1 files changed, 318 insertions, 209 deletions

diff --git a/fs/ocfs2/quota_global.c b/fs/ocfs2/quota_global.c
index 1ed0f7c8686..b990a62cff5 100644
--- a/fs/ocfs2/quota_global.c
+++ b/fs/ocfs2/quota_global.c
@@ -3,14 +3,15 @@
  */
 #include <linux/spinlock.h>
 #include <linux/fs.h>
+#include <linux/slab.h>
 #include <linux/quota.h>
 #include <linux/quotaops.h>
 #include <linux/dqblk_qtree.h>
 #include <linux/jiffies.h>
 #include <linux/writeback.h>
 #include <linux/workqueue.h>
+#include <linux/llist.h>
 
-#define MLOG_MASK_PREFIX ML_QUOTA
 #include <cluster/masklog.h>
 
 #include "ocfs2_fs.h"
@@ -23,9 +24,45 @@
 #include "sysfile.h"
 #include "dlmglue.h"
 #include "uptodate.h"
+#include "super.h"
+#include "buffer_head_io.h"
 #include "quota.h"
+#include "ocfs2_trace.h"
 
-static struct workqueue_struct *ocfs2_quota_wq = NULL;
+/*
+ * Locking of quotas with OCFS2 is rather complex. Here are rules that
+ * should be obeyed by all the functions:
+ * - any write of quota structure (either to local or global file) is protected
+ *   by dqio_mutex or dquot->dq_lock.
+ * - any modification of global quota file holds inode cluster lock, i_mutex,
+ *   and ip_alloc_sem of the global quota file (achieved by
+ *   ocfs2_lock_global_qf). It also has to hold qinfo_lock.
+ * - an allocation of new blocks for local quota file is protected by
+ *   its ip_alloc_sem
+ *
+ * A rough sketch of locking dependencies (lf = local file, gf = global file):
+ * Normal filesystem operation:
+ *   start_trans -> dqio_mutex -> write to lf
+ * Syncing of local and global file:
+ *   ocfs2_lock_global_qf -> start_trans -> dqio_mutex -> qinfo_lock ->
+ *     write to gf
+ *						       -> write to lf
+ * Acquire dquot for the first time:
+ *   dq_lock -> ocfs2_lock_global_qf -> qinfo_lock -> read from gf
+ *				     -> alloc space for gf
+ *				     -> start_trans -> qinfo_lock -> write to gf
+ *	     -> ip_alloc_sem of lf -> alloc space for lf
+ *	     -> write to lf
+ * Release last reference to dquot:
+ *   dq_lock -> ocfs2_lock_global_qf -> start_trans -> qinfo_lock -> write to gf
+ *	     -> write to lf
+ * Note that all the above operations also hold the inode cluster lock of lf.
+ * Recovery:
+ *   inode cluster lock of recovered lf
+ *     -> read bitmaps -> ip_alloc_sem of lf
+ *     -> ocfs2_lock_global_qf -> start_trans -> dqio_mutex -> qinfo_lock ->
+ *        write to gf
+ */
 
 static void qsync_work_fn(struct work_struct *work);
 
@@ -59,7 +96,7 @@ static void ocfs2_global_mem2diskdqb(void *dp, struct dquot *dquot)
 	struct ocfs2_global_disk_dqblk *d = dp;
 	struct mem_dqblk *m = &dquot->dq_dqb;
 
-	d->dqb_id = cpu_to_le32(dquot->dq_id);
+	d->dqb_id = cpu_to_le32(from_kqid(&init_user_ns, dquot->dq_id));
 	d->dqb_use_count = cpu_to_le32(OCFS2_DQUOT(dquot)->dq_use_count);
 	d->dqb_ihardlimit = cpu_to_le64(m->dqb_ihardlimit);
 	d->dqb_isoftlimit = cpu_to_le64(m->dqb_isoftlimit);
@@ -69,17 +106,21 @@ static void ocfs2_global_mem2diskdqb(void *dp, struct dquot *dquot)
 	d->dqb_curspace = cpu_to_le64(m->dqb_curspace);
 	d->dqb_btime = cpu_to_le64(m->dqb_btime);
 	d->dqb_itime = cpu_to_le64(m->dqb_itime);
+	d->dqb_pad1 = d->dqb_pad2 = 0;
 }
 
 static int ocfs2_global_is_id(void *dp, struct dquot *dquot)
 {
 	struct ocfs2_global_disk_dqblk *d = dp;
 	struct ocfs2_mem_dqinfo *oinfo =
-			sb_dqinfo(dquot->dq_sb, dquot->dq_type)->dqi_priv;
+			sb_dqinfo(dquot->dq_sb, dquot->dq_id.type)->dqi_priv;
 
 	if (qtree_entry_unused(&oinfo->dqi_gi, dp))
 		return 0;
-	return le32_to_cpu(d->dqb_id) == dquot->dq_id;
+
+	return qid_eq(make_kqid(&init_user_ns, dquot->dq_id.type,
+				le32_to_cpu(d->dqb_id)),
+		      dquot->dq_id);
 }
 
 struct qtree_fmt_operations ocfs2_global_ops = {
@@ -88,14 +129,12 @@ struct qtree_fmt_operations ocfs2_global_ops = {
 	.is_id = ocfs2_global_is_id,
 };
 
