ext2 reservations

Val's cross-port of the ext3 reservations code into ext2.

[mbligh@mbligh.org: Small type error for printk
[akpm@linux-foundation.org: fix types, sync with ext3]
[mbligh@mbligh.org: Bring ext2 reservations code in line with latest ext3]
[akpm@linux-foundation.org: kill noisy printk]
[akpm@linux-foundation.org: remember to dirty the gdp's block]
[akpm@linux-foundation.org: cross-port the missed 5dea5176]
[akpm@linux-foundation.org: cross-port e6022603]
[akpm@linux-foundation.org: Port the omitted 08fb306f]
[akpm@linux-foundation.org: Backport the missed 20acaa18]
[akpm@linux-foundation.org: fixes]
[cmm@us.ibm.com: fix reservation extension]
[bunk@stusta.de: make ext2_get_blocks() static]
[hugh@veritas.com: fix hang]
[hugh@veritas.com: ext2_new_blocks should reset the reservation window size]
[hugh@veritas.com: ext2 balloc: fix off-by-one against rsv_end]
[hugh@veritas.com: grp_goal 0 is a genuine goal (unlike -1), so ext2_try_to_allocate_with_rsv should treat it as such]
[hugh@veritas.com: rbtree usage cleanup]
[pbadari@us.ibm.com: Fix for ext2 reservation]
[bunk@kernel.org: remove fs/ext2/balloc.c:reserve_blocks()]
[hugh@veritas.com: ext2 balloc: use io_error label]
Cc: "Martin J. Bligh" <mbligh@mbligh.org>
Cc: Valerie Henson <val_henson@linux.intel.com>
Cc: Mingming Cao <cmm@us.ibm.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NHugh Dickins <hugh@veritas.com>
Signed-off-by: NBadari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

fs/ext2/balloc.c

@@ -133,86 +133,332 @@ read_block_bitmap(struct super_block *sb, unsigned int block_group)
 	return NULL;
 }
 
-/*
- * Set sb->s_dirt here because the superblock was "logically" altered.  We
- * need to recalculate its free blocks count and flush it out.
- */
-static int reserve_blocks(struct super_block *sb, int count)
+static void release_blocks(struct super_block *sb, int count)
 {
+	if (count) {
+		struct ext2_sb_info *sbi = EXT2_SB(sb);
+
+		percpu_counter_add(&sbi->s_freeblocks_counter, count);
+		sb->s_dirt = 1;
+	}
+}
+
+static void group_adjust_blocks(struct super_block *sb, int group_no,
+	struct ext2_group_desc *desc, struct buffer_head *bh, int count)
+{
+	if (count) {
 		struct ext2_sb_info *sbi = EXT2_SB(sb);
-	struct ext2_super_block *es = sbi->s_es;
 		unsigned free_blocks;
-	unsigned root_blocks;
 
-	free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
-	root_blocks = le32_to_cpu(es->s_r_blocks_count);
+		spin_lock(sb_bgl_lock(sbi, group_no));
+		free_blocks = le16_to_cpu(desc->bg_free_blocks_count);
+		desc->bg_free_blocks_count = cpu_to_le16(free_blocks + count);
+		spin_unlock(sb_bgl_lock(sbi, group_no));
+		sb->s_dirt = 1;
+		mark_buffer_dirty(bh);
+	}
+}
 
-	if (free_blocks < count)
-		count = free_blocks;
+/*
+ * The reservation window structure operations
+ * --------------------------------------------
+ * Operations include:
+ * dump, find, add, remove, is_empty, find_next_reservable_window, etc.
+ *
+ * We use a red-black tree to represent per-filesystem reservation
+ * windows.
+ *
+ */
 
-	if (free_blocks < root_blocks + count && !capable(CAP_SYS_RESOURCE) &&
-	    sbi->s_resuid != current->fsuid &&
-	    (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))) {
-		/*
-		 * We are too close to reserve and we are not privileged.
-		 * Can we allocate anything at all?
+/**
+ * __rsv_window_dump() -- Dump the filesystem block allocation reservation map
+ * @rb_root:		root of per-filesystem reservation rb tree
+ * @verbose:		verbose mode
+ * @fn:			function which wishes to dump the reservation map
+ *
+ * If verbose is turned on, it will print the whole block reservation
+ * windows(start, end). Otherwise, it will only print out the "bad" windows,
+ * those windows that overlap with their immediate neighbors.
  */
-		if (free_blocks > root_blocks)
-			count = free_blocks - root_blocks;
-		else
-			return 0;
+#if 1
+static void __rsv_window_dump(struct rb_root *root, int verbose,
+			      const char *fn)
+{
+	struct rb_node *n;
+	struct ext2_reserve_window_node *rsv, *prev;
+	int bad;
+
+restart:
+	n = rb_first(root);
+	bad = 0;
+	prev = NULL;
+
+	printk("Block Allocation Reservation Windows Map (%s):\n", fn);
+	while (n) {
+		rsv = rb_entry(n, struct ext2_reserve_window_node, rsv_node);
+		if (verbose)
+			printk("reservation window 0x%p "
+				"start: %lu, end: %lu\n",
+				rsv, rsv->rsv_start, rsv->rsv_end);
+		if (rsv->rsv_start && rsv->rsv_start >= rsv->rsv_end) {
+			printk("Bad reservation %p (start >= end)\n",
+			       rsv);
+			bad = 1;
+		}
+		if (prev && prev->rsv_end >= rsv->rsv_start) {
+			printk("Bad reservation %p (prev->end >= start)\n",
+			       rsv);
+			bad = 1;
+		}
+		if (bad) {
+			if (!verbose) {
+				printk("Restarting reservation walk in verbose mode\n");
+				verbose = 1;
+				goto restart;
 			}
+		}
+		n = rb_next(n);
+		prev = rsv;
+	}
+	printk("Window map complete.\n");
+	if (bad)
+		BUG();
+}
+#define rsv_window_dump(root, verbose) \
+	__rsv_window_dump((root), (verbose), __FUNCTION__)
+#else
+#define rsv_window_dump(root, verbose) do {} while (0)
+#endif
 
