Merge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw

Pull GFS2 updates from Steven Whitehouse:
 "The major feature this time is the "rbm" conversion in the resource
  group code.  The new struct gfs2_rbm specifies the location of an
  allocatable block in (resource group, bitmap, offset) form.  There are
  a number of added helper functions, and later patches then rewrite
  some of the resource group code in terms of this new structure.  Not
  only does this give us a nice code clean up, but it also removes some
  of the previous restrictions where extents could not cross bitmap
  boundaries, for example.

  In addition to that, there are a few bug fixes and clean ups, but the
  rbm work is by far the majority of this patch set in terms of number
  of changed lines."

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw: (27 commits)
  GFS2: Write out dirty inode metadata in delayed deletes
  GFS2: fix s_writers.counter imbalance in gfs2_ail_empty_gl
  GFS2: Fix infinite loop in rbm_find
  GFS2: Consolidate free block searching functions
  GFS2: Get rid of I_MUTEX_QUOTA usage
  GFS2: Stop block extents at the end of bitmaps
  GFS2: Fix unclaimed_blocks() wrapping bug and clean up
  GFS2: Improve block reservation tracing
  GFS2: Fall back to ignoring reservations, if there are no other blocks left
  GFS2: Fix ->show_options() for statfs slow
  GFS2: Use rbm for gfs2_setbit()
  GFS2: Use rbm for gfs2_testbit()
  GFS2: Eliminate unnecessary check for state > 3 in bitfit
  GFS2: Eliminate redundant calls to may_grant
  GFS2: Combine functions gfs2_glock_dq_wait and wait_on_demote
  GFS2: Combine functions gfs2_glock_wait and wait_on_holder
  GFS2: inline __gfs2_glock_schedule_for_reclaim
  GFS2: change function gfs2_direct_IO to use a normal gfs2_glock_dq
  GFS2: rbm code cleanup
  GFS2: Fix case where reservation finished at end of rgrp
  ...

fs/gfs2/aops.c

@@ -612,6 +612,7 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 	struct gfs2_sbd *sdp = GFS2_SB(mapping->host);
 	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
 	unsigned int data_blocks = 0, ind_blocks = 0, rblocks;
+	unsigned requested = 0;
 	int alloc_required;
 	int error = 0;
 	pgoff_t index = pos >> PAGE_CACHE_SHIFT;
@@ -641,7 +642,8 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 		if (error)
 			goto out_unlock;
 
-		error = gfs2_inplace_reserve(ip, data_blocks + ind_blocks);
+		requested = data_blocks + ind_blocks;
+		error = gfs2_inplace_reserve(ip, requested);
 		if (error)
 			goto out_qunlock;
 	}
@@ -654,7 +656,7 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
 	if (&ip->i_inode == sdp->sd_rindex)
 		rblocks += 2 * RES_STATFS;
 	if (alloc_required)
-		rblocks += gfs2_rg_blocks(ip);
+		rblocks += gfs2_rg_blocks(ip, requested);
 
 	error = gfs2_trans_begin(sdp, rblocks,
 				 PAGE_CACHE_SIZE/sdp->sd_sb.sb_bsize);
@@ -868,8 +870,7 @@ static int gfs2_write_end(struct file *file, struct address_space *mapping,
 	brelse(dibh);
 failed:
 	gfs2_trans_end(sdp);
-	if (gfs2_mb_reserved(ip))
-		gfs2_inplace_release(ip);
+	gfs2_inplace_release(ip);
 	if (ip->i_res->rs_qa_qd_num)
 		gfs2_quota_unlock(ip);
 	if (inode == sdp->sd_rindex) {
@@ -1023,7 +1024,7 @@ static ssize_t gfs2_direct_IO(int rw, struct kiocb *iocb,
 				  offset, nr_segs, gfs2_get_block_direct,
 				  NULL, NULL, 0);
 out:
-	gfs2_glock_dq_m(1, &gh);
+	gfs2_glock_dq(&gh);
 	gfs2_holder_uninit(&gh);
 	return rv;
 }

fs/gfs2/bmap.c

@@ -786,7 +786,7 @@ static int do_strip(struct gfs2_inode *ip, struct buffer_head *dibh,
 		goto out_rlist;
 
 	if (gfs2_rs_active(ip->i_res)) /* needs to be done with the rgrp glock held */
-		gfs2_rs_deltree(ip->i_res);
+		gfs2_rs_deltree(ip, ip->i_res);
 
 	error = gfs2_trans_begin(sdp, rg_blocks + RES_DINODE +
 				 RES_INDIRECT + RES_STATFS + RES_QUOTA,

fs/gfs2/file.c

@@ -441,7 +441,7 @@ static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 		rblocks += data_blocks ? data_blocks : 1;
 	if (ind_blocks || data_blocks) {
 		rblocks += RES_STATFS + RES_QUOTA;
-		rblocks += gfs2_rg_blocks(ip);
+		rblocks += gfs2_rg_blocks(ip, data_blocks + ind_blocks);
 	}
 	ret = gfs2_trans_begin(sdp, rblocks, 0);
 	if (ret)
@@ -845,7 +845,7 @@ static long gfs2_fallocate(struct file *file, int mode, loff_t offset,
 				&max_bytes, &data_blocks, &ind_blocks);
 
 		rblocks = RES_DINODE + ind_blocks + RES_STATFS + RES_QUOTA +
-			  RES_RG_HDR + gfs2_rg_blocks(ip);
+			  RES_RG_HDR + gfs2_rg_blocks(ip, data_blocks + ind_blocks);
 		if (gfs2_is_jdata(ip))
 			rblocks += data_blocks ? data_blocks : 1;
 

fs/gfs2/glock.c

@@ -185,20 +185,6 @@ static void gfs2_glock_remove_from_lru(struct gfs2_glock *gl)
 	spin_unlock(&lru_lock);
 }
 
-/**
- * __gfs2_glock_schedule_for_reclaim - Add a glock to the reclaim list
- * @gl: the glock
- *
- * If the glock is demotable, then we add it (or move it) to the end
- * of the glock LRU list.
- */
-
-static void __gfs2_glock_schedule_for_reclaim(struct gfs2_glock *gl)
-{
-	if (demote_ok(gl))
-		gfs2_glock_add_to_lru(gl);
-}
-
 /**
  * gfs2_glock_put_nolock() - Decrement reference count on glock
  * @gl: The glock to put
@@ -883,7 +869,14 @@ static int gfs2_glock_demote_wait(void *word)
 	return 0;
 }
 
-static void wait_on_holder(struct gfs2_holder *gh)
+/**
+ * gfs2_glock_wait - wait on a glock acquisition
+ * @gh: the glock holder
+ *
+ * Returns: 0 on success
+ */
+
+int gfs2_glock_wait(struct gfs2_holder *gh)
 {
 	unsigned long time1 = jiffies;
 
@@ -894,12 +887,7 @@ static void wait_on_holder(struct gfs2_holder *gh)
 		gh->gh_gl->gl_hold_time = min(gh->gh_gl->gl_hold_time +
 					      GL_GLOCK_HOLD_INCR,
 					      GL_GLOCK_MAX_HOLD);
-}
-
-static void wait_on_demote(struct gfs2_glock *gl)
-{
-	might_sleep();
-	wait_on_bit(&gl->gl_flags, GLF_DEMOTE, gfs2_glock_demote_wait, TASK_UNINTERRUPTIBLE);
+	return gh->gh_error;
 }
 
 /**
@@ -929,19 +917,6 @@ static void handle_callback(struct gfs2_glock *gl, unsigned int state,
 	trace_gfs2_demote_rq(gl);
 }
 
-/**
- * gfs2_glock_wait - wait on a glock acquisition
- * @gh: the glock holder
- *
- * Returns: 0 on success
- */
-
-int gfs2_glock_wait(struct gfs2_holder *gh)
-{
-	wait_on_holder(gh);
-	return gh->gh_error;
-}
-
 void gfs2_print_dbg(struct seq_file *seq, const char *fmt, ...)
 {
 	struct va_format vaf;
@@ -979,7 +954,7 @@ __acquires(&gl->gl_spin)
 	struct gfs2_sbd *sdp = gl->gl_sbd;
 	struct list_head *insert_pt = NULL;
 	struct gfs2_holder *gh2;
-	int try_lock = 0;
+	int try_futile = 0;
 
 	BUG_ON(gh->gh_owner_pid == NULL);
 	if (test_and_set_bit(HIF_WAIT, &gh->gh_iflags))
@@ -987,7 +962,7 @@ __acquires(&gl->gl_spin)
 
 	if (gh->gh_flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB)) {
 		if (test_bit(GLF_LOCK, &gl->gl_flags))
-			try_lock = 1;
+			try_futile = !may_grant(gl, gh);
 		if (test_bit(GLF_INVALIDATE_IN_PROGRESS, &gl->gl_flags))
 			goto fail;
 	}
@@ -996,9 +971,8 @@ __acquires(&gl->gl_spin)
 		if (unlikely(gh2->gh_owner_pid == gh->gh_owner_pid &&
 		    (gh->gh_gl->gl_ops->go_type != LM_TYPE_FLOCK)))
 			goto trap_recursive;
-		if (try_lock &&
-		    !(gh2->gh_flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB)) &&
-		    !may_grant(gl, gh)) {
+		if (try_futile &&
+		    !(gh2->gh_flags & (LM_FLAG_TRY | LM_FLAG_TRY_1CB))) {
 fail:
 			gh->gh_error = GLR_TRYFAILED;
 			gfs2_holder_wake(gh);
@@ -1121,8 +1095,9 @@ void gfs2_glock_dq(struct gfs2_holder *gh)
 		    !test_bit(GLF_DEMOTE, &gl->gl_flags))
 			fast_path = 1;
 	}
-	if (!test_bit(GLF_LFLUSH, &gl->gl_flags))
-		__gfs2_glock_schedule_for_reclaim(gl);
+	if (!test_bit(GLF_LFLUSH, &gl->gl_flags) && demote_ok(gl))
+		gfs2_glock_add_to_lru(gl);
+
 	trace_gfs2_glock_queue(gh, 0);
 	spin_unlock(&gl->gl_spin);
 	if (likely(fast_path))
@@ -1141,7 +1116,8 @@ void gfs2_glock_dq_wait(struct gfs2_holder *gh)
 {
 	struct gfs2_glock *gl = gh->gh_gl;
 	gfs2_glock_dq(gh);
-	wait_on_demote(gl);
+	might_sleep();
+	wait_on_bit(&gl->gl_flags, GLF_DEMOTE, gfs2_glock_demote_wait, TASK_UNINTERRUPTIBLE);
 }
 
 /**

fs/gfs2/glops.c

@@ -94,6 +94,7 @@ static void gfs2_ail_empty_gl(struct gfs2_glock *gl)
 	/* A shortened, inline version of gfs2_trans_begin() */
 	tr.tr_reserved = 1 + gfs2_struct2blk(sdp, tr.tr_revokes, sizeof(u64));
 	tr.tr_ip = (unsigned long)__builtin_return_address(0);
+	sb_start_intwrite(sdp->sd_vfs);
 	gfs2_log_reserve(sdp, tr.tr_reserved);
 	BUG_ON(current->journal_info);
 	current->journal_info = &tr;

fs/gfs2/incore.h

@@ -99,9 +99,26 @@ struct gfs2_rgrpd {
 #define GFS2_RDF_MASK		0xf0000000 /* mask for internal flags */
 	spinlock_t rd_rsspin;           /* protects reservation related vars */
 	struct rb_root rd_rstree;       /* multi-block reservation tree */
-	u32 rd_rs_cnt;                  /* count of current reservations */
 };
 
+struct gfs2_rbm {
+	struct gfs2_rgrpd *rgd;
+	struct gfs2_bitmap *bi;	/* Bitmap must belong to the rgd */
+	u32 offset;		/* The offset is bitmap relative */
+};
+
+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;
+}
+
+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) && 
+	       (rbm1->offset == rbm2->offset);
+}
+
 enum gfs2_state_bits {
 	BH_Pinned = BH_PrivateStart,
 	BH_Escaped = BH_PrivateStart + 1,
@@ -250,18 +267,11 @@ struct gfs2_blkreserv {
 	/* components used during write (step 1): */
 	atomic_t rs_sizehint;         /* hint of the write size */
 
