/* Find the roots of the per-AG btrees described in btree_info. */
int
xrep_find_ag_btree_roots(
struct xfs_scrub *sc,
struct xfs_buf *agf_bp,
struct xrep_find_ag_btree *btree_info,
struct xfs_buf *agfl_bp)
{
struct xfs_mount *mp = sc->mp;
struct xrep_findroot ri;
struct xrep_find_ag_btree *fab;
struct xfs_btree_cur *cur;
int error;
ASSERT(xfs_buf_islocked(agf_bp));
ASSERT(agfl_bp == NULL || xfs_buf_islocked(agfl_bp));
ri.sc = sc;
ri.btree_info = btree_info;
ri.agf = XFS_BUF_TO_AGF(agf_bp);
ri.agfl_bp = agfl_bp;
for (fab = btree_info; fab->buf_ops; fab++) {
ASSERT(agfl_bp || fab->rmap_owner != XFS_RMAP_OWN_AG);
ASSERT(XFS_RMAP_NON_INODE_OWNER(fab->rmap_owner));
fab->root = NULLAGBLOCK;
fab->height = 0;
}
cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
error = xfs_rmap_query_all(cur, xrep_findroot_rmap, &ri);
xfs_btree_del_cursor(cur, error);
return error;
}
/*
* Read in the in-core dquot's on-disk metadata and return the buffer.
* Returns ENOENT to signal a hole.
*/
STATIC int
xfs_dquot_disk_read(
struct xfs_mount *mp,
struct xfs_dquot *dqp,
struct xfs_buf **bpp)
{
struct xfs_bmbt_irec map;
struct xfs_buf *bp;
struct xfs_inode *quotip = xfs_quota_inode(mp, dqp->dq_flags);
uint lock_mode;
int nmaps = 1;
int error;
lock_mode = xfs_ilock_data_map_shared(quotip);
if (!xfs_this_quota_on(mp, dqp->dq_flags)) {
/*
* Return if this type of quotas is turned off while we
* didn't have the quota inode lock.
*/
xfs_iunlock(quotip, lock_mode);
return -ESRCH;
}
/*
* Find the block map; no allocations yet
*/
error = xfs_bmapi_read(quotip, dqp->q_fileoffset,
XFS_DQUOT_CLUSTER_SIZE_FSB, &map, &nmaps, 0);
xfs_iunlock(quotip, lock_mode);
if (error)
return error;
ASSERT(nmaps == 1);
ASSERT(map.br_blockcount >= 1);
ASSERT(map.br_startblock != DELAYSTARTBLOCK);
if (map.br_startblock == HOLESTARTBLOCK)
return -ENOENT;
trace_xfs_dqtobp_read(dqp);
/*
* store the blkno etc so that we don't have to do the
* mapping all the time
*/
dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);
error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp, dqp->q_blkno,
mp->m_quotainfo->qi_dqchunklen, 0, &bp,
&xfs_dquot_buf_ops);
if (error) {
ASSERT(bp == NULL);
return error;
}
ASSERT(xfs_buf_islocked(bp));
xfs_buf_set_ref(bp, XFS_DQUOT_REF);
*bpp = bp;
return 0;
}
STATIC int
xfs_qm_dqrepair(
struct xfs_mount *mp,
struct xfs_trans *tp,
struct xfs_dquot *dqp,
xfs_dqid_t firstid,
struct xfs_buf **bpp)
{
int error;
struct xfs_disk_dquot *ddq;
struct xfs_dqblk *d;
int i;
/*
* Read the buffer without verification so we get the corrupted
* buffer returned to us. make sure we verify it on write, though.
