xfs_dquot.c

linux-2.4.29操作系统的源码

/*
 * Copyright (c) 2000-2003 Silicon Graphics, Inc.  All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of version 2 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.
 *
 * Further, this software is distributed without any warranty that it is
 * free of the rightful claim of any third person regarding infringement
 * or the like.	 Any license provided herein, whether implied or
 * otherwise, applies only to this software file.  Patent licenses, if
 * any, provided herein do not apply to combinations of this program with
 * other software, or any other product whatsoever.
 *
 * You should have received a copy of the GNU General Public License along
 * with this program; if not, write the Free Software Foundation, Inc., 59
 * Temple Place - Suite 330, Boston MA 02111-1307, USA.
 *
 * Contact information: Silicon Graphics, Inc., 1600 Amphitheatre Pkwy,
 * Mountain View, CA  94043, or:
 *
 * http://www.sgi.com
 *
 * For further information regarding this notice, see:
 *
 * http://oss.sgi.com/projects/GenInfo/SGIGPLNoticeExplan/
 */

#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_inum.h"
#include "xfs_log.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_alloc_btree.h"
#include "xfs_bmap_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_btree.h"
#include "xfs_ialloc.h"
#include "xfs_attr_sf.h"
#include "xfs_dir_sf.h"
#include "xfs_dir2_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_bmap.h"
#include "xfs_bit.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_trans_space.h"
#include "xfs_trans_priv.h"

#include "xfs_qm.h"


/*
   LOCK ORDER

   inode lock		    (ilock)
   dquot hash-chain lock    (hashlock)
   xqm dquot freelist lock  (freelistlock
   mount's dquot list lock  (mplistlock)
   user dquot lock - lock ordering among dquots is based on the uid or gid
   group dquot lock - similar to udquots. Between the two dquots, the udquot
		      has to be locked first.
   pin lock - the dquot lock must be held to take this lock.
   flush lock - ditto.
*/

STATIC void		xfs_qm_dqflush_done(xfs_buf_t *, xfs_dq_logitem_t *);

#ifdef DEBUG
xfs_buftarg_t *xfs_dqerror_target;
int xfs_do_dqerror;
int xfs_dqreq_num;
int xfs_dqerror_mod = 33;
#endif

/*
 * Allocate and initialize a dquot. We don't always allocate fresh memory;
 * we try to reclaim a free dquot if the number of incore dquots are above
 * a threshold.
 * The only field inside the core that gets initialized at this point
 * is the d_id field. The idea is to fill in the entire q_core
 * when we read in the on disk dquot.
 */
xfs_dquot_t *
xfs_qm_dqinit(
	xfs_mount_t  *mp,
	xfs_dqid_t   id,
	uint	     type)
{
	xfs_dquot_t	*dqp;
	boolean_t	brandnewdquot;

	brandnewdquot = xfs_qm_dqalloc_incore(&dqp);
	dqp->dq_flags = type;
	INT_SET(dqp->q_core.d_id, ARCH_CONVERT, id);
	dqp->q_mount = mp;

	/*
	 * No need to re-initialize these if this is a reclaimed dquot.
	 */
	if (brandnewdquot) {
		dqp->dq_flnext = dqp->dq_flprev = dqp;
		mutex_init(&dqp->q_qlock,  MUTEX_DEFAULT, "xdq");
		initnsema(&dqp->q_flock, 1, "fdq");
		sv_init(&dqp->q_pinwait, SV_DEFAULT, "pdq");

#ifdef XFS_DQUOT_TRACE
		dqp->q_trace = ktrace_alloc(DQUOT_TRACE_SIZE, KM_SLEEP);
		xfs_dqtrace_entry(dqp, "DQINIT");
#endif
	} else {
		/*
		 * Only the q_core portion was zeroed in dqreclaim_one().
		 * So, we need to reset others.
		 */
		 dqp->q_nrefs = 0;
		 dqp->q_blkno = 0;
		 dqp->MPL_NEXT = dqp->HL_NEXT = NULL;
		 dqp->HL_PREVP = dqp->MPL_PREVP = NULL;
		 dqp->q_bufoffset = 0;
		 dqp->q_fileoffset = 0;
		 dqp->q_transp = NULL;
		 dqp->q_gdquot = NULL;
		 dqp->q_res_bcount = 0;
		 dqp->q_res_icount = 0;
		 dqp->q_res_rtbcount = 0;
		 dqp->q_pincount = 0;
		 dqp->q_hash = NULL;
		 ASSERT(dqp->dq_flnext == dqp->dq_flprev);