-static int ocfs2_validate_quota_block(struct super_block *sb,
-				      struct buffer_head *bh)
+int ocfs2_validate_quota_block(struct super_block *sb, struct buffer_head *bh)
 {
 	struct ocfs2_disk_dqtrailer *dqt =
 		ocfs2_block_dqtrailer(sb->s_blocksize, bh->b_data);
 
-	mlog(0, "Validating quota block %llu\n",
-	     (unsigned long long)bh->b_blocknr);
+	trace_ocfs2_validate_quota_block((unsigned long long)bh->b_blocknr);
 
 	BUG_ON(!buffer_uptodate(bh));
 
@@ -107,45 +146,19 @@ static int ocfs2_validate_quota_block(struct super_block *sb,
 	return ocfs2_validate_meta_ecc(sb, bh->b_data, &dqt->dq_check);
 }
 
-int ocfs2_read_quota_block(struct inode *inode, u64 v_block,
-			   struct buffer_head **bh)
+int ocfs2_read_quota_phys_block(struct inode *inode, u64 p_block,
+				struct buffer_head **bhp)
 {
-	int rc = 0;
-	struct buffer_head *tmp = *bh;
+	int rc;
 
-	rc = ocfs2_read_virt_blocks(inode, v_block, 1, &tmp, 0,
-				    ocfs2_validate_quota_block);
+	*bhp = NULL;
+	rc = ocfs2_read_blocks(INODE_CACHE(inode), p_block, 1, bhp, 0,
+			       ocfs2_validate_quota_block);
 	if (rc)
 		mlog_errno(rc);
-
-	/* If ocfs2_read_virt_blocks() got us a new bh, pass it up. */
-	if (!rc && !*bh)
-		*bh = tmp;
-
 	return rc;
 }
 
-static int ocfs2_get_quota_block(struct inode *inode, int block,
-				 struct buffer_head **bh)
-{
-	u64 pblock, pcount;
-	int err;
-
-	down_read(&OCFS2_I(inode)->ip_alloc_sem);
-	err = ocfs2_extent_map_get_blocks(inode, block, &pblock, &pcount, NULL);
-	up_read(&OCFS2_I(inode)->ip_alloc_sem);
-	if (err) {
-		mlog_errno(err);
-		return err;
-	}
-	*bh = sb_getblk(inode->i_sb, pblock);
-	if (!*bh) {
-		err = -EIO;
-		mlog_errno(err);
-	}
-	return err;;
-}
-
 /* Read data from global quotafile - avoid pagecache and such because we cannot
  * afford acquiring the locks... We use quota cluster lock to serialize
  * operations. Caller is responsible for acquiring it. */
@@ -160,6 +173,7 @@ ssize_t ocfs2_quota_read(struct super_block *sb, int type, char *data,
 	int err = 0;
 	struct buffer_head *bh;
 	size_t toread, tocopy;
+	u64 pblock = 0, pcount = 0;
 
 	if (off > i_size)
 		return 0;
@@ -168,8 +182,19 @@ ssize_t ocfs2_quota_read(struct super_block *sb, int type, char *data,
 	toread = len;
 	while (toread > 0) {
 		tocopy = min_t(size_t, (sb->s_blocksize - offset), toread);
+		if (!pcount) {
+			err = ocfs2_extent_map_get_blocks(gqinode, blk, &pblock,
+							  &pcount, NULL);
+			if (err) {
+				mlog_errno(err);
+				return err;
+			}
+		} else {
+			pcount--;
+			pblock++;
+		}
 		bh = NULL;
-		err = ocfs2_read_quota_block(gqinode, blk, &bh);
+		err = ocfs2_read_quota_phys_block(gqinode, pblock, &bh);
 		if (err) {
 			mlog_errno(err);
 			return err;
@@ -197,6 +222,7 @@ ssize_t ocfs2_quota_write(struct super_block *sb, int type,
 	int err = 0, new = 0, ja_type;
 	struct buffer_head *bh = NULL;
 	handle_t *handle = journal_current_handle();
+	u64 pblock, pcount;
 
 	if (!handle) {
 		mlog(ML_ERROR, "Quota write (off=%llu, len=%llu) cancelled "
@@ -209,32 +235,37 @@ ssize_t ocfs2_quota_write(struct super_block *sb, int type,
 		len = sb->s_blocksize - OCFS2_QBLK_RESERVED_SPACE - offset;
 	}
 