*/
error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, dqp->q_blkno,
mp->m_quotainfo->qi_dqchunklen,
0, bpp, NULL);
if (error) {
ASSERT(*bpp == NULL);
return XFS_ERROR(error);
}
(*bpp)->b_ops = &xfs_dquot_buf_ops;
ASSERT(xfs_buf_islocked(*bpp));
d = (struct xfs_dqblk *)(*bpp)->b_addr;
/* Do the actual repair of dquots in this buffer */
for (i = 0; i < mp->m_quotainfo->qi_dqperchunk; i++) {
ddq = &d[i].dd_diskdq;
error = xfs_dqcheck(mp, ddq, firstid + i,
dqp->dq_flags & XFS_DQ_ALLTYPES,
XFS_QMOPT_DQREPAIR, "xfs_qm_dqrepair");
if (error) {
/* repair failed, we're screwed */
xfs_trans_brelse(tp, *bpp);
return XFS_ERROR(EIO);
}
}
return 0;
}
/*
* 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)
{
struct xfs_quotainfo *q = mp->m_quotainfo;
xfs_dqblk_t *d;
int curid, i;
ASSERT(tp);
ASSERT(xfs_buf_islocked(bp));
d = bp->b_addr;
/*
* ID of the first dquot in the block - id's are zero based.
*/
curid = id - (id % q->qi_dqperchunk);
ASSERT(curid >= 0);
memset(d, 0, BBTOB(q->qi_dqchunklen));
for (i = 0; i < q->qi_dqperchunk; i++, d++, curid++) {
d->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
d->dd_diskdq.d_version = XFS_DQUOT_VERSION;
d->dd_diskdq.d_id = cpu_to_be32(curid);
d->dd_diskdq.d_flags = type;
if (xfs_sb_version_hascrc(&mp->m_sb)) {
uuid_copy(&d->dd_uuid, &mp->m_sb.sb_uuid);
xfs_update_cksum((char *)d, sizeof(struct xfs_dqblk),
XFS_DQUOT_CRC_OFF);
}
}
xfs_trans_dquot_buf(tp, bp,
(type & XFS_DQ_USER ? XFS_BLF_UDQUOT_BUF :
((type & XFS_DQ_PROJ) ? XFS_BLF_PDQUOT_BUF :
XFS_BLF_GDQUOT_BUF)));
xfs_trans_log_buf(tp, bp, 0, BBTOB(q->qi_dqchunklen) - 1);
}
/*
* Read in the ondisk dquot using dqtobp() then copy it to an incore version,
* and release the buffer immediately. If @can_alloc is true, fill any
* holes in the on-disk metadata.
*/
static int
xfs_qm_dqread(
struct xfs_mount *mp,
xfs_dqid_t id,
uint type,
bool can_alloc,
struct xfs_dquot **dqpp)
{
struct xfs_dquot *dqp;
struct xfs_buf *bp;
int error;
dqp = xfs_dquot_alloc(mp, id, type);
trace_xfs_dqread(dqp);
/* Try to read the buffer, allocating if necessary. */
error = xfs_dquot_disk_read(mp, dqp, &bp);
if (error == -ENOENT && can_alloc)
error = xfs_qm_dqread_alloc(mp, dqp, &bp);
if (error)
goto err;
/*
* At this point we should have a clean locked buffer. Copy the data
* to the incore dquot and release the buffer since the incore dquot
* has its own locking protocol so we needn't tie up the buffer any
* further.
*/
ASSERT(xfs_buf_islocked(bp));
xfs_dquot_from_disk(dqp, bp);
xfs_buf_relse(bp);
*dqpp = dqp;
return error;
err:
trace_xfs_dqread_fail(dqp);
xfs_qm_dqdestroy(dqp);
*dqpp = NULL;
return error;
}
/*
* Add the given log item with its callback to the list of callbacks
* to be called when the buffer's I/O completes. If it is not set
* already, set the buffer's b_iodone() routine to be
* xfs_buf_iodone_callbacks() and link the log item into the list of
* items rooted at b_fsprivate. Items are always added as the second
* entry in the list if there is a first, because the buf item code
* assumes that the buf log item is first.
