Merge branch 'punch-hole' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git

7 files changed, 431 insertions, 198 deletions

diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
index d5f0d96169c5..2ec11981e694 100644
--- a/fs/gfs2/bmap.c
+++ b/fs/gfs2/bmap.c
@@ -279,14 +279,13 @@ static inline __be64 *metapointer(unsigned int height, const struct metapath *mp
 	return p + mp->mp_list[height];
 }
 
-static void gfs2_metapath_ra(struct gfs2_glock *gl,
-			     const struct buffer_head *bh, const __be64 *pos)
+static void gfs2_metapath_ra(struct gfs2_glock *gl, __be64 *start, __be64 *end)
 {
-	struct buffer_head *rabh;
-	const __be64 *endp = (const __be64 *)(bh->b_data + bh->b_size);
 	const __be64 *t;
 
-	for (t = pos; t < endp; t++) {
+	for (t = start; t < end; t++) {
+		struct buffer_head *rabh;
+
 		if (!*t)
 			continue;
 
@@ -305,21 +304,22 @@ static void gfs2_metapath_ra(struct gfs2_glock *gl,
 	}
 }
 
-/**
- * lookup_mp_height - helper function for lookup_metapath
- * @ip: the inode
- * @mp: the metapath
- * @h: the height which needs looking up
- */
-static int lookup_mp_height(struct gfs2_inode *ip, struct metapath *mp, int h)
+static int __fillup_metapath(struct gfs2_inode *ip, struct metapath *mp,
+			     unsigned int x, unsigned int h)
 {
-	__be64 *ptr = metapointer(h, mp);
-	u64 dblock = be64_to_cpu(*ptr);
-
-	if (!dblock)
-		return h + 1;
+	for (; x < h; x++) {
+		__be64 *ptr = metapointer(x, mp);
+		u64 dblock = be64_to_cpu(*ptr);
+		int ret;
 
-	return gfs2_meta_indirect_buffer(ip, h + 1, dblock, &mp->mp_bh[h + 1]);
+		if (!dblock)
+			break;
+		ret = gfs2_meta_indirect_buffer(ip, x + 1, dblock, &mp->mp_bh[x + 1]);
+		if (ret)
+			return ret;
+	}
+	mp->mp_aheight = x + 1;
+	return 0;
 }
 
 /**
@@ -336,25 +336,12 @@ static int lookup_mp_height(struct gfs2_inode *ip, struct metapath *mp, int h)
  * at which it found the unallocated block. Blocks which are found are
  * added to the mp->mp_bh[] list.
  *
- * Returns: error or height of metadata tree
+ * Returns: error
  */
 
 static int lookup_metapath(struct gfs2_inode *ip, struct metapath *mp)
 {
-	unsigned int end_of_metadata = ip->i_height - 1;
-	unsigned int x;
-	int ret;
-
-	for (x = 0; x < end_of_metadata; x++) {
-		ret = lookup_mp_height(ip, mp, x);
-		if (ret)
-			goto out;
-	}
-
-	ret = ip->i_height;
-out:
-	mp->mp_aheight = ret;
-	return ret;
+	return __fillup_metapath(ip, mp, 0, ip->i_height - 1);
 }
 
 /**
@@ -365,25 +352,25 @@ out:
  *
  * Similar to lookup_metapath, but does lookups for a range of heights
  *
- * Returns: error or height of metadata tree
+ * Returns: error or the number of buffers filled
  */
 
 static int fillup_metapath(struct gfs2_inode *ip, struct metapath *mp, int h)
 {
-	unsigned int start_h = h - 1;
+	unsigned int x = 0;
 	int ret;
 
 	if (h) {
 		/* find the first buffer we need to look up. */
-		while (start_h > 0 && mp->mp_bh[start_h] == NULL)
-			start_h--;
-		for (; start_h < h; start_h++) {
-			ret = lookup_mp_height(ip, mp, start_h);
-			if (ret)
-				return ret;
+		for (x = h - 1; x > 0; x--) {
+			if (mp->mp_bh[x])
+				break;
 		}
 	}
-	return ip->i_height;
+	ret = __fillup_metapath(ip, mp, x, h);
+	if (ret)
+		return ret;
+	return mp->mp_aheight - x - 1;
 }
 
 static inline void release_metapath(struct metapath *mp)
@@ -474,13 +461,6 @@ enum alloc_state {
 	/* ALLOC_UNSTUFF = 3,   TBD and rather complicated */
 };
 
-static inline unsigned int hptrs(struct gfs2_sbd *sdp, const unsigned int hgt)
-{
-	if (hgt)
-		return sdp->sd_inptrs;
-	return sdp->sd_diptrs;
-}
-
 /**
  * gfs2_bmap_alloc - Build a metadata tree of the requested height
  * @inode: The GFS2 inode
@@ -788,7 +768,7 @@ int gfs2_iomap_begin(struct inode *inode, loff_t pos, loff_t length,
 		goto do_alloc;
 
 	ret = lookup_metapath(ip, &mp);
-	if (ret < 0)
+	if (ret)
 		goto out_release;
 
 	if (mp.mp_aheight != ip->i_height)
@@ -913,17 +893,18 @@ int gfs2_extent_map(struct inode *inode, u64 lblock, int *new, u64 *dblock, unsi
 }
 