-	percpu_counter_sub(&sbi->s_freeblocks_counter, count);
-	sb->s_dirt = 1;
-	return count;
+/**
+ * goal_in_my_reservation()
+ * @rsv:		inode's reservation window
+ * @grp_goal:		given goal block relative to the allocation block group
+ * @group:		the current allocation block group
+ * @sb:			filesystem super block
+ *
+ * Test if the given goal block (group relative) is within the file's
+ * own block reservation window range.
+ *
+ * If the reservation window is outside the goal allocation group, return 0;
+ * grp_goal (given goal block) could be -1, which means no specific
+ * goal block. In this case, always return 1.
+ * If the goal block is within the reservation window, return 1;
+ * otherwise, return 0;
+ */
+static int
+goal_in_my_reservation(struct ext2_reserve_window *rsv, ext2_grpblk_t grp_goal,
+			unsigned int group, struct super_block * sb)
+{
+	ext2_fsblk_t group_first_block, group_last_block;
+
+	group_first_block = ext2_group_first_block_no(sb, group);
+	group_last_block = group_first_block + EXT2_BLOCKS_PER_GROUP(sb) - 1;
+
+	if ((rsv->_rsv_start > group_last_block) ||
+	    (rsv->_rsv_end < group_first_block))
+		return 0;
+	if ((grp_goal >= 0) && ((grp_goal + group_first_block < rsv->_rsv_start)
+		|| (grp_goal + group_first_block > rsv->_rsv_end)))
+		return 0;
+	return 1;
 }
 
-static void release_blocks(struct super_block *sb, int count)
+/**
+ * search_reserve_window()
+ * @rb_root:		root of reservation tree
+ * @goal:		target allocation block
+ *
+ * Find the reserved window which includes the goal, or the previous one
+ * if the goal is not in any window.
+ * Returns NULL if there are no windows or if all windows start after the goal.
+ */
+static struct ext2_reserve_window_node *
+search_reserve_window(struct rb_root *root, ext2_fsblk_t goal)
 {
-	if (count) {
-		struct ext2_sb_info *sbi = EXT2_SB(sb);
+	struct rb_node *n = root->rb_node;
+	struct ext2_reserve_window_node *rsv;
 
-		percpu_counter_add(&sbi->s_freeblocks_counter, count);
-		sb->s_dirt = 1;
+	if (!n)
+		return NULL;
+
+	do {
+		rsv = rb_entry(n, struct ext2_reserve_window_node, rsv_node);
+
+		if (goal < rsv->rsv_start)
+			n = n->rb_left;
+		else if (goal > rsv->rsv_end)
+			n = n->rb_right;
+		else
+			return rsv;
+	} while (n);
+	/*
+	 * We've fallen off the end of the tree: the goal wasn't inside
+	 * any particular node.  OK, the previous node must be to one
+	 * side of the interval containing the goal.  If it's the RHS,
+	 * we need to back up one.
+	 */
+	if (rsv->rsv_start > goal) {
+		n = rb_prev(&rsv->rsv_node);
+		rsv = rb_entry(n, struct ext2_reserve_window_node, rsv_node);
 	}
+	return rsv;
 }
 
-static int group_reserve_blocks(struct ext2_sb_info *sbi, int group_no,
-	struct ext2_group_desc *desc, struct buffer_head *bh, int count)
+/*
+ * ext2_rsv_window_add() -- Insert a window to the block reservation rb tree.
+ * @sb:			super block
+ * @rsv:		reservation window to add
+ *
+ * Must be called with rsv_lock held.
+ */
+void ext2_rsv_window_add(struct super_block *sb,
+		    struct ext2_reserve_window_node *rsv)
 {
-	unsigned free_blocks;
+	struct rb_root *root = &EXT2_SB(sb)->s_rsv_window_root;
+	struct rb_node *node = &rsv->rsv_node;
+	ext2_fsblk_t start = rsv->rsv_start;
+
+	struct rb_node ** p = &root->rb_node;
+	struct rb_node * parent = NULL;
+	struct ext2_reserve_window_node *this;
+
+	while (*p)
+	{
+		parent = *p;
+		this = rb_entry(parent, struct ext2_reserve_window_node, rsv_node);
+
+		if (start < this->rsv_start)
+			p = &(*p)->rb_left;
+		else if (start > this->rsv_end)
+			p = &(*p)->rb_right;
+		else {
+			rsv_window_dump(root, 1);
+			BUG();
+		}
+	}
 