*/
void
xfs_buf_attach_iodone(
xfs_buf_t *bp,
void (*cb)(xfs_buf_t *, xfs_log_item_t *),
xfs_log_item_t *lip)
{
xfs_log_item_t *head_lip;
ASSERT(xfs_buf_islocked(bp));
lip->li_cb = cb;
head_lip = bp->b_fspriv;
if (head_lip) {
lip->li_bio_list = head_lip->li_bio_list;
head_lip->li_bio_list = lip;
} else {
bp->b_fspriv = lip;
}
ASSERT(bp->b_iodone == NULL ||
bp->b_iodone == xfs_buf_iodone_callbacks);
bp->b_iodone = xfs_buf_iodone_callbacks;
}
/*
* This is called to unpin the buffer associated with the buf log
* item which was previously pinned with a call to xfs_buf_item_pin().
*
* Also drop the reference to the buf item for the current transaction.
* If the XFS_BLI_STALE flag is set and we are the last reference,
* then free up the buf log item and unlock the buffer.
*
* If the remove flag is set we are called from uncommit in the
* forced-shutdown path. If that is true and the reference count on
* the log item is going to drop to zero we need to free the item's
* descriptor in the transaction.
*/
STATIC void
xfs_buf_item_unpin(
struct xfs_log_item *lip,
int remove)
{
struct xfs_buf_log_item *bip = BUF_ITEM(lip);
xfs_buf_t *bp = bip->bli_buf;
struct xfs_ail *ailp = lip->li_ailp;
int stale = bip->bli_flags & XFS_BLI_STALE;
int freed;
ASSERT(bp->b_fspriv == bip);
ASSERT(atomic_read(&bip->bli_refcount) > 0);
trace_xfs_buf_item_unpin(bip);
freed = atomic_dec_and_test(&bip->bli_refcount);
if (atomic_dec_and_test(&bp->b_pin_count))
wake_up_all(&bp->b_waiters);
if (freed && stale) {
ASSERT(bip->bli_flags & XFS_BLI_STALE);
ASSERT(xfs_buf_islocked(bp));
ASSERT(XFS_BUF_ISSTALE(bp));
ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
trace_xfs_buf_item_unpin_stale(bip);
if (remove) {
/*
* If we are in a transaction context, we have to
* remove the log item from the transaction as we are
* about to release our reference to the buffer. If we
* don't, the unlock that occurs later in
* xfs_trans_uncommit() will try to reference the
* buffer which we no longer have a hold on.
*/
if (lip->li_desc)
xfs_trans_del_item(lip);
/*
* Since the transaction no longer refers to the buffer,
* the buffer should no longer refer to the transaction.
*/
bp->b_transp = NULL;
}
/*
* If we get called here because of an IO error, we may
* or may not have the item on the AIL. xfs_trans_ail_delete()
* will take care of that situation.
* xfs_trans_ail_delete() drops the AIL lock.
*/
if (bip->bli_flags & XFS_BLI_STALE_INODE) {
xfs_buf_do_callbacks(bp);
bp->b_fspriv = NULL;
bp->b_iodone = NULL;
} else {
spin_lock(&ailp->xa_lock);
xfs_trans_ail_delete(ailp, lip, SHUTDOWN_LOG_IO_ERROR);
xfs_buf_item_relse(bp);
ASSERT(bp->b_fspriv == NULL);
}
xfs_buf_relse(bp);
} else if (freed && remove) {
/*
* There are currently two references to the buffer - the active
* LRU reference and the buf log item. What we are about to do
* here - simulate a failed IO completion - requires 3
* references.
*
* The LRU reference is removed by the xfs_buf_stale() call. The
* buf item reference is removed by the xfs_buf_iodone()
* callback that is run by xfs_buf_do_callbacks() during ioend
* processing (via the bp->b_iodone callback), and then finally
* the ioend processing will drop the IO reference if the buffer
* is marked XBF_ASYNC.