 /**
- * gfs2_block_truncate_page - Deal with zeroing out data for truncate
+ * gfs2_block_zero_range - Deal with zeroing out data
  *
  * This is partly borrowed from ext3.
  */
-static int gfs2_block_truncate_page(struct address_space *mapping, loff_t from)
+static int gfs2_block_zero_range(struct inode *inode, loff_t from,
+				 unsigned int length)
 {
-	struct inode *inode = mapping->host;
+	struct address_space *mapping = inode->i_mapping;
 	struct gfs2_inode *ip = GFS2_I(inode);
 	unsigned long index = from >> PAGE_SHIFT;
 	unsigned offset = from & (PAGE_SIZE-1);
-	unsigned blocksize, iblock, length, pos;
+	unsigned blocksize, iblock, pos;
 	struct buffer_head *bh;
 	struct page *page;
 	int err;
@@ -933,7 +914,6 @@ static int gfs2_block_truncate_page(struct address_space *mapping, loff_t from)
 		return 0;
 
 	blocksize = inode->i_sb->s_blocksize;
-	length = blocksize - (offset & (blocksize - 1));
 	iblock = index << (PAGE_SHIFT - inode->i_sb->s_blocksize_bits);
 
 	if (!page_has_buffers(page))
@@ -1003,11 +983,24 @@ static int gfs2_journaled_truncate(struct inode *inode, u64 oldsize, u64 newsize
 	int error;
 
 	while (oldsize != newsize) {
+		struct gfs2_trans *tr;
+		unsigned int offs;
+
 		chunk = oldsize - newsize;
 		if (chunk > max_chunk)
 			chunk = max_chunk;
+
+		offs = oldsize & ~PAGE_MASK;
+		if (offs && chunk > PAGE_SIZE)
+			chunk = offs + ((chunk - offs) & PAGE_MASK);
+
 		truncate_pagecache(inode, oldsize - chunk);
 		oldsize -= chunk;
+
+		tr = current->journal_info;
+		if (!test_bit(TR_TOUCHED, &tr->tr_flags))
+			continue;
+
 		gfs2_trans_end(sdp);
 		error = gfs2_trans_begin(sdp, RES_DINODE, GFS2_JTRUNC_REVOKES);
 		if (error)
@@ -1017,13 +1010,13 @@ static int gfs2_journaled_truncate(struct inode *inode, u64 oldsize, u64 newsize
 	return 0;
 }
 
-static int trunc_start(struct inode *inode, u64 oldsize, u64 newsize)
+static int trunc_start(struct inode *inode, u64 newsize)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 	struct gfs2_sbd *sdp = GFS2_SB(inode);
-	struct address_space *mapping = inode->i_mapping;
-	struct buffer_head *dibh;
+	struct buffer_head *dibh = NULL;
 	int journaled = gfs2_is_jdata(ip);
+	u64 oldsize = inode->i_size;
 	int error;
 
 	if (journaled)
@@ -1042,10 +1035,13 @@ static int trunc_start(struct inode *inode, u64 oldsize, u64 newsize)
 	if (gfs2_is_stuffed(ip)) {
 		gfs2_buffer_clear_tail(dibh, sizeof(struct gfs2_dinode) + newsize);
 	} else {
-		if (newsize & (u64)(sdp->sd_sb.sb_bsize - 1)) {
-			error = gfs2_block_truncate_page(mapping, newsize);
+		unsigned int blocksize = i_blocksize(inode);
+		unsigned int offs = newsize & (blocksize - 1);
+		if (offs) {
+			error = gfs2_block_zero_range(inode, newsize,
+						      blocksize - offs);
 			if (error)
-				goto out_brelse;
+				goto out;
 		}
 		ip->i_diskflags |= GFS2_DIF_TRUNC_IN_PROG;
 	}
@@ -1059,15 +1055,10 @@ static int trunc_start(struct inode *inode, u64 oldsize, u64 newsize)
 	else
 		truncate_pagecache(inode, newsize);
 
-	if (error) {
-		brelse(dibh);
-		return error;
-	}
-
-out_brelse:
-	brelse(dibh);
 out:
-	gfs2_trans_end(sdp);
+	brelse(dibh);
+	if (current->journal_info)
+		gfs2_trans_end(sdp);
 	return error;
 }
 
@@ -1075,10 +1066,11 @@ out:
  * sweep_bh_for_rgrps - find an rgrp in a meta buffer and free blocks therein
  * @ip: inode
  * @rg_gh: holder of resource group glock
- * @mp: current metapath fully populated with buffers
+ * @bh: buffer head to sweep
+ * @start: starting point in bh
+ * @end: end point in bh
+ * @meta: true if bh points to metadata (rather than data)
  * @btotal: place to keep count of total blocks freed
- * @hgt: height we're processing
- * @first: true if this is the first call to this function for this height
  *
  * We sweep a metadata buffer (provided by the metapath) for blocks we need to
  * free, and free them all. However, we do it one rgrp at a time. If this
@@ -1093,47 +1085,46 @@ out:
  *          *btotal has the total number of blocks freed
  */
 static int sweep_bh_for_rgrps(struct gfs2_inode *ip, struct gfs2_holder *rd_gh,
-			      const struct metapath *mp, u32 *btotal, int hgt,
-			      bool preserve1)
+			      struct buffer_head *bh, __be64 *start, __be64 *end,
+			      bool meta, u32 *btotal)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
 	struct gfs2_rgrpd *rgd;
 	struct gfs2_trans *tr;
-	struct buffer_head *bh = mp->mp_bh[hgt];
-	__be64 *top, *bottom, *p;
+	__be64 *p;
 	int blks_outside_rgrp;
 	u64 bn, bstart, isize_blks;
 	s64 blen; /* needs to be s64 or gfs2_add_inode_blocks breaks */
-	int meta = ((hgt != ip->i_height - 1) ? 1 : 0);
 	int ret = 0;
 	bool buf_in_tr = false; /* buffer was added to transaction */
 
