xfs_super.c

linux-2.6.15.6

/*
 * Copyright (c) 2000-2005 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_bit.h"
#include "xfs_log.h"
#include "xfs_clnt.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_dir.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_dir_sf.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_btree.h"
#include "xfs_ialloc.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_cap.h"
#include "xfs_mac.h"
#include "xfs_attr.h"
#include "xfs_buf_item.h"
#include "xfs_utils.h"
#include "xfs_version.h"

#include <linux/namei.h>
#include <linux/init.h>
#include <linux/mount.h>
#include <linux/mempool.h>
#include <linux/writeback.h>
#include <linux/kthread.h>

STATIC struct quotactl_ops linvfs_qops;
STATIC struct super_operations linvfs_sops;
STATIC kmem_zone_t *xfs_vnode_zone;
STATIC kmem_zone_t *xfs_ioend_zone;
mempool_t *xfs_ioend_pool;

STATIC struct xfs_mount_args *
xfs_args_allocate(
	struct super_block	*sb)
{
	struct xfs_mount_args	*args;

	args = kmem_zalloc(sizeof(struct xfs_mount_args), KM_SLEEP);
	args->logbufs = args->logbufsize = -1;
	strncpy(args->fsname, sb->s_id, MAXNAMELEN);

	/* Copy the already-parsed mount(2) flags we're interested in */
	if (sb->s_flags & MS_NOATIME)
		args->flags |= XFSMNT_NOATIME;
	if (sb->s_flags & MS_DIRSYNC)
		args->flags |= XFSMNT_DIRSYNC;
	if (sb->s_flags & MS_SYNCHRONOUS)
		args->flags |= XFSMNT_WSYNC;

	/* Default to 32 bit inodes on Linux all the time */
	args->flags |= XFSMNT_32BITINODES;

	return args;
}

__uint64_t
xfs_max_file_offset(
	unsigned int		blockshift)
{
	unsigned int		pagefactor = 1;
	unsigned int		bitshift = BITS_PER_LONG - 1;

	/* Figure out maximum filesize, on Linux this can depend on
	 * the filesystem blocksize (on 32 bit platforms).
	 * __block_prepare_write does this in an [unsigned] long...
	 *      page->index << (PAGE_CACHE_SHIFT - bbits)
	 * So, for page sized blocks (4K on 32 bit platforms),
	 * this wraps at around 8Tb (hence MAX_LFS_FILESIZE which is
	 *      (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
	 * but for smaller blocksizes it is less (bbits = log2 bsize).
	 * Note1: get_block_t takes a long (implicit cast from above)
	 * Note2: The Large Block Device (LBD and HAVE_SECTOR_T) patch
	 * can optionally convert the [unsigned] long from above into
	 * an [unsigned] long long.
	 */

#if BITS_PER_LONG == 32
# if defined(CONFIG_LBD)
	ASSERT(sizeof(sector_t) == 8);
	pagefactor = PAGE_CACHE_SIZE;
	bitshift = BITS_PER_LONG;
# else
	pagefactor = PAGE_CACHE_SIZE >> (PAGE_CACHE_SHIFT - blockshift);
# endif
#endif

	return (((__uint64_t)pagefactor) << bitshift) - 1;
}

STATIC __inline__ void
xfs_set_inodeops(
	struct inode		*inode)
{
	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_op = &linvfs_file_inode_operations;
		inode->i_fop = &linvfs_file_operations;
		inode->i_mapping->a_ops = &linvfs_aops;
		break;
	case S_IFDIR:
		inode->i_op = &linvfs_dir_inode_operations;
		inode->i_fop = &linvfs_dir_operations;
		break;
	case S_IFLNK:
		inode->i_op = &linvfs_symlink_inode_operations;
		if (inode->i_blocks)
			inode->i_mapping->a_ops = &linvfs_aops;
		break;
	default:
		inode->i_op = &linvfs_file_inode_operations;
		init_special_inode(inode, inode->i_mode, inode->i_rdev);
		break;
	}
}