-	if (!desc->bg_free_blocks_count)
-		return 0;
+	rb_link_node(node, parent, p);
+	rb_insert_color(node, root);
+}
 
-	spin_lock(sb_bgl_lock(sbi, group_no));
-	free_blocks = le16_to_cpu(desc->bg_free_blocks_count);
-	if (free_blocks < count)
-		count = free_blocks;
-	desc->bg_free_blocks_count = cpu_to_le16(free_blocks - count);
-	spin_unlock(sb_bgl_lock(sbi, group_no));
-	mark_buffer_dirty(bh);
-	return count;
+/**
+ * rsv_window_remove() -- unlink a window from the reservation rb tree
+ * @sb:			super block
+ * @rsv:		reservation window to remove
+ *
+ * Mark the block reservation window as not allocated, and unlink it
+ * from the filesystem reservation window rb tree. Must be called with
+ * rsv_lock held.
+ */
+static void rsv_window_remove(struct super_block *sb,
+			      struct ext2_reserve_window_node *rsv)
+{
+	rsv->rsv_start = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+	rsv->rsv_end = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+	rsv->rsv_alloc_hit = 0;
+	rb_erase(&rsv->rsv_node, &EXT2_SB(sb)->s_rsv_window_root);
 }
 
-static void group_release_blocks(struct super_block *sb, int group_no,
-	struct ext2_group_desc *desc, struct buffer_head *bh, int count)
+/*
+ * rsv_is_empty() -- Check if the reservation window is allocated.
+ * @rsv:		given reservation window to check
+ *
+ * returns 1 if the end block is EXT2_RESERVE_WINDOW_NOT_ALLOCATED.
+ */
+static inline int rsv_is_empty(struct ext2_reserve_window *rsv)
 {
-	if (count) {
-		struct ext2_sb_info *sbi = EXT2_SB(sb);
-		unsigned free_blocks;
+	/* a valid reservation end block could not be 0 */
+	return (rsv->_rsv_end == EXT2_RESERVE_WINDOW_NOT_ALLOCATED);
+}
 
-		spin_lock(sb_bgl_lock(sbi, group_no));
-		free_blocks = le16_to_cpu(desc->bg_free_blocks_count);
-		desc->bg_free_blocks_count = cpu_to_le16(free_blocks + count);
-		spin_unlock(sb_bgl_lock(sbi, group_no));
-		sb->s_dirt = 1;
-		mark_buffer_dirty(bh);
+/**
+ * ext2_init_block_alloc_info()
+ * @inode:		file inode structure
+ *
+ * Allocate and initialize the  reservation window structure, and
+ * link the window to the ext2 inode structure at last
+ *
+ * The reservation window structure is only dynamically allocated
+ * and linked to ext2 inode the first time the open file
+ * needs a new block. So, before every ext2_new_block(s) call, for
+ * regular files, we should check whether the reservation window
+ * structure exists or not. In the latter case, this function is called.
+ * Fail to do so will result in block reservation being turned off for that
+ * open file.
+ *
+ * This function is called from ext2_get_blocks_handle(), also called
+ * when setting the reservation window size through ioctl before the file
+ * is open for write (needs block allocation).
+ *
+ * Needs truncate_mutex protection prior to calling this function.
+ */
+void ext2_init_block_alloc_info(struct inode *inode)
+{
+	struct ext2_inode_info *ei = EXT2_I(inode);
+	struct ext2_block_alloc_info *block_i = ei->i_block_alloc_info;
+	struct super_block *sb = inode->i_sb;
+
+	block_i = kmalloc(sizeof(*block_i), GFP_NOFS);
+	if (block_i) {
+		struct ext2_reserve_window_node *rsv = &block_i->rsv_window_node;
+
+		rsv->rsv_start = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+		rsv->rsv_end = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+
+	 	/*
+		 * if filesystem is mounted with NORESERVATION, the goal
+		 * reservation window size is set to zero to indicate
+		 * block reservation is off
+		 */
+		if (!test_opt(sb, RESERVATION))
+			rsv->rsv_goal_size = 0;
+		else
+			rsv->rsv_goal_size = EXT2_DEFAULT_RESERVE_BLOCKS;
+		rsv->rsv_alloc_hit = 0;
+		block_i->last_alloc_logical_block = 0;
+		block_i->last_alloc_physical_block = 0;
 	}
+	ei->i_block_alloc_info = block_i;
 }
 