-	mutex_lock_nested(&gqinode->i_mutex, I_MUTEX_QUOTA);
-	if (gqinode->i_size < off + len) {
-		down_write(&OCFS2_I(gqinode)->ip_alloc_sem);
-		err = ocfs2_extend_no_holes(gqinode, off + len, off);
-		up_write(&OCFS2_I(gqinode)->ip_alloc_sem);
-		if (err < 0)
-			goto out;
+	if (i_size_read(gqinode) < off + len) {
+		loff_t rounded_end =
+				ocfs2_align_bytes_to_blocks(sb, off + len);
+
+		/* Space is already allocated in ocfs2_acquire_dquot() */
 		err = ocfs2_simple_size_update(gqinode,
 					       oinfo->dqi_gqi_bh,
-					       off + len);
+					       rounded_end);
 		if (err < 0)
 			goto out;
 		new = 1;
 	}
+	err = ocfs2_extent_map_get_blocks(gqinode, blk, &pblock, &pcount, NULL);
+	if (err) {
+		mlog_errno(err);
+		goto out;
+	}
 	/* Not rewriting whole block? */
 	if ((offset || len < sb->s_blocksize - OCFS2_QBLK_RESERVED_SPACE) &&
 	    !new) {
-		err = ocfs2_read_quota_block(gqinode, blk, &bh);
+		err = ocfs2_read_quota_phys_block(gqinode, pblock, &bh);
 		ja_type = OCFS2_JOURNAL_ACCESS_WRITE;
 	} else {
-		err = ocfs2_get_quota_block(gqinode, blk, &bh);
+		bh = sb_getblk(sb, pblock);
+		if (!bh)
+			err = -ENOMEM;
 		ja_type = OCFS2_JOURNAL_ACCESS_CREATE;
 	}
 	if (err) {
 		mlog_errno(err);
-		return err;
+		goto out;
 	}
 	lock_buffer(bh);
 	if (new)
@@ -243,25 +274,22 @@ ssize_t ocfs2_quota_write(struct super_block *sb, int type,
 	flush_dcache_page(bh->b_page);
 	set_buffer_uptodate(bh);
 	unlock_buffer(bh);
-	ocfs2_set_buffer_uptodate(gqinode, bh);
-	err = ocfs2_journal_access_dq(handle, gqinode, bh, ja_type);
+	ocfs2_set_buffer_uptodate(INODE_CACHE(gqinode), bh);
+	err = ocfs2_journal_access_dq(handle, INODE_CACHE(gqinode), bh,
+				      ja_type);
 	if (err < 0) {
 		brelse(bh);
 		goto out;
 	}
-	err = ocfs2_journal_dirty(handle, bh);
+	ocfs2_journal_dirty(handle, bh);
 	brelse(bh);
-	if (err < 0)
-		goto out;
 out:
 	if (err) {
-		mutex_unlock(&gqinode->i_mutex);
 		mlog_errno(err);
 		return err;
 	}
 	gqinode->i_version++;
 	ocfs2_mark_inode_dirty(handle, gqinode, oinfo->dqi_gqi_bh);
-	mutex_unlock(&gqinode->i_mutex);
 	return len;
 }
 
@@ -279,11 +307,23 @@ int ocfs2_lock_global_qf(struct ocfs2_mem_dqinfo *oinfo, int ex)
 	else
 		WARN_ON(bh != oinfo->dqi_gqi_bh);
 	spin_unlock(&dq_data_lock);
+	if (ex) {
+		mutex_lock(&oinfo->dqi_gqinode->i_mutex);
+		down_write(&OCFS2_I(oinfo->dqi_gqinode)->ip_alloc_sem);
+	} else {
+		down_read(&OCFS2_I(oinfo->dqi_gqinode)->ip_alloc_sem);
+	}
 	return 0;
 }
 
 void ocfs2_unlock_global_qf(struct ocfs2_mem_dqinfo *oinfo, int ex)
 {
+	if (ex) {
+		up_write(&OCFS2_I(oinfo->dqi_gqinode)->ip_alloc_sem);
+		mutex_unlock(&oinfo->dqi_gqinode->i_mutex);
+	} else {
+		up_read(&OCFS2_I(oinfo->dqi_gqinode)->ip_alloc_sem);
+	}
 	ocfs2_inode_unlock(oinfo->dqi_gqinode, ex);
 	brelse(oinfo->dqi_gqi_bh);
 	spin_lock(&dq_data_lock);
@@ -301,10 +341,9 @@ int ocfs2_global_read_info(struct super_block *sb, int type)
 	struct ocfs2_global_disk_dqinfo dinfo;
 	struct mem_dqinfo *info = sb_dqinfo(sb, type);
 	struct ocfs2_mem_dqinfo *oinfo = info->dqi_priv;
+	u64 pcount;
 	int status;
 