#ifdef XFS_DQUOT_TRACE
		 ASSERT(dqp->q_trace);
		 xfs_dqtrace_entry(dqp, "DQRECLAIMED_INIT");
#endif
	 }

	/*
	 * log item gets initialized later
	 */
	return (dqp);
}

/*
 * This is called to free all the memory associated with a dquot
 */
void
xfs_qm_dqdestroy(
	xfs_dquot_t	*dqp)
{
	ASSERT(! XFS_DQ_IS_ON_FREELIST(dqp));

	mutex_destroy(&dqp->q_qlock);
	freesema(&dqp->q_flock);
	sv_destroy(&dqp->q_pinwait);

#ifdef XFS_DQUOT_TRACE
	if (dqp->q_trace)
	     ktrace_free(dqp->q_trace);
	dqp->q_trace = NULL;
#endif
	kmem_zone_free(xfs_Gqm->qm_dqzone, dqp);
	atomic_dec(&xfs_Gqm->qm_totaldquots);
}

/*
 * This is what a 'fresh' dquot inside a dquot chunk looks like on disk.
 */
STATIC void
xfs_qm_dqinit_core(
	xfs_dqid_t	id,
	uint		type,
	xfs_dqblk_t	*d)
{
	/*
	 * Caller has zero'd the entire dquot 'chunk' already.
	 */
	INT_SET(d->dd_diskdq.d_magic, ARCH_CONVERT, XFS_DQUOT_MAGIC);
	INT_SET(d->dd_diskdq.d_version, ARCH_CONVERT, XFS_DQUOT_VERSION);
	INT_SET(d->dd_diskdq.d_id, ARCH_CONVERT, id);
	INT_SET(d->dd_diskdq.d_flags, ARCH_CONVERT, type);
}


#ifdef XFS_DQUOT_TRACE
/*
 * Dquot tracing for debugging.
 */
/* ARGSUSED */
void
__xfs_dqtrace_entry(
	xfs_dquot_t	*dqp,
	char		*func,
	void		*retaddr,
	xfs_inode_t	*ip)
{
	xfs_dquot_t	*udqp = NULL;
	xfs_ino_t	ino = 0;

	ASSERT(dqp->q_trace);
	if (ip) {
		ino = ip->i_ino;
		udqp = ip->i_udquot;
	}
	ktrace_enter(dqp->q_trace,
		     (void *)(__psint_t)DQUOT_KTRACE_ENTRY,
		     (void *)func,
		     (void *)(__psint_t)dqp->q_nrefs,
		     (void *)(__psint_t)dqp->dq_flags,
		     (void *)(__psint_t)dqp->q_res_bcount,
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_bcount,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_icount,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_blk_hardlimit,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_blk_softlimit,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_ino_hardlimit,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_ino_softlimit,
						ARCH_CONVERT),
		     (void *)(__psint_t)INT_GET(dqp->q_core.d_id, ARCH_CONVERT),
		     (void *)(__psint_t)current_pid(),
		     (void *)(__psint_t)ino,
		     (void *)(__psint_t)retaddr,
		     (void *)(__psint_t)udqp);
	return;
}
#endif


/*
 * If default limits are in force, push them into the dquot now.
 * We overwrite the dquot limits only if they are zero and this
 * is not the root dquot.
 */
void
xfs_qm_adjust_dqlimits(
	xfs_mount_t		*mp,
	xfs_disk_dquot_t	*d)
{
	xfs_quotainfo_t		*q = mp->m_quotainfo;