STATIC __inline__ void
xfs_revalidate_inode(
	xfs_mount_t		*mp,
	vnode_t			*vp,
	xfs_inode_t		*ip)
{
	struct inode		*inode = LINVFS_GET_IP(vp);

	inode->i_mode	= ip->i_d.di_mode;
	inode->i_nlink	= ip->i_d.di_nlink;
	inode->i_uid	= ip->i_d.di_uid;
	inode->i_gid	= ip->i_d.di_gid;

	switch (inode->i_mode & S_IFMT) {
	case S_IFBLK:
	case S_IFCHR:
		inode->i_rdev =
			MKDEV(sysv_major(ip->i_df.if_u2.if_rdev) & 0x1ff,
			      sysv_minor(ip->i_df.if_u2.if_rdev));
		break;
	default:
		inode->i_rdev = 0;
		break;
	}

	inode->i_blksize = xfs_preferred_iosize(mp);
	inode->i_generation = ip->i_d.di_gen;
	i_size_write(inode, ip->i_d.di_size);
	inode->i_blocks =
		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);
	inode->i_atime.tv_sec	= ip->i_d.di_atime.t_sec;
	inode->i_atime.tv_nsec	= ip->i_d.di_atime.t_nsec;
	inode->i_mtime.tv_sec	= ip->i_d.di_mtime.t_sec;
	inode->i_mtime.tv_nsec	= ip->i_d.di_mtime.t_nsec;
	inode->i_ctime.tv_sec	= ip->i_d.di_ctime.t_sec;
	inode->i_ctime.tv_nsec	= ip->i_d.di_ctime.t_nsec;
	if (ip->i_d.di_flags & XFS_DIFLAG_IMMUTABLE)
		inode->i_flags |= S_IMMUTABLE;
	else
		inode->i_flags &= ~S_IMMUTABLE;
	if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
		inode->i_flags |= S_APPEND;
	else
		inode->i_flags &= ~S_APPEND;
	if (ip->i_d.di_flags & XFS_DIFLAG_SYNC)
		inode->i_flags |= S_SYNC;
	else
		inode->i_flags &= ~S_SYNC;
	if (ip->i_d.di_flags & XFS_DIFLAG_NOATIME)
		inode->i_flags |= S_NOATIME;
	else
		inode->i_flags &= ~S_NOATIME;
	vp->v_flag &= ~VMODIFIED;
}

void
xfs_initialize_vnode(
	bhv_desc_t		*bdp,
	vnode_t			*vp,
	bhv_desc_t		*inode_bhv,
	int			unlock)
{
	xfs_inode_t		*ip = XFS_BHVTOI(inode_bhv);
	struct inode		*inode = LINVFS_GET_IP(vp);

	if (!inode_bhv->bd_vobj) {
		vp->v_vfsp = bhvtovfs(bdp);
		bhv_desc_init(inode_bhv, ip, vp, &xfs_vnodeops);
		bhv_insert(VN_BHV_HEAD(vp), inode_bhv);
	}

	/*
	 * We need to set the ops vectors, and unlock the inode, but if
	 * we have been called during the new inode create process, it is
	 * too early to fill in the Linux inode.  We will get called a
	 * second time once the inode is properly set up, and then we can
	 * finish our work.
	 */
	if (ip->i_d.di_mode != 0 && unlock && (inode->i_state & I_NEW)) {
		xfs_revalidate_inode(XFS_BHVTOM(bdp), vp, ip);
		xfs_set_inodeops(inode);
	
		ip->i_flags &= ~XFS_INEW;
		barrier();

		unlock_new_inode(inode);
	}
}

int
xfs_blkdev_get(
	xfs_mount_t		*mp,
	const char		*name,
	struct block_device	**bdevp)
{
	int			error = 0;