-	mlog_entry_void();
-
 	/* Read global header */
 	gqinode = ocfs2_get_system_file_inode(OCFS2_SB(sb), ino[type],
 			OCFS2_INVALID_SLOT);
@@ -327,9 +366,19 @@ int ocfs2_global_read_info(struct super_block *sb, int type)
 		mlog_errno(status);
 		goto out_err;
 	}
+
+	status = ocfs2_extent_map_get_blocks(gqinode, 0, &oinfo->dqi_giblk,
+					     &pcount, NULL);
+	if (status < 0)
+		goto out_unlock;
+
+	status = ocfs2_qinfo_lock(oinfo, 0);
+	if (status < 0)
+		goto out_unlock;
 	status = sb->s_op->quota_read(sb, type, (char *)&dinfo,
 				      sizeof(struct ocfs2_global_disk_dqinfo),
 				      OCFS2_GLOBAL_INFO_OFF);
+	ocfs2_qinfo_unlock(oinfo, 0);
 	ocfs2_unlock_global_qf(oinfo, 0);
 	if (status != sizeof(struct ocfs2_global_disk_dqinfo)) {
 		mlog(ML_ERROR, "Cannot read global quota info (%d).\n",
@@ -342,7 +391,6 @@ int ocfs2_global_read_info(struct super_block *sb, int type)
 	info->dqi_bgrace = le32_to_cpu(dinfo.dqi_bgrace);
 	info->dqi_igrace = le32_to_cpu(dinfo.dqi_igrace);
 	oinfo->dqi_syncms = le32_to_cpu(dinfo.dqi_syncms);
-	oinfo->dqi_syncjiff = msecs_to_jiffies(oinfo->dqi_syncms);
 	oinfo->dqi_gi.dqi_blocks = le32_to_cpu(dinfo.dqi_blocks);
 	oinfo->dqi_gi.dqi_free_blk = le32_to_cpu(dinfo.dqi_free_blk);
 	oinfo->dqi_gi.dqi_free_entry = le32_to_cpu(dinfo.dqi_free_entry);
@@ -351,12 +399,15 @@ int ocfs2_global_read_info(struct super_block *sb, int type)
 						OCFS2_QBLK_RESERVED_SPACE;
 	oinfo->dqi_gi.dqi_qtree_depth = qtree_depth(&oinfo->dqi_gi);
 	INIT_DELAYED_WORK(&oinfo->dqi_sync_work, qsync_work_fn);
-	queue_delayed_work(ocfs2_quota_wq, &oinfo->dqi_sync_work,
-			   oinfo->dqi_syncjiff);
+	schedule_delayed_work(&oinfo->dqi_sync_work,
+			      msecs_to_jiffies(oinfo->dqi_syncms));
 
 out_err:
-	mlog_exit(status);
 	return status;
+out_unlock:
+	ocfs2_unlock_global_qf(oinfo, 0);
+	mlog_errno(status);
+	goto out_err;
 }
 
 /* Write information to global quota file. Expects exlusive lock on quota
@@ -402,44 +453,23 @@ int ocfs2_global_write_info(struct super_block *sb, int type)
 	return err;
 }
 
-/* Read in information from global quota file and acquire a reference to it.
- * dquot_acquire() has already started the transaction and locked quota file */
-int ocfs2_global_read_dquot(struct dquot *dquot)
+static int ocfs2_global_qinit_alloc(struct super_block *sb, int type)
 {
-	int err, err2, ex = 0;
-	struct ocfs2_mem_dqinfo *info =
-			sb_dqinfo(dquot->dq_sb, dquot->dq_type)->dqi_priv;
+	struct ocfs2_mem_dqinfo *oinfo = sb_dqinfo(sb, type)->dqi_priv;
 
-	err = ocfs2_qinfo_lock(info, 0);
-	if (err < 0)
-		goto out;
-	err = qtree_read_dquot(&info->dqi_gi, dquot);
-	if (err < 0)
-		goto out_qlock;
-	OCFS2_DQUOT(dquot)->dq_use_count++;
-	OCFS2_DQUOT(dquot)->dq_origspace = dquot->dq_dqb.dqb_curspace;
-	OCFS2_DQUOT(dquot)->dq_originodes = dquot->dq_dqb.dqb_curinodes;
-	if (!dquot->dq_off) {	/* No real quota entry? */
-		/* Upgrade to exclusive lock for allocation */
-		err = ocfs2_qinfo_lock(info, 1);
-		if (err < 0)
-			goto out_qlock;
-		ex = 1;
-	}
-	err = qtree_write_dquot(&info->dqi_gi, dquot);
-	if (ex && info_dirty(sb_dqinfo(dquot->dq_sb, dquot->dq_type))) {
-		err2 = __ocfs2_global_write_info(dquot->dq_sb, dquot->dq_type);
-		if (!err)
-			err = err2;
-	}
-out_qlock:
-	if (ex)
-		ocfs2_qinfo_unlock(info, 1);
-	ocfs2_qinfo_unlock(info, 0);
-out:
-	if (err < 0)
-		mlog_errno(err);
-	return err;
+	/*
+	 * We may need to allocate tree blocks and a leaf block but not the
+	 * root block
+	 */
+	return oinfo->dqi_gi.dqi_qtree_depth;
+}
+
+static int ocfs2_calc_global_qinit_credits(struct super_block *sb, int type)
+{
+	/* We modify all the allocated blocks, tree root, info block and
+	 * the inode */
+	return (ocfs2_global_qinit_alloc(sb, type) + 2) *
+			OCFS2_QUOTA_BLOCK_UPDATE_CREDITS + 1;
 }
 