-/* Free given blocks, update quota and i_blocks field */
+/**
+ * ext2_discard_reservation()
+ * @inode:		inode
+ *
+ * Discard(free) block reservation window on last file close, or truncate
+ * or at last iput().
+ *
+ * It is being called in three cases:
+ * 	ext2_release_file(): last writer closes the file
+ * 	ext2_clear_inode(): last iput(), when nobody links to this file.
+ * 	ext2_truncate(): when the block indirect map is about to change.
+ */
+void ext2_discard_reservation(struct inode *inode)
+{
+	struct ext2_inode_info *ei = EXT2_I(inode);
+	struct ext2_block_alloc_info *block_i = ei->i_block_alloc_info;
+	struct ext2_reserve_window_node *rsv;
+	spinlock_t *rsv_lock = &EXT2_SB(inode->i_sb)->s_rsv_window_lock;
+
+	if (!block_i)
+		return;
+
+	rsv = &block_i->rsv_window_node;
+	if (!rsv_is_empty(&rsv->rsv_window)) {
+		spin_lock(rsv_lock);
+		if (!rsv_is_empty(&rsv->rsv_window))
+			rsv_window_remove(inode->i_sb, rsv);
+		spin_unlock(rsv_lock);
+	}
+}
+
+/**
+ * ext2_free_blocks_sb() -- Free given blocks and update quota and i_blocks
+ * @inode:		inode
+ * @block:		start physcial block to free
+ * @count:		number of blocks to free
+ */
 void ext2_free_blocks (struct inode * inode, unsigned long block,
 		       unsigned long count)
 {
@@ -287,7 +533,7 @@ void ext2_free_blocks (struct inode * inode, unsigned long block,
 	if (sb->s_flags & MS_SYNCHRONOUS)
 		sync_dirty_buffer(bitmap_bh);
 
-	group_release_blocks(sb, block_group, desc, bh2, group_freed);
+	group_adjust_blocks(sb, block_group, desc, bh2, group_freed);
 	freed += group_freed;
 
 	if (overflow) {
@@ -301,16 +547,46 @@ void ext2_free_blocks (struct inode * inode, unsigned long block,
 	DQUOT_FREE_BLOCK(inode, freed);
 }
 
-static int grab_block(spinlock_t *lock, char *map, unsigned size, int goal)
+/**
+ * bitmap_search_next_usable_block()
+ * @start:		the starting block (group relative) of the search
+ * @bh:			bufferhead contains the block group bitmap
+ * @maxblocks:		the ending block (group relative) of the reservation
+ *
+ * The bitmap search --- search forward through the actual bitmap on disk until
+ * we find a bit free.
+ */
+static ext2_grpblk_t
+bitmap_search_next_usable_block(ext2_grpblk_t start, struct buffer_head *bh,
+					ext2_grpblk_t maxblocks)
+{
+	ext2_grpblk_t next;
+
+	next = ext2_find_next_zero_bit(bh->b_data, maxblocks, start);
+	if (next >= maxblocks)
+		return -1;
+	return next;
+}
+
+/**
+ * find_next_usable_block()
+ * @start:		the starting block (group relative) to find next
+ * 			allocatable block in bitmap.
+ * @bh:			bufferhead contains the block group bitmap
+ * @maxblocks:		the ending block (group relative) for the search
+ *
+ * Find an allocatable block in a bitmap.  We perform the "most
+ * appropriate allocation" algorithm of looking for a free block near
+ * the initial goal; then for a free byte somewhere in the bitmap;
+ * then for any free bit in the bitmap.
+ */
+static ext2_grpblk_t
+find_next_usable_block(int start, struct buffer_head *bh, int maxblocks)
 {
-	int k;
+	ext2_grpblk_t here, next;
 	char *p, *r;
 
