_rtdev_targp;
		} else {
			iomapp->iomap_target = mp->m_ddev_targp;
		}
		start_block = imap->br_startblock;
		if (start_block == HOLESTARTBLOCK) {
			iomapp->iomap_bn = IOMAP_DADDR_NULL;
			iomapp->iomap_flags |= IOMAP_HOLE;
		} else if (start_block == DELAYSTARTBLOCK) {
			iomapp->iomap_bn = IOMAP_DADDR_NULL;
			iomapp->iomap_flags |= IOMAP_DELAY;
		} else {
			iomapp->iomap_bn = XFS_FSB_TO_DB_IO(io, start_block);
			if (ISUNWRITTEN(imap))
				iomapp->iomap_flags |= IOMAP_UNWRITTEN;
		}

		offset += iomapp->iomap_bsize - iomapp->iomap_delta;
	}
	return pbm;	/* Return the number filled */
}

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);
}

STATIC int
xfs_iomap_eof_align_last_fsb(
	xfs_mount_t	*mp,
	xfs_iocore_t	*io,
	xfs_fsize_t	isize,
	xfs_extlen_t	extsize,
	xfs_fileoff_t	*last_fsb)
{
	xfs_fileoff_t	new_last_fsb = 0;
	xfs_extlen_t	align;
	int		eof, error;

	if (io->io_flags & XFS_IOCORE_RT)
		;
	/*
	 * If mounted with the "-o swalloc" option, roundup the allocation
	 * request to a stripe width boundary if the file size is >=
	 * stripe width and we are allocating past the allocation eof.
	 */
	else if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC) &&
	        (isize >= XFS_FSB_TO_B(mp, mp->m_swidth)))
		new_last_fsb = roundup_64(*last_fsb, mp->m_swidth);
	/*
	 * Roundup the allocation request to a stripe unit (m_dalign) boundary
	 * if the file size is >= stripe unit size, and we are allocating past
	 * the allocation eof.
	 */
	else if (mp->m_dalign && (isize >= XFS_FSB_TO_B(mp, mp->m_dalign)))
		new_last_fsb = roundup_64(*last_fsb, mp->m_dalign);

	/*
	 * Always round up the allocation request to an extent boundary
	 * (when file on a real-time subvolume or has di_extsize hint).
	 */
	if (extsize) {
		if (new_last_fsb)
			align = roundup_64(new_last_fsb, extsize);
		else
			align = extsize;
		new_last_fsb = roundup_64(*last_fsb, align);
	}

	if (new_last_fsb) {
		error = XFS_BMAP_EOF(mp, io, new_last_fsb, XFS_DATA_FORK, &eof);
		if (error)
			return error;
		if (eof)
			*last_fsb = new_last_fsb;
	}
	return 0;
}

STATIC int
xfs_flush_space(
	xfs_inode_t	*ip,
	int		*fsynced,
	int		*ioflags)
{
	switch (*fsynced) {
	case 0:
		if (ip->i_delayed_blks) {
			xfs_iunlock(ip, XFS_ILOCK_EXCL);
			xfs_flush_inode(ip);
			xfs_ilock(ip, XFS_ILOCK_EXCL);
			*fsynced = 1;
		} else {
			*ioflags |= BMAPI_SYNC;
			*fsynced = 2;
		}
		return 0;
	case 1:
		*fsynced = 2;
		*ioflags |= BMAPI_SYNC;
		return 0;
	case 2:
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
		xfs_flush_device(ip);
		xfs_ilock(ip, XFS_ILOCK_EXCL);
		*fsynced = 3;
		return 0;
	}
	return 1;
}

STATIC int
xfs_cmn_err_fsblock_zero(
	xfs_inode_t	*ip,
	xfs_bmbt_irec_t	*imap)
{
	xfs_cmn_err(XFS_PTAG_FSBLOCK_ZERO, CE_ALERT, ip->i_mount,
			"Access to block zero in inode %llu "
			"start_block: %llx start_off: %llx "
			"blkcnt: %llx extent-state: %x\n",
		(unsigned long long)ip->i_ino,
		(unsigned long long)imap->br_startblock,
		(unsigned long long)imap->br_startoff,
		(unsigned long long)imap->br_blockcount,
		imap->br_state);
	return EFSCORRUPTED;
}

int
xfs_iomap_write_direct(
	xfs_inode_t	*ip,
	xfs_off_t	offset,
	size_t		count,
	int		flags,
	xfs_bmbt_irec_t *ret_imap,
	int		*nmaps,
	int		found)
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_iocore_t	*io = &ip->i_iocore;
	xfs_fileoff_t	offset_fsb;
	xfs_fileoff_t	last_fsb;
	xfs_filblks_t	count_fsb, resaligned;
	xfs_fsblock_t	firstfsb;
	xfs_extlen_t	extsz, temp;
	xfs_fsize_t	isize;
	int		nimaps;
	int		bmapi_flag;
	int		quota_flag;
	int		rt;
	xfs_trans_t	*tp;
	xfs_bmbt_irec_t imap;
	xfs_bmap_free_t free_list;
	uint		qblocks, resblks, resrtextents;
	int		committed;
	int		error;

	/*
	 * Make sure that the dquots are there. This doesn't hold
	 * the ilock across a disk read.
	 */
	error = XFS_QM_DQATTACH(ip->i_mount, ip, XFS_QMOPT_ILOCKED);
	if (error)
		return XFS_ERROR(error);

	rt = XFS_IS_REALTIME_INODE(ip);
	extsz = xfs_get_extsz_hint(ip);

	isize = ip->i_size;
	if (io->io_new_size > isize)
		isize = io->io_new_size;

	offset_fsb = XFS_B_TO_FSBT(mp, offset);
	last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
	if ((offset + count) > isize) {
		error = xfs_iomap_eof_align_last_fsb(mp, io, isize, extsz,
							&last_fsb);
		if (error)
			goto error_out;
	} else {
		if (found && (ret_imap->br_startblock == HOLESTARTBLOCK))
			last_fsb = MIN(last_fsb, (xfs_fileoff_t)
					ret_imap->br_blockcount +
					ret_imap->br_startoff);
	}
	count_fsb = last_fsb - offset_fsb;
	ASSERT(count_fsb > 0);

	resaligned = count_fsb;
	if (unlikely(extsz)) {
		if ((temp = do_mod(offset_fsb, extsz)))
			resaligned += temp;
		if ((temp = do_mod(resaligned, extsz)))
			resaligned += extsz - temp;
	}

	if (unlikely(rt)) {
		resrtextents = qblocks = resaligned;