-	/* components used during inplace_reserve (step 2): */
-	u32 rs_requested; /* Filled in by caller of gfs2_inplace_reserve() */
-
-	/* components used during get_local_rgrp (step 3): */
-	struct gfs2_rgrpd *rs_rgd;    /* pointer to the gfs2_rgrpd */
 	struct gfs2_holder rs_rgd_gh; /* Filled in by get_local_rgrp */
 	struct rb_node rs_node;       /* link to other block reservations */
-
-	/* components used during block searches and assignments (step 4): */
-	struct gfs2_bitmap *rs_bi;    /* bitmap for the current allocation */
-	u32 rs_biblk;                 /* start block relative to the bi */
+	struct gfs2_rbm rs_rbm;       /* Start of reservation */
 	u32 rs_free;                  /* how many blocks are still free */
+	u64 rs_inum;                  /* Inode number for reservation */
 
 	/* ancillary quota stuff */
 	struct gfs2_quota_data *rs_qa_qd[2 * MAXQUOTAS];

fs/gfs2/inode.c

@@ -712,14 +712,9 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry,
 	if (error)
 		goto fail_gunlock2;
 
-	/* The newly created inode needs a reservation so it can allocate
-	   xattrs. At the same time, we want new blocks allocated to the new
-	   dinode to be as contiguous as possible. Since we allocated the
-	   dinode block under the directory's reservation, we transfer
-	   ownership of that reservation to the new inode. The directory
-	   doesn't need a reservation unless it needs a new allocation. */
-	ip->i_res = dip->i_res;
-	dip->i_res = NULL;
+	error = gfs2_rs_alloc(ip);
+	if (error)
+		goto fail_gunlock2;
 
 	error = gfs2_acl_create(dip, inode);
 	if (error)
@@ -737,10 +732,7 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry,
 		brelse(bh);
 
 	gfs2_trans_end(sdp);
-	/* Check if we reserved space in the rgrp. Function link_dinode may
-	   not, depending on whether alloc is required. */
-	if (gfs2_mb_reserved(dip))
-		gfs2_inplace_release(dip);
+	gfs2_inplace_release(dip);
 	gfs2_quota_unlock(dip);
 	mark_inode_dirty(inode);
 	gfs2_glock_dq_uninit_m(2, ghs);
@@ -897,7 +889,7 @@ static int gfs2_link(struct dentry *old_dentry, struct inode *dir,
 			goto out_gunlock_q;
 
 		error = gfs2_trans_begin(sdp, sdp->sd_max_dirres +
-					 gfs2_rg_blocks(dip) +
+					 gfs2_rg_blocks(dip, sdp->sd_max_dirres) +
 					 2 * RES_DINODE + RES_STATFS +
 					 RES_QUOTA, 0);
 		if (error)
@@ -1378,7 +1370,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
 			goto out_gunlock_q;
 
 		error = gfs2_trans_begin(sdp, sdp->sd_max_dirres +
-					 gfs2_rg_blocks(ndip) +
+					 gfs2_rg_blocks(ndip, sdp->sd_max_dirres) +
 					 4 * RES_DINODE + 4 * RES_LEAF +
 					 RES_STATFS + RES_QUOTA + 4, 0);
 		if (error)

fs/gfs2/ops_fstype.c

@@ -19,6 +19,7 @@
 #include <linux/mount.h>
 #include <linux/gfs2_ondisk.h>
 #include <linux/quotaops.h>
+#include <linux/lockdep.h>
 
 #include "gfs2.h"
 #include "incore.h"
@@ -766,6 +767,7 @@ static int init_journal(struct gfs2_sbd *sdp, int undo)
 	return error;
 }
 
+static struct lock_class_key gfs2_quota_imutex_key;
 
 static int init_inodes(struct gfs2_sbd *sdp, int undo)
 {
@@ -803,6 +805,12 @@ static int init_inodes(struct gfs2_sbd *sdp, int undo)
 		fs_err(sdp, "can't get quota file inode: %d\n", error);
 		goto fail_rindex;
 	}
+	/*
+	 * i_mutex on quota files is special. Since this inode is hidden system
+	 * file, we are safe to define locking ourselves.
+	 */
+	lockdep_set_class(&sdp->sd_quota_inode->i_mutex,
+			  &gfs2_quota_imutex_key);
 
 	error = gfs2_rindex_update(sdp);
 	if (error)

fs/gfs2/quota.c

@@ -765,6 +765,7 @@ static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
 	struct gfs2_holder *ghs, i_gh;
 	unsigned int qx, x;
 	struct gfs2_quota_data *qd;
+	unsigned reserved;
 	loff_t offset;
 	unsigned int nalloc = 0, blocks;
 	int error;
@@ -781,7 +782,7 @@ static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
 		return -ENOMEM;
 
 	sort(qda, num_qd, sizeof(struct gfs2_quota_data *), sort_qd, NULL);
-	mutex_lock_nested(&ip->i_inode.i_mutex, I_MUTEX_QUOTA);
+	mutex_lock(&ip->i_inode.i_mutex);
 	for (qx = 0; qx < num_qd; qx++) {
 		error = gfs2_glock_nq_init(qda[qx]->qd_gl, LM_ST_EXCLUSIVE,
 					   GL_NOCACHE, &ghs[qx]);
@@ -811,13 +812,13 @@ static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
 	 * two blocks need to be updated instead of 1 */
 	blocks = num_qd * data_blocks + RES_DINODE + num_qd + 3;
 
-	error = gfs2_inplace_reserve(ip, 1 +
-				     (nalloc * (data_blocks + ind_blocks)));
+	reserved = 1 + (nalloc * (data_blocks + ind_blocks));
+	error = gfs2_inplace_reserve(ip, reserved);
 	if (error)
 		goto out_alloc;
 
 	if (nalloc)
-		blocks += gfs2_rg_blocks(ip) + nalloc * ind_blocks + RES_STATFS;
+		blocks += gfs2_rg_blocks(ip, reserved) + nalloc * ind_blocks + RES_STATFS;
 
 	error = gfs2_trans_begin(sdp, blocks, 0);
 	if (error)
@@ -1598,7 +1599,7 @@ static int gfs2_set_dqblk(struct super_block *sb, int type, qid_t id,
 		error = gfs2_inplace_reserve(ip, blocks);
 		if (error)
 			goto out_i;
-		blocks += gfs2_rg_blocks(ip);
+		blocks += gfs2_rg_blocks(ip, blocks);
 	}
 
 	/* Some quotas span block boundaries and can update two blocks,

fs/gfs2/rgrp.c

@@ -35,9 +35,6 @@
 #define BFITNOENT ((u32)~0)
 #define NO_BLOCK ((u64)~0)
 
-#define RSRV_CONTENTION_FACTOR 4
-#define RGRP_RSRV_MAX_CONTENDERS 2
-
 #if BITS_PER_LONG == 32
 #define LBITMASK   (0x55555555UL)
 #define LBITSKIP55 (0x55555555UL)
@@ -67,53 +64,48 @@ static const char valid_change[16] = {
 	        1, 0, 0, 0
 };
 
-static u32 rgblk_search(struct gfs2_rgrpd *rgd, u32 goal,
-			unsigned char old_state,
-			struct gfs2_bitmap **rbi);
+static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 minext,
+                         const struct gfs2_inode *ip, bool nowrap);
+
 
 /**
  * gfs2_setbit - Set a bit in the bitmaps
- * @rgd: the resource group descriptor
- * @buf2: the clone buffer that holds the bitmaps
- * @bi: the bitmap structure
- * @block: the block to set
+ * @rbm: The position of the bit to set
+ * @do_clone: Also set the clone bitmap, if it exists
  * @new_state: the new state of the block
  *
  */
 
-static inline void gfs2_setbit(struct gfs2_rgrpd *rgd, unsigned char *buf2,
-			       struct gfs2_bitmap *bi, u32 block,
+static inline void gfs2_setbit(const struct gfs2_rbm *rbm, bool do_clone,
 			       unsigned char new_state)
 {
 	unsigned char *byte1, *byte2, *end, cur_state;
-	unsigned int buflen = bi->bi_len;
-	const unsigned int bit = (block % GFS2_NBBY) * GFS2_BIT_SIZE;
+	unsigned int buflen = rbm->bi->bi_len;
+	const unsigned int bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE;
 
-	byte1 = bi->bi_bh->b_data + bi->bi_offset + (block / GFS2_NBBY);
-	end = bi->bi_bh->b_data + bi->bi_offset + buflen;
+	byte1 = rbm->bi->bi_bh->b_data + rbm->bi->bi_offset + (rbm->offset / GFS2_NBBY);
+	end = rbm->bi->bi_bh->b_data + rbm->bi->bi_offset + buflen;
 
 	BUG_ON(byte1 >= end);
 
 	cur_state = (*byte1 >> bit) & GFS2_BIT_MASK;
 
 	if (unlikely(!valid_change[new_state * 4 + cur_state])) {
-		printk(KERN_WARNING "GFS2: buf_blk = 0x%llx old_state=%d, "
-		       "new_state=%d\n",
-		       (unsigned long long)block, cur_state, new_state);
-		printk(KERN_WARNING "GFS2: rgrp=0x%llx bi_start=0x%lx\n",
-		       (unsigned long long)rgd->rd_addr,
-		       (unsigned long)bi->bi_start);
-		printk(KERN_WARNING "GFS2: bi_offset=0x%lx bi_len=0x%lx\n",
-		       (unsigned long)bi->bi_offset,
-		       (unsigned long)bi->bi_len);
+		printk(KERN_WARNING "GFS2: buf_blk = 0x%x old_state=%d, "
+		       "new_state=%d\n", rbm->offset, cur_state, new_state);
+		printk(KERN_WARNING "GFS2: rgrp=0x%llx bi_start=0x%x\n",
+		       (unsigned long long)rbm->rgd->rd_addr,
+		       rbm->bi->bi_start);
+		printk(KERN_WARNING "GFS2: bi_offset=0x%x bi_len=0x%x\n",
+		       rbm->bi->bi_offset, rbm->bi->bi_len);
 		dump_stack();
-		gfs2_consist_rgrpd(rgd);
+		gfs2_consist_rgrpd(rbm->rgd);
 		return;
 	}
 	*byte1 ^= (cur_state ^ new_state) << bit;
 
-	if (buf2) {
-		byte2 = buf2 + bi->bi_offset + (block / GFS2_NBBY);
+	if (do_clone && rbm->bi->bi_clone) {
+		byte2 = rbm->bi->bi_clone + rbm->bi->bi_offset + (rbm->offset / GFS2_NBBY);
 		cur_state = (*byte2 >> bit) & GFS2_BIT_MASK;
 		*byte2 ^= (cur_state ^ new_state) << bit;
 	}
@@ -121,30 +113,21 @@ static inline void gfs2_setbit(struct gfs2_rgrpd *rgd, unsigned char *buf2,
 
 /**
  * gfs2_testbit - test a bit in the bitmaps
- * @rgd: the resource group descriptor
- * @buffer: the buffer that holds the bitmaps
- * @buflen: the length (in bytes) of the buffer
- * @block: the block to read
+ * @rbm: The bit to test
  *
+ * Returns: The two bit block state of the requested bit
  */
 
-static inline unsigned char gfs2_testbit(struct gfs2_rgrpd *rgd,
-					 const unsigned char *buffer,
-					 unsigned int buflen, u32 block)
+static inline u8 gfs2_testbit(const struct gfs2_rbm *rbm)
 {
-	const unsigned char *byte, *end;
-	unsigned char cur_state;
+	const u8 *buffer = rbm->bi->bi_bh->b_data + rbm->bi->bi_offset;
+	const u8 *byte;
 	unsigned int bit;
 