-	if (!ext2_test_bit(goal, map))
-		goto got_it;
-
-repeat:
-	if (goal) {
+	if (start > 0) {
 		/*
 		 * The goal was occupied; search forward for a free 
 		 * block within the next XX blocks.
@@ -319,244 +595,807 @@ static int grab_block(spinlock_t *lock, char *map, unsigned size, int goal)
 		 * less than EXT2_BLOCKS_PER_GROUP. Aligning up to the
 		 * next 64-bit boundary is simple..
 		 */
-		k = (goal + 63) & ~63;
-		goal = ext2_find_next_zero_bit(map, k, goal);
-		if (goal < k)
-			goto got_it;
+		ext2_grpblk_t end_goal = (start + 63) & ~63;
+		if (end_goal > maxblocks)
+			end_goal = maxblocks;
+		here = ext2_find_next_zero_bit(bh->b_data, end_goal, start);
+		if (here < end_goal)
+			return here;
+		ext2_debug("Bit not found near goal\n");
+	}
+
+	here = start;
+	if (here < 0)
+		here = 0;
+
+	p = ((char *)bh->b_data) + (here >> 3);
+	r = memscan(p, 0, ((maxblocks + 7) >> 3) - (here >> 3));
+	next = (r - ((char *)bh->b_data)) << 3;
+
+	if (next < maxblocks && next >= here)
+		return next;
+
+	here = bitmap_search_next_usable_block(here, bh, maxblocks);
+	return here;
+}
+
+/*
+ * ext2_try_to_allocate()
+ * @sb:			superblock
+ * @handle:		handle to this transaction
+ * @group:		given allocation block group
+ * @bitmap_bh:		bufferhead holds the block bitmap
+ * @grp_goal:		given target block within the group
+ * @count:		target number of blocks to allocate
+ * @my_rsv:		reservation window
+ *
+ * Attempt to allocate blocks within a give range. Set the range of allocation
+ * first, then find the first free bit(s) from the bitmap (within the range),
+ * and at last, allocate the blocks by claiming the found free bit as allocated.
+ *
+ * To set the range of this allocation:
+ * 	if there is a reservation window, only try to allocate block(s)
+ * 	from the file's own reservation window;
+ * 	Otherwise, the allocation range starts from the give goal block,
+ * 	ends at the block group's last block.
+ *
+ * If we failed to allocate the desired block then we may end up crossing to a
+ * new bitmap.
+ */
+static int
+ext2_try_to_allocate(struct super_block *sb, int group,
+			struct buffer_head *bitmap_bh, ext2_grpblk_t grp_goal,
+			unsigned long *count,
+			struct ext2_reserve_window *my_rsv)
+{
+	ext2_fsblk_t group_first_block;
+       	ext2_grpblk_t start, end;
+	unsigned long num = 0;
+
+	/* we do allocation within the reservation window if we have a window */
+	if (my_rsv) {
+		group_first_block = ext2_group_first_block_no(sb, group);
+		if (my_rsv->_rsv_start >= group_first_block)
+			start = my_rsv->_rsv_start - group_first_block;
+		else
+			/* reservation window cross group boundary */
+			start = 0;
+		end = my_rsv->_rsv_end - group_first_block + 1;
+		if (end > EXT2_BLOCKS_PER_GROUP(sb))
+			/* reservation window crosses group boundary */
+			end = EXT2_BLOCKS_PER_GROUP(sb);
+		if ((start <= grp_goal) && (grp_goal < end))
+			start = grp_goal;
+		else
+			grp_goal = -1;
+	} else {
+		if (grp_goal > 0)
+			start = grp_goal;
+		else
+			start = 0;
+		end = EXT2_BLOCKS_PER_GROUP(sb);
+	}
+
+	BUG_ON(start > EXT2_BLOCKS_PER_GROUP(sb));
+
+repeat:
+	if (grp_goal < 0) {
+		grp_goal = find_next_usable_block(start, bitmap_bh, end);
+		if (grp_goal < 0)
+			goto fail_access;
+		if (!my_rsv) {
+			int i;
+
+			for (i = 0; i < 7 && grp_goal > start &&
+					!ext2_test_bit(grp_goal - 1,
+					     		bitmap_bh->b_data);
+			     		i++, grp_goal--)
+				;
+		}
+	}
+	start = grp_goal;
+
+	if (ext2_set_bit_atomic(sb_bgl_lock(EXT2_SB(sb), group), grp_goal,
+			       				bitmap_bh->b_data)) {
 		/*
-		 * Search in the remainder of the current group.
+		 * The block was allocated by another thread, or it was
+		 * allocated and then freed by another thread
 		 */
+		start++;
+		grp_goal++;
+		if (start >= end)
+			goto fail_access;
+		goto repeat;
+	}
+	num++;
+	grp_goal++;
+	while (num < *count && grp_goal < end
+		&& !ext2_set_bit_atomic(sb_bgl_lock(EXT2_SB(sb), group),
+					grp_goal, bitmap_bh->b_data)) {
+		num++;
+		grp_goal++;
 	}
+	*count = num;
+	return grp_goal - num;
+fail_access:
+	*count = num;
+	return -1;
+}
+
+/**
+ * 	find_next_reservable_window():
+ *		find a reservable space within the given range.
+ *		It does not allocate the reservation window for now:
+ *		alloc_new_reservation() will do the work later.
+ *
+ * 	@search_head: the head of the searching list;
+ *		This is not necessarily the list head of the whole filesystem
+ *
+ *		We have both head and start_block to assist the search
+ *		for the reservable space. The list starts from head,
+ *		but we will shift to the place where start_block is,
+ *		then start from there, when looking for a reservable space.
+ *
+ * 	@size: the target new reservation window size
+ *
+ * 	@group_first_block: the first block we consider to start
+ *			the real search from
+ *
+ * 	@last_block:
+ *		the maximum block number that our goal reservable space
+ *		could start from. This is normally the last block in this
+ *		group. The search will end when we found the start of next
+ *		possible reservable space is out of this boundary.
+ *		This could handle the cross boundary reservation window
+ *		request.
+ *
+ * 	basically we search from the given range, rather than the whole
+ * 	reservation double linked list, (start_block, last_block)
+ * 	to find a free region that is of my size and has not
+ * 	been reserved.
+ *
+ */
+static int find_next_reservable_window(
+				struct ext2_reserve_window_node *search_head,
+				struct ext2_reserve_window_node *my_rsv,
+				struct super_block * sb,
+				ext2_fsblk_t start_block,
+				ext2_fsblk_t last_block)
+{
+	struct rb_node *next;
+	struct ext2_reserve_window_node *rsv, *prev;
+	ext2_fsblk_t cur;
+	int size = my_rsv->rsv_goal_size;
+
+	/* TODO: make the start of the reservation window byte-aligned */
+	/* cur = *start_block & ~7;*/
+	cur = start_block;
+	rsv = search_head;
+	if (!rsv)
+		return -1;
+
+	while (1) {
+		if (cur <= rsv->rsv_end)
+			cur = rsv->rsv_end + 1;
+
+		/* TODO?
+		 * in the case we could not find a reservable space
+		 * that is what is expected, during the re-search, we could
+		 * remember what's the largest reservable space we could have
+		 * and return that one.
+		 *
+		 * For now it will fail if we could not find the reservable
+		 * space with expected-size (or more)...
+		 */
+		if (cur > last_block)
+			return -1;		/* fail */
+
+		prev = rsv;
+		next = rb_next(&rsv->rsv_node);
+		rsv = rb_entry(next,struct ext2_reserve_window_node,rsv_node);
 