-	if (gfs2_metatype_check(sdp, bh,
-				(hgt ? GFS2_METATYPE_IN : GFS2_METATYPE_DI)))
-		return -EIO;
-
 more_rgrps:
+	rgd = NULL;
+	if (gfs2_holder_initialized(rd_gh)) {
+		rgd = gfs2_glock2rgrp(rd_gh->gh_gl);
+		gfs2_assert_withdraw(sdp,
+			     gfs2_glock_is_locked_by_me(rd_gh->gh_gl));
+	}
 	blks_outside_rgrp = 0;
 	bstart = 0;
 	blen = 0;
-	top = metapointer(hgt, mp); /* first ptr from metapath */
-	/* If we're keeping some data at the truncation point, we've got to
-	   preserve the metadata tree by adding 1 to the starting metapath. */
-	if (preserve1)
-		top++;
 
-	bottom = (__be64 *)(bh->b_data + bh->b_size);
-
-	for (p = top; p < bottom; p++) {
+	for (p = start; p < end; p++) {
 		if (!*p)
 			continue;
 		bn = be64_to_cpu(*p);
-		if (gfs2_holder_initialized(rd_gh)) {
-			rgd = gfs2_glock2rgrp(rd_gh->gh_gl);
-			gfs2_assert_withdraw(sdp,
-				     gfs2_glock_is_locked_by_me(rd_gh->gh_gl));
+
+		if (rgd) {
+			if (!rgrp_contains_block(rgd, bn)) {
+				blks_outside_rgrp++;
+				continue;
+			}
 		} else {
-			rgd = gfs2_blk2rgrpd(sdp, bn, false);
+			rgd = gfs2_blk2rgrpd(sdp, bn, true);
+			if (unlikely(!rgd)) {
+				ret = -EIO;
+				goto out;
+			}
 			ret = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_EXCLUSIVE,
 						 0, rd_gh);
 			if (ret)
@@ -1145,11 +1136,6 @@ more_rgrps:
 				gfs2_rs_deltree(&ip->i_res);
 		}
 
-		if (!rgrp_contains_block(rgd, bn)) {
-			blks_outside_rgrp++;
-			continue;
-		}
-
 		/* The size of our transactions will be unknown until we
 		   actually process all the metadata blocks that relate to
 		   the rgrp. So we estimate. We know it can't be more than
@@ -1168,7 +1154,7 @@ more_rgrps:
 				jblocks_rqsted += isize_blks;
 			revokes = jblocks_rqsted;
 			if (meta)
-				revokes += hptrs(sdp, hgt);
+				revokes += end - start;
 			else if (ip->i_depth)
 				revokes += sdp->sd_inptrs;
 			ret = gfs2_trans_begin(sdp, jblocks_rqsted, revokes);
@@ -1226,7 +1212,11 @@ out_unlock:
 					    outside the rgrp we just processed,
 					    do it all over again. */
 		if (current->journal_info) {
-			struct buffer_head *dibh = mp->mp_bh[0];
+			struct buffer_head *dibh;
+
+			ret = gfs2_meta_inode_buffer(ip, &dibh);
+			if (ret)
+				goto out;
 
 			/* Every transaction boundary, we rewrite the dinode
 			   to keep its di_blocks current in case of failure. */
@@ -1234,6 +1224,7 @@ out_unlock:
 				current_time(&ip->i_inode);
 			gfs2_trans_add_meta(ip->i_gl, dibh);
 			gfs2_dinode_out(ip, dibh->b_data);
+			brelse(dibh);
 			up_write(&ip->i_rw_mutex);
 			gfs2_trans_end(sdp);
 		}
@@ -1245,38 +1236,48 @@ out:
 	return ret;
 }
 
+static bool mp_eq_to_hgt(struct metapath *mp, __u16 *list, unsigned int h)
+{
+	if (memcmp(mp->mp_list, list, h * sizeof(mp->mp_list[0])))
+		return false;
+	return true;
+}
+
 /**
  * find_nonnull_ptr - find a non-null pointer given a metapath and height
- * assumes the metapath is valid (with buffers) out to height h
  * @mp: starting metapath
  * @h: desired height to search
  *
+ * Assumes the metapath is valid (with buffers) out to height h.
  * Returns: true if a non-null pointer was found in the metapath buffer
  *          false if all remaining pointers are NULL in the buffer
  */
 static bool find_nonnull_ptr(struct gfs2_sbd *sdp, struct metapath *mp,
-			     unsigned int h)
+			     unsigned int h,
+			     __u16 *end_list, unsigned int end_aligned)
 {
-	__be64 *ptr;
-	unsigned int ptrs = hptrs(sdp, h) - 1;
+	struct buffer_head *bh = mp->mp_bh[h];
+	__be64 *first, *ptr, *end;
+
+	first = metaptr1(h, mp);
+	ptr = first + mp->mp_list[h];
+	end = (__be64 *)(bh->b_data + bh->b_size);
+	if (end_list && mp_eq_to_hgt(mp, end_list, h)) {
+		bool keep_end = h < end_aligned;
+		end = first + end_list[h] + keep_end;
+	}
 