 /* Sync local information about quota modifications with global quota file.
@@ -449,7 +479,7 @@ int __ocfs2_sync_dquot(struct dquot *dquot, int freeing)
 {
 	int err, err2;
 	struct super_block *sb = dquot->dq_sb;
-	int type = dquot->dq_type;
+	int type = dquot->dq_id.type;
 	struct ocfs2_mem_dqinfo *info = sb_dqinfo(sb, type)->dqi_priv;
 	struct ocfs2_global_disk_dqblk dqblk;
 	s64 spacechange, inodechange;
@@ -478,9 +508,11 @@ int __ocfs2_sync_dquot(struct dquot *dquot, int freeing)
 	olditime = dquot->dq_dqb.dqb_itime;
 	oldbtime = dquot->dq_dqb.dqb_btime;
 	ocfs2_global_disk2memdqb(dquot, &dqblk);
-	mlog(0, "Syncing global dquot %u space %lld+%lld, inodes %lld+%lld\n",
-	     dquot->dq_id, dquot->dq_dqb.dqb_curspace, (long long)spacechange,
-	     dquot->dq_dqb.dqb_curinodes, (long long)inodechange);
+	trace_ocfs2_sync_dquot(from_kqid(&init_user_ns, dquot->dq_id),
+			       dquot->dq_dqb.dqb_curspace,
+			       (long long)spacechange,
+			       dquot->dq_dqb.dqb_curinodes,
+			       (long long)inodechange);
 	if (!test_bit(DQ_LASTSET_B + QIF_SPACE_B, &dquot->dq_flags))
 		dquot->dq_dqb.dqb_curspace += spacechange;
 	if (!test_bit(DQ_LASTSET_B + QIF_INODES_B, &dquot->dq_flags))
@@ -527,9 +559,9 @@ int __ocfs2_sync_dquot(struct dquot *dquot, int freeing)
 	spin_unlock(&dq_data_lock);
 	err = ocfs2_qinfo_lock(info, freeing);
 	if (err < 0) {
-		mlog(ML_ERROR, "Failed to lock quota info, loosing quota write"
-			       " (type=%d, id=%u)\n", dquot->dq_type,
-			       (unsigned)dquot->dq_id);
+		mlog(ML_ERROR, "Failed to lock quota info, losing quota write"
+			       " (type=%d, id=%u)\n", dquot->dq_id.type,
+			       (unsigned)from_kqid(&init_user_ns, dquot->dq_id));
 		goto out;
 	}
 	if (freeing)
@@ -564,9 +596,10 @@ static int ocfs2_sync_dquot_helper(struct dquot *dquot, unsigned long type)
 	struct ocfs2_super *osb = OCFS2_SB(sb);
 	int status = 0;
 
-	mlog_entry("id=%u qtype=%u type=%lu device=%s\n", dquot->dq_id,
-		   dquot->dq_type, type, sb->s_id);
-	if (type != dquot->dq_type)
+	trace_ocfs2_sync_dquot_helper(from_kqid(&init_user_ns, dquot->dq_id),
+				      dquot->dq_id.type,
+				      type, sb->s_id);
+	if (type != dquot->dq_id.type)
 		goto out;
 	status = ocfs2_lock_global_qf(oinfo, 1);
 	if (status < 0)
@@ -580,19 +613,17 @@ static int ocfs2_sync_dquot_helper(struct dquot *dquot, unsigned long type)
 	}
 	mutex_lock(&sb_dqopt(sb)->dqio_mutex);
 	status = ocfs2_sync_dquot(dquot);
-	mutex_unlock(&sb_dqopt(sb)->dqio_mutex);
 	if (status < 0)
 		mlog_errno(status);
 	/* We have to write local structure as well... */
-	dquot_mark_dquot_dirty(dquot);
-	status = dquot_commit(dquot);
+	status = ocfs2_local_write_dquot(dquot);
 	if (status < 0)
 		mlog_errno(status);
+	mutex_unlock(&sb_dqopt(sb)->dqio_mutex);
 	ocfs2_commit_trans(osb, handle);
 out_ilock:
 	ocfs2_unlock_global_qf(oinfo, 1);
 out:
-	mlog_exit(status);
 	return status;
 }
 
@@ -604,8 +635,8 @@ static void qsync_work_fn(struct work_struct *work)
 	struct super_block *sb = oinfo->dqi_gqinode->i_sb;
 
 	dquot_scan_active(sb, ocfs2_sync_dquot_helper, oinfo->dqi_type);
-	queue_delayed_work(ocfs2_quota_wq, &oinfo->dqi_sync_work,
-			   oinfo->dqi_syncjiff);
+	schedule_delayed_work(&oinfo->dqi_sync_work,
+			      msecs_to_jiffies(oinfo->dqi_syncms));
 }
 
 /*
@@ -618,7 +649,8 @@ static int ocfs2_write_dquot(struct dquot *dquot)
 	struct ocfs2_super *osb = OCFS2_SB(dquot->dq_sb);
 	int status = 0;
 
-	mlog_entry("id=%u, type=%d", dquot->dq_id, dquot->dq_type);
+	trace_ocfs2_write_dquot(from_kqid(&init_user_ns, dquot->dq_id),
+				dquot->dq_id.type);
 
 	handle = ocfs2_start_trans(osb, OCFS2_QWRITE_CREDITS);
 	if (IS_ERR(handle)) {
@@ -626,106 +658,205 @@ static int ocfs2_write_dquot(struct dquot *dquot)
 		mlog_errno(status);
 		goto out;
 	}
-	status = dquot_commit(dquot);
+	mutex_lock(&sb_dqopt(dquot->dq_sb)->dqio_mutex);
+	status = ocfs2_local_write_dquot(dquot);
+	mutex_unlock(&sb_dqopt(dquot->dq_sb)->dqio_mutex);
 	ocfs2_commit_trans(osb, handle);
 out:
-	mlog_exit(status);
 	return status;
 }
 