-	byte = buffer + (block / GFS2_NBBY);
-	bit = (block % GFS2_NBBY) * GFS2_BIT_SIZE;
-	end = buffer + buflen;
-
-	gfs2_assert(rgd->rd_sbd, byte < end);
+	byte = buffer + (rbm->offset / GFS2_NBBY);
+	bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE;
 
-	cur_state = (*byte >> bit) & GFS2_BIT_MASK;
-
-	return cur_state;
+	return (*byte >> bit) & GFS2_BIT_MASK;
 }
 
 /**
@@ -192,7 +175,7 @@ static inline u64 gfs2_bit_search(const __le64 *ptr, u64 mask, u8 state)
  */
 static inline int rs_cmp(u64 blk, u32 len, struct gfs2_blkreserv *rs)
 {
-	u64 startblk = gfs2_rs_startblk(rs);
+	u64 startblk = gfs2_rbm_to_block(&rs->rs_rbm);
 
 	if (blk >= startblk + rs->rs_free)
 		return 1;
@@ -201,36 +184,6 @@ static inline int rs_cmp(u64 blk, u32 len, struct gfs2_blkreserv *rs)
 	return 0;
 }
 
-/**
- * rs_find - Find a rgrp multi-block reservation that contains a given block
- * @rgd: The rgrp
- * @rgblk: The block we're looking for, relative to the rgrp
- */
-static struct gfs2_blkreserv *rs_find(struct gfs2_rgrpd *rgd, u32 rgblk)
-{
-	struct rb_node **newn;
-	int rc;
-	u64 fsblk = rgblk + rgd->rd_data0;
-
-	spin_lock(&rgd->rd_rsspin);
-	newn = &rgd->rd_rstree.rb_node;
-	while (*newn) {
-		struct gfs2_blkreserv *cur =
-			rb_entry(*newn, struct gfs2_blkreserv, rs_node);
-		rc = rs_cmp(fsblk, 1, cur);
-		if (rc < 0)
-			newn = &((*newn)->rb_left);
-		else if (rc > 0)
-			newn = &((*newn)->rb_right);
-		else {
-			spin_unlock(&rgd->rd_rsspin);
-			return cur;
-		}
-	}
-	spin_unlock(&rgd->rd_rsspin);
-	return NULL;
-}
-
 /**
  * gfs2_bitfit - Search an rgrp's bitmap buffer to find a bit-pair representing
  *       a block in a given allocation state.
@@ -262,8 +215,6 @@ static u32 gfs2_bitfit(const u8 *buf, const unsigned int len,
 	u64 mask = 0x5555555555555555ULL;
 	u32 bit;
 
-	BUG_ON(state > 3);
-
 	/* Mask off bits we don't care about at the start of the search */
 	mask <<= spoint;
 	tmp = gfs2_bit_search(ptr, mask, state);
@@ -284,6 +235,131 @@ static u32 gfs2_bitfit(const u8 *buf, const unsigned int len,
 	return (((const unsigned char *)ptr - buf) * GFS2_NBBY) + bit;
 }
 
+/**
+ * gfs2_rbm_from_block - Set the rbm based upon rgd and block number
+ * @rbm: The rbm with rgd already set correctly
+ * @block: The block number (filesystem relative)
+ *
+ * This sets the bi and offset members of an rbm based on a
+ * resource group and a filesystem relative block number. The
+ * resource group must be set in the rbm on entry, the bi and
+ * offset members will be set by this function.
+ *
+ * Returns: 0 on success, or an error code
+ */
+
+static int gfs2_rbm_from_block(struct gfs2_rbm *rbm, u64 block)
+{
+	u64 rblock = block - rbm->rgd->rd_data0;
+	u32 goal = (u32)rblock;
+	int x;
+
+	if (WARN_ON_ONCE(rblock > UINT_MAX))
+		return -EINVAL;
+	if (block >= rbm->rgd->rd_data0 + rbm->rgd->rd_data)
+		return -E2BIG;
+
+	for (x = 0; x < rbm->rgd->rd_length; x++) {
+		rbm->bi = rbm->rgd->rd_bits + x;
+		if (goal < (rbm->bi->bi_start + rbm->bi->bi_len) * GFS2_NBBY) {
+			rbm->offset = goal - (rbm->bi->bi_start * GFS2_NBBY);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * gfs2_unaligned_extlen - Look for free blocks which are not byte aligned
+ * @rbm: Position to search (value/result)
+ * @n_unaligned: Number of unaligned blocks to check
+ * @len: Decremented for each block found (terminate on zero)
+ *
+ * Returns: true if a non-free block is encountered
+ */
+
+static bool gfs2_unaligned_extlen(struct gfs2_rbm *rbm, u32 n_unaligned, u32 *len)
+{
+	u64 block;
+	u32 n;
+	u8 res;
+
+	for (n = 0; n < n_unaligned; n++) {
+		res = gfs2_testbit(rbm);
+		if (res != GFS2_BLKST_FREE)
+			return true;
+		(*len)--;
+		if (*len == 0)
+			return true;
+		block = gfs2_rbm_to_block(rbm);
+		if (gfs2_rbm_from_block(rbm, block + 1))
+			return true;
+	}
+
+	return false;
+}
+
+/**
+ * gfs2_free_extlen - Return extent length of free blocks
+ * @rbm: Starting position
+ * @len: Max length to check
+ *
+ * Starting at the block specified by the rbm, see how many free blocks
+ * there are, not reading more than len blocks ahead. This can be done
+ * using memchr_inv when the blocks are byte aligned, but has to be done
+ * on a block by block basis in case of unaligned blocks. Also this
+ * function can cope with bitmap boundaries (although it must stop on
+ * a resource group boundary)
+ *
+ * Returns: Number of free blocks in the extent
+ */
+
+static u32 gfs2_free_extlen(const struct gfs2_rbm *rrbm, u32 len)
+{
+	struct gfs2_rbm rbm = *rrbm;
+	u32 n_unaligned = rbm.offset & 3;
+	u32 size = len;
+	u32 bytes;
+	u32 chunk_size;
+	u8 *ptr, *start, *end;
+	u64 block;
+
+	if (n_unaligned &&
+	    gfs2_unaligned_extlen(&rbm, 4 - n_unaligned, &len))
+		goto out;
+
+	n_unaligned = len & 3;
+	/* Start is now byte aligned */
+	while (len > 3) {
+		start = rbm.bi->bi_bh->b_data;
+		if (rbm.bi->bi_clone)
+			start = rbm.bi->bi_clone;
+		end = start + rbm.bi->bi_bh->b_size;
+		start += rbm.bi->bi_offset;
+		BUG_ON(rbm.offset & 3);
+		start += (rbm.offset / GFS2_NBBY);
+		bytes = min_t(u32, len / GFS2_NBBY, (end - start));
+		ptr = memchr_inv(start, 0, bytes);
+		chunk_size = ((ptr == NULL) ? bytes : (ptr - start));
+		chunk_size *= GFS2_NBBY;
+		BUG_ON(len < chunk_size);
+		len -= chunk_size;
+		block = gfs2_rbm_to_block(&rbm);
+		gfs2_rbm_from_block(&rbm, block + chunk_size);
+		n_unaligned = 3;
+		if (ptr)
+			break;
+		n_unaligned = len & 3;
+	}
+
+	/* Deal with any bits left over at the end */
+	if (n_unaligned)
+		gfs2_unaligned_extlen(&rbm, n_unaligned, &len);
+out:
+	return size - len;
+}
+
 /**
  * gfs2_bitcount - count the number of bits in a certain state
  * @rgd: the resource group descriptor
@@ -487,6 +563,8 @@ int gfs2_rs_alloc(struct gfs2_inode *ip)
 	if (!res)
 		error = -ENOMEM;
 
+	RB_CLEAR_NODE(&res->rs_node);
+
 	down_write(&ip->i_rw_mutex);
 	if (ip->i_res)
 		kmem_cache_free(gfs2_rsrv_cachep, res);
@@ -496,11 +574,12 @@ int gfs2_rs_alloc(struct gfs2_inode *ip)
 	return error;
 }
 
-static void dump_rs(struct seq_file *seq, struct gfs2_blkreserv *rs)
+static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs)
 {
-	gfs2_print_dbg(seq, "  r: %llu s:%llu b:%u f:%u\n",
-		       rs->rs_rgd->rd_addr, gfs2_rs_startblk(rs), rs->rs_biblk,
-		       rs->rs_free);
+	gfs2_print_dbg(seq, "  B: n:%llu s:%llu b:%u f:%u\n",
+		       (unsigned long long)rs->rs_inum,
+		       (unsigned long long)gfs2_rbm_to_block(&rs->rs_rbm),
+		       rs->rs_rbm.offset, rs->rs_free);
 }
 
 /**
@@ -508,41 +587,26 @@ static void dump_rs(struct seq_file *seq, struct gfs2_blkreserv *rs)
  * @rs: The reservation to remove
  *
  */
-static void __rs_deltree(struct gfs2_blkreserv *rs)
+static void __rs_deltree(struct gfs2_inode *ip, struct gfs2_blkreserv *rs)
 {
 	struct gfs2_rgrpd *rgd;
 
 	if (!gfs2_rs_active(rs))
 		return;
 
-	rgd = rs->rs_rgd;
-	/* We can't do this: The reason is that when the rgrp is invalidated,
-	   it's in the "middle" of acquiring the glock, but the HOLDER bit
-	   isn't set yet:
-	   BUG_ON(!gfs2_glock_is_locked_by_me(rs->rs_rgd->rd_gl));*/
-	trace_gfs2_rs(NULL, rs, TRACE_RS_TREEDEL);
-
-	if (!RB_EMPTY_ROOT(&rgd->rd_rstree))
-		rb_erase(&rs->rs_node, &rgd->rd_rstree);
-	BUG_ON(!rgd->rd_rs_cnt);
-	rgd->rd_rs_cnt--;
+	rgd = rs->rs_rbm.rgd;
+	trace_gfs2_rs(rs, TRACE_RS_TREEDEL);
+	rb_erase(&rs->rs_node, &rgd->rd_rstree);
+	RB_CLEAR_NODE(&rs->rs_node);
 
 	if (rs->rs_free) {
 		/* return reserved blocks to the rgrp and the ip */
-		BUG_ON(rs->rs_rgd->rd_reserved < rs->rs_free);
-		rs->rs_rgd->rd_reserved -= rs->rs_free;
+		BUG_ON(rs->rs_rbm.rgd->rd_reserved < rs->rs_free);
+		rs->rs_rbm.rgd->rd_reserved -= rs->rs_free;
 		rs->rs_free = 0;
-		clear_bit(GBF_FULL, &rs->rs_bi->bi_flags);
+		clear_bit(GBF_FULL, &rs->rs_rbm.bi->bi_flags);
 		smp_mb__after_clear_bit();
 	}
-	/* We can't change any of the step 1 or step 2 components of the rs.
-	   E.g. We can't set rs_rgd to NULL because the rgd glock is held and
-	   dequeued through this pointer.
-	   Can't: atomic_set(&rs->rs_sizehint, 0);
-	   Can't: rs->rs_requested = 0;
-	   Can't: rs->rs_rgd = NULL;*/
-	rs->rs_bi = NULL;
-	rs->rs_biblk = 0;
 }
 
 /**
@@ -550,17 +614,16 @@ static void __rs_deltree(struct gfs2_blkreserv *rs)
  * @rs: The reservation to remove
  *
  */
-void gfs2_rs_deltree(struct gfs2_blkreserv *rs)
+void gfs2_rs_deltree(struct gfs2_inode *ip, struct gfs2_blkreserv *rs)
 {
 	struct gfs2_rgrpd *rgd;
 
-	if (!gfs2_rs_active(rs))
-		return;
-
-	rgd = rs->rs_rgd;
-	spin_lock(&rgd->rd_rsspin);
-	__rs_deltree(rs);
-	spin_unlock(&rgd->rd_rsspin);
+	rgd = rs->rs_rbm.rgd;
+	if (rgd) {
+		spin_lock(&rgd->rd_rsspin);
+		__rs_deltree(ip, rs);
+		spin_unlock(&rgd->rd_rsspin);
+	}
 }
 