-	while (true) {
-		ptr = metapointer(h, mp);
+	while (ptr < end) {
 		if (*ptr) { /* if we have a non-null pointer */
-			/* Now zero the metapath after the current height. */
+			mp->mp_list[h] = ptr - first;
 			h++;
 			if (h < GFS2_MAX_META_HEIGHT)
-				memset(&mp->mp_list[h], 0,
-				       (GFS2_MAX_META_HEIGHT - h) *
-				       sizeof(mp->mp_list[0]));
+				mp->mp_list[h] = 0;
 			return true;
 		}
-
-		if (mp->mp_list[h] < ptrs)
-			mp->mp_list[h]++;
-		else
-			return false; /* no more pointers in this buffer */
+		ptr++;
 	}
+	return false;
 }
 
 enum dealloc_states {
@@ -1286,49 +1287,126 @@ enum dealloc_states {
 	DEALLOC_DONE = 3,       /* process complete */
 };
 
-static bool mp_eq_to_hgt(struct metapath *mp, __u16 *nbof, unsigned int h)
+static inline void
+metapointer_range(struct metapath *mp, int height,
+		  __u16 *start_list, unsigned int start_aligned,
+		  __u16 *end_list, unsigned int end_aligned,
+		  __be64 **start, __be64 **end)
 {
-	if (memcmp(mp->mp_list, nbof, h * sizeof(mp->mp_list[0])))
-		return false;
-	return true;
+	struct buffer_head *bh = mp->mp_bh[height];
+	__be64 *first;
+
+	first = metaptr1(height, mp);
+	*start = first;
+	if (mp_eq_to_hgt(mp, start_list, height)) {
+		bool keep_start = height < start_aligned;
+		*start = first + start_list[height] + keep_start;
+	}
+	*end = (__be64 *)(bh->b_data + bh->b_size);
+	if (end_list && mp_eq_to_hgt(mp, end_list, height)) {
+		bool keep_end = height < end_aligned;
+		*end = first + end_list[height] + keep_end;
+	}
+}
+
+static inline bool walk_done(struct gfs2_sbd *sdp,
+			     struct metapath *mp, int height,
+			     __u16 *end_list, unsigned int end_aligned)
+{
+	__u16 end;
+
+	if (end_list) {
+		bool keep_end = height < end_aligned;
+		if (!mp_eq_to_hgt(mp, end_list, height))
+			return false;
+		end = end_list[height] + keep_end;
+	} else
+		end = (height > 0) ? sdp->sd_inptrs : sdp->sd_diptrs;
+	return mp->mp_list[height] >= end;
 }
 
 /**
- * trunc_dealloc - truncate a file down to a desired size
+ * punch_hole - deallocate blocks in a file
  * @ip: inode to truncate
- * @newsize: The desired size of the file
+ * @offset: the start of the hole
+ * @length: the size of the hole (or 0 for truncate)
+ *
+ * Punch a hole into a file or truncate a file at a given position.  This
+ * function operates in whole blocks (@offset and @length are rounded
+ * accordingly); partially filled blocks must be cleared otherwise.
  *
- * This function truncates a file to newsize. It works from the
- * bottom up, and from the right to the left. In other words, it strips off
- * the highest layer (data) before stripping any of the metadata. Doing it
- * this way is best in case the operation is interrupted by power failure, etc.
- * The dinode is rewritten in every transaction to guarantee integrity.
+ * This function works from the bottom up, and from the right to the left. In
+ * other words, it strips off the highest layer (data) before stripping any of
+ * the metadata. Doing it this way is best in case the operation is interrupted
+ * by power failure, etc.  The dinode is rewritten in every transaction to
+ * guarantee integrity.
  */
-static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
+static int punch_hole(struct gfs2_inode *ip, u64 offset, u64 length)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
-	struct metapath mp;
+	struct metapath mp = {};
 	struct buffer_head *dibh, *bh;
 	struct gfs2_holder rd_gh;
-	u64 lblock;
-	__u16 nbof[GFS2_MAX_META_HEIGHT]; /* new beginning of truncation */
+	unsigned int bsize_shift = sdp->sd_sb.sb_bsize_shift;
+	u64 lblock = (offset + (1 << bsize_shift) - 1) >> bsize_shift;
+	__u16 start_list[GFS2_MAX_META_HEIGHT];
+	__u16 __end_list[GFS2_MAX_META_HEIGHT], *end_list = NULL;
+	unsigned int start_aligned, uninitialized_var(end_aligned);
 	unsigned int strip_h = ip->i_height - 1;
 	u32 btotal = 0;
 	int ret, state;
 	int mp_h; /* metapath buffers are read in to this height */
-	sector_t last_ra = 0;
 	u64 prev_bnr = 0;
-	bool preserve1; /* need to preserve the first meta pointer? */
+	__be64 *start, *end;
 