-int ocfs2_calc_qdel_credits(struct super_block *sb, int type)
+static int ocfs2_calc_qdel_credits(struct super_block *sb, int type)
 {
-	struct ocfs2_mem_dqinfo *oinfo;
-	int features[MAXQUOTAS] = { OCFS2_FEATURE_RO_COMPAT_USRQUOTA,
-				    OCFS2_FEATURE_RO_COMPAT_GRPQUOTA };
+	struct ocfs2_mem_dqinfo *oinfo = sb_dqinfo(sb, type)->dqi_priv;
+	/*
+	 * We modify tree, leaf block, global info, local chunk header,
+	 * global and local inode; OCFS2_QINFO_WRITE_CREDITS already
+	 * accounts for inode update
+	 */
+	return (oinfo->dqi_gi.dqi_qtree_depth + 2) *
+	       OCFS2_QUOTA_BLOCK_UPDATE_CREDITS +
+	       OCFS2_QINFO_WRITE_CREDITS +
+	       OCFS2_INODE_UPDATE_CREDITS;
+}
 
-	if (!OCFS2_HAS_RO_COMPAT_FEATURE(sb, features[type]))
-		return 0;
+void ocfs2_drop_dquot_refs(struct work_struct *work)
+{
+	struct ocfs2_super *osb = container_of(work, struct ocfs2_super,
+					       dquot_drop_work);
+	struct llist_node *list;
+	struct ocfs2_dquot *odquot, *next_odquot;
 
-	oinfo = sb_dqinfo(sb, type)->dqi_priv;
-	/* We modify tree, leaf block, global info, local chunk header,
-	 * global and local inode */
-	return oinfo->dqi_gi.dqi_qtree_depth + 2 + 1 +
-	       2 * OCFS2_INODE_UPDATE_CREDITS;
+	list = llist_del_all(&osb->dquot_drop_list);
+	llist_for_each_entry_safe(odquot, next_odquot, list, list) {
+		/* Drop the reference we acquired in ocfs2_dquot_release() */
+		dqput(&odquot->dq_dquot);
+	}
 }
 
+/*
+ * Called when the last reference to dquot is dropped. If we are called from
+ * downconvert thread, we cannot do all the handling here because grabbing
+ * quota lock could deadlock (the node holding the quota lock could need some
+ * other cluster lock to proceed but with blocked downconvert thread we cannot
+ * release any lock).
+ */
 static int ocfs2_release_dquot(struct dquot *dquot)
 {
 	handle_t *handle;
 	struct ocfs2_mem_dqinfo *oinfo =
-			sb_dqinfo(dquot->dq_sb, dquot->dq_type)->dqi_priv;
+			sb_dqinfo(dquot->dq_sb, dquot->dq_id.type)->dqi_priv;
 	struct ocfs2_super *osb = OCFS2_SB(dquot->dq_sb);
 	int status = 0;
 
-	mlog_entry("id=%u, type=%d", dquot->dq_id, dquot->dq_type);
+	trace_ocfs2_release_dquot(from_kqid(&init_user_ns, dquot->dq_id),
+				  dquot->dq_id.type);
 
+	mutex_lock(&dquot->dq_lock);
+	/* Check whether we are not racing with some other dqget() */
+	if (atomic_read(&dquot->dq_count) > 1)
+		goto out;
+	/* Running from downconvert thread? Postpone quota processing to wq */
+	if (current == osb->dc_task) {
+		/*
+		 * Grab our own reference to dquot and queue it for delayed
+		 * dropping.  Quota code rechecks after calling
+		 * ->release_dquot() and won't free dquot structure.
+		 */
+		dqgrab(dquot);
+		/* First entry on list -> queue work */
+		if (llist_add(&OCFS2_DQUOT(dquot)->list, &osb->dquot_drop_list))
+			queue_work(ocfs2_wq, &osb->dquot_drop_work);
+		goto out;
+	}
 	status = ocfs2_lock_global_qf(oinfo, 1);
 	if (status < 0)
 		goto out;
 	handle = ocfs2_start_trans(osb,
-		ocfs2_calc_qdel_credits(dquot->dq_sb, dquot->dq_type));
+		ocfs2_calc_qdel_credits(dquot->dq_sb, dquot->dq_id.type));
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
 		goto out_ilock;
 	}
-	status = dquot_release(dquot);
+
+	status = ocfs2_global_release_dquot(dquot);
+	if (status < 0) {
+		mlog_errno(status);
+		goto out_trans;
+	}
+	status = ocfs2_local_release_dquot(handle, dquot);
+	/*
+	 * If we fail here, we cannot do much as global structure is
+	 * already released. So just complain...
+	 */
+	if (status < 0)
+		mlog_errno(status);
+	/*
+	 * Clear dq_off so that we search for the structure in quota file next
+	 * time we acquire it. The structure might be deleted and reallocated
+	 * elsewhere by another node while our dquot structure is on freelist.
+	 */
+	dquot->dq_off = 0;
+	clear_bit(DQ_ACTIVE_B, &dquot->dq_flags);
+out_trans:
 	ocfs2_commit_trans(osb, handle);
 out_ilock:
 	ocfs2_unlock_global_qf(oinfo, 1);
 out:
-	mlog_exit(status);
+	mutex_unlock(&dquot->dq_lock);
+	if (status)
+		mlog_errno(status);
 	return status;
 }
 
