Merge tag 'gfs2-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw

Pull gfs2 updates from Steven Whitehouse:
 "The main feature of interest this time is quota updates.  There are
  some clean ups and some patches to use the new generic lru list code.

  There is still plenty of scope for some further changes in due course -
  faster lookups of quota structures is very much on the todo list.
  Also, a start has been made towards the more tricky issue of using the
  generic lru code with glocks, but that will have to be completed in a
  subsequent merge window.

  The other, more minor feature, is that there have been a number of
  performance patches which relate to block allocation.  In particular
  they will improve performance when the disk is nearly full"

* tag 'gfs2-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw:
  GFS2: Use generic list_lru for quota
  GFS2: Rename quota qd_lru_lock qd_lock
  GFS2: Use reflink for quota data cache
  GFS2: Use lockref for glocks
  GFS2: Protect quota sync generation
  GFS2: Inline qd_trylock into gfs2_quota_unlock
  GFS2: Make two similar quota code fragments into a function
  GFS2: Remove obsolete quota tunable
  GFS2: Move gfs2_icbit_munge into quota.c
  GFS2: Speed up starting point selection for block allocation
  GFS2: Add allocation parameters structure
  GFS2: Clean up reservation removal
  GFS2: fix dentry leaks
  GFS2: new function gfs2_rbm_incr
  GFS2: Introduce rbm field bii
  GFS2: Do not reset flags on active reservations
  GFS2: introduce bi_blocks for optimization
  GFS2: optimize rbm_from_block wrt bi_start
  GFS2: d_splice_alias() can't return error

19 files changed, 448 insertions, 339 deletions

diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c
index 1f7d8057ea6..b7fc035a694 100644
--- a/fs/gfs2/aops.c
+++ b/fs/gfs2/aops.c
@@ -611,12 +611,14 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 		gfs2_write_calc_reserv(ip, len, &data_blocks, &ind_blocks);
 
 	if (alloc_required) {
+		struct gfs2_alloc_parms ap = { .aflags = 0, };
 		error = gfs2_quota_lock_check(ip);
 		if (error)
 			goto out_unlock;
 
 		requested = data_blocks + ind_blocks;
-		error = gfs2_inplace_reserve(ip, requested, 0);
+		ap.target = requested;
+		error = gfs2_inplace_reserve(ip, &ap);
 		if (error)
 			goto out_qunlock;
 	}
diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
index 62a65fc448d..fe0500c0af7 100644
--- a/fs/gfs2/bmap.c
+++ b/fs/gfs2/bmap.c
@@ -1216,6 +1216,7 @@ static int do_grow(struct inode *inode, u64 size)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	struct gfs2_alloc_parms ap = { .target = 1, };
 	struct buffer_head *dibh;
 	int error;
 	int unstuff = 0;
@@ -1226,7 +1227,7 @@ static int do_grow(struct inode *inode, u64 size)
 		if (error)
 			return error;
 
-		error = gfs2_inplace_reserve(ip, 1, 0);
+		error = gfs2_inplace_reserve(ip, &ap);
 		if (error)
 			goto do_grow_qunlock;
 		unstuff = 1;
@@ -1279,6 +1280,7 @@ do_grow_qunlock:
 
 int gfs2_setattr_size(struct inode *inode, u64 newsize)
 {
+	struct gfs2_inode *ip = GFS2_I(inode);
 	int ret;
 	u64 oldsize;
 
@@ -1294,7 +1296,7 @@ int gfs2_setattr_size(struct inode *inode, u64 newsize)
 
 	inode_dio_wait(inode);
 
-	ret = gfs2_rs_alloc(GFS2_I(inode));
+	ret = gfs2_rs_alloc(ip);
 	if (ret)
 		goto out;
 
@@ -1304,6 +1306,7 @@ int gfs2_setattr_size(struct inode *inode, u64 newsize)
 		goto out;
 	}
 
+	gfs2_rs_deltree(ip->i_res);
 	ret = do_shrink(inode, oldsize, newsize);
 out:
 	put_write_access(inode);
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 0621b46d474..efc078f0ee4 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -383,6 +383,7 @@ static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	struct inode *inode = file_inode(vma->vm_file);
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	struct gfs2_alloc_parms ap = { .aflags = 0, };
 	unsigned long last_index;
 	u64 pos = page->index << PAGE_CACHE_SHIFT;
 	unsigned int data_blocks, ind_blocks, rblocks;
@@ -430,7 +431,8 @@ static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	if (ret)
 		goto out_unlock;
 	gfs2_write_calc_reserv(ip, PAGE_CACHE_SIZE, &data_blocks, &ind_blocks);
-	ret = gfs2_inplace_reserve(ip, data_blocks + ind_blocks, 0);
+	ap.target = data_blocks + ind_blocks;
+	ret = gfs2_inplace_reserve(ip, &ap);
 	if (ret)
 		goto out_quota_unlock;
 