 /**
@@ -572,8 +635,7 @@ void gfs2_rs_delete(struct gfs2_inode *ip)
 {
 	down_write(&ip->i_rw_mutex);
 	if (ip->i_res) {
-		gfs2_rs_deltree(ip->i_res);
-		trace_gfs2_rs(ip, ip->i_res, TRACE_RS_DELETE);
+		gfs2_rs_deltree(ip, ip->i_res);
 		BUG_ON(ip->i_res->rs_free);
 		kmem_cache_free(gfs2_rsrv_cachep, ip->i_res);
 		ip->i_res = NULL;
@@ -597,7 +659,7 @@ static void return_all_reservations(struct gfs2_rgrpd *rgd)
 	spin_lock(&rgd->rd_rsspin);
 	while ((n = rb_first(&rgd->rd_rstree))) {
 		rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
-		__rs_deltree(rs);
+		__rs_deltree(NULL, rs);
 	}
 	spin_unlock(&rgd->rd_rsspin);
 }
@@ -1270,211 +1332,276 @@ int gfs2_fitrim(struct file *filp, void __user *argp)
 
 /**
  * rs_insert - insert a new multi-block reservation into the rgrp's rb_tree
- * @bi: the bitmap with the blocks
  * @ip: the inode structure
- * @biblk: the 32-bit block number relative to the start of the bitmap
- * @amount: the number of blocks to reserve
  *
- * Returns: NULL - reservation was already taken, so not inserted
- *          pointer to the inserted reservation
  */
-static struct gfs2_blkreserv *rs_insert(struct gfs2_bitmap *bi,
-				       struct gfs2_inode *ip, u32 biblk,
-				       int amount)
+static void rs_insert(struct gfs2_inode *ip)
 {
 	struct rb_node **newn, *parent = NULL;
 	int rc;
 	struct gfs2_blkreserv *rs = ip->i_res;
-	struct gfs2_rgrpd *rgd = rs->rs_rgd;
-	u64 fsblock = gfs2_bi2rgd_blk(bi, biblk) + rgd->rd_data0;
+	struct gfs2_rgrpd *rgd = rs->rs_rbm.rgd;
+	u64 fsblock = gfs2_rbm_to_block(&rs->rs_rbm);
+
+	BUG_ON(gfs2_rs_active(rs));
 
 	spin_lock(&rgd->rd_rsspin);
 	newn = &rgd->rd_rstree.rb_node;
-	BUG_ON(!ip->i_res);
-	BUG_ON(gfs2_rs_active(rs));
-	/* Figure out where to put new node */
-	/*BUG_ON(!gfs2_glock_is_locked_by_me(rgd->rd_gl));*/
 	while (*newn) {
 		struct gfs2_blkreserv *cur =
 			rb_entry(*newn, struct gfs2_blkreserv, rs_node);
 
 		parent = *newn;
-		rc = rs_cmp(fsblock, amount, cur);
+		rc = rs_cmp(fsblock, rs->rs_free, cur);
 		if (rc > 0)
 			newn = &((*newn)->rb_right);
 		else if (rc < 0)
 			newn = &((*newn)->rb_left);
 		else {
 			spin_unlock(&rgd->rd_rsspin);
-			return NULL; /* reservation already in use */
+			WARN_ON(1);
+			return;
 		}
 	}
 
-	/* Do our reservation work */
-	rs = ip->i_res;
-	rs->rs_free = amount;
-	rs->rs_biblk = biblk;
-	rs->rs_bi = bi;
 	rb_link_node(&rs->rs_node, parent, newn);
 	rb_insert_color(&rs->rs_node, &rgd->rd_rstree);
 
-	/* Do our inode accounting for the reservation */
-	/*BUG_ON(!gfs2_glock_is_locked_by_me(ip->i_gl));*/
-
 	/* Do our rgrp accounting for the reservation */
-	rgd->rd_reserved += amount; /* blocks reserved */
-	rgd->rd_rs_cnt++; /* number of in-tree reservations */
+	rgd->rd_reserved += rs->rs_free; /* blocks reserved */
 	spin_unlock(&rgd->rd_rsspin);
-	trace_gfs2_rs(ip, rs, TRACE_RS_INSERT);
-	return rs;
-}
-
-/**
- * unclaimed_blocks - return number of blocks that aren't spoken for
- */
-static u32 unclaimed_blocks(struct gfs2_rgrpd *rgd)
-{
-	return rgd->rd_free_clone - rgd->rd_reserved;
+	trace_gfs2_rs(rs, TRACE_RS_INSERT);
 }
 
 /**
- * rg_mblk_search - find a group of multiple free blocks
+ * rg_mblk_search - find a group of multiple free blocks to form a reservation
  * @rgd: the resource group descriptor
- * @rs: the block reservation
  * @ip: pointer to the inode for which we're reserving blocks
+ * @requested: number of blocks required for this allocation
  *
- * This is very similar to rgblk_search, except we're looking for whole
- * 64-bit words that represent a chunk of 32 free blocks. I'm only focusing
- * on aligned dwords for speed's sake.
- *
- * Returns: 0 if successful or BFITNOENT if there isn't enough free space
  */
 
-static int rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip)
+static void rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip,
+			   unsigned requested)
 {
-	struct gfs2_bitmap *bi = rgd->rd_bits;
-	const u32 length = rgd->rd_length;
-	u32 blk;
-	unsigned int buf, x, search_bytes;
-	u8 *buffer = NULL;
-	u8 *ptr, *end, *nonzero;
-	u32 goal, rsv_bytes;
-	struct gfs2_blkreserv *rs;
-	u32 best_rs_bytes, unclaimed;
-	int best_rs_blocks;
+	struct gfs2_rbm rbm = { .rgd = rgd, };
+	u64 goal;
+	struct gfs2_blkreserv *rs = ip->i_res;
+	u32 extlen;
+	u32 free_blocks = rgd->rd_free_clone - rgd->rd_reserved;
+	int ret;
+
+	extlen = max_t(u32, atomic_read(&rs->rs_sizehint), requested);
+	extlen = clamp(extlen, RGRP_RSRV_MINBLKS, free_blocks);
+	if ((rgd->rd_free_clone < rgd->rd_reserved) || (free_blocks < extlen))
+		return;
 
 	/* Find bitmap block that contains bits for goal block */
 	if (rgrp_contains_block(rgd, ip->i_goal))
-		goal = ip->i_goal - rgd->rd_data0;
+		goal = ip->i_goal;
 	else
-		goal = rgd->rd_last_alloc;
-	for (buf = 0; buf < length; buf++) {
-		bi = rgd->rd_bits + buf;
-		/* Convert scope of "goal" from rgrp-wide to within
-		   found bit block */
-		if (goal < (bi->bi_start + bi->bi_len) * GFS2_NBBY) {
-			goal -= bi->bi_start * GFS2_NBBY;
-			goto do_search;
-		}
+		goal = rgd->rd_last_alloc + rgd->rd_data0;
+
+	if (WARN_ON(gfs2_rbm_from_block(&rbm, goal)))
+		return;
+
+	ret = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, extlen, ip, true);
+	if (ret == 0) {
+		rs->rs_rbm = rbm;
+		rs->rs_free = extlen;
+		rs->rs_inum = ip->i_no_addr;
+		rs_insert(ip);
 	}
-	buf = 0;
-	goal = 0;
-
-do_search:
-	best_rs_blocks = max_t(int, atomic_read(&ip->i_res->rs_sizehint),
-			       (RGRP_RSRV_MINBLKS * rgd->rd_length));
-	best_rs_bytes = (best_rs_blocks *
-			 (1 + (RSRV_CONTENTION_FACTOR * rgd->rd_rs_cnt))) /
-		GFS2_NBBY; /* 1 + is for our not-yet-created reservation */
-	best_rs_bytes = ALIGN(best_rs_bytes, sizeof(u64));
-	unclaimed = unclaimed_blocks(rgd);
-	if (best_rs_bytes * GFS2_NBBY > unclaimed)
-		best_rs_bytes = unclaimed >> GFS2_BIT_SIZE;
-
-	for (x = 0; x <= length; x++) {
-		bi = rgd->rd_bits + buf;
+}
+
+/**
+ * gfs2_next_unreserved_block - Return next block that is not reserved
+ * @rgd: The resource group
+ * @block: The starting block
+ * @length: The required length
+ * @ip: Ignore any reservations for this inode
+ *
+ * If the block does not appear in any reservation, then return the
+ * block number unchanged. If it does appear in the reservation, then
+ * keep looking through the tree of reservations in order to find the
+ * first block number which is not reserved.
+ */
 
-		if (test_bit(GBF_FULL, &bi->bi_flags))
-			goto skip;
+static u64 gfs2_next_unreserved_block(struct gfs2_rgrpd *rgd, u64 block,
+				      u32 length,
+				      const struct gfs2_inode *ip)
+{
+	struct gfs2_blkreserv *rs;
+	struct rb_node *n;
+	int rc;
 
-		WARN_ON(!buffer_uptodate(bi->bi_bh));
-		if (bi->bi_clone)
-			buffer = bi->bi_clone + bi->bi_offset;
+	spin_lock(&rgd->rd_rsspin);
+	n = rgd->rd_rstree.rb_node;
+	while (n) {
+		rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
+		rc = rs_cmp(block, length, rs);
+		if (rc < 0)
+			n = n->rb_left;
+		else if (rc > 0)
+			n = n->rb_right;
 		else
-			buffer = bi->bi_bh->b_data + bi->bi_offset;
-
-		/* We have to keep the reservations aligned on u64 boundaries
-		   otherwise we could get situations where a byte can't be
-		   used because it's after a reservation, but a free bit still
-		   is within the reservation's area. */
-		ptr = buffer + ALIGN(goal >> GFS2_BIT_SIZE, sizeof(u64));
-		end = (buffer + bi->bi_len);
-		while (ptr < end) {
-			rsv_bytes = 0;
-			if ((ptr + best_rs_bytes) <= end)
-				search_bytes = best_rs_bytes;
-			else
-				search_bytes = end - ptr;
-			BUG_ON(!search_bytes);
-			nonzero = memchr_inv(ptr, 0, search_bytes);
-			/* If the lot is all zeroes, reserve the whole size. If
-			   there's enough zeroes to satisfy the request, use
-			   what we can. If there's not enough, keep looking. */
-			if (nonzero == NULL)
-				rsv_bytes = search_bytes;
-			else if ((nonzero - ptr) * GFS2_NBBY >=
-				 ip->i_res->rs_requested)
-				rsv_bytes = (nonzero - ptr);
-
-			if (rsv_bytes) {
-				blk = ((ptr - buffer) * GFS2_NBBY);
-				BUG_ON(blk >= bi->bi_len * GFS2_NBBY);
-				rs = rs_insert(bi, ip, blk,
-					       rsv_bytes * GFS2_NBBY);
-				if (IS_ERR(rs))
-					return PTR_ERR(rs);
-				if (rs)
-					return 0;
-			}
-			ptr += ALIGN(search_bytes, sizeof(u64));
+			break;
+	}
+
+	if (n) {
+		while ((rs_cmp(block, length, rs) == 0) && (ip->i_res != rs)) {
+			block = gfs2_rbm_to_block(&rs->rs_rbm) + rs->rs_free;
+			n = n->rb_right;
+			if (n == NULL)
+				break;
+			rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
 		}
-skip:
-		/* Try next bitmap block (wrap back to rgrp header
-		   if at end) */
-		buf++;
-		buf %= length;
-		goal = 0;
 	}
 