-	p = map + (goal >> 3);
-	r = memscan(p, 0, (size - goal + 7) >> 3);
-	k = (r - map) << 3;
-	if (k < size) {
 		/*
-		 * We have succeeded in finding a free byte in the block
-		 * bitmap.  Now search backwards to find the start of this
-		 * group of free blocks - won't take more than 7 iterations.
+		 * Reached the last reservation, we can just append to the
+		 * previous one.
 		 */
-		for (goal = k; goal && !ext2_test_bit (goal - 1, map); goal--)
-			;
-		goto got_it;
+		if (!next)
+			break;
+
+		if (cur + size <= rsv->rsv_start) {
+			/*
+			 * Found a reserveable space big enough.  We could
+			 * have a reservation across the group boundary here
+		 	 */
+			break;
 		}
+	}
+	/*
+	 * we come here either :
+	 * when we reach the end of the whole list,
+	 * and there is empty reservable space after last entry in the list.
+	 * append it to the end of the list.
+	 *
+	 * or we found one reservable space in the middle of the list,
+	 * return the reservation window that we could append to.
+	 * succeed.
+	 */
+
+	if ((prev != my_rsv) && (!rsv_is_empty(&my_rsv->rsv_window)))
+		rsv_window_remove(sb, my_rsv);
+
+	/*
+	 * Let's book the whole avaliable window for now.  We will check the
+	 * disk bitmap later and then, if there are free blocks then we adjust
+	 * the window size if it's larger than requested.
+	 * Otherwise, we will remove this node from the tree next time
+	 * call find_next_reservable_window.
+	 */
+	my_rsv->rsv_start = cur;
+	my_rsv->rsv_end = cur + size - 1;
+	my_rsv->rsv_alloc_hit = 0;
+
+	if (prev != my_rsv)
+		ext2_rsv_window_add(sb, my_rsv);
 