	*bdevp = open_bdev_excl(name, 0, mp);
	if (IS_ERR(*bdevp)) {
		error = PTR_ERR(*bdevp);
		printk("XFS: Invalid device [%s], error=%d\n", name, error);
	}

	return -error;
}

void
xfs_blkdev_put(
	struct block_device	*bdev)
{
	if (bdev)
		close_bdev_excl(bdev);
}

/*
 * Try to write out the superblock using barriers.
 */
STATIC int
xfs_barrier_test(
	xfs_mount_t	*mp)
{
	xfs_buf_t	*sbp = xfs_getsb(mp, 0);
	int		error;

	XFS_BUF_UNDONE(sbp);
	XFS_BUF_UNREAD(sbp);
	XFS_BUF_UNDELAYWRITE(sbp);
	XFS_BUF_WRITE(sbp);
	XFS_BUF_UNASYNC(sbp);
	XFS_BUF_ORDERED(sbp);

	xfsbdstrat(mp, sbp);
	error = xfs_iowait(sbp);

	/*
	 * Clear all the flags we set and possible error state in the
	 * buffer.  We only did the write to try out whether barriers
	 * worked and shouldn't leave any traces in the superblock
	 * buffer.
	 */
	XFS_BUF_DONE(sbp);
	XFS_BUF_ERROR(sbp, 0);
	XFS_BUF_UNORDERED(sbp);

	xfs_buf_relse(sbp);
	return error;
}

void
xfs_mountfs_check_barriers(xfs_mount_t *mp)
{
	int error;

	if (mp->m_logdev_targp != mp->m_ddev_targp) {
		xfs_fs_cmn_err(CE_NOTE, mp,
		  "Disabling barriers, not supported with external log device");
		mp->m_flags &= ~XFS_MOUNT_BARRIER;
	}

	if (mp->m_ddev_targp->pbr_bdev->bd_disk->queue->ordered ==
					QUEUE_ORDERED_NONE) {
		xfs_fs_cmn_err(CE_NOTE, mp,
		  "Disabling barriers, not supported by the underlying device");
		mp->m_flags &= ~XFS_MOUNT_BARRIER;
	}

	error = xfs_barrier_test(mp);
	if (error) {
		xfs_fs_cmn_err(CE_NOTE, mp,
		  "Disabling barriers, trial barrier write failed");
		mp->m_flags &= ~XFS_MOUNT_BARRIER;
	}
}

void
xfs_blkdev_issue_flush(
	xfs_buftarg_t		*buftarg)
{
	blkdev_issue_flush(buftarg->pbr_bdev, NULL);
}

STATIC struct inode *
linvfs_alloc_inode(
	struct super_block	*sb)
{
	vnode_t			*vp;

	vp = kmem_cache_alloc(xfs_vnode_zone, kmem_flags_convert(KM_SLEEP));
	if (!vp)
		return NULL;
	return LINVFS_GET_IP(vp);
}

STATIC void
linvfs_destroy_inode(
	struct inode		*inode)
{
	kmem_zone_free(xfs_vnode_zone, LINVFS_GET_VP(inode));
}

STATIC void
linvfs_inode_init_once(
	void			*data,
	kmem_cache_t		*cachep,
	unsigned long		flags)
{
	vnode_t			*vp = (vnode_t *)data;

	if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
	    SLAB_CTOR_CONSTRUCTOR)
		inode_init_once(LINVFS_GET_IP(vp));
}

STATIC int
linvfs_init_zones(void)
{
	xfs_vnode_zone = kmem_cache_create("xfs_vnode",
				sizeof(vnode_t), 0, SLAB_RECLAIM_ACCOUNT,
				linvfs_inode_init_once, NULL);
	if (!xfs_vnode_zone)
		goto out;

	xfs_ioend_zone = kmem_zone_init(sizeof(xfs_ioend_t), "xfs_ioend");
	if (!xfs_ioend_zone)
		goto out_destroy_vnode_zone;

	xfs_ioend_pool = mempool_create(4 * MAX_BUF_PER_PAGE,
			mempool_alloc_slab, mempool_free_slab,
			xfs_ioend_zone);
	if (!xfs_ioend_pool)
		goto out_free_ioend_zone;

	return 0;


 out_free_ioend_zone:
	kmem_zone_destroy(xfs_ioend_zone);
 out_destroy_vnode_zone:
	kmem_zone_destroy(xfs_vnode_zone);
 out:
	return -ENOMEM;
}