-	return BFITNOENT;
+	spin_unlock(&rgd->rd_rsspin);
+	return block;
 }
 
 /**
- * try_rgrp_fit - See if a given reservation will fit in a given RG
- * @rgd: the RG data
- * @ip: the inode
+ * gfs2_reservation_check_and_update - Check for reservations during block alloc
+ * @rbm: The current position in the resource group
+ * @ip: The inode for which we are searching for blocks
+ * @minext: The minimum extent length
  *
- * If there's room for the requested blocks to be allocated from the RG:
- * This will try to get a multi-block reservation first, and if that doesn't
- * fit, it will take what it can.
+ * This checks the current position in the rgrp to see whether there is
+ * a reservation covering this block. If not then this function is a
+ * no-op. If there is, then the position is moved to the end of the
+ * contiguous reservation(s) so that we are pointing at the first
+ * non-reserved block.
  *
- * Returns: 1 on success (it fits), 0 on failure (it doesn't fit)
+ * Returns: 0 if no reservation, 1 if @rbm has changed, otherwise an error
  */
 
-static int try_rgrp_fit(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip)
+static int gfs2_reservation_check_and_update(struct gfs2_rbm *rbm,
+					     const struct gfs2_inode *ip,
+					     u32 minext)
 {
-	struct gfs2_blkreserv *rs = ip->i_res;
+	u64 block = gfs2_rbm_to_block(rbm);
+	u32 extlen = 1;
+	u64 nblock;
+	int ret;
 
-	if (rgd->rd_flags & (GFS2_RGF_NOALLOC | GFS2_RDF_ERROR))
+	/*
+	 * If we have a minimum extent length, then skip over any extent
+	 * which is less than the min extent length in size.
+	 */
+	if (minext) {
+		extlen = gfs2_free_extlen(rbm, minext);
+		nblock = block + extlen;
+		if (extlen < minext)
+			goto fail;
+	}
+
+	/*
+	 * Check the extent which has been found against the reservations
+	 * and skip if parts of it are already reserved
+	 */
+	nblock = gfs2_next_unreserved_block(rbm->rgd, block, extlen, ip);
+	if (nblock == block)
 		return 0;
-	/* Look for a multi-block reservation. */
-	if (unclaimed_blocks(rgd) >= RGRP_RSRV_MINBLKS &&
-	    rg_mblk_search(rgd, ip) != BFITNOENT)
-		return 1;
-	if (unclaimed_blocks(rgd) >= rs->rs_requested)
-		return 1;
+fail:
+	ret = gfs2_rbm_from_block(rbm, nblock);
+	if (ret < 0)
+		return ret;
+	return 1;
+}
 
-	return 0;
+/**
+ * gfs2_rbm_find - Look for blocks of a particular state
+ * @rbm: Value/result starting position and final position
+ * @state: The state which we want to find
+ * @minext: The requested extent length (0 for a single block)
+ * @ip: If set, check for reservations
+ * @nowrap: Stop looking at the end of the rgrp, rather than wrapping
+ *          around until we've reached the starting point.
+ *
+ * Side effects:
+ * - If looking for free blocks, we set GBF_FULL on each bitmap which
+ *   has no free blocks in it.
+ *
+ * Returns: 0 on success, -ENOSPC if there is no block of the requested state
+ */
+
+static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 minext,
+			 const struct gfs2_inode *ip, bool nowrap)
+{
+	struct buffer_head *bh;
+	struct gfs2_bitmap *initial_bi;
+	u32 initial_offset;
+	u32 offset;
+	u8 *buffer;
+	int index;
+	int n = 0;
+	int iters = rbm->rgd->rd_length;
+	int ret;
+
+	/* If we are not starting at the beginning of a bitmap, then we
+	 * need to add one to the bitmap count to ensure that we search
+	 * the starting bitmap twice.
+	 */
+	if (rbm->offset != 0)
+		iters++;
+
+	while(1) {
+		if (test_bit(GBF_FULL, &rbm->bi->bi_flags) &&
+		    (state == GFS2_BLKST_FREE))
+			goto next_bitmap;
+
+		bh = rbm->bi->bi_bh;
+		buffer = bh->b_data + rbm->bi->bi_offset;
+		WARN_ON(!buffer_uptodate(bh));
+		if (state != GFS2_BLKST_UNLINKED && rbm->bi->bi_clone)
+			buffer = rbm->bi->bi_clone + rbm->bi->bi_offset;
+		initial_offset = rbm->offset;
+		offset = gfs2_bitfit(buffer, rbm->bi->bi_len, rbm->offset, state);
+		if (offset == BFITNOENT)
+			goto bitmap_full;
+		rbm->offset = offset;
+		if (ip == NULL)
+			return 0;
+
+		initial_bi = rbm->bi;
+		ret = gfs2_reservation_check_and_update(rbm, ip, minext);
+		if (ret == 0)
+			return 0;
+		if (ret > 0) {
+			n += (rbm->bi - initial_bi);
+			goto next_iter;
+		}
+		if (ret == -E2BIG) {
+			index = 0;
+			rbm->offset = 0;
+			n += (rbm->bi - initial_bi);
+			goto res_covered_end_of_rgrp;
+		}
+		return ret;
+
+bitmap_full:	/* Mark bitmap as full and fall through */
+		if ((state == GFS2_BLKST_FREE) && initial_offset == 0)
+			set_bit(GBF_FULL, &rbm->bi->bi_flags);
+
+next_bitmap:	/* Find next bitmap in the rgrp */
+		rbm->offset = 0;
+		index = rbm->bi - rbm->rgd->rd_bits;
+		index++;
+		if (index == rbm->rgd->rd_length)
+			index = 0;
+res_covered_end_of_rgrp:
+		rbm->bi = &rbm->rgd->rd_bits[index];
+		if ((index == 0) && nowrap)
+			break;
+		n++;
+next_iter:
+		if (n >= iters)
+			break;
+	}
+
+	return -ENOSPC;
 }
 
 /**
@@ -1489,34 +1616,33 @@ static int try_rgrp_fit(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip)
 
 static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip)
 {
-	u32 goal = 0, block;
-	u64 no_addr;
+	u64 block;
 	struct gfs2_sbd *sdp = rgd->rd_sbd;
 	struct gfs2_glock *gl;
 	struct gfs2_inode *ip;
 	int error;
 	int found = 0;
-	struct gfs2_bitmap *bi;
+	struct gfs2_rbm rbm = { .rgd = rgd, .bi = rgd->rd_bits, .offset = 0 };
 
-	while (goal < rgd->rd_data) {
+	while (1) {
 		down_write(&sdp->sd_log_flush_lock);
-		block = rgblk_search(rgd, goal, GFS2_BLKST_UNLINKED, &bi);
+		error = gfs2_rbm_find(&rbm, GFS2_BLKST_UNLINKED, 0, NULL, true);
 		up_write(&sdp->sd_log_flush_lock);
-		if (block == BFITNOENT)
+		if (error == -ENOSPC)
+			break;
+		if (WARN_ON_ONCE(error))
 			break;
 
-		block = gfs2_bi2rgd_blk(bi, block);
-		/* rgblk_search can return a block < goal, so we need to
-		   keep it marching forward. */
-		no_addr = block + rgd->rd_data0;
-		goal = max(block + 1, goal + 1);
-		if (*last_unlinked != NO_BLOCK && no_addr <= *last_unlinked)
+		block = gfs2_rbm_to_block(&rbm);
+		if (gfs2_rbm_from_block(&rbm, block + 1))
+			break;
+		if (*last_unlinked != NO_BLOCK && block <= *last_unlinked)
 			continue;
-		if (no_addr == skip)
+		if (block == skip)
 			continue;
-		*last_unlinked = no_addr;
+		*last_unlinked = block;
 
-		error = gfs2_glock_get(sdp, no_addr, &gfs2_inode_glops, CREATE, &gl);
+		error = gfs2_glock_get(sdp, block, &gfs2_inode_glops, CREATE, &gl);
 		if (error)
 			continue;
 
@@ -1543,6 +1669,19 @@ static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip
 	return;
 }
 
+static bool gfs2_select_rgrp(struct gfs2_rgrpd **pos, const struct gfs2_rgrpd *begin)
+{
+	struct gfs2_rgrpd *rgd = *pos;
+
+	rgd = gfs2_rgrpd_get_next(rgd);
+	if (rgd == NULL)
+		rgd = gfs2_rgrpd_get_next(NULL);
+	*pos = rgd;
+	if (rgd != begin) /* If we didn't wrap */
+		return true;
+	return false;
+}
+
 /**
  * gfs2_inplace_reserve - Reserve space in the filesystem
  * @ip: the inode to reserve space for
@@ -1562,103 +1701,96 @@ int gfs2_inplace_reserve(struct gfs2_inode *ip, u32 requested)
 
 	if (sdp->sd_args.ar_rgrplvb)
 		flags |= GL_SKIP;
-	rs->rs_requested = requested;
-	if (gfs2_assert_warn(sdp, requested)) {
-		error = -EINVAL;
-		goto out;
-	}
+	if (gfs2_assert_warn(sdp, requested))
+		return -EINVAL;
 	if (gfs2_rs_active(rs)) {
-		begin = rs->rs_rgd;
+		begin = rs->rs_rbm.rgd;
 		flags = 0; /* Yoda: Do or do not. There is no try */
 	} else if (ip->i_rgd && rgrp_contains_block(ip->i_rgd, ip->i_goal)) {
-		rs->rs_rgd = begin = ip->i_rgd;
+		rs->rs_rbm.rgd = begin = ip->i_rgd;
 	} else {
-		rs->rs_rgd = begin = gfs2_blk2rgrpd(sdp, ip->i_goal, 1);
+		rs->rs_rbm.rgd = begin = gfs2_blk2rgrpd(sdp, ip->i_goal, 1);
 	}
-	if (rs->rs_rgd == NULL)
+	if (rs->rs_rbm.rgd == NULL)
 		return -EBADSLT;
 
 	while (loops < 3) {
-		rg_locked = 0;
-
-		if (gfs2_glock_is_locked_by_me(rs->rs_rgd->rd_gl)) {
-			rg_locked = 1;
-			error = 0;
-		} else if (!loops && !gfs2_rs_active(rs) &&
-			   rs->rs_rgd->rd_rs_cnt > RGRP_RSRV_MAX_CONTENDERS) {
-			/* If the rgrp already is maxed out for contenders,
-			   we can eliminate it as a "first pass" without even
-			   requesting the rgrp glock. */
-			error = GLR_TRYFAILED;
-		} else {
-			error = gfs2_glock_nq_init(rs->rs_rgd->rd_gl,
+		rg_locked = 1;
+
+		if (!gfs2_glock_is_locked_by_me(rs->rs_rbm.rgd->rd_gl)) {
+			rg_locked = 0;
+			error = gfs2_glock_nq_init(rs->rs_rbm.rgd->rd_gl,
 						   LM_ST_EXCLUSIVE, flags,
 						   &rs->rs_rgd_gh);
-			if (!error && sdp->sd_args.ar_rgrplvb) {
-				error = update_rgrp_lvb(rs->rs_rgd);
-				if (error) {
+			if (error == GLR_TRYFAILED)
+				goto next_rgrp;
+			if (unlikely(error))
+				return error;
+			if (sdp->sd_args.ar_rgrplvb) {
+				error = update_rgrp_lvb(rs->rs_rbm.rgd);
+				if (unlikely(error)) {
 					gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
 					return error;
 				}
 			}
 		}
-		switch (error) {
-		case 0:
-			if (gfs2_rs_active(rs)) {
-				if (unclaimed_blocks(rs->rs_rgd) +
-				    rs->rs_free >= rs->rs_requested) {
-					ip->i_rgd = rs->rs_rgd;
-					return 0;
-				}
-				/* We have a multi-block reservation, but the
-				   rgrp doesn't have enough free blocks to
-				   satisfy the request. Free the reservation
-				   and look for a suitable rgrp. */
-				gfs2_rs_deltree(rs);
-			}
-			if (try_rgrp_fit(rs->rs_rgd, ip)) {
-				if (sdp->sd_args.ar_rgrplvb)
-					gfs2_rgrp_bh_get(rs->rs_rgd);
-				ip->i_rgd = rs->rs_rgd;
-				return 0;
-			}
-			if (rs->rs_rgd->rd_flags & GFS2_RDF_CHECK) {
-				if (sdp->sd_args.ar_rgrplvb)
-					gfs2_rgrp_bh_get(rs->rs_rgd);
-				try_rgrp_unlink(rs->rs_rgd, &last_unlinked,
-						ip->i_no_addr);
-			}
-			if (!rg_locked)
-				gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
-			/* fall through */
-		case GLR_TRYFAILED:
-			rs->rs_rgd = gfs2_rgrpd_get_next(rs->rs_rgd);
-			rs->rs_rgd = rs->rs_rgd ? : begin; /* if NULL, wrap */
-			if (rs->rs_rgd != begin) /* If we didn't wrap */
-				break;
 