-	if (!newsize)
-		lblock = 0;
-	else
-		lblock = (newsize - 1) >> sdp->sd_sb.sb_bsize_shift;
+	/*
+	 * The start position of the hole is defined by lblock, start_list, and
+	 * start_aligned.  The end position of the hole is defined by lend,
+	 * end_list, and end_aligned.
+	 *
+	 * start_aligned and end_aligned define down to which height the start
+	 * and end positions are aligned to the metadata tree (i.e., the
+	 * position is a multiple of the metadata granularity at the height
+	 * above).  This determines at which heights additional meta pointers
+	 * needs to be preserved for the remaining data.
+	 */
+
+	if (length) {
+		u64 maxsize = sdp->sd_heightsize[ip->i_height];
+		u64 end_offset = offset + length;
+		u64 lend;
+
+		/*
+		 * Clip the end at the maximum file size for the given height:
+		 * that's how far the metadata goes; files bigger than that
+		 * will have additional layers of indirection.
+		 */
+		if (end_offset > maxsize)
+			end_offset = maxsize;
+		lend = end_offset >> bsize_shift;
+
+		if (lblock >= lend)
+			return 0;
+
+		find_metapath(sdp, lend, &mp, ip->i_height);
+		end_list = __end_list;
+		memcpy(end_list, mp.mp_list, sizeof(mp.mp_list));
+
+		for (mp_h = ip->i_height - 1; mp_h > 0; mp_h--) {
+			if (end_list[mp_h])
+				break;
+		}
+		end_aligned = mp_h;
+	}
 
-	memset(&mp, 0, sizeof(mp));
 	find_metapath(sdp, lblock, &mp, ip->i_height);
+	memcpy(start_list, mp.mp_list, sizeof(start_list));
 
-	memcpy(&nbof, &mp.mp_list, sizeof(nbof));
+	for (mp_h = ip->i_height - 1; mp_h > 0; mp_h--) {
+		if (start_list[mp_h])
+			break;
+	}
+	start_aligned = mp_h;
 
 	ret = gfs2_meta_inode_buffer(ip, &dibh);
 	if (ret)
@@ -1336,7 +1414,17 @@ static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
 
 	mp.mp_bh[0] = dibh;
 	ret = lookup_metapath(ip, &mp);
-	if (ret == ip->i_height)
+	if (ret)
+		goto out_metapath;
+
+	/* issue read-ahead on metadata */
+	for (mp_h = 0; mp_h < mp.mp_aheight - 1; mp_h++) {
+		metapointer_range(&mp, mp_h, start_list, start_aligned,
+				  end_list, end_aligned, &start, &end);
+		gfs2_metapath_ra(ip->i_gl, start, end);
+	}
+
+	if (mp.mp_aheight == ip->i_height)
 		state = DEALLOC_MP_FULL; /* We have a complete metapath */
 	else
 		state = DEALLOC_FILL_MP; /* deal with partial metapath */
@@ -1357,20 +1445,6 @@ static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
 		/* Truncate a full metapath at the given strip height.
 		 * Note that strip_h == mp_h in order to be in this state. */
 		case DEALLOC_MP_FULL:
-			if (mp_h > 0) { /* issue read-ahead on metadata */
-				__be64 *top;
-
-				bh = mp.mp_bh[mp_h - 1];
-				if (bh->b_blocknr != last_ra) {
-					last_ra = bh->b_blocknr;
-					top = metaptr1(mp_h - 1, &mp);
-					gfs2_metapath_ra(ip->i_gl, bh, top);
-				}
-			}
-			/* If we're truncating to a non-zero size and the mp is
-			   at the beginning of file for the strip height, we
-			   need to preserve the first metadata pointer. */
-			preserve1 = (newsize && mp_eq_to_hgt(&mp, nbof, mp_h));
 			bh = mp.mp_bh[mp_h];
 			gfs2_assert_withdraw(sdp, bh);
 			if (gfs2_assert_withdraw(sdp,
@@ -1382,8 +1456,28 @@ static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
 				       prev_bnr, ip->i_height, strip_h, mp_h);
 			}
 			prev_bnr = bh->b_blocknr;
-			ret = sweep_bh_for_rgrps(ip, &rd_gh, &mp, &btotal,
-						 mp_h, preserve1);
+
+			if (gfs2_metatype_check(sdp, bh,
+						(mp_h ? GFS2_METATYPE_IN :
+							GFS2_METATYPE_DI))) {
+				ret = -EIO;
+				goto out;
+			}
+
+			/*
+			 * Below, passing end_aligned as 0 gives us the
+			 * metapointer range excluding the end point: the end
+			 * point is the first metapath we must not deallocate!
+			 */
+
+			metapointer_range(&mp, mp_h, start_list, start_aligned,
+					  end_list, 0 /* end_aligned */,
+					  &start, &end);
+			ret = sweep_bh_for_rgrps(ip, &rd_gh, mp.mp_bh[mp_h],
+						 start, end,
+						 mp_h != ip->i_height - 1,
+						 &btotal);
+
 			/* If we hit an error or just swept dinode buffer,
 			   just exit. */
 			if (ret || !mp_h) {
@@ -1407,20 +1501,20 @@ static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
 			   stripping the previous level of metadata. */
 			if (mp_h == 0) {
 				strip_h--;
-				memcpy(&mp.mp_list, &nbof, sizeof(nbof));
+				memcpy(mp.mp_list, start_list, sizeof(start_list));
 				mp_h = strip_h;
 				state = DEALLOC_FILL_MP;
 				break;
 			}
 			mp.mp_list[mp_h] = 0;
 			mp_h--; /* search one metadata height down */
-			if (mp.mp_list[mp_h] >= hptrs(sdp, mp_h) - 1)
-				break; /* loop around in the same state */
 			mp.mp_list[mp_h]++;
+			if (walk_done(sdp, &mp, mp_h, end_list, end_aligned))
+				break;
 			/* Here we've found a part of the metapath that is not
 			 * allocated. We need to search at that height for the
 			 * next non-null pointer. */
-			if (find_nonnull_ptr(sdp, &mp, mp_h)) {
+			if (find_nonnull_ptr(sdp, &mp, mp_h, end_list, end_aligned)) {
 				state = DEALLOC_FILL_MP;
 				mp_h++;
 			}
@@ -1435,18 +1529,29 @@ static int trunc_dealloc(struct gfs2_inode *ip, u64 newsize)
 			if (ret < 0)
 				goto out;
 