*
* Hence we need to take an additional reference here so that IO
* completion processing doesn't free the buffer prematurely.
*/
xfs_buf_lock(bp);
xfs_buf_hold(bp);
bp->b_flags |= XBF_ASYNC;
xfs_buf_ioerror(bp, EIO);
XFS_BUF_UNDONE(bp);
xfs_buf_stale(bp);
xfs_buf_ioend(bp, 0);
}
}
/*
* Read in the ondisk dquot using dqtobp() then copy it to an incore version,
* and release the buffer immediately.
*
* If XFS_QMOPT_DQALLOC is set, allocate a dquot on disk if it needed.
*/
int
xfs_qm_dqread(
struct xfs_mount *mp,
xfs_dqid_t id,
uint type,
uint flags,
struct xfs_dquot **O_dqpp)
{
struct xfs_dquot *dqp;
struct xfs_disk_dquot *ddqp;
struct xfs_buf *bp;
struct xfs_trans *tp = NULL;
int error;
int cancelflags = 0;
dqp = kmem_zone_zalloc(xfs_qm_dqzone, KM_SLEEP);
dqp->dq_flags = type;
dqp->q_core.d_id = cpu_to_be32(id);
dqp->q_mount = mp;
INIT_LIST_HEAD(&dqp->q_lru);
mutex_init(&dqp->q_qlock);
init_waitqueue_head(&dqp->q_pinwait);
/*
* Because we want to use a counting completion, complete
* the flush completion once to allow a single access to
* the flush completion without blocking.
*/
init_completion(&dqp->q_flush);
complete(&dqp->q_flush);
/*
* Make sure group quotas have a different lock class than user
* quotas.
*/
switch (type) {
case XFS_DQ_USER:
/* uses the default lock class */
break;
case XFS_DQ_GROUP:
lockdep_set_class(&dqp->q_qlock, &xfs_dquot_group_class);
break;
case XFS_DQ_PROJ:
lockdep_set_class(&dqp->q_qlock, &xfs_dquot_project_class);
break;
default:
ASSERT(0);
break;
}
XFS_STATS_INC(xs_qm_dquot);
trace_xfs_dqread(dqp);
if (flags & XFS_QMOPT_DQALLOC) {
tp = xfs_trans_alloc(mp, XFS_TRANS_QM_DQALLOC);
error = xfs_trans_reserve(tp, &M_RES(mp)->tr_qm_dqalloc,
XFS_QM_DQALLOC_SPACE_RES(mp), 0);
if (error)
goto error1;
cancelflags = XFS_TRANS_RELEASE_LOG_RES;
}
/*
* get a pointer to the on-disk dquot and the buffer containing it
* dqp already knows its own type (GROUP/USER).
*/
error = xfs_qm_dqtobp(&tp, dqp, &ddqp, &bp, flags);
if (error) {
/*
* This can happen if quotas got turned off (ESRCH),
* or if the dquot didn't exist on disk and we ask to
* allocate (ENOENT).
*/
trace_xfs_dqread_fail(dqp);
cancelflags |= XFS_TRANS_ABORT;
goto error1;
}
/* copy everything from disk dquot to the incore dquot */
memcpy(&dqp->q_core, ddqp, sizeof(xfs_disk_dquot_t));
xfs_qm_dquot_logitem_init(dqp);
/*
* Reservation counters are defined as reservation plus current usage
* to avoid having to add every time.
*/
dqp->q_res_bcount = be64_to_cpu(ddqp->d_bcount);
dqp->q_res_icount = be64_to_cpu(ddqp->d_icount);
dqp->q_res_rtbcount = be64_to_cpu(ddqp->d_rtbcount);
/* initialize the dquot speculative prealloc thresholds */
xfs_dquot_set_prealloc_limits(dqp);
/* Mark the buf so that this will stay incore a little longer */
xfs_buf_set_ref(bp, XFS_DQUOT_REF);
/*
* We got the buffer with a xfs_trans_read_buf() (in dqtobp())
* So we need to release with xfs_trans_brelse().