	ASSERT(!INT_ISZERO(d->d_id, ARCH_CONVERT));

	if (q->qi_bsoftlimit && INT_ISZERO(d->d_blk_softlimit, ARCH_CONVERT))
		INT_SET(d->d_blk_softlimit, ARCH_CONVERT, q->qi_bsoftlimit);
	if (q->qi_bhardlimit && INT_ISZERO(d->d_blk_hardlimit, ARCH_CONVERT))
		INT_SET(d->d_blk_hardlimit, ARCH_CONVERT, q->qi_bhardlimit);
	if (q->qi_isoftlimit && INT_ISZERO(d->d_ino_softlimit, ARCH_CONVERT))
		INT_SET(d->d_ino_softlimit, ARCH_CONVERT, q->qi_isoftlimit);
	if (q->qi_ihardlimit && INT_ISZERO(d->d_ino_hardlimit, ARCH_CONVERT))
		INT_SET(d->d_ino_hardlimit, ARCH_CONVERT, q->qi_ihardlimit);
	if (q->qi_rtbsoftlimit &&
	    INT_ISZERO(d->d_rtb_softlimit, ARCH_CONVERT))
		INT_SET(d->d_rtb_softlimit, ARCH_CONVERT, q->qi_rtbsoftlimit);
	if (q->qi_rtbhardlimit &&
	    INT_ISZERO(d->d_rtb_hardlimit, ARCH_CONVERT))
		INT_SET(d->d_rtb_hardlimit, ARCH_CONVERT, q->qi_rtbhardlimit);
}

/*
 * Check the limits and timers of a dquot and start or reset timers
 * if necessary.
 * This gets called even when quota enforcement is OFF, which makes our
 * life a little less complicated. (We just don't reject any quota
 * reservations in that case, when enforcement is off).
 * We also return 0 as the values of the timers in Q_GETQUOTA calls, when
 * enforcement's off.
 * In contrast, warnings are a little different in that they don't
 * 'automatically' get started when limits get exceeded.
 */
void
xfs_qm_adjust_dqtimers(
	xfs_mount_t		*mp,
	xfs_disk_dquot_t	*d)
{
	ASSERT(!INT_ISZERO(d->d_id, ARCH_CONVERT));

#ifdef QUOTADEBUG
	if (INT_GET(d->d_blk_hardlimit, ARCH_CONVERT))
		ASSERT(INT_GET(d->d_blk_softlimit, ARCH_CONVERT) <=
			INT_GET(d->d_blk_hardlimit, ARCH_CONVERT));
	if (INT_GET(d->d_ino_hardlimit, ARCH_CONVERT))
		ASSERT(INT_GET(d->d_ino_softlimit, ARCH_CONVERT) <=
			INT_GET(d->d_ino_hardlimit, ARCH_CONVERT));
	if (INT_GET(d->d_rtb_hardlimit, ARCH_CONVERT))
		ASSERT(INT_GET(d->d_rtb_softlimit, ARCH_CONVERT) <=
			INT_GET(d->d_rtb_hardlimit, ARCH_CONVERT));
#endif
	if (INT_ISZERO(d->d_btimer, ARCH_CONVERT)) {
		if ((INT_GET(d->d_blk_softlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_bcount, ARCH_CONVERT) >=
				INT_GET(d->d_blk_softlimit, ARCH_CONVERT))) ||
		    (INT_GET(d->d_blk_hardlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_bcount, ARCH_CONVERT) >=
				INT_GET(d->d_blk_hardlimit, ARCH_CONVERT)))) {
			INT_SET(d->d_btimer, ARCH_CONVERT,
				get_seconds() + XFS_QI_BTIMELIMIT(mp));
		}
	} else {
		if ((INT_ISZERO(d->d_blk_softlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_bcount, ARCH_CONVERT) <
				INT_GET(d->d_blk_softlimit, ARCH_CONVERT))) &&
		    (INT_ISZERO(d->d_blk_hardlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_bcount, ARCH_CONVERT) <
				INT_GET(d->d_blk_hardlimit, ARCH_CONVERT)))) {
			INT_ZERO(d->d_btimer, ARCH_CONVERT);
		}
	}

	if (INT_ISZERO(d->d_itimer, ARCH_CONVERT)) {
		if ((INT_GET(d->d_ino_softlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_icount, ARCH_CONVERT) >=
				INT_GET(d->d_ino_softlimit, ARCH_CONVERT))) ||
		    (INT_GET(d->d_ino_hardlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_icount, ARCH_CONVERT) >=
				INT_GET(d->d_ino_hardlimit, ARCH_CONVERT)))) {
			INT_SET(d->d_itimer, ARCH_CONVERT,
				get_seconds() + XFS_QI_ITIMELIMIT(mp));
		}
	} else {
		if ((INT_ISZERO(d->d_ino_softlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_icount, ARCH_CONVERT) <
				INT_GET(d->d_ino_softlimit, ARCH_CONVERT)))  &&
		    (INT_ISZERO(d->d_ino_hardlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_icount, ARCH_CONVERT) <
				INT_GET(d->d_ino_hardlimit, ARCH_CONVERT)))) {
			INT_ZERO(d->d_itimer, ARCH_CONVERT);
		}
	}