@@ -620,7 +622,7 @@ static int gfs2_release(struct inode *inode, struct file *file)
 	if (!(file->f_mode & FMODE_WRITE))
 		return 0;
 
-	gfs2_rs_delete(ip);
+	gfs2_rs_delete(ip, &inode->i_writecount);
 	return 0;
 }
 
@@ -800,6 +802,7 @@ static long gfs2_fallocate(struct file *file, int mode, loff_t offset,
 	struct inode *inode = file_inode(file);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
 	struct gfs2_inode *ip = GFS2_I(inode);
+	struct gfs2_alloc_parms ap = { .aflags = 0, };
 	unsigned int data_blocks = 0, ind_blocks = 0, rblocks;
 	loff_t bytes, max_bytes;
 	int error;
@@ -850,7 +853,8 @@ static long gfs2_fallocate(struct file *file, int mode, loff_t offset,
 retry:
 		gfs2_write_calc_reserv(ip, bytes, &data_blocks, &ind_blocks);
 
-		error = gfs2_inplace_reserve(ip, data_blocks + ind_blocks, 0);
+		ap.target = data_blocks + ind_blocks;
+		error = gfs2_inplace_reserve(ip, &ap);
 		if (error) {
 			if (error == -ENOSPC && bytes > sdp->sd_sb.sb_bsize) {
 				bytes >>= 1;
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index c2f41b4d00b..e66a8009aff 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -31,6 +31,7 @@
 #include <linux/bit_spinlock.h>
 #include <linux/percpu.h>
 #include <linux/list_sort.h>
+#include <linux/lockref.h>
 
 #include "gfs2.h"
 #include "incore.h"
@@ -129,10 +130,10 @@ void gfs2_glock_free(struct gfs2_glock *gl)
  *
  */
 
-void gfs2_glock_hold(struct gfs2_glock *gl)
+static void gfs2_glock_hold(struct gfs2_glock *gl)
 {
-	GLOCK_BUG_ON(gl, atomic_read(&gl->gl_ref) == 0);
-	atomic_inc(&gl->gl_ref);
+	GLOCK_BUG_ON(gl, __lockref_is_dead(&gl->gl_lockref));
+	lockref_get(&gl->gl_lockref);
 }
 
 /**
@@ -187,20 +188,6 @@ static void gfs2_glock_remove_from_lru(struct gfs2_glock *gl)
 }
 
 /**
- * gfs2_glock_put_nolock() - Decrement reference count on glock
- * @gl: The glock to put
- *
- * This function should only be used if the caller has its own reference
- * to the glock, in addition to the one it is dropping.
- */
-
-void gfs2_glock_put_nolock(struct gfs2_glock *gl)
-{
-	if (atomic_dec_and_test(&gl->gl_ref))
-		GLOCK_BUG_ON(gl, 1);
-}
-
-/**
  * gfs2_glock_put() - Decrement reference count on glock
  * @gl: The glock to put
  *
@@ -211,17 +198,22 @@ void gfs2_glock_put(struct gfs2_glock *gl)
 	struct gfs2_sbd *sdp = gl->gl_sbd;
 	struct address_space *mapping = gfs2_glock2aspace(gl);
 
-	if (atomic_dec_and_lock(&gl->gl_ref, &lru_lock)) {
-		__gfs2_glock_remove_from_lru(gl);
-		spin_unlock(&lru_lock);
-		spin_lock_bucket(gl->gl_hash);
-		hlist_bl_del_rcu(&gl->gl_list);
-		spin_unlock_bucket(gl->gl_hash);
-		GLOCK_BUG_ON(gl, !list_empty(&gl->gl_holders));
-		GLOCK_BUG_ON(gl, mapping && mapping->nrpages);
-		trace_gfs2_glock_put(gl);
-		sdp->sd_lockstruct.ls_ops->lm_put_lock(gl);
-	}
+	if (lockref_put_or_lock(&gl->gl_lockref))
+		return;
+
+	lockref_mark_dead(&gl->gl_lockref);
+
+	spin_lock(&lru_lock);
+	__gfs2_glock_remove_from_lru(gl);
+	spin_unlock(&lru_lock);
+	spin_unlock(&gl->gl_lockref.lock);
+	spin_lock_bucket(gl->gl_hash);
+	hlist_bl_del_rcu(&gl->gl_list);
+	spin_unlock_bucket(gl->gl_hash);
+	GLOCK_BUG_ON(gl, !list_empty(&gl->gl_holders));
+	GLOCK_BUG_ON(gl, mapping && mapping->nrpages);
+	trace_gfs2_glock_put(gl);
+	sdp->sd_lockstruct.ls_ops->lm_put_lock(gl);
 }
 