STATIC void
linvfs_destroy_zones(void)
{
	mempool_destroy(xfs_ioend_pool);
	kmem_zone_destroy(xfs_vnode_zone);
	kmem_zone_destroy(xfs_ioend_zone);
}

/*
 * Attempt to flush the inode, this will actually fail
 * if the inode is pinned, but we dirty the inode again
 * at the point when it is unpinned after a log write,
 * since this is when the inode itself becomes flushable. 
 */
STATIC int
linvfs_write_inode(
	struct inode		*inode,
	int			sync)
{
	vnode_t			*vp = LINVFS_GET_VP(inode);
	int			error = 0, flags = FLUSH_INODE;

	if (vp) {
		vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
		if (sync)
			flags |= FLUSH_SYNC;
		VOP_IFLUSH(vp, flags, error);
		if (error == EAGAIN) {
			if (sync)
				VOP_IFLUSH(vp, flags | FLUSH_LOG, error);
			else
				error = 0;
		}
	}

	return -error;
}

STATIC void
linvfs_clear_inode(
	struct inode		*inode)
{
	vnode_t			*vp = LINVFS_GET_VP(inode);
	int			error, cache;

	vn_trace_entry(vp, "clear_inode", (inst_t *)__return_address);

	XFS_STATS_INC(vn_rele);
	XFS_STATS_INC(vn_remove);
	XFS_STATS_INC(vn_reclaim);
	XFS_STATS_DEC(vn_active);

	/*
	 * This can happen because xfs_iget_core calls xfs_idestroy if we
	 * find an inode with di_mode == 0 but without IGET_CREATE set.
	 */
	if (vp->v_fbhv)
		VOP_INACTIVE(vp, NULL, cache);

	VN_LOCK(vp);
	vp->v_flag &= ~VMODIFIED;
	VN_UNLOCK(vp, 0);

	if (vp->v_fbhv) {
		VOP_RECLAIM(vp, error);
		if (error)
			panic("vn_purge: cannot reclaim");
	}

	ASSERT(vp->v_fbhv == NULL);

#ifdef XFS_VNODE_TRACE
	ktrace_free(vp->v_trace);
#endif
}

/*
 * Enqueue a work item to be picked up by the vfs xfssyncd thread.
 * Doing this has two advantages:
 * - It saves on stack space, which is tight in certain situations
 * - It can be used (with care) as a mechanism to avoid deadlocks.
 * Flushing while allocating in a full filesystem requires both.
 */
STATIC void
xfs_syncd_queue_work(
	struct vfs	*vfs,
	void		*data,
	void		(*syncer)(vfs_t *, void *))
{
	vfs_sync_work_t	*work;

	work = kmem_alloc(sizeof(struct vfs_sync_work), KM_SLEEP);
	INIT_LIST_HEAD(&work->w_list);
	work->w_syncer = syncer;
	work->w_data = data;
	work->w_vfs = vfs;
	spin_lock(&vfs->vfs_sync_lock);
	list_add_tail(&work->w_list, &vfs->vfs_sync_list);
	spin_unlock(&vfs->vfs_sync_lock);
	wake_up_process(vfs->vfs_sync_task);
}

/*
 * Flush delayed allocate data, attempting to free up reserved space
 * from existing allocations.  At this point a new allocation attempt
 * has failed with ENOSPC and we are in the process of scratching our
 * heads, looking about for more room...
 */
STATIC void
xfs_flush_inode_work(
	vfs_t		*vfs,
	void		*inode)