-			flags &= ~LM_FLAG_TRY;
-			loops++;
-			/* Check that fs hasn't grown if writing to rindex */
-			if (ip == GFS2_I(sdp->sd_rindex) &&
-			    !sdp->sd_rindex_uptodate) {
-				error = gfs2_ri_update(ip);
-				if (error)
-					goto out;
-			} else if (loops == 2)
-				/* Flushing the log may release space */
-				gfs2_log_flush(sdp, NULL);
-			break;
-		default:
-			goto out;
+		/* Skip unuseable resource groups */
+		if (rs->rs_rbm.rgd->rd_flags & (GFS2_RGF_NOALLOC | GFS2_RDF_ERROR))
+			goto skip_rgrp;
+
+		if (sdp->sd_args.ar_rgrplvb)
+			gfs2_rgrp_bh_get(rs->rs_rbm.rgd);
+
+		/* Get a reservation if we don't already have one */
+		if (!gfs2_rs_active(rs))
+			rg_mblk_search(rs->rs_rbm.rgd, ip, requested);
+
+		/* Skip rgrps when we can't get a reservation on first pass */
+		if (!gfs2_rs_active(rs) && (loops < 1))
+			goto check_rgrp;
+
+		/* If rgrp has enough free space, use it */
+		if (rs->rs_rbm.rgd->rd_free_clone >= requested) {
+			ip->i_rgd = rs->rs_rbm.rgd;
+			return 0;
+		}
+
+		/* Drop reservation, if we couldn't use reserved rgrp */
+		if (gfs2_rs_active(rs))
+			gfs2_rs_deltree(ip, rs);
+check_rgrp:
+		/* Check for unlinked inodes which can be reclaimed */
+		if (rs->rs_rbm.rgd->rd_flags & GFS2_RDF_CHECK)
+			try_rgrp_unlink(rs->rs_rbm.rgd, &last_unlinked,
+					ip->i_no_addr);
+skip_rgrp:
+		/* Unlock rgrp if required */
+		if (!rg_locked)
+			gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
+next_rgrp:
+		/* Find the next rgrp, and continue looking */
+		if (gfs2_select_rgrp(&rs->rs_rbm.rgd, begin))
+			continue;
+
+		/* If we've scanned all the rgrps, but found no free blocks
+		 * then this checks for some less likely conditions before
+		 * trying again.
+		 */
+		flags &= ~LM_FLAG_TRY;
+		loops++;
+		/* Check that fs hasn't grown if writing to rindex */
+		if (ip == GFS2_I(sdp->sd_rindex) && !sdp->sd_rindex_uptodate) {
+			error = gfs2_ri_update(ip);
+			if (error)
+				return error;
 		}
+		/* Flushing the log may release space */
+		if (loops == 2)
+			gfs2_log_flush(sdp, NULL);
 	}
-	error = -ENOSPC;
 
-out:
-	if (error)
-		rs->rs_requested = 0;
-	return error;
+	return -ENOSPC;
 }
 
 /**
@@ -1672,15 +1804,8 @@ void gfs2_inplace_release(struct gfs2_inode *ip)
 {
 	struct gfs2_blkreserv *rs = ip->i_res;
 
-	if (!rs)
-		return;
-
-	if (!rs->rs_free)
-		gfs2_rs_deltree(rs);
-
 	if (rs->rs_rgd_gh.gh_gl)
 		gfs2_glock_dq_uninit(&rs->rs_rgd_gh);
-	rs->rs_requested = 0;
 }
 
 /**
@@ -1693,173 +1818,47 @@ void gfs2_inplace_release(struct gfs2_inode *ip)
 
 static unsigned char gfs2_get_block_type(struct gfs2_rgrpd *rgd, u64 block)
 {
-	struct gfs2_bitmap *bi = NULL;
-	u32 length, rgrp_block, buf_block;
-	unsigned int buf;
-	unsigned char type;
-
-	length = rgd->rd_length;
-	rgrp_block = block - rgd->rd_data0;
-
-	for (buf = 0; buf < length; buf++) {
-		bi = rgd->rd_bits + buf;
-		if (rgrp_block < (bi->bi_start + bi->bi_len) * GFS2_NBBY)
-			break;
-	}
-
-	gfs2_assert(rgd->rd_sbd, buf < length);
-	buf_block = rgrp_block - bi->bi_start * GFS2_NBBY;
+	struct gfs2_rbm rbm = { .rgd = rgd, };
+	int ret;
 
-	type = gfs2_testbit(rgd, bi->bi_bh->b_data + bi->bi_offset,
-			   bi->bi_len, buf_block);
+	ret = gfs2_rbm_from_block(&rbm, block);
+	WARN_ON_ONCE(ret != 0);
 
-	return type;
+	return gfs2_testbit(&rbm);
 }
 
-/**
- * rgblk_search - find a block in @state
- * @rgd: the resource group descriptor
- * @goal: the goal block within the RG (start here to search for avail block)
- * @state: GFS2_BLKST_XXX the before-allocation state to find
- * @rbi: address of the pointer to the bitmap containing the block found
- *
- * Walk rgrp's bitmap to find bits that represent a block in @state.
- *
- * This function never fails, because we wouldn't call it unless we
- * know (from reservation results, etc.) that a block is available.
- *
- * Scope of @goal is just within rgrp, not the whole filesystem.
- * Scope of @returned block is just within bitmap, not the whole filesystem.
- *
- * Returns: the block number found relative to the bitmap rbi
- */
-
-static u32 rgblk_search(struct gfs2_rgrpd *rgd, u32 goal, unsigned char state,
-			struct gfs2_bitmap **rbi)
-{
-	struct gfs2_bitmap *bi = NULL;
-	const u32 length = rgd->rd_length;
-	u32 biblk = BFITNOENT;
-	unsigned int buf, x;
-	const u8 *buffer = NULL;
-
-	*rbi = NULL;
-	/* Find bitmap block that contains bits for goal block */
-	for (buf = 0; buf < length; buf++) {
-		bi = rgd->rd_bits + buf;
-		/* Convert scope of "goal" from rgrp-wide to within found bit block */
-		if (goal < (bi->bi_start + bi->bi_len) * GFS2_NBBY) {
-			goal -= bi->bi_start * GFS2_NBBY;
-			goto do_search;
-		}
-	}
-	buf = 0;
-	goal = 0;
-
-do_search:
-	/* Search (up to entire) bitmap in this rgrp for allocatable block.
-	   "x <= length", instead of "x < length", because we typically start
-	   the search in the middle of a bit block, but if we can't find an
-	   allocatable block anywhere else, we want to be able wrap around and
-	   search in the first part of our first-searched bit block.  */
-	for (x = 0; x <= length; x++) {
-		bi = rgd->rd_bits + buf;
-
-		if (test_bit(GBF_FULL, &bi->bi_flags) &&
-		    (state == GFS2_BLKST_FREE))
-			goto skip;
-
-		/* The GFS2_BLKST_UNLINKED state doesn't apply to the clone
-		   bitmaps, so we must search the originals for that. */
-		buffer = bi->bi_bh->b_data + bi->bi_offset;
-		WARN_ON(!buffer_uptodate(bi->bi_bh));
-		if (state != GFS2_BLKST_UNLINKED && bi->bi_clone)
-			buffer = bi->bi_clone + bi->bi_offset;
-
-		while (1) {
-			struct gfs2_blkreserv *rs;
-			u32 rgblk;
-
-			biblk = gfs2_bitfit(buffer, bi->bi_len, goal, state);
-			if (biblk == BFITNOENT)
-				break;
-			/* Check if this block is reserved() */
-			rgblk = gfs2_bi2rgd_blk(bi, biblk);
-			rs = rs_find(rgd, rgblk);
-			if (rs == NULL)
-				break;
-
-			BUG_ON(rs->rs_bi != bi);
-			biblk = BFITNOENT;
-			/* This should jump to the first block after the
-			   reservation. */
-			goal = rs->rs_biblk + rs->rs_free;
-			if (goal >= bi->bi_len * GFS2_NBBY)
-				break;
-		}
-		if (biblk != BFITNOENT)
-			break;
-
-		if ((goal == 0) && (state == GFS2_BLKST_FREE))
-			set_bit(GBF_FULL, &bi->bi_flags);
-
-		/* Try next bitmap block (wrap back to rgrp header if at end) */
-skip:
-		buf++;
-		buf %= length;
-		goal = 0;
-	}
-
-	if (biblk != BFITNOENT)
-		*rbi = bi;
-
-	return biblk;
-}
 
 /**
  * gfs2_alloc_extent - allocate an extent from a given bitmap
- * @rgd: the resource group descriptor
- * @bi: the bitmap within the rgrp
- * @blk: the block within the bitmap
+ * @rbm: the resource group information
  * @dinode: TRUE if the first block we allocate is for a dinode
- * @n: The extent length
+ * @n: The extent length (value/result)
  *
- * Add the found bitmap buffer to the transaction.
+ * Add the bitmap buffer to the transaction.
  * Set the found bits to @new_state to change block's allocation state.
- * Returns: starting block number of the extent (fs scope)
  */
-static u64 gfs2_alloc_extent(struct gfs2_rgrpd *rgd, struct gfs2_bitmap *bi,
-			     u32 blk, bool dinode, unsigned int *n)
+static void gfs2_alloc_extent(const struct gfs2_rbm *rbm, bool dinode,
+			     unsigned int *n)
 {
+	struct gfs2_rbm pos = { .rgd = rbm->rgd, };
 	const unsigned int elen = *n;
-	u32 goal, rgblk;
-	const u8 *buffer = NULL;
-	struct gfs2_blkreserv *rs;
-
-	*n = 0;
-	buffer = bi->bi_bh->b_data + bi->bi_offset;
-	gfs2_trans_add_bh(rgd->rd_gl, bi->bi_bh, 1);
-	gfs2_setbit(rgd, bi->bi_clone, bi, blk,
-		    dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
-	(*n)++;
-	goal = blk;
+	u64 block;
+	int ret;
+
+	*n = 1;
+	block = gfs2_rbm_to_block(rbm);
+	gfs2_trans_add_bh(rbm->rgd->rd_gl, rbm->bi->bi_bh, 1);
+	gfs2_setbit(rbm, true, dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
+	block++;
 	while (*n < elen) {
-		goal++;
-		if (goal >= (bi->bi_len * GFS2_NBBY))
-			break;
-		rgblk = gfs2_bi2rgd_blk(bi, goal);
-		rs = rs_find(rgd, rgblk);
-		if (rs) /* Oops, we bumped into someone's reservation */
-			break;
-		if (gfs2_testbit(rgd, buffer, bi->bi_len, goal) !=
-		    GFS2_BLKST_FREE)
+		ret = gfs2_rbm_from_block(&pos, block);
+		if (ret || gfs2_testbit(&pos) != GFS2_BLKST_FREE)
 			break;
-		gfs2_setbit(rgd, bi->bi_clone, bi, goal, GFS2_BLKST_USED);
+		gfs2_trans_add_bh(pos.rgd->rd_gl, pos.bi->bi_bh, 1);
+		gfs2_setbit(&pos, true, GFS2_BLKST_USED);
 		(*n)++;
+		block++;
 	}
-	blk = gfs2_bi2rgd_blk(bi, blk);
-	rgd->rd_last_alloc = blk + *n - 1;
-	return rgd->rd_data0 + blk;
 }
 