 /**
@@ -244,7 +236,7 @@ static struct gfs2_glock *search_bucket(unsigned int hash,
 			continue;
 		if (gl->gl_sbd != sdp)
 			continue;
-		if (atomic_inc_not_zero(&gl->gl_ref))
+		if (lockref_get_not_dead(&gl->gl_lockref))
 			return gl;
 	}
 
@@ -396,10 +388,11 @@ static void state_change(struct gfs2_glock *gl, unsigned int new_state)
 	held2 = (new_state != LM_ST_UNLOCKED);
 
 	if (held1 != held2) {
+		GLOCK_BUG_ON(gl, __lockref_is_dead(&gl->gl_lockref));
 		if (held2)
-			gfs2_glock_hold(gl);
+			gl->gl_lockref.count++;
 		else
-			gfs2_glock_put_nolock(gl);
+			gl->gl_lockref.count--;
 	}
 	if (held1 && held2 && list_empty(&gl->gl_holders))
 		clear_bit(GLF_QUEUED, &gl->gl_flags);
@@ -626,9 +619,9 @@ out:
 out_sched:
 	clear_bit(GLF_LOCK, &gl->gl_flags);
 	smp_mb__after_clear_bit();
-	gfs2_glock_hold(gl);
+	gl->gl_lockref.count++;
 	if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
-		gfs2_glock_put_nolock(gl);
+		gl->gl_lockref.count--;
 	return;
 
 out_unlock:
@@ -754,7 +747,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
 	gl->gl_sbd = sdp;
 	gl->gl_flags = 0;
 	gl->gl_name = name;
-	atomic_set(&gl->gl_ref, 1);
+	gl->gl_lockref.count = 1;
 	gl->gl_state = LM_ST_UNLOCKED;
 	gl->gl_target = LM_ST_UNLOCKED;
 	gl->gl_demote_state = LM_ST_EXCLUSIVE;
@@ -1356,10 +1349,10 @@ void gfs2_glock_complete(struct gfs2_glock *gl, int ret)
 		}
 	}
 
-	spin_unlock(&gl->gl_spin);
+	gl->gl_lockref.count++;
 	set_bit(GLF_REPLY_PENDING, &gl->gl_flags);
-	smp_wmb();
-	gfs2_glock_hold(gl);
+	spin_unlock(&gl->gl_spin);
+
 	if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
 		gfs2_glock_put(gl);
 }
@@ -1404,15 +1397,19 @@ __acquires(&lru_lock)
 	while(!list_empty(list)) {
 		gl = list_entry(list->next, struct gfs2_glock, gl_lru);
 		list_del_init(&gl->gl_lru);
+		if (!spin_trylock(&gl->gl_spin)) {
+			list_add(&gl->gl_lru, &lru_list);
+			atomic_inc(&lru_count);
+			continue;
+		}
 		clear_bit(GLF_LRU, &gl->gl_flags);
-		gfs2_glock_hold(gl);
 		spin_unlock(&lru_lock);
-		spin_lock(&gl->gl_spin);
+		gl->gl_lockref.count++;
 		if (demote_ok(gl))
 			handle_callback(gl, LM_ST_UNLOCKED, 0, false);
 		WARN_ON(!test_and_clear_bit(GLF_LOCK, &gl->gl_flags));
 		if (queue_delayed_work(glock_workqueue, &gl->gl_work, 0) == 0)
-			gfs2_glock_put_nolock(gl);
+			gl->gl_lockref.count--;
 		spin_unlock(&gl->gl_spin);
 		spin_lock(&lru_lock);
 	}
@@ -1493,7 +1490,7 @@ static void examine_bucket(glock_examiner examiner, const struct gfs2_sbd *sdp,
 
 	rcu_read_lock();
 	hlist_bl_for_each_entry_rcu(gl, pos, head, gl_list) {
-		if ((gl->gl_sbd == sdp) && atomic_inc_not_zero(&gl->gl_ref))
+		if ((gl->gl_sbd == sdp) && lockref_get_not_dead(&gl->gl_lockref))
 			examiner(gl);
 	}
 	rcu_read_unlock();
@@ -1746,7 +1743,7 @@ int gfs2_dump_glock(struct seq_file *seq, const struct gfs2_glock *gl)
 		  state2str(gl->gl_demote_state), dtime,
 		  atomic_read(&gl->gl_ail_count),
 		  atomic_read(&gl->gl_revokes),
-		  atomic_read(&gl->gl_ref), gl->gl_hold_time);
+		  (int)gl->gl_lockref.count, gl->gl_hold_time);
 
 	list_for_each_entry(gh, &gl->gl_holders, gh_list) {
 		error = dump_holder(seq, gh);
@@ -1902,7 +1899,7 @@ static int gfs2_glock_iter_next(struct gfs2_glock_iter *gi)
 			gi->nhash = 0;
 		}
 	/* Skip entries for other sb and dead entries */
-	} while (gi->sdp != gi->gl->gl_sbd || atomic_read(&gi->gl->gl_ref) == 0);
+	} while (gi->sdp != gi->gl->gl_sbd || __lockref_is_dead(&gl->gl_lockref));
 