	if (INT_ISZERO(d->d_rtbtimer, ARCH_CONVERT)) {
		if ((INT_GET(d->d_rtb_softlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_rtbcount, ARCH_CONVERT) >=
				INT_GET(d->d_rtb_softlimit, ARCH_CONVERT))) ||
		    (INT_GET(d->d_rtb_hardlimit, ARCH_CONVERT) &&
		    (INT_GET(d->d_rtbcount, ARCH_CONVERT) >=
				INT_GET(d->d_rtb_hardlimit, ARCH_CONVERT)))) {
			INT_SET(d->d_rtbtimer, ARCH_CONVERT,
				get_seconds() + XFS_QI_RTBTIMELIMIT(mp));
		}
	} else {
		if ((INT_ISZERO(d->d_rtb_softlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_rtbcount, ARCH_CONVERT) <
				INT_GET(d->d_rtb_softlimit, ARCH_CONVERT))) &&
		    (INT_ISZERO(d->d_rtb_hardlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_rtbcount, ARCH_CONVERT) <
				INT_GET(d->d_rtb_hardlimit, ARCH_CONVERT)))) {
			INT_ZERO(d->d_rtbtimer, ARCH_CONVERT);
		}
	}
}

/*
 * Increment or reset warnings of a given dquot.
 */
int
xfs_qm_dqwarn(
	xfs_disk_dquot_t	*d,
	uint			flags)
{
	int	warned;

	/*
	 * root's limits are not real limits.
	 */
	if (INT_ISZERO(d->d_id, ARCH_CONVERT))
		return (0);

	warned = 0;
	if (INT_GET(d->d_blk_softlimit, ARCH_CONVERT) &&
	    (INT_GET(d->d_bcount, ARCH_CONVERT) >=
	     INT_GET(d->d_blk_softlimit, ARCH_CONVERT))) {
		if (flags & XFS_QMOPT_DOWARN) {
			INT_MOD(d->d_bwarns, ARCH_CONVERT, +1);
			warned++;
		}
	} else {
		if (INT_ISZERO(d->d_blk_softlimit, ARCH_CONVERT) ||
		    (INT_GET(d->d_bcount, ARCH_CONVERT) <
		     INT_GET(d->d_blk_softlimit, ARCH_CONVERT))) {
			INT_ZERO(d->d_bwarns, ARCH_CONVERT);
		}
	}

	if (INT_GET(d->d_ino_softlimit, ARCH_CONVERT) > 0 &&
	    (INT_GET(d->d_icount, ARCH_CONVERT) >=
	     INT_GET(d->d_ino_softlimit, ARCH_CONVERT))) {
		if (flags & XFS_QMOPT_DOWARN) {
			INT_MOD(d->d_iwarns, ARCH_CONVERT, +1);
			warned++;
		}
	} else {
		if ((INT_ISZERO(d->d_ino_softlimit, ARCH_CONVERT)) ||
		    (INT_GET(d->d_icount, ARCH_CONVERT) <
		     INT_GET(d->d_ino_softlimit, ARCH_CONVERT))) {
			INT_ZERO(d->d_iwarns, ARCH_CONVERT);
		}
	}
#ifdef QUOTADEBUG
	if (INT_GET(d->d_iwarns, ARCH_CONVERT))
		cmn_err(CE_DEBUG,
			"--------@@Inode warnings running : %Lu >= %Lu",
			INT_GET(d->d_icount, ARCH_CONVERT),
			INT_GET(d->d_ino_softlimit, ARCH_CONVERT));
	if (INT_GET(d->d_bwarns, ARCH_CONVERT))
		cmn_err(CE_DEBUG,
			"--------@@Blks warnings running : %Lu >= %Lu",
			INT_GET(d->d_bcount, ARCH_CONVERT),
			INT_GET(d->d_blk_softlimit, ARCH_CONVERT));
#endif
	return (warned);
}