+			/* issue read-ahead on metadata */
+			if (mp.mp_aheight > 1) {
+				for (; ret > 1; ret--) {
+					metapointer_range(&mp, mp.mp_aheight - ret,
+							  start_list, start_aligned,
+							  end_list, end_aligned,
+							  &start, &end);
+					gfs2_metapath_ra(ip->i_gl, start, end);
+				}
+			}
+
 			/* If buffers found for the entire strip height */
-			if ((ret == ip->i_height) && (mp_h == strip_h)) {
+			if (mp.mp_aheight - 1 == strip_h) {
 				state = DEALLOC_MP_FULL;
 				break;
 			}
-			if (ret < ip->i_height) /* We have a partial height */
-				mp_h = ret - 1;
+			if (mp.mp_aheight < ip->i_height) /* We have a partial height */
+				mp_h = mp.mp_aheight - 1;
 
 			/* If we find a non-null block pointer, crawl a bit
 			   higher up in the metapath and try again, otherwise
 			   we need to look lower for a new starting point. */
-			if (find_nonnull_ptr(sdp, &mp, mp_h))
+			if (find_nonnull_ptr(sdp, &mp, mp_h, end_list, end_aligned))
 				mp_h++;
 			else
 				state = DEALLOC_MP_LOWER;
@@ -1524,7 +1629,6 @@ out:
 /**
  * do_shrink - make a file smaller
  * @inode: the inode
- * @oldsize: the current inode size
  * @newsize: the size to make the file
  *
  * Called with an exclusive lock on @inode. The @size must
@@ -1533,18 +1637,18 @@ out:
  * Returns: errno
  */
 
-static int do_shrink(struct inode *inode, u64 oldsize, u64 newsize)
+static int do_shrink(struct inode *inode, u64 newsize)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 	int error;
 
-	error = trunc_start(inode, oldsize, newsize);
+	error = trunc_start(inode, newsize);
 	if (error < 0)
 		return error;
 	if (gfs2_is_stuffed(ip))
 		return 0;
 
-	error = trunc_dealloc(ip, newsize);
+	error = punch_hole(ip, newsize, 0);
 	if (error == 0)
 		error = trunc_end(ip);
 
@@ -1553,10 +1657,9 @@ static int do_shrink(struct inode *inode, u64 oldsize, u64 newsize)
 
 void gfs2_trim_blocks(struct inode *inode)
 {
-	u64 size = inode->i_size;
 	int ret;
 
-	ret = do_shrink(inode, size, size);
+	ret = do_shrink(inode, inode->i_size);
 	WARN_ON(ret != 0);
 }
 
@@ -1650,7 +1753,6 @@ int gfs2_setattr_size(struct inode *inode, u64 newsize)
 {
 	struct gfs2_inode *ip = GFS2_I(inode);
 	int ret;
-	u64 oldsize;
 
 	BUG_ON(!S_ISREG(inode->i_mode));
 
@@ -1664,13 +1766,12 @@ int gfs2_setattr_size(struct inode *inode, u64 newsize)
 	if (ret)
 		goto out;
 
-	oldsize = inode->i_size;
-	if (newsize >= oldsize) {
+	if (newsize >= inode->i_size) {
 		ret = do_grow(inode, newsize);
 		goto out;
 	}
 
-	ret = do_shrink(inode, oldsize, newsize);
+	ret = do_shrink(inode, newsize);
 out:
 	gfs2_rsqa_delete(ip, NULL);
 	return ret;
@@ -1679,7 +1780,7 @@ out:
 int gfs2_truncatei_resume(struct gfs2_inode *ip)
 {
 	int error;
-	error = trunc_dealloc(ip, i_size_read(&ip->i_inode));
+	error = punch_hole(ip, i_size_read(&ip->i_inode), 0);
 	if (!error)
 		error = trunc_end(ip);
 	return error;
@@ -1687,7 +1788,7 @@ int gfs2_truncatei_resume(struct gfs2_inode *ip)
 
 int gfs2_file_dealloc(struct gfs2_inode *ip)
 {
-	return trunc_dealloc(ip, 0);
+	return punch_hole(ip, 0, 0);
 }
 
 /**
@@ -1855,3 +1956,123 @@ int gfs2_write_alloc_required(struct gfs2_inode *ip, u64 offset,
 	return 0;
 }
 