 	return 0;
 }
diff --git a/fs/gfs2/glock.h b/fs/gfs2/glock.h
index 69f66e3d22b..6647d77366b 100644
--- a/fs/gfs2/glock.h
+++ b/fs/gfs2/glock.h
@@ -181,8 +181,6 @@ static inline struct address_space *gfs2_glock2aspace(struct gfs2_glock *gl)
 extern int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
 			  const struct gfs2_glock_operations *glops,
 			  int create, struct gfs2_glock **glp);
-extern void gfs2_glock_hold(struct gfs2_glock *gl);
-extern void gfs2_glock_put_nolock(struct gfs2_glock *gl);
 extern void gfs2_glock_put(struct gfs2_glock *gl);
 extern void gfs2_holder_init(struct gfs2_glock *gl, unsigned int state,
 			     unsigned flags, struct gfs2_holder *gh);
diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c
index e2e0a90396e..db908f69713 100644
--- a/fs/gfs2/glops.c
+++ b/fs/gfs2/glops.c
@@ -525,9 +525,9 @@ static void iopen_go_callback(struct gfs2_glock *gl, bool remote)
 
 	if (gl->gl_demote_state == LM_ST_UNLOCKED &&
 	    gl->gl_state == LM_ST_SHARED && ip) {
-		gfs2_glock_hold(gl);
+		gl->gl_lockref.count++;
 		if (queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0)
-			gfs2_glock_put_nolock(gl);
+			gl->gl_lockref.count--;
 	}
 }
 
diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h
index 26aabd7caba..ba1ea67f4ee 100644
--- a/fs/gfs2/incore.h
+++ b/fs/gfs2/incore.h
@@ -21,6 +21,7 @@
 #include <linux/rbtree.h>
 #include <linux/ktime.h>
 #include <linux/percpu.h>
+#include <linux/lockref.h>
 
 #define DIO_WAIT	0x00000010
 #define DIO_METADATA	0x00000020
@@ -71,6 +72,7 @@ struct gfs2_bitmap {
 	u32 bi_offset;
 	u32 bi_start;
 	u32 bi_len;
+	u32 bi_blocks;
 };
 
 struct gfs2_rgrpd {
@@ -101,19 +103,25 @@ struct gfs2_rgrpd {
 
 struct gfs2_rbm {
 	struct gfs2_rgrpd *rgd;
-	struct gfs2_bitmap *bi;	/* Bitmap must belong to the rgd */
 	u32 offset;		/* The offset is bitmap relative */
+	int bii;		/* Bitmap index */
 };
 
+static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
+{
+	return rbm->rgd->rd_bits + rbm->bii;
+}
+
 static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm)
 {
-	return rbm->rgd->rd_data0 + (rbm->bi->bi_start * GFS2_NBBY) + rbm->offset;
+	return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) +
+		rbm->offset;
 }
 
 static inline bool gfs2_rbm_eq(const struct gfs2_rbm *rbm1,
 			       const struct gfs2_rbm *rbm2)
 {
-	return (rbm1->rgd == rbm2->rgd) && (rbm1->bi == rbm2->bi) && 
+	return (rbm1->rgd == rbm2->rgd) && (rbm1->bii == rbm2->bii) &&
 	       (rbm1->offset == rbm2->offset);
 }
 
@@ -278,6 +286,20 @@ struct gfs2_blkreserv {
 	unsigned int rs_qa_qd_num;
 };
 
+/*
+ * Allocation parameters
+ * @target: The number of blocks we'd ideally like to allocate
+ * @aflags: The flags (e.g. Orlov flag)
+ *
+ * The intent is to gradually expand this structure over time in
+ * order to give more information, e.g. alignment, min extent size
+ * to the allocation code.
+ */
+struct gfs2_alloc_parms {
+	u32 target;
+	u32 aflags;
+};
+
 enum {
 	GLF_LOCK			= 1,
 	GLF_DEMOTE			= 3,
@@ -300,9 +322,9 @@ struct gfs2_glock {
 	struct gfs2_sbd *gl_sbd;
 	unsigned long gl_flags;		/* GLF_... */
 	struct lm_lockname gl_name;
-	atomic_t gl_ref;
 
-	spinlock_t gl_spin;
+	struct lockref gl_lockref;
+#define gl_spin gl_lockref.lock
 
 	/* State fields protected by gl_spin */
 	unsigned int gl_state:2,	/* Current state */
@@ -398,11 +420,10 @@ enum {
 
 struct gfs2_quota_data {
 	struct list_head qd_list;
-	struct list_head qd_reclaim;
-
-	atomic_t qd_count;
-
 	struct kqid qd_id;
+	struct lockref qd_lockref;
+	struct list_head qd_lru;
+
 	unsigned long qd_flags;		/* QDF_... */
 
 	s64 qd_change;
@@ -516,7 +537,6 @@ struct gfs2_tune {
 
 	unsigned int gt_logd_secs;
 