/*
 * initialize a buffer full of dquots and log the whole thing
 */
STATIC void
xfs_qm_init_dquot_blk(
	xfs_trans_t	*tp,
	xfs_mount_t	*mp,
	xfs_dqid_t	id,
	uint		type,
	xfs_buf_t	*bp)
{
	xfs_dqblk_t	*d;
	int		curid, i;

	ASSERT(tp);
	ASSERT(XFS_BUF_ISBUSY(bp));
	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);

	d = (xfs_dqblk_t *)XFS_BUF_PTR(bp);

	/*
	 * ID of the first dquot in the block - id's are zero based.
	 */
	curid = id - (id % XFS_QM_DQPERBLK(mp));
	ASSERT(curid >= 0);
	memset(d, 0, BBTOB(XFS_QI_DQCHUNKLEN(mp)));
	for (i = 0; i < XFS_QM_DQPERBLK(mp); i++, d++, curid++)
		xfs_qm_dqinit_core(curid, type, d);
	xfs_trans_dquot_buf(tp, bp,
			    type & XFS_DQ_USER ?
			    XFS_BLI_UDQUOT_BUF :
			    XFS_BLI_GDQUOT_BUF);
	xfs_trans_log_buf(tp, bp, 0, BBTOB(XFS_QI_DQCHUNKLEN(mp)) - 1);
}



/*
 * Allocate a block and fill it with dquots.
 * This is called when the bmapi finds a hole.
 */
STATIC int
xfs_qm_dqalloc(
	xfs_trans_t	*tp,
	xfs_mount_t	*mp,
	xfs_dquot_t	*dqp,
	xfs_inode_t	*quotip,
	xfs_fileoff_t	offset_fsb,
	xfs_buf_t	**O_bpp)
{
	xfs_fsblock_t	firstblock;
	xfs_bmap_free_t flist;
	xfs_bmbt_irec_t map;
	int		nmaps, error, committed;
	xfs_buf_t	*bp;

	ASSERT(tp != NULL);
	xfs_dqtrace_entry(dqp, "DQALLOC");

	/*
	 * Initialize the bmap freelist prior to calling bmapi code.
	 */
	XFS_BMAP_INIT(&flist, &firstblock);
	xfs_ilock(quotip, XFS_ILOCK_EXCL);
	/*
	 * Return if this type of quotas is turned off while we didn't
	 * have an inode lock
	 */
	if (XFS_IS_THIS_QUOTA_OFF(dqp)) {
		xfs_iunlock(quotip, XFS_ILOCK_EXCL);
		return (ESRCH);
	}

	/*
	 * xfs_trans_commit normally decrements the vnode ref count
	 * when it unlocks the inode. Since we want to keep the quota
	 * inode around, we bump the vnode ref count now.
	 */
	VN_HOLD(XFS_ITOV(quotip));

	xfs_trans_ijoin(tp, quotip, XFS_ILOCK_EXCL);
	nmaps = 1;
	if ((error = xfs_bmapi(tp, quotip,
			      offset_fsb, XFS_DQUOT_CLUSTER_SIZE_FSB,
			      XFS_BMAPI_METADATA | XFS_BMAPI_WRITE,
			      &firstblock,
			      XFS_QM_DQALLOC_SPACE_RES(mp),
			      &map, &nmaps, &flist))) {
		goto error0;
	}
	ASSERT(map.br_blockcount == XFS_DQUOT_CLUSTER_SIZE_FSB);
	ASSERT(nmaps == 1);
	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
	       (map.br_startblock != HOLESTARTBLOCK));

	/*
	 * Keep track of the blkno to save a lookup later
	 */
	dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);

	/* now we can just get the buffer (there's nothing to read yet) */
	bp = xfs_trans_get_buf(tp, mp->m_ddev_targp,
			       dqp->q_blkno,
			       XFS_QI_DQCHUNKLEN(mp),
			       0);
	if (!bp || (error = XFS_BUF_GETERROR(bp)))
		goto error1;
	/*
	 * Make a chunk of dquots out of this buffer and log
	 * the entire thing.
	 */
	xfs_qm_init_dquot_blk(tp, mp, INT_GET(dqp->q_core.d_id, ARCH_CONVERT),
			      dqp->dq_flags & (XFS_DQ_USER|XFS_DQ_GROUP),
			      bp);