 /**
@@ -1875,46 +1874,30 @@ static u64 gfs2_alloc_extent(struct gfs2_rgrpd *rgd, struct gfs2_bitmap *bi,
 static struct gfs2_rgrpd *rgblk_free(struct gfs2_sbd *sdp, u64 bstart,
 				     u32 blen, unsigned char new_state)
 {
-	struct gfs2_rgrpd *rgd;
-	struct gfs2_bitmap *bi = NULL;
-	u32 length, rgrp_blk, buf_blk;
-	unsigned int buf;
+	struct gfs2_rbm rbm;
 
-	rgd = gfs2_blk2rgrpd(sdp, bstart, 1);
-	if (!rgd) {
+	rbm.rgd = gfs2_blk2rgrpd(sdp, bstart, 1);
+	if (!rbm.rgd) {
 		if (gfs2_consist(sdp))
 			fs_err(sdp, "block = %llu\n", (unsigned long long)bstart);
 		return NULL;
 	}
 
-	length = rgd->rd_length;
-
-	rgrp_blk = bstart - rgd->rd_data0;
-
 	while (blen--) {
-		for (buf = 0; buf < length; buf++) {
-			bi = rgd->rd_bits + buf;
-			if (rgrp_blk < (bi->bi_start + bi->bi_len) * GFS2_NBBY)
-				break;
+		gfs2_rbm_from_block(&rbm, bstart);
+		bstart++;
+		if (!rbm.bi->bi_clone) {
+			rbm.bi->bi_clone = kmalloc(rbm.bi->bi_bh->b_size,
+						   GFP_NOFS | __GFP_NOFAIL);
+			memcpy(rbm.bi->bi_clone + rbm.bi->bi_offset,
+			       rbm.bi->bi_bh->b_data + rbm.bi->bi_offset,
+			       rbm.bi->bi_len);
 		}
-
-		gfs2_assert(rgd->rd_sbd, buf < length);
-
-		buf_blk = rgrp_blk - bi->bi_start * GFS2_NBBY;
-		rgrp_blk++;
-
-		if (!bi->bi_clone) {
-			bi->bi_clone = kmalloc(bi->bi_bh->b_size,
-					       GFP_NOFS | __GFP_NOFAIL);
-			memcpy(bi->bi_clone + bi->bi_offset,
-			       bi->bi_bh->b_data + bi->bi_offset,
-			       bi->bi_len);
-		}
-		gfs2_trans_add_bh(rgd->rd_gl, bi->bi_bh, 1);
-		gfs2_setbit(rgd, NULL, bi, buf_blk, new_state);
+		gfs2_trans_add_bh(rbm.rgd->rd_gl, rbm.bi->bi_bh, 1);
+		gfs2_setbit(&rbm, false, new_state);
 	}
 
-	return rgd;
+	return rbm.rgd;
 }
 
 /**
@@ -1956,46 +1939,41 @@ static void gfs2_rgrp_error(struct gfs2_rgrpd *rgd)
 }
 
 /**
- * claim_reserved_blks - Claim previously reserved blocks
- * @ip: the inode that's claiming the reservation
- * @dinode: 1 if this block is a dinode block, otherwise data block
- * @nblocks: desired extent length
+ * gfs2_adjust_reservation - Adjust (or remove) a reservation after allocation
+ * @ip: The inode we have just allocated blocks for
+ * @rbm: The start of the allocated blocks
+ * @len: The extent length
  *
- * Lay claim to previously reserved blocks.
- * Returns: Starting block number of the blocks claimed.
- * Sets *nblocks to the actual extent length allocated.
+ * Adjusts a reservation after an allocation has taken place. If the
+ * reservation does not match the allocation, or if it is now empty
+ * then it is removed.
  */
-static u64 claim_reserved_blks(struct gfs2_inode *ip, bool dinode,
-			       unsigned int *nblocks)
+
+static void gfs2_adjust_reservation(struct gfs2_inode *ip,
+				    const struct gfs2_rbm *rbm, unsigned len)
 {
 	struct gfs2_blkreserv *rs = ip->i_res;
-	struct gfs2_rgrpd *rgd = rs->rs_rgd;
-	struct gfs2_bitmap *bi;
-	u64 start_block = gfs2_rs_startblk(rs);
-	const unsigned int elen = *nblocks;
+	struct gfs2_rgrpd *rgd = rbm->rgd;
+	unsigned rlen;
+	u64 block;
+	int ret;
 
-	bi = rs->rs_bi;
-	gfs2_trans_add_bh(rgd->rd_gl, bi->bi_bh, 1);
-
-	for (*nblocks = 0; *nblocks < elen && rs->rs_free; (*nblocks)++) {
-		if (gfs2_testbit(rgd, bi->bi_bh->b_data + bi->bi_offset,
-				 bi->bi_len, rs->rs_biblk) != GFS2_BLKST_FREE)
-			break;
-		gfs2_setbit(rgd, bi->bi_clone, bi, rs->rs_biblk,
-			    dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
-		rs->rs_biblk++;
-		rs->rs_free--;
-
-		BUG_ON(!rgd->rd_reserved);
-		rgd->rd_reserved--;
-		dinode = false;
+	spin_lock(&rgd->rd_rsspin);
+	if (gfs2_rs_active(rs)) {
+		if (gfs2_rbm_eq(&rs->rs_rbm, rbm)) {
+			block = gfs2_rbm_to_block(rbm);
+			ret = gfs2_rbm_from_block(&rs->rs_rbm, block + len);
+			rlen = min(rs->rs_free, len);
+			rs->rs_free -= rlen;
+			rgd->rd_reserved -= rlen;
+			trace_gfs2_rs(rs, TRACE_RS_CLAIM);
+			if (rs->rs_free && !ret)
+				goto out;
+		}
+		__rs_deltree(ip, rs);
 	}
-
-	trace_gfs2_rs(ip, rs, TRACE_RS_CLAIM);
-	if (!rs->rs_free || *nblocks != elen)
-		gfs2_rs_deltree(rs);
-
-	return start_block;
+out:
+	spin_unlock(&rgd->rd_rsspin);
 }
 
 /**
@@ -2014,47 +1992,40 @@ int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks,
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
 	struct buffer_head *dibh;
-	struct gfs2_rgrpd *rgd;
+	struct gfs2_rbm rbm = { .rgd = ip->i_rgd, };
 	unsigned int ndata;
-	u32 goal, blk; /* block, within the rgrp scope */
+	u64 goal;
 	u64 block; /* block, within the file system scope */
 	int error;
-	struct gfs2_bitmap *bi;
 
-	/* Only happens if there is a bug in gfs2, return something distinctive
-	 * to ensure that it is noticed.
-	 */
-	if (ip->i_res->rs_requested == 0)
-		return -ECANCELED;
-
-	/* If we have a reservation, claim blocks from it. */
-	if (gfs2_rs_active(ip->i_res)) {
-		BUG_ON(!ip->i_res->rs_free);
-		rgd = ip->i_res->rs_rgd;
-		block = claim_reserved_blks(ip, dinode, nblocks);
-		if (*nblocks)
-			goto found_blocks;
-	}
-
-	rgd = ip->i_rgd;
-
-	if (!dinode && rgrp_contains_block(rgd, ip->i_goal))
-		goal = ip->i_goal - rgd->rd_data0;
+	if (gfs2_rs_active(ip->i_res))
+		goal = gfs2_rbm_to_block(&ip->i_res->rs_rbm);
+	else if (!dinode && rgrp_contains_block(rbm.rgd, ip->i_goal))
+		goal = ip->i_goal;
 	else
-		goal = rgd->rd_last_alloc;
+		goal = rbm.rgd->rd_last_alloc + rbm.rgd->rd_data0;
 
-	blk = rgblk_search(rgd, goal, GFS2_BLKST_FREE, &bi);
+	gfs2_rbm_from_block(&rbm, goal);
+	error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, 0, ip, false);
+
+	if (error == -ENOSPC) {
+		gfs2_rbm_from_block(&rbm, goal);
+		error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, 0, NULL, false);
+	}
 
 	/* Since all blocks are reserved in advance, this shouldn't happen */
-	if (blk == BFITNOENT) {
-		printk(KERN_WARNING "BFITNOENT, nblocks=%u\n", *nblocks);
-		printk(KERN_WARNING "FULL=%d\n",
-		       test_bit(GBF_FULL, &rgd->rd_bits->bi_flags));
+	if (error) {
+		fs_warn(sdp, "inum=%llu error=%d, nblocks=%u, full=%d\n",
+			(unsigned long long)ip->i_no_addr, error, *nblocks,
+			test_bit(GBF_FULL, &rbm.rgd->rd_bits->bi_flags));
 		goto rgrp_error;
 	}
 
-	block = gfs2_alloc_extent(rgd, bi, blk, dinode, nblocks);
-found_blocks:
+	gfs2_alloc_extent(&rbm, dinode, nblocks);
+	block = gfs2_rbm_to_block(&rbm);
+	rbm.rgd->rd_last_alloc = block - rbm.rgd->rd_data0;
+	if (gfs2_rs_active(ip->i_res))
+		gfs2_adjust_reservation(ip, &rbm, *nblocks);
 	ndata = *nblocks;
 	if (dinode)
 		ndata--;
@@ -2071,22 +2042,22 @@ int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks,
 			brelse(dibh);
 		}
 	}
-	if (rgd->rd_free < *nblocks) {
+	if (rbm.rgd->rd_free < *nblocks) {
 		printk(KERN_WARNING "nblocks=%u\n", *nblocks);
 		goto rgrp_error;
 	}
 
-	rgd->rd_free -= *nblocks;
+	rbm.rgd->rd_free -= *nblocks;
 	if (dinode) {
-		rgd->rd_dinodes++;
-		*generation = rgd->rd_igeneration++;
+		rbm.rgd->rd_dinodes++;
+		*generation = rbm.rgd->rd_igeneration++;
 		if (*generation == 0)
-			*generation = rgd->rd_igeneration++;
+			*generation = rbm.rgd->rd_igeneration++;
 	}
 
-	gfs2_trans_add_bh(rgd->rd_gl, rgd->rd_bits[0].bi_bh, 1);
-	gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data);
-	gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, rgd->rd_bits[0].bi_bh->b_data);
+	gfs2_trans_add_bh(rbm.rgd->rd_gl, rbm.rgd->rd_bits[0].bi_bh, 1);
+	gfs2_rgrp_out(rbm.rgd, rbm.rgd->rd_bits[0].bi_bh->b_data);
+	gfs2_rgrp_ondisk2lvb(rbm.rgd->rd_rgl, rbm.rgd->rd_bits[0].bi_bh->b_data);
 
 	gfs2_statfs_change(sdp, 0, -(s64)*nblocks, dinode ? 1 : 0);
 	if (dinode)
@@ -2100,14 +2071,14 @@ int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks,
 		gfs2_quota_change(ip, ndata, ip->i_inode.i_uid,
 				  ip->i_inode.i_gid);
 
-	rgd->rd_free_clone -= *nblocks;
-	trace_gfs2_block_alloc(ip, rgd, block, *nblocks,
+	rbm.rgd->rd_free_clone -= *nblocks;
+	trace_gfs2_block_alloc(ip, rbm.rgd, block, *nblocks,
 			       dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
 	*bn = block;
 	return 0;
 
 rgrp_error:
-	gfs2_rgrp_error(rgd);
+	gfs2_rgrp_error(rbm.rgd);
 	return -EIO;
 }
 

fs/gfs2/rgrp.h

@@ -46,7 +46,7 @@ extern int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *n,
 			     bool dinode, u64 *generation);
 
 extern int gfs2_rs_alloc(struct gfs2_inode *ip);
-extern void gfs2_rs_deltree(struct gfs2_blkreserv *rs);
+extern void gfs2_rs_deltree(struct gfs2_inode *ip, struct gfs2_blkreserv *rs);
 extern void gfs2_rs_delete(struct gfs2_inode *ip);
 extern void __gfs2_free_blocks(struct gfs2_inode *ip, u64 bstart, u32 blen, int meta);
 extern void gfs2_free_meta(struct gfs2_inode *ip, u64 bstart, u32 blen);
@@ -73,30 +73,10 @@ extern int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset,
 				   const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed);
 extern int gfs2_fitrim(struct file *filp, void __user *argp);
 