+static int stuffed_zero_range(struct inode *inode, loff_t offset, loff_t length)
+{
+	struct gfs2_inode *ip = GFS2_I(inode);
+	struct buffer_head *dibh;
+	int error;
+
+	if (offset >= inode->i_size)
+		return 0;
+	if (offset + length > inode->i_size)
+		length = inode->i_size - offset;
+
+	error = gfs2_meta_inode_buffer(ip, &dibh);
+	if (error)
+		return error;
+	gfs2_trans_add_meta(ip->i_gl, dibh);
+	memset(dibh->b_data + sizeof(struct gfs2_dinode) + offset, 0,
+	       length);
+	brelse(dibh);
+	return 0;
+}
+
+static int gfs2_journaled_truncate_range(struct inode *inode, loff_t offset,
+					 loff_t length)
+{
+	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	loff_t max_chunk = GFS2_JTRUNC_REVOKES * sdp->sd_vfs->s_blocksize;
+	int error;
+
+	while (length) {
+		struct gfs2_trans *tr;
+		loff_t chunk;
+		unsigned int offs;
+
+		chunk = length;
+		if (chunk > max_chunk)
+			chunk = max_chunk;
+
+		offs = offset & ~PAGE_MASK;
+		if (offs && chunk > PAGE_SIZE)
+			chunk = offs + ((chunk - offs) & PAGE_MASK);
+
+		truncate_pagecache_range(inode, offset, chunk);
+		offset += chunk;
+		length -= chunk;
+
+		tr = current->journal_info;
+		if (!test_bit(TR_TOUCHED, &tr->tr_flags))
+			continue;
+
+		gfs2_trans_end(sdp);
+		error = gfs2_trans_begin(sdp, RES_DINODE, GFS2_JTRUNC_REVOKES);
+		if (error)
+			return error;
+	}
+	return 0;
+}
+
+int __gfs2_punch_hole(struct file *file, loff_t offset, loff_t length)
+{
+	struct inode *inode = file_inode(file);
+	struct gfs2_inode *ip = GFS2_I(inode);
+	struct gfs2_sbd *sdp = GFS2_SB(inode);
+	int error;
+
+	if (gfs2_is_jdata(ip))
+		error = gfs2_trans_begin(sdp, RES_DINODE + 2 * RES_JDATA,
+					 GFS2_JTRUNC_REVOKES);
+	else
+		error = gfs2_trans_begin(sdp, RES_DINODE, 0);
+	if (error)
+		return error;
+
+	if (gfs2_is_stuffed(ip)) {
+		error = stuffed_zero_range(inode, offset, length);
+		if (error)
+			goto out;
+	} else {
+		unsigned int start_off, end_off, blocksize;
+
+		blocksize = i_blocksize(inode);
+		start_off = offset & (blocksize - 1);
+		end_off = (offset + length) & (blocksize - 1);
+		if (start_off) {
+			unsigned int len = length;
+			if (length > blocksize - start_off)
+				len = blocksize - start_off;
+			error = gfs2_block_zero_range(inode, offset, len);
+			if (error)
+				goto out;
+			if (start_off + length < blocksize)
+				end_off = 0;
+		}
+		if (end_off) {
+			error = gfs2_block_zero_range(inode,
+				offset + length - end_off, end_off);
+			if (error)
+				goto out;
+		}
+	}
+
+	if (gfs2_is_jdata(ip)) {
+		BUG_ON(!current->journal_info);
+		gfs2_journaled_truncate_range(inode, offset, length);
+	} else
+		truncate_pagecache_range(inode, offset, offset + length - 1);
+
+	file_update_time(file);
+	mark_inode_dirty(inode);
+
+	if (current->journal_info)
+		gfs2_trans_end(sdp);
+
+	if (!gfs2_is_stuffed(ip))
+		error = punch_hole(ip, offset, length);
+
+out:
+	if (current->journal_info)
+		gfs2_trans_end(sdp);
+	return error;
+}
diff --git a/fs/gfs2/bmap.h b/fs/gfs2/bmap.h
index 443cc182cf18..c3402fe00653 100644
--- a/fs/gfs2/bmap.h
+++ b/fs/gfs2/bmap.h
@@ -61,5 +61,6 @@ extern int gfs2_write_alloc_required(struct gfs2_inode *ip, u64 offset,
 				     unsigned int len);
 extern int gfs2_map_journal_extents(struct gfs2_sbd *sdp, struct gfs2_jdesc *jd);
 extern void gfs2_free_journal_extents(struct gfs2_jdesc *jd);
+extern int __gfs2_punch_hole(struct file *file, loff_t offset, loff_t length);
 
 #endif /* __BMAP_DOT_H__ */
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 58705ef8643a..bd60dc682676 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -924,7 +924,7 @@ static long gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t le
 	struct gfs2_holder gh;
 	int ret;
 
-	if (mode & ~FALLOC_FL_KEEP_SIZE)
+	if (mode & ~(FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE))
 		return -EOPNOTSUPP;
 	/* fallocate is needed by gfs2_grow to reserve space in the rindex */
 	if (gfs2_is_jdata(ip) && inode != sdp->sd_rindex)
@@ -948,13 +948,18 @@ static long gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t le
 	if (ret)
 		goto out_unlock;
 
-	ret = gfs2_rsqa_alloc(ip);
-	if (ret)
-		goto out_putw;
+	if (mode & FALLOC_FL_PUNCH_HOLE) {
+		ret = __gfs2_punch_hole(file, offset, len);
+	} else {
+		ret = gfs2_rsqa_alloc(ip);
+		if (ret)
+			goto out_putw;
 
-	ret = __gfs2_fallocate(file, mode, offset, len);
-	if (ret)
-		gfs2_rs_deltree(&ip->i_res);
+		ret = __gfs2_fallocate(file, mode, offset, len);
+
+		if (ret)
+			gfs2_rs_deltree(&ip->i_res);
+	}
 
 out_putw:
 	put_write_access(inode);
diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index a3711f543405..ad55eb86a250 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -1065,15 +1065,15 @@ static int fill_super(struct super_block *sb, struct gfs2_args *args, int silent
 	sdp->sd_args = *args;
 
 	if (sdp->sd_args.ar_spectator) {
-                sb->s_flags |= MS_RDONLY;
+                sb->s_flags |= SB_RDONLY;
 		set_bit(SDF_RORECOVERY, &sdp->sd_flags);
 	}
 	if (sdp->sd_args.ar_posix_acl)
-		sb->s_flags |= MS_POSIXACL;
+		sb->s_flags |= SB_POSIXACL;
 	if (sdp->sd_args.ar_nobarrier)
 		set_bit(SDF_NOBARRIERS, &sdp->sd_flags);
 