-int ocfs2_calc_qinit_credits(struct super_block *sb, int type)
-{
-	struct ocfs2_mem_dqinfo *oinfo;
-	int features[MAXQUOTAS] = { OCFS2_FEATURE_RO_COMPAT_USRQUOTA,
-				    OCFS2_FEATURE_RO_COMPAT_GRPQUOTA };
-	struct ocfs2_dinode *lfe, *gfe;
-
-	if (!OCFS2_HAS_RO_COMPAT_FEATURE(sb, features[type]))
-		return 0;
-
-	oinfo = sb_dqinfo(sb, type)->dqi_priv;
-	gfe = (struct ocfs2_dinode *)oinfo->dqi_gqi_bh->b_data;
-	lfe = (struct ocfs2_dinode *)oinfo->dqi_lqi_bh->b_data;
-	/* We can extend local file + global file. In local file we
-	 * can modify info, chunk header block and dquot block. In
-	 * global file we can modify info, tree and leaf block */
-	return ocfs2_calc_extend_credits(sb, &lfe->id2.i_list, 0) +
-	       ocfs2_calc_extend_credits(sb, &gfe->id2.i_list, 0) +
-	       3 + oinfo->dqi_gi.dqi_qtree_depth + 2;
-}
-
+/*
+ * Read global dquot structure from disk or create it if it does
+ * not exist. Also update use count of the global structure and
+ * create structure in node-local quota file.
+ */
 static int ocfs2_acquire_dquot(struct dquot *dquot)
 {
+	int status = 0, err;
+	int ex = 0;
+	struct super_block *sb = dquot->dq_sb;
+	struct ocfs2_super *osb = OCFS2_SB(sb);
+	int type = dquot->dq_id.type;
+	struct ocfs2_mem_dqinfo *info = sb_dqinfo(sb, type)->dqi_priv;
+	struct inode *gqinode = info->dqi_gqinode;
+	int need_alloc = ocfs2_global_qinit_alloc(sb, type);
 	handle_t *handle;
-	struct ocfs2_mem_dqinfo *oinfo =
-			sb_dqinfo(dquot->dq_sb, dquot->dq_type)->dqi_priv;
-	struct ocfs2_super *osb = OCFS2_SB(dquot->dq_sb);
-	int status = 0;
 
-	mlog_entry("id=%u, type=%d", dquot->dq_id, dquot->dq_type);
-	/* We need an exclusive lock, because we're going to update use count
-	 * and instantiate possibly new dquot structure */
-	status = ocfs2_lock_global_qf(oinfo, 1);
+	trace_ocfs2_acquire_dquot(from_kqid(&init_user_ns, dquot->dq_id),
+				  type);
+	mutex_lock(&dquot->dq_lock);
+	/*
+	 * We need an exclusive lock, because we're going to update use count
+	 * and instantiate possibly new dquot structure
+	 */
+	status = ocfs2_lock_global_qf(info, 1);
 	if (status < 0)
 		goto out;
+	status = ocfs2_qinfo_lock(info, 0);
+	if (status < 0)
+		goto out_dq;
+	/*
+	 * We always want to read dquot structure from disk because we don't
+	 * know what happened with it while it was on freelist.
+	 */
+	status = qtree_read_dquot(&info->dqi_gi, dquot);
+	ocfs2_qinfo_unlock(info, 0);
+	if (status < 0)
+		goto out_dq;
+
+	OCFS2_DQUOT(dquot)->dq_use_count++;
+	OCFS2_DQUOT(dquot)->dq_origspace = dquot->dq_dqb.dqb_curspace;
+	OCFS2_DQUOT(dquot)->dq_originodes = dquot->dq_dqb.dqb_curinodes;
+	if (!dquot->dq_off) {	/* No real quota entry? */
+		ex = 1;
+		/*
+		 * Add blocks to quota file before we start a transaction since
+		 * locking allocators ranks above a transaction start
+		 */
+		WARN_ON(journal_current_handle());
+		status = ocfs2_extend_no_holes(gqinode, NULL,
+			i_size_read(gqinode) + (need_alloc << sb->s_blocksize_bits),
+			i_size_read(gqinode));
+		if (status < 0)
+			goto out_dq;
+	}
+
 	handle = ocfs2_start_trans(osb,
-		ocfs2_calc_qinit_credits(dquot->dq_sb, dquot->dq_type));
+				   ocfs2_calc_global_qinit_credits(sb, type));
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
-		mlog_errno(status);
-		goto out_ilock;
+		goto out_dq;
 	}
-	status = dquot_acquire(dquot);
+	status = ocfs2_qinfo_lock(info, ex);
+	if (status < 0)
+		goto out_trans;
+	status = qtree_write_dquot(&info->dqi_gi, dquot);
+	if (ex && info_dirty(sb_dqinfo(sb, type))) {
+		err = __ocfs2_global_write_info(sb, type);
+		if (!status)
+			status = err;
+	}
+	ocfs2_qinfo_unlock(info, ex);
+out_trans:
 	ocfs2_commit_trans(osb, handle);
-out_ilock:
-	ocfs2_unlock_global_qf(oinfo, 1);
+out_dq:
+	ocfs2_unlock_global_qf(info, 1);
+	if (status < 0)
+		goto out;
+
+	status = ocfs2_create_local_dquot(dquot);
+	if (status < 0)
+		goto out;
+	set_bit(DQ_ACTIVE_B, &dquot->dq_flags);
 out:
-	mlog_exit(status);
+	mutex_unlock(&dquot->dq_lock);
+	if (status)
+		mlog_errno(status);
 	return status;
 }
 