-	k = ext2_find_next_zero_bit ((u32 *)map, size, goal);
-	if (k < size) {
-		goal = k;
-		goto got_it;
+	return 0;
+}
+
+/**
+ * 	alloc_new_reservation()--allocate a new reservation window
+ *
+ *		To make a new reservation, we search part of the filesystem
+ *		reservation list (the list that inside the group). We try to
+ *		allocate a new reservation window near the allocation goal,
+ *		or the beginning of the group, if there is no goal.
+ *
+ *		We first find a reservable space after the goal, then from
+ *		there, we check the bitmap for the first free block after
+ *		it. If there is no free block until the end of group, then the
+ *		whole group is full, we failed. Otherwise, check if the free
+ *		block is inside the expected reservable space, if so, we
+ *		succeed.
+ *		If the first free block is outside the reservable space, then
+ *		start from the first free block, we search for next available
+ *		space, and go on.
+ *
+ *	on succeed, a new reservation will be found and inserted into the list
+ *	It contains at least one free block, and it does not overlap with other
+ *	reservation windows.
+ *
+ *	failed: we failed to find a reservation window in this group
+ *
+ *	@rsv: the reservation
+ *
+ *	@grp_goal: The goal (group-relative).  It is where the search for a
+ *		free reservable space should start from.
+ *		if we have a goal(goal >0 ), then start from there,
+ *		no goal(goal = -1), we start from the first block
+ *		of the group.
+ *
+ *	@sb: the super block
+ *	@group: the group we are trying to allocate in
+ *	@bitmap_bh: the block group block bitmap
+ *
+ */
+static int alloc_new_reservation(struct ext2_reserve_window_node *my_rsv,
+		ext2_grpblk_t grp_goal, struct super_block *sb,
+		unsigned int group, struct buffer_head *bitmap_bh)
+{
+	struct ext2_reserve_window_node *search_head;
+	ext2_fsblk_t group_first_block, group_end_block, start_block;
+	ext2_grpblk_t first_free_block;
+	struct rb_root *fs_rsv_root = &EXT2_SB(sb)->s_rsv_window_root;
+	unsigned long size;
+	int ret;
+	spinlock_t *rsv_lock = &EXT2_SB(sb)->s_rsv_window_lock;
+
+	group_first_block = ext2_group_first_block_no(sb, group);
+	group_end_block = group_first_block + (EXT2_BLOCKS_PER_GROUP(sb) - 1);
+
+	if (grp_goal < 0)
+		start_block = group_first_block;
+	else
+		start_block = grp_goal + group_first_block;
+
+	size = my_rsv->rsv_goal_size;
+
+	if (!rsv_is_empty(&my_rsv->rsv_window)) {
+		/*
+		 * if the old reservation is cross group boundary
+		 * and if the goal is inside the old reservation window,
+		 * we will come here when we just failed to allocate from
+		 * the first part of the window. We still have another part
+		 * that belongs to the next group. In this case, there is no
+		 * point to discard our window and try to allocate a new one
+		 * in this group(which will fail). we should
+		 * keep the reservation window, just simply move on.
+		 *
+		 * Maybe we could shift the start block of the reservation
+		 * window to the first block of next group.
+		 */
+
+		if ((my_rsv->rsv_start <= group_end_block) &&
+				(my_rsv->rsv_end > group_end_block) &&
+				(start_block >= my_rsv->rsv_start))
+			return -1;
+
+		if ((my_rsv->rsv_alloc_hit >
+		     (my_rsv->rsv_end - my_rsv->rsv_start + 1) / 2)) {
+			/*
+			 * if the previously allocation hit ratio is
+			 * greater than 1/2, then we double the size of
+			 * the reservation window the next time,
+			 * otherwise we keep the same size window
+			 */
+			size = size * 2;
+			if (size > EXT2_MAX_RESERVE_BLOCKS)
+				size = EXT2_MAX_RESERVE_BLOCKS;
+			my_rsv->rsv_goal_size= size;
 		}
+	}
+
+	spin_lock(rsv_lock);
+	/*
+	 * shift the search start to the window near the goal block
+	 */
+	search_head = search_reserve_window(fs_rsv_root, start_block);
+
+	/*
+	 * find_next_reservable_window() simply finds a reservable window
+	 * inside the given range(start_block, group_end_block).
+	 *
+	 * To make sure the reservation window has a free bit inside it, we
+	 * need to check the bitmap after we found a reservable window.
+	 */
+retry:
+	ret = find_next_reservable_window(search_head, my_rsv, sb,
+						start_block, group_end_block);
+
+	if (ret == -1) {
+		if (!rsv_is_empty(&my_rsv->rsv_window))
+			rsv_window_remove(sb, my_rsv);
+		spin_unlock(rsv_lock);
 		return -1;
-got_it:
-	if (ext2_set_bit_atomic(lock, goal, (void *) map)) 
-		goto repeat;	
-	return goal;
+	}
+
+	/*
+	 * On success, find_next_reservable_window() returns the
+	 * reservation window where there is a reservable space after it.
+	 * Before we reserve this reservable space, we need
+	 * to make sure there is at least a free block inside this region.
+	 *
+	 * Search the first free bit on the block bitmap.  Search starts from
+	 * the start block of the reservable space we just found.
+	 */
+	spin_unlock(rsv_lock);
+	first_free_block = bitmap_search_next_usable_block(
+			my_rsv->rsv_start - group_first_block,
+			bitmap_bh, group_end_block - group_first_block + 1);
+
+	if (first_free_block < 0) {
+		/*
+		 * no free block left on the bitmap, no point
+		 * to reserve the space. return failed.
+		 */
+		spin_lock(rsv_lock);
+		if (!rsv_is_empty(&my_rsv->rsv_window))
+			rsv_window_remove(sb, my_rsv);
+		spin_unlock(rsv_lock);
+		return -1;		/* failed */
+	}
+