-	sb->s_flags |= MS_NOSEC;
+	sb->s_flags |= SB_NOSEC;
 	sb->s_magic = GFS2_MAGIC;
 	sb->s_op = &gfs2_super_ops;
 	sb->s_d_op = &gfs2_dops;
@@ -1257,7 +1257,7 @@ static struct dentry *gfs2_mount(struct file_system_type *fs_type, int flags,
 	struct gfs2_args args;
 	struct gfs2_sbd *sdp;
 
-	if (!(flags & MS_RDONLY))
+	if (!(flags & SB_RDONLY))
 		mode |= FMODE_WRITE;
 
 	bdev = blkdev_get_by_path(dev_name, mode, fs_type);
@@ -1313,15 +1313,15 @@ static struct dentry *gfs2_mount(struct file_system_type *fs_type, int flags,
 
 	if (s->s_root) {
 		error = -EBUSY;
-		if ((flags ^ s->s_flags) & MS_RDONLY)
+		if ((flags ^ s->s_flags) & SB_RDONLY)
 			goto error_super;
 	} else {
 		snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev);
 		sb_set_blocksize(s, block_size(bdev));
-		error = fill_super(s, &args, flags & MS_SILENT ? 1 : 0);
+		error = fill_super(s, &args, flags & SB_SILENT ? 1 : 0);
 		if (error)
 			goto error_super;
-		s->s_flags |= MS_ACTIVE;
+		s->s_flags |= SB_ACTIVE;
 		bdev->bd_super = s;
 	}
 
@@ -1365,7 +1365,7 @@ static struct dentry *gfs2_mount_meta(struct file_system_type *fs_type,
 		pr_warn("gfs2 mount does not exist\n");
 		return ERR_CAST(s);
 	}
-	if ((flags ^ s->s_flags) & MS_RDONLY) {
+	if ((flags ^ s->s_flags) & SB_RDONLY) {
 		deactivate_locked_super(s);
 		return ERR_PTR(-EBUSY);
 	}
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index e8aba6fa1472..6dea72f49316 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -490,6 +490,13 @@ void gfs2_rgrp_verify(struct gfs2_rgrpd *rgd)
  * @blk: The data block number
  * @exact: True if this needs to be an exact match
  *
+ * The @exact argument should be set to true by most callers. The exception
+ * is when we need to match blocks which are not represented by the rgrp
+ * bitmap, but which are part of the rgrp (i.e. padding blocks) which are
+ * there for alignment purposes. Another way of looking at it is that @exact
+ * matches only valid data/metadata blocks, but with @exact false, it will
+ * match any block within the extent of the rgrp.
+ *
  * Returns: The resource group, or NULL if not found
  */
 
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 9cb5c9a97d69..d81d46e19726 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -1256,10 +1256,10 @@ static int gfs2_remount_fs(struct super_block *sb, int *flags, char *data)
 		return -EINVAL;
 
 	if (sdp->sd_args.ar_spectator)
-		*flags |= MS_RDONLY;
+		*flags |= SB_RDONLY;
 
-	if ((sb->s_flags ^ *flags) & MS_RDONLY) {
-		if (*flags & MS_RDONLY)
+	if ((sb->s_flags ^ *flags) & SB_RDONLY) {
+		if (*flags & SB_RDONLY)
 			error = gfs2_make_fs_ro(sdp);
 		else
 			error = gfs2_make_fs_rw(sdp);
@@ -1269,9 +1269,9 @@ static int gfs2_remount_fs(struct super_block *sb, int *flags, char *data)
 
 	sdp->sd_args = args;
 	if (sdp->sd_args.ar_posix_acl)
-		sb->s_flags |= MS_POSIXACL;
+		sb->s_flags |= SB_POSIXACL;
 	else
-		sb->s_flags &= ~MS_POSIXACL;
+		sb->s_flags &= ~SB_POSIXACL;
 	if (sdp->sd_args.ar_nobarrier)
 		set_bit(SDF_NOBARRIERS, &sdp->sd_flags);
 	else
diff --git a/fs/gfs2/trans.c b/fs/gfs2/trans.c
index a85ca8b2c9ba..b95ebd166cac 100644
--- a/fs/gfs2/trans.c
+++ b/fs/gfs2/trans.c
@@ -92,7 +92,6 @@ void gfs2_trans_end(struct gfs2_sbd *sdp)
 	s64 nbuf;
 	int alloced = test_bit(TR_ALLOCED, &tr->tr_flags);
 
-	BUG_ON(!tr);
 	current->journal_info = NULL;
 
 	if (!test_bit(TR_TOUCHED, &tr->tr_flags)) {
@@ -117,7 +116,7 @@ void gfs2_trans_end(struct gfs2_sbd *sdp)
 		kfree(tr);
 	up_read(&sdp->sd_log_flush_lock);
 
-	if (sdp->sd_vfs->s_flags & MS_SYNCHRONOUS)
+	if (sdp->sd_vfs->s_flags & SB_SYNCHRONOUS)
 		gfs2_log_flush(sdp, NULL, NORMAL_FLUSH);
 	if (alloced)
 		sb_end_intwrite(sdp->sd_vfs);