* The strategy here is identical to that of inodes; we lock
* the dquot in xfs_qm_dqget() before making it accessible to
* others. This is because dquots, like inodes, need a good level of
* concurrency, and we don't want to take locks on the entire buffers
* for dquot accesses.
* Note also that the dquot buffer may even be dirty at this point, if
* this particular dquot was repaired. We still aren't afraid to
* brelse it because we have the changes incore.
*/
ASSERT(xfs_buf_islocked(bp));
xfs_trans_brelse(tp, bp);
if (tp) {
error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
if (error)
goto error0;
}
*O_dqpp = dqp;
return error;
error1:
if (tp)
xfs_trans_cancel(tp, cancelflags);
error0:
xfs_qm_dqdestroy(dqp);
*O_dqpp = NULL;
return error;
}
/*
* Maps a dquot to the buffer containing its on-disk version.
* This returns a ptr to the buffer containing the on-disk dquot
* in the bpp param, and a ptr to the on-disk dquot within that buffer
*/
STATIC int
xfs_qm_dqtobp(
xfs_trans_t **tpp,
xfs_dquot_t *dqp,
xfs_disk_dquot_t **O_ddpp,
xfs_buf_t **O_bpp,
uint flags)
{
struct xfs_bmbt_irec map;
int nmaps = 1, error;
struct xfs_buf *bp;
struct xfs_inode *quotip = xfs_dq_to_quota_inode(dqp);
struct xfs_mount *mp = dqp->q_mount;
xfs_dqid_t id = be32_to_cpu(dqp->q_core.d_id);
struct xfs_trans *tp = (tpp ? *tpp : NULL);
uint lock_mode;
dqp->q_fileoffset = (xfs_fileoff_t)id / mp->m_quotainfo->qi_dqperchunk;
lock_mode = xfs_ilock_data_map_shared(quotip);
if (!xfs_this_quota_on(dqp->q_mount, dqp->dq_flags)) {
/*
* Return if this type of quotas is turned off while we
* didn't have the quota inode lock.
*/
xfs_iunlock(quotip, lock_mode);
return ESRCH;
}
/*
* Find the block map; no allocations yet
*/
error = xfs_bmapi_read(quotip, dqp->q_fileoffset,
XFS_DQUOT_CLUSTER_SIZE_FSB, &map, &nmaps, 0);
xfs_iunlock(quotip, lock_mode);
if (error)
return error;
ASSERT(nmaps == 1);
ASSERT(map.br_blockcount == 1);
/*
* Offset of dquot in the (fixed sized) dquot chunk.
*/
dqp->q_bufoffset = (id % mp->m_quotainfo->qi_dqperchunk) *
sizeof(xfs_dqblk_t);
ASSERT(map.br_startblock != DELAYSTARTBLOCK);
if (map.br_startblock == HOLESTARTBLOCK) {
/*
* We don't allocate unless we're asked to
*/
if (!(flags & XFS_QMOPT_DQALLOC))
return ENOENT;
ASSERT(tp);
error = xfs_qm_dqalloc(tpp, mp, dqp, quotip,
dqp->q_fileoffset, &bp);
if (error)
return error;
tp = *tpp;
} else {
trace_xfs_dqtobp_read(dqp);
/*
* store the blkno etc so that we don't have to do the
* mapping all the time
*/
dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);
error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
dqp->q_blkno,
mp->m_quotainfo->qi_dqchunklen,
0, &bp, &xfs_dquot_buf_ops);
if (error == EFSCORRUPTED && (flags & XFS_QMOPT_DQREPAIR)) {
xfs_dqid_t firstid = (xfs_dqid_t)map.br_startoff *
mp->m_quotainfo->qi_dqperchunk;
ASSERT(bp == NULL);
error = xfs_qm_dqrepair(mp, tp, dqp, firstid, &bp);
}
if (error) {
ASSERT(bp == NULL);
return XFS_ERROR(error);
}
}
ASSERT(xfs_buf_islocked(bp));
*O_bpp = bp;
*O_ddpp = bp->b_addr + dqp->q_bufoffset;
return (0);
}
/*
* Maps a dquot to the buffer containing its on-disk version.