-/* This is how to tell if a multi-block reservation is "inplace" reserved: */
-static inline int gfs2_mb_reserved(struct gfs2_inode *ip)
+/* This is how to tell if a reservation is in the rgrp tree: */
+static inline bool gfs2_rs_active(struct gfs2_blkreserv *rs)
 {
-	if (ip->i_res && ip->i_res->rs_requested)
-		return 1;
-	return 0;
-}
-
-/* This is how to tell if a multi-block reservation is in the rgrp tree: */
-static inline int gfs2_rs_active(struct gfs2_blkreserv *rs)
-{
-	if (rs && rs->rs_bi)
-		return 1;
-	return 0;
-}
-
-static inline u32 gfs2_bi2rgd_blk(const struct gfs2_bitmap *bi, u32 blk)
-{
-	return (bi->bi_start * GFS2_NBBY) + blk;
-}
-
-static inline u64 gfs2_rs_startblk(const struct gfs2_blkreserv *rs)
-{
-	return gfs2_bi2rgd_blk(rs->rs_bi, rs->rs_biblk) + rs->rs_rgd->rd_data0;
+	return rs && !RB_EMPTY_NODE(&rs->rs_node);
 }
 
 #endif /* __RGRP_DOT_H__ */

fs/gfs2/super.c

@@ -1366,6 +1366,8 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root)
 	val = sdp->sd_tune.gt_statfs_quantum;
 	if (val != 30)
 		seq_printf(s, ",statfs_quantum=%d", val);
+	else if (sdp->sd_tune.gt_statfs_slow)
+		seq_puts(s, ",statfs_quantum=0");
 	val = sdp->sd_tune.gt_quota_quantum;
 	if (val != 60)
 		seq_printf(s, ",quota_quantum=%d", val);
@@ -1543,6 +1545,11 @@ static void gfs2_evict_inode(struct inode *inode)
 
 out_truncate:
 	gfs2_log_flush(sdp, ip->i_gl);
+	if (test_bit(GLF_DIRTY, &ip->i_gl->gl_flags)) {
+		struct address_space *metamapping = gfs2_glock2aspace(ip->i_gl);
+		filemap_fdatawrite(metamapping);
+		filemap_fdatawait(metamapping);
+	}
 	write_inode_now(inode, 1);
 	gfs2_ail_flush(ip->i_gl, 0);
 
@@ -1557,7 +1564,7 @@ static void gfs2_evict_inode(struct inode *inode)
 out_unlock:
 	/* Error path for case 1 */
 	if (gfs2_rs_active(ip->i_res))
-		gfs2_rs_deltree(ip->i_res);
+		gfs2_rs_deltree(ip, ip->i_res);
 
 	if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags))
 		gfs2_glock_dq(&ip->i_iopen_gh);

fs/gfs2/trace_gfs2.h

@@ -509,10 +509,9 @@ TRACE_EVENT(gfs2_block_alloc,
 /* Keep track of multi-block reservations as they are allocated/freed */
 TRACE_EVENT(gfs2_rs,
 
-	TP_PROTO(const struct gfs2_inode *ip, const struct gfs2_blkreserv *rs,
-		 u8 func),
+	TP_PROTO(const struct gfs2_blkreserv *rs, u8 func),
 
-	TP_ARGS(ip, rs, func),
+	TP_ARGS(rs, func),
 
 	TP_STRUCT__entry(
 		__field(        dev_t,  dev                     )
@@ -526,18 +525,17 @@ TRACE_EVENT(gfs2_rs,
 	),
 
 	TP_fast_assign(
-		__entry->dev		= rs->rs_rgd ? rs->rs_rgd->rd_sbd->sd_vfs->s_dev : 0;
-		__entry->rd_addr	= rs->rs_rgd ? rs->rs_rgd->rd_addr : 0;
-		__entry->rd_free_clone	= rs->rs_rgd ? rs->rs_rgd->rd_free_clone : 0;
-		__entry->rd_reserved	= rs->rs_rgd ? rs->rs_rgd->rd_reserved : 0;
-		__entry->inum		= ip ? ip->i_no_addr : 0;
-		__entry->start		= gfs2_rs_startblk(rs);
+		__entry->dev		= rs->rs_rbm.rgd->rd_sbd->sd_vfs->s_dev;
+		__entry->rd_addr	= rs->rs_rbm.rgd->rd_addr;
+		__entry->rd_free_clone	= rs->rs_rbm.rgd->rd_free_clone;
+		__entry->rd_reserved	= rs->rs_rbm.rgd->rd_reserved;
+		__entry->inum		= rs->rs_inum;
+		__entry->start		= gfs2_rbm_to_block(&rs->rs_rbm);
 		__entry->free		= rs->rs_free;
 		__entry->func		= func;
 	),
 
-	TP_printk("%u,%u bmap %llu resrv %llu rg:%llu rf:%lu rr:%lu %s "
-		  "f:%lu",
+	TP_printk("%u,%u bmap %llu resrv %llu rg:%llu rf:%lu rr:%lu %s f:%lu",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  (unsigned long long)__entry->inum,
 		  (unsigned long long)__entry->start,

fs/gfs2/trans.h

@@ -28,11 +28,10 @@ struct gfs2_glock;
 
 /* reserve either the number of blocks to be allocated plus the rg header
  * block, or all of the blocks in the rg, whichever is smaller */
-static inline unsigned int gfs2_rg_blocks(const struct gfs2_inode *ip)
+static inline unsigned int gfs2_rg_blocks(const struct gfs2_inode *ip, unsigned requested)
 {
-	const struct gfs2_blkreserv *rs = ip->i_res;
-	if (rs && rs->rs_requested < ip->i_rgd->rd_length)
-		return rs->rs_requested + 1;
+	if (requested < ip->i_rgd->rd_length)
+		return requested + 1;
 	return ip->i_rgd->rd_length;
 }
 

fs/gfs2/xattr.c

@@ -448,17 +448,18 @@ ssize_t gfs2_listxattr(struct dentry *dentry, char *buffer, size_t size)
 }
 
 /**
- * ea_get_unstuffed - actually copies the unstuffed data into the
- *                    request buffer
+ * ea_iter_unstuffed - copies the unstuffed xattr data to/from the
+ *                     request buffer
  * @ip: The GFS2 inode
  * @ea: The extended attribute header structure
- * @data: The data to be copied
+ * @din: The data to be copied in
+ * @dout: The data to be copied out (one of din,dout will be NULL)
  *
  * Returns: errno
  */
 
-static int ea_get_unstuffed(struct gfs2_inode *ip, struct gfs2_ea_header *ea,
-			    char *data)
+static int gfs2_iter_unstuffed(struct gfs2_inode *ip, struct gfs2_ea_header *ea,
+			       const char *din, char *dout)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
 	struct buffer_head **bh;
@@ -467,6 +468,8 @@ static int ea_get_unstuffed(struct gfs2_inode *ip, struct gfs2_ea_header *ea,
 	__be64 *dataptrs = GFS2_EA2DATAPTRS(ea);
 	unsigned int x;
 	int error = 0;
+	unsigned char *pos;
+	unsigned cp_size;
 
 	bh = kcalloc(nptrs, sizeof(struct buffer_head *), GFP_NOFS);
 	if (!bh)
@@ -497,12 +500,21 @@ static int ea_get_unstuffed(struct gfs2_inode *ip, struct gfs2_ea_header *ea,
 			goto out;
 		}
 
-		memcpy(data, bh[x]->b_data + sizeof(struct gfs2_meta_header),
-		       (sdp->sd_jbsize > amount) ? amount : sdp->sd_jbsize);
+		pos = bh[x]->b_data + sizeof(struct gfs2_meta_header);
+		cp_size = (sdp->sd_jbsize > amount) ? amount : sdp->sd_jbsize;
 
-		amount -= sdp->sd_jbsize;
-		data += sdp->sd_jbsize;
+		if (dout) {
+			memcpy(dout, pos, cp_size);
+			dout += sdp->sd_jbsize;
+		}
+
+		if (din) {
+			gfs2_trans_add_bh(ip->i_gl, bh[x], 1);
+			memcpy(pos, din, cp_size);
+			din += sdp->sd_jbsize;
+		}
 
+		amount -= sdp->sd_jbsize;
 		brelse(bh[x]);
 	}
 
@@ -523,7 +535,7 @@ static int gfs2_ea_get_copy(struct gfs2_inode *ip, struct gfs2_ea_location *el,
 		memcpy(data, GFS2_EA2DATA(el->el_ea), len);
 		return len;
 	}
-	ret = ea_get_unstuffed(ip, el->el_ea, data);
+	ret = gfs2_iter_unstuffed(ip, el->el_ea, NULL, data);
 	if (ret < 0)
 		return ret;
 	return len;
@@ -727,7 +739,7 @@ static int ea_alloc_skeleton(struct gfs2_inode *ip, struct gfs2_ea_request *er,
 		goto out_gunlock_q;
 
 	error = gfs2_trans_begin(GFS2_SB(&ip->i_inode),
-				 blks + gfs2_rg_blocks(ip) +
+				 blks + gfs2_rg_blocks(ip, blks) +
 				 RES_DINODE + RES_STATFS + RES_QUOTA, 0);
 	if (error)
 		goto out_ipres;
@@ -1220,69 +1232,23 @@ static int gfs2_xattr_set(struct dentry *dentry, const char *name,
 				size, flags, type);
 }
 
+
 static int ea_acl_chmod_unstuffed(struct gfs2_inode *ip,
 				  struct gfs2_ea_header *ea, char *data)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
-	struct buffer_head **bh;
 	unsigned int amount = GFS2_EA_DATA_LEN(ea);
 	unsigned int nptrs = DIV_ROUND_UP(amount, sdp->sd_jbsize);
-	__be64 *dataptrs = GFS2_EA2DATAPTRS(ea);
-	unsigned int x;
-	int error;
-
-	bh = kcalloc(nptrs, sizeof(struct buffer_head *), GFP_NOFS);
-	if (!bh)
-		return -ENOMEM;
-
-	error = gfs2_trans_begin(sdp, nptrs + RES_DINODE, 0);
-	if (error)
-		goto out;
-
-	for (x = 0; x < nptrs; x++) {
-		error = gfs2_meta_read(ip->i_gl, be64_to_cpu(*dataptrs), 0,
-				       bh + x);
-		if (error) {
-			while (x--)
-				brelse(bh[x]);
-			goto fail;
-		}
-		dataptrs++;
-	}
-
-	for (x = 0; x < nptrs; x++) {
-		error = gfs2_meta_wait(sdp, bh[x]);
-		if (error) {
-			for (; x < nptrs; x++)
-				brelse(bh[x]);
-			goto fail;
-		}
-		if (gfs2_metatype_check(sdp, bh[x], GFS2_METATYPE_ED)) {
-			for (; x < nptrs; x++)
-				brelse(bh[x]);
-			error = -EIO;
-			goto fail;
-		}
-
-		gfs2_trans_add_bh(ip->i_gl, bh[x], 1);
-
-		memcpy(bh[x]->b_data + sizeof(struct gfs2_meta_header), data,
-		       (sdp->sd_jbsize > amount) ? amount : sdp->sd_jbsize);
-
-		amount -= sdp->sd_jbsize;
-		data += sdp->sd_jbsize;
-
-		brelse(bh[x]);
-	}
+	int ret;
 
-out:
-	kfree(bh);
-	return error;
+	ret = gfs2_trans_begin(sdp, nptrs + RES_DINODE, 0);
+	if (ret)
+		return ret;
 
-fail:
+	ret = gfs2_iter_unstuffed(ip, ea, data, NULL);
 	gfs2_trans_end(sdp);
-	kfree(bh);
-	return error;
+
+	return ret;
 }
 
 int gfs2_xattr_acl_chmod(struct gfs2_inode *ip, struct iattr *attr, char *data)