@@ -740,13 +871,13 @@ static int ocfs2_mark_dquot_dirty(struct dquot *dquot)
 	int sync = 0;
 	int status;
 	struct super_block *sb = dquot->dq_sb;
-	int type = dquot->dq_type;
+	int type = dquot->dq_id.type;
 	struct ocfs2_mem_dqinfo *oinfo = sb_dqinfo(sb, type)->dqi_priv;
 	handle_t *handle;
 	struct ocfs2_super *osb = OCFS2_SB(sb);
 
-	mlog_entry("id=%u, type=%d", dquot->dq_id, type);
-	dquot_mark_dquot_dirty(dquot);
+	trace_ocfs2_mark_dquot_dirty(from_kqid(&init_user_ns, dquot->dq_id),
+				     type);
 
 	/* In case user set some limits, sync dquot immediately to global
 	 * quota file so that information propagates quicker */
@@ -769,19 +900,22 @@ static int ocfs2_mark_dquot_dirty(struct dquot *dquot)
 		mlog_errno(status);
 		goto out_ilock;
 	}
+	mutex_lock(&sb_dqopt(sb)->dqio_mutex);
 	status = ocfs2_sync_dquot(dquot);
 	if (status < 0) {
 		mlog_errno(status);
-		goto out_trans;
+		goto out_dlock;
 	}
 	/* Now write updated local dquot structure */
-	status = dquot_commit(dquot);
-out_trans:
+	status = ocfs2_local_write_dquot(dquot);
+out_dlock:
+	mutex_unlock(&sb_dqopt(sb)->dqio_mutex);
 	ocfs2_commit_trans(osb, handle);
 out_ilock:
 	ocfs2_unlock_global_qf(oinfo, 1);
 out:
-	mlog_exit(status);
+	if (status)
+		mlog_errno(status);
 	return status;
 }
 
@@ -792,8 +926,6 @@ static int ocfs2_write_info(struct super_block *sb, int type)
 	int status = 0;
 	struct ocfs2_mem_dqinfo *oinfo = sb_dqinfo(sb, type)->dqi_priv;
 
-	mlog_entry_void();
-
 	status = ocfs2_lock_global_qf(oinfo, 1);
 	if (status < 0)
 		goto out;
@@ -808,7 +940,8 @@ static int ocfs2_write_info(struct super_block *sb, int type)
 out_ilock:
 	ocfs2_unlock_global_qf(oinfo, 1);
 out:
-	mlog_exit(status);
+	if (status)
+		mlog_errno(status);
 	return status;
 }
 
@@ -827,15 +960,8 @@ static void ocfs2_destroy_dquot(struct dquot *dquot)
 	kmem_cache_free(ocfs2_dquot_cachep, dquot);
 }
 
-struct dquot_operations ocfs2_quota_operations = {
-	.initialize	= dquot_initialize,
-	.drop		= dquot_drop,
-	.alloc_space	= dquot_alloc_space,
-	.alloc_inode	= dquot_alloc_inode,
-	.free_space	= dquot_free_space,
-	.free_inode	= dquot_free_inode,
-	.transfer	= dquot_transfer,
-	.write_dquot	= ocfs2_write_dquot,
+const struct dquot_operations ocfs2_quota_operations = {
+	/* We never make dquot dirty so .write_dquot is never called */
 	.acquire_dquot	= ocfs2_acquire_dquot,
 	.release_dquot	= ocfs2_release_dquot,
 	.mark_dirty	= ocfs2_mark_dquot_dirty,
@@ -843,20 +969,3 @@ struct dquot_operations ocfs2_quota_operations = {
 	.alloc_dquot	= ocfs2_alloc_dquot,
 	.destroy_dquot	= ocfs2_destroy_dquot,
 };
-
-int ocfs2_quota_setup(void)
-{
-	ocfs2_quota_wq = create_workqueue("o2quot");
-	if (!ocfs2_quota_wq)
-		return -ENOMEM;
-	return 0;
-}
-
-void ocfs2_quota_shutdown(void)
-{
-	if (ocfs2_quota_wq) {
-		flush_workqueue(ocfs2_quota_wq);
-		destroy_workqueue(ocfs2_quota_wq);
-		ocfs2_quota_wq = NULL;
-	}
-}