-	unsigned int gt_quota_simul_sync; /* Max quotavals to sync at once */
 	unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */
 	unsigned int gt_quota_scale_num; /* Numerator */
 	unsigned int gt_quota_scale_den; /* Denominator */
@@ -694,6 +714,7 @@ struct gfs2_sbd {
 	struct list_head sd_quota_list;
 	atomic_t sd_quota_count;
 	struct mutex sd_quota_mutex;
+	struct mutex sd_quota_sync_mutex;
 	wait_queue_head_t sd_quota_wait;
 	struct list_head sd_trunc_list;
 	spinlock_t sd_trunc_lock;
diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index ced3257f06e..109ce9325b7 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -379,6 +379,7 @@ static void munge_mode_uid_gid(const struct gfs2_inode *dip,
 static int alloc_dinode(struct gfs2_inode *ip, u32 flags)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
+	struct gfs2_alloc_parms ap = { .target = RES_DINODE, .aflags = flags, };
 	int error;
 	int dblocks = 1;
 
@@ -386,7 +387,7 @@ static int alloc_dinode(struct gfs2_inode *ip, u32 flags)
 	if (error)
 		goto out;
 
-	error = gfs2_inplace_reserve(ip, RES_DINODE, flags);
+	error = gfs2_inplace_reserve(ip, &ap);
 	if (error)
 		goto out_quota;
 
@@ -472,6 +473,7 @@ static int link_dinode(struct gfs2_inode *dip, const struct qstr *name,
 		       struct gfs2_inode *ip, int arq)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
+	struct gfs2_alloc_parms ap = { .target = sdp->sd_max_dirres, };
 	int error;
 
 	if (arq) {
@@ -479,7 +481,7 @@ static int link_dinode(struct gfs2_inode *dip, const struct qstr *name,
 		if (error)
 			goto fail_quota_locks;
 
-		error = gfs2_inplace_reserve(dip, sdp->sd_max_dirres, 0);
+		error = gfs2_inplace_reserve(dip, &ap);
 		if (error)
 			goto fail_quota_locks;
 
@@ -584,17 +586,17 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry,
 	if (!IS_ERR(inode)) {
 		d = d_splice_alias(inode, dentry);
 		error = 0;
-		if (file && !IS_ERR(d)) {
-			if (d == NULL)
-				d = dentry;
-			if (S_ISREG(inode->i_mode))
-				error = finish_open(file, d, gfs2_open_common, opened);
-			else
+		if (file) {
+			if (S_ISREG(inode->i_mode)) {
+				WARN_ON(d != NULL);
+				error = finish_open(file, dentry, gfs2_open_common, opened);
+			} else {
 				error = finish_no_open(file, d);
+			}
+		} else {
+			dput(d);
 		}
 		gfs2_glock_dq_uninit(ghs);
-		if (IS_ERR(d))
-			return PTR_ERR(d);
 		return error;
 	} else if (error != -ENOENT) {
 		goto fail_gunlock;
@@ -713,7 +715,7 @@ fail_gunlock2:
 fail_free_inode:
 	if (ip->i_gl)
 		gfs2_glock_put(ip->i_gl);
-	gfs2_rs_delete(ip);
+	gfs2_rs_delete(ip, NULL);
 	free_inode_nonrcu(inode);
 	inode = NULL;
 fail_gunlock:
@@ -781,8 +783,10 @@ static struct dentry *__gfs2_lookup(struct inode *dir, struct dentry *dentry,
 		error = finish_open(file, dentry, gfs2_open_common, opened);
 
 	gfs2_glock_dq_uninit(&gh);
-	if (error)
+	if (error) {
+		dput(d);
 		return ERR_PTR(error);
+	}
 	return d;
 }
 
@@ -874,11 +878,12 @@ static int gfs2_link(struct dentry *old_dentry, struct inode *dir,
 	error = 0;
 
 	if (alloc_required) {
+		struct gfs2_alloc_parms ap = { .target = sdp->sd_max_dirres, };
 		error = gfs2_quota_lock_check(dip);
 		if (error)
 			goto out_gunlock;
 
-		error = gfs2_inplace_reserve(dip, sdp->sd_max_dirres, 0);
+		error = gfs2_inplace_reserve(dip, &ap);
 		if (error)
 			goto out_gunlock_q;
 
@@ -1163,14 +1168,16 @@ static int gfs2_atomic_open(struct inode *dir, struct dentry *dentry,
 	d = __gfs2_lookup(dir, dentry, file, opened);
 	if (IS_ERR(d))
 		return PTR_ERR(d);
-	if (d == NULL)
-		d = dentry;
-	if (d->d_inode) {
+	if (d != NULL)
+		dentry = d;
+	if (dentry->d_inode) {
 		if (!(*opened & FILE_OPENED))
-			return finish_no_open(file, d);
+			return finish_no_open(file, dentry);
+		dput(d);
 		return 0;
 	}
 