* This returns a ptr to the buffer containing the on-disk dquot
* in the bpp param, and a ptr to the on-disk dquot within that buffer
*/
STATIC int
xfs_qm_dqtobp(
xfs_trans_t **tpp,
xfs_dquot_t *dqp,
xfs_disk_dquot_t **O_ddpp,
xfs_buf_t **O_bpp,
uint flags)
{
xfs_bmbt_irec_t map;
int nmaps = 1, error;
xfs_buf_t *bp;
xfs_inode_t *quotip = XFS_DQ_TO_QIP(dqp);
xfs_mount_t *mp = dqp->q_mount;
xfs_disk_dquot_t *ddq;
xfs_dqid_t id = be32_to_cpu(dqp->q_core.d_id);
xfs_trans_t *tp = (tpp ? *tpp : NULL);
dqp->q_fileoffset = (xfs_fileoff_t)id / mp->m_quotainfo->qi_dqperchunk;
xfs_ilock(quotip, XFS_ILOCK_SHARED);
if (!xfs_this_quota_on(dqp->q_mount, dqp->dq_flags)) {
/*
* Return if this type of quotas is turned off while we
* didn't have the quota inode lock.
*/
xfs_iunlock(quotip, XFS_ILOCK_SHARED);
return ESRCH;
}
/*
* Find the block map; no allocations yet
*/
error = xfs_bmapi_read(quotip, dqp->q_fileoffset,
XFS_DQUOT_CLUSTER_SIZE_FSB, &map, &nmaps, 0);
xfs_iunlock(quotip, XFS_ILOCK_SHARED);
if (error)
return error;
ASSERT(nmaps == 1);
ASSERT(map.br_blockcount == 1);
/*
* Offset of dquot in the (fixed sized) dquot chunk.
*/
dqp->q_bufoffset = (id % mp->m_quotainfo->qi_dqperchunk) *
sizeof(xfs_dqblk_t);
ASSERT(map.br_startblock != DELAYSTARTBLOCK);
if (map.br_startblock == HOLESTARTBLOCK) {
/*
* We don't allocate unless we're asked to
*/
if (!(flags & XFS_QMOPT_DQALLOC))
return ENOENT;
ASSERT(tp);
error = xfs_qm_dqalloc(tpp, mp, dqp, quotip,
dqp->q_fileoffset, &bp);
if (error)
return error;
tp = *tpp;
} else {
trace_xfs_dqtobp_read(dqp);
/*
* store the blkno etc so that we don't have to do the
* mapping all the time
*/
dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);
error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
dqp->q_blkno,
mp->m_quotainfo->qi_dqchunklen,
0, &bp);
if (error || !bp)
return XFS_ERROR(error);
}
ASSERT(xfs_buf_islocked(bp));
/*
* calculate the location of the dquot inside the buffer.
*/
ddq = bp->b_addr + dqp->q_bufoffset;
/*
* A simple sanity check in case we got a corrupted dquot...
*/
error = xfs_qm_dqcheck(mp, ddq, id, dqp->dq_flags & XFS_DQ_ALLTYPES,
flags & (XFS_QMOPT_DQREPAIR|XFS_QMOPT_DOWARN),
"dqtobp");
if (error) {
if (!(flags & XFS_QMOPT_DQREPAIR)) {
xfs_trans_brelse(tp, bp);
return XFS_ERROR(EIO);
}
}
*O_bpp = bp;
*O_ddpp = ddq;
return (0);
}