+	BUG_ON(d != NULL);
 	if (!(flags & O_CREAT))
 		return -ENOENT;
 
@@ -1385,11 +1392,12 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 		goto out_gunlock;
 
 	if (alloc_required) {
+		struct gfs2_alloc_parms ap = { .target = sdp->sd_max_dirres, };
 		error = gfs2_quota_lock_check(ndip);
 		if (error)
 			goto out_gunlock;
 
-		error = gfs2_inplace_reserve(ndip, sdp->sd_max_dirres, 0);
+		error = gfs2_inplace_reserve(ndip, &ap);
 		if (error)
 			goto out_gunlock_q;
 
diff --git a/fs/gfs2/main.c b/fs/gfs2/main.c
index 351586e24e3..0650db2541e 100644
--- a/fs/gfs2/main.c
+++ b/fs/gfs2/main.c
@@ -31,12 +31,6 @@
 
 struct workqueue_struct *gfs2_control_wq;
 
-static struct shrinker qd_shrinker = {
-	.count_objects = gfs2_qd_shrink_count,
-	.scan_objects = gfs2_qd_shrink_scan,
-	.seeks = DEFAULT_SEEKS,
-};
-
 static void gfs2_init_inode_once(void *foo)
 {
 	struct gfs2_inode *ip = foo;
@@ -87,6 +81,10 @@ static int __init init_gfs2_fs(void)
 	if (error)
 		return error;
 
+	error = list_lru_init(&gfs2_qd_lru);
+	if (error)
+		goto fail_lru;
+
 	error = gfs2_glock_init();
 	if (error)
 		goto fail;
@@ -139,7 +137,7 @@ static int __init init_gfs2_fs(void)
 	if (!gfs2_rsrv_cachep)
 		goto fail;
 
-	register_shrinker(&qd_shrinker);
+	register_shrinker(&gfs2_qd_shrinker);
 
 	error = register_filesystem(&gfs2_fs_type);
 	if (error)
@@ -179,7 +177,9 @@ fail_wq:
 fail_unregister:
 	unregister_filesystem(&gfs2_fs_type);
 fail:
-	unregister_shrinker(&qd_shrinker);
+	list_lru_destroy(&gfs2_qd_lru);
+fail_lru:
+	unregister_shrinker(&gfs2_qd_shrinker);
 	gfs2_glock_exit();
 
 	if (gfs2_rsrv_cachep)
@@ -214,13 +214,14 @@ fail:
 
 static void __exit exit_gfs2_fs(void)
 {
-	unregister_shrinker(&qd_shrinker);
+	unregister_shrinker(&gfs2_qd_shrinker);
 	gfs2_glock_exit();
 	gfs2_unregister_debugfs();
 	unregister_filesystem(&gfs2_fs_type);
 	unregister_filesystem(&gfs2meta_fs_type);
 	destroy_workqueue(gfs_recovery_wq);
 	destroy_workqueue(gfs2_control_wq);
+	list_lru_destroy(&gfs2_qd_lru);
 
 	rcu_barrier();
 
diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index 19ff5e8c285..82303b47495 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -51,7 +51,6 @@ static void gfs2_tune_init(struct gfs2_tune *gt)
 {
 	spin_lock_init(&gt->gt_spin);
 
-	gt->gt_quota_simul_sync = 64;
 	gt->gt_quota_warn_period = 10;
 	gt->gt_quota_scale_num = 1;
 	gt->gt_quota_scale_den = 1;
@@ -94,6 +93,7 @@ static struct gfs2_sbd *init_sbd(struct super_block *sb)
 
 	INIT_LIST_HEAD(&sdp->sd_quota_list);
 	mutex_init(&sdp->sd_quota_mutex);
+	mutex_init(&sdp->sd_quota_sync_mutex);
 	init_waitqueue_head(&sdp->sd_quota_wait);
 	INIT_LIST_HEAD(&sdp->sd_trunc_list);
 	spin_lock_init(&sdp->sd_trunc_lock);
diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c
index db441359ee8..453b50eadde 100644
--- a/fs/gfs2/quota.c
+++ b/fs/gfs2/quota.c
@@ -50,6 +50,8 @@
 #include <linux/freezer.h>
 #include <linux/quota.h>
 #include <linux/dqblk_xfs.h>
+#include <linux/lockref.h>
+#include <linux/list_lru.h>
 
 #include "gfs2.h"
 #include "incore.h"
@@ -71,29 +73,25 @@ struct gfs2_quota_change_host {
 	struct kqid qc_id;
 };
 
-static LIST_HEAD(qd_lru_list);
-static atomic_t qd_lru_count = ATOMIC_INIT(0);
-static DEFINE_SPINLOCK(qd_lru_lock);
+/* Lock order: qd_lock -> qd->lockref.lock -> lru lock */
+static DEFINE_SPINLOCK(qd_lock);
+struct list_lru gfs2_qd_lru;
 
-unsigned long gfs2_qd_shrink_scan(struct shrinker *shrink,
-				  struct shrink_control *sc)
+static void gfs2_qd_dispose(struct list_head *list)
 {
 	struct gfs2_quota_data *qd;
 	struct gfs2_sbd *sdp;
-	int nr_to_scan = sc->nr_to_scan;
-	long freed = 0;
 
-	if (!(sc->gfp_mask & __GFP_FS))
-		return SHRINK_STOP;
-
-	spin_lock(&qd_lru_lock);
-	while (nr_to_scan && !list_empty(&qd_lru_list)) {
-		qd = list_entry(qd_lru_list.next,
-				struct gfs2_quota_data, qd_reclaim);
+	while (!list_empty(list)) {
+		qd = list_entry(list->next, struct gfs2_quota_data, qd_lru);
 		sdp = qd->qd_gl->gl_sbd;
 
+		list_del(&qd->qd_lru);
+
 		/* Free from the filesystem-specific list */
+		spin_lock(&qd_lock);
 		list_del(&qd->qd_list);
+		spin_unlock(&qd_lock);
 
 		gfs2_assert_warn(sdp, !qd->qd_change);
 		gfs2_assert_warn(sdp, !qd->qd_slot_count);
@@ -103,24 +101,59 @@ unsigned long gfs2_qd_shrink_scan(struct shrinker *shrink,
 		atomic_dec(&sdp->sd_quota_count);
 
 		/* Delete it from the common reclaim list */
-		list_del_init(&qd->qd_reclaim);
-		atomic_dec(&qd_lru_count);
-		spin_unlock(&qd_lru_lock);
 		kmem_cache_free(gfs2_quotad_cachep, qd);
-		spin_lock(&qd_lru_lock);
-		nr_to_scan--;
-		freed++;
 	}
-	spin_unlock(&qd_lru_lock);
+}
+
+
+static enum lru_status gfs2_qd_isolate(struct list_head *item, spinlock_t *lock, void *arg)
+{
+	struct list_head *dispose = arg;
+	struct gfs2_quota_data *qd = list_entry(item, struct gfs2_quota_data, qd_lru);
+
+	if (!spin_trylock(&qd->qd_lockref.lock))
+		return LRU_SKIP;
+
+	if (qd->qd_lockref.count == 0) {
+		lockref_mark_dead(&qd->qd_lockref);
+		list_move(&qd->qd_lru, dispose);
+	}
+
+	spin_unlock(&qd->qd_lockref.lock);
+	return LRU_REMOVED;
+}
+
+static unsigned long gfs2_qd_shrink_scan(struct shrinker *shrink,
+					 struct shrink_control *sc)
+{
+	LIST_HEAD(dispose);
+	unsigned long freed;
+
+	if (!(sc->gfp_mask & __GFP_FS))
+		return SHRINK_STOP;
+
+	freed = list_lru_walk_node(&gfs2_qd_lru, sc->nid, gfs2_qd_isolate,
+				   &dispose, &sc->nr_to_scan);
+
+	gfs2_qd_dispose(&dispose);
+
 	return freed;
 }
 
-unsigned long gfs2_qd_shrink_count(struct shrinker *shrink,
-				   struct shrink_control *sc)
+static unsigned long gfs2_qd_shrink_count(struct shrinker *shrink,
+					  struct shrink_control *sc)
 {
-	return vfs_pressure_ratio(atomic_read(&qd_lru_count));
+	return vfs_pressure_ratio(list_lru_count_node(&gfs2_qd_lru, sc->nid));
 }
 
+struct shrinker gfs2_qd_shrinker = {
+	.count_objects = gfs2_qd_shrink_count,
+	.scan_objects = gfs2_qd_shrink_scan,
+	.seeks = DEFAULT_SEEKS,
+	.flags = SHRINKER_NUMA_AWARE,
+};
+
+
 static u64 qd2index(struct gfs2_quota_data *qd)
 {
 	struct kqid qid = qd->qd_id;
@@ -148,10 +181,11 @@ static int qd_alloc(struct gfs2_sbd *sdp, struct kqid qid,
 	if (!qd)
 		return -ENOMEM;
 
-	atomic_set(&qd->qd_count, 1);
+	qd->qd_lockref.count = 1;
+	spin_lock_init(&qd->qd_lockref.lock);
 	qd->qd_id = qid;
 	qd->qd_slot = -1;
-	INIT_LIST_HEAD(&qd->qd_reclaim);
+	INIT_LIST_HEAD(&qd->qd_lru);
 
 	error = gfs2_glock_get(sdp, qd2index(qd),
 			      &gfs2_quota_glops, CREATE, &qd->qd_gl);
@@ -177,16 +211,11 @@ static int qd_get(struct gfs2_sbd *sdp, struct kqid qid,
 
 	for (;;) {
 		found = 0;
-		spin_lock(&qd_lru_lock);
+		spin_lock(&qd_lock);
 		list_for_each_entry(qd, &sdp->sd_quota_list, qd_list) {
-			if (qid_eq(qd->qd_id, qid)) {
-				if (!atomic_read(&qd->qd_count) &&
-				    !list_empty(&qd->qd_reclaim)) {
-					/* Remove it from reclaim list */
-					list_del_init(&qd->qd_reclaim);
-					atomic_dec(&qd_lru_count);
-				}
-				atomic_inc(&qd->qd_count);
+			if (qid_eq(qd->qd_id, qid) &&
+			    lockref_get_not_dead(&qd->qd_lockref)) {
+				list_lru_del(&gfs2_qd_lru, &qd->qd_lru);
 				found = 1;
 				break;
 			}
@@ -202,7 +231,7 @@ static int qd_get(struct gfs2_sbd *sdp, struct kqid qid,
 			new_qd = NULL;
 		}
 
-		spin_unlock(&qd_lru_lock);
+		spin_unlock(&qd_lock);
 
 		if (qd) {
 			if (new_qd) {
@@ -222,18 +251,19 @@ static int qd_get(struct gfs2_sbd *sdp, struct kqid qid,
 static void qd_hold(struct gfs2_quota_data *qd)
 {
 	struct gfs2_sbd *sdp = qd->qd_gl->gl_sbd;
-	gfs2_assert(sdp, atomic_read(&qd->qd_count));
-	atomic_inc(&qd->qd_count);
+	gfs2_assert(sdp, !__lockref_is_dead(&qd->qd_lockref));
+	lockref_get(&qd->qd_lockref);
 }
 
 static void qd_put(struct gfs2_quota_data *qd)
 {
-	if (atomic_dec_and_lock(&qd->qd_count, &qd_lru_lock)) {
-		/* Add to the reclaim list */
-		list_add_tail(&qd->qd_reclaim, &qd_lru_list);
-		atomic_inc(&qd_lru_count);
-		spin_unlock(&qd_lru_lock);
-	}
+	if (lockref_put_or_lock(&qd->qd_lockref))
+		return;
+
+	qd->qd_lockref.count = 0;
+	list_lru_add(&gfs2_qd_lru, &qd->qd_lru);
+	spin_unlock(&qd->qd_lockref.lock);
+
 }
 
 static int slot_get(struct gfs2_quota_data *qd)
@@ -242,10 +272,10 @@ static int slot_get(struct gfs2_quota_data *qd)
 	unsigned int c, o = 0, b;
 	unsigned char byte = 0;
 
-	spin_lock(&qd_lru_lock);
+	spin_lock(&qd_lock);
 
 	if (qd->qd_slot_count++) {
-		spin_unlock(&qd_lru_lock);
+		spin_unlock(&qd_lock);
 		return 0;
 	}
 
@@ -269,13 +299,13 @@ found:
 
 	sdp->sd_quota_bitmap[c][o] |= 1 << b;
 
-	spin_unlock(&qd_lru_lock);
+	spin_unlock(&qd_lock);
 
 	return 0;
 
 fail:
 	qd->qd_slot_count--;
-	spin_unlock(&qd_lru_lock);
+	spin_unlock(&qd_lock);
 	return -ENOSPC;
 }
 
@@ -283,23 +313,43 @@ static void slot_hold(struct gfs2_quota_data *qd)
 {
 	struct gfs2_sbd *sdp = qd->qd_gl->gl_sbd;
 
-	spin_lock(&qd_lru_lock);
+	spin_lock(&qd_lock);
 	gfs2_assert(sdp, qd->qd_slot_count);
 	qd->qd_slot_count++;
-	spin_unlock(&qd_lru_lock);
+	spin_unlock(&qd_lock);
+}
+
+static void gfs2_icbit_munge(struct gfs2_sbd *sdp, unsigned char **bitmap,
+			     unsigned int bit, int new_value)
+{
+	unsigned int c, o, b = bit;
+	int old_value;
+
+	c = b / (8 * PAGE_SIZE);
+	b %= 8 * PAGE_SIZE;
+	o = b / 8;
+	b %= 8;
+
+	old_value = (bitmap[c][o] & (1 << b));
+	gfs2_assert_withdraw(sdp, !old_value != !new_value);
+
+	if (new_value)
+		bitmap[c][o] |= 1 << b;
+	else
+		bitmap[c][o] &= ~(1 << b);
 }
 
 static void slot_put(struct gfs2_quota_data *qd)
 {
 	struct gfs2_sbd *sdp = qd->qd_gl->gl_sbd;
 
-	spin_lock(&qd_lru_lock);
+	spin_lock(&qd_lock);