+	start_block = first_free_block + group_first_block;
+	/*
+	 * check if the first free block is within the
+	 * free space we just reserved
+	 */
+	if (start_block >= my_rsv->rsv_start && start_block <= my_rsv->rsv_end)
+		return 0;		/* success */
+	/*
+	 * if the first free bit we found is out of the reservable space
+	 * continue search for next reservable space,
+	 * start from where the free block is,
+	 * we also shift the list head to where we stopped last time
+	 */
+	search_head = my_rsv;
+	spin_lock(rsv_lock);
+	goto retry;
+}
+
+/**
+ * try_to_extend_reservation()
+ * @my_rsv:		given reservation window
+ * @sb:			super block
+ * @size:		the delta to extend
+ *
+ * Attempt to expand the reservation window large enough to have
+ * required number of free blocks
+ *
+ * Since ext2_try_to_allocate() will always allocate blocks within
+ * the reservation window range, if the window size is too small,
+ * multiple blocks allocation has to stop at the end of the reservation
+ * window. To make this more efficient, given the total number of
+ * blocks needed and the current size of the window, we try to
+ * expand the reservation window size if necessary on a best-effort
+ * basis before ext2_new_blocks() tries to allocate blocks.
+ */
+static void try_to_extend_reservation(struct ext2_reserve_window_node *my_rsv,
+			struct super_block *sb, int size)
+{
+	struct ext2_reserve_window_node *next_rsv;
+	struct rb_node *next;
+	spinlock_t *rsv_lock = &EXT2_SB(sb)->s_rsv_window_lock;
+
+	if (!spin_trylock(rsv_lock))
+		return;
+
+	next = rb_next(&my_rsv->rsv_node);
+
+	if (!next)
+		my_rsv->rsv_end += size;
+	else {
+		next_rsv = rb_entry(next, struct ext2_reserve_window_node, rsv_node);
+
+		if ((next_rsv->rsv_start - my_rsv->rsv_end - 1) >= size)
+			my_rsv->rsv_end += size;
+		else
+			my_rsv->rsv_end = next_rsv->rsv_start - 1;
+	}
+	spin_unlock(rsv_lock);
+}
+
+/**
+ * ext2_try_to_allocate_with_rsv()
+ * @sb:			superblock
+ * @group:		given allocation block group
+ * @bitmap_bh:		bufferhead holds the block bitmap
+ * @grp_goal:		given target block within the group
+ * @count:		target number of blocks to allocate
+ * @my_rsv:		reservation window
+ *
+ * This is the main function used to allocate a new block and its reservation
+ * window.
+ *
+ * Each time when a new block allocation is need, first try to allocate from
+ * its own reservation.  If it does not have a reservation window, instead of
+ * looking for a free bit on bitmap first, then look up the reservation list to
+ * see if it is inside somebody else's reservation window, we try to allocate a
+ * reservation window for it starting from the goal first. Then do the block
+ * allocation within the reservation window.
+ *
+ * This will avoid keeping on searching the reservation list again and
+ * again when somebody is looking for a free block (without
+ * reservation), and there are lots of free blocks, but they are all
+ * being reserved.
+ *
+ * We use a red-black tree for the per-filesystem reservation list.
+ */
+static ext2_grpblk_t
+ext2_try_to_allocate_with_rsv(struct super_block *sb, unsigned int group,
+			struct buffer_head *bitmap_bh, ext2_grpblk_t grp_goal,
+			struct ext2_reserve_window_node * my_rsv,
+			unsigned long *count)
+{
+	ext2_fsblk_t group_first_block, group_last_block;
+	ext2_grpblk_t ret = 0;
+	unsigned long num = *count;
+
+	/*
+	 * we don't deal with reservation when
+	 * filesystem is mounted without reservation
+	 * or the file is not a regular file
+	 * or last attempt to allocate a block with reservation turned on failed
+	 */
+	if (my_rsv == NULL) {
+		return ext2_try_to_allocate(sb, group, bitmap_bh,
+						grp_goal, count, NULL);
+	}
+	/*
+	 * grp_goal is a group relative block number (if there is a goal)
+	 * 0 <= grp_goal < EXT2_BLOCKS_PER_GROUP(sb)
+	 * first block is a filesystem wide block number
+	 * first block is the block number of the first block in this group
+	 */
+	group_first_block = ext2_group_first_block_no(sb, group);
+	group_last_block = group_first_block + (EXT2_BLOCKS_PER_GROUP(sb) - 1);
+
+	/*
+	 * Basically we will allocate a new block from inode's reservation
+	 * window.
+	 *
+	 * We need to allocate a new reservation window, if:
+	 * a) inode does not have a reservation window; or
+	 * b) last attempt to allocate a block from existing reservation
+	 *    failed; or
+	 * c) we come here with a goal and with a reservation window
+	 *
+	 * We do not need to allocate a new reservation window if we come here
+	 * at the beginning with a goal and the goal is inside the window, or
+	 * we don't have a goal but already have a reservation window.
+	 * then we could go to allocate from the reservation window directly.
+	 */
+	while (1) {
+		if (rsv_is_empty(&my_rsv->rsv_window) || (ret < 0) ||
+			!goal_in_my_reservation(&my_rsv->rsv_window,
+						grp_goal, group, sb)) {
+			if (my_rsv->rsv_goal_size < *count)
+				my_rsv->rsv_goal_size = *count;
+			ret = alloc_new_reservation(my_rsv, grp_goal, sb,
+							group, bitmap_bh);
+			if (ret < 0)
+				break;			/* failed */
+
+			if (!goal_in_my_reservation(&my_rsv->rsv_window,
+							grp_goal, group, sb))
+				grp_goal = -1;
+		} else if (grp_goal >= 0) {
+			int curr = my_rsv->rsv_end -
+					(grp_goal + group_first_block) + 1;
+
+			if (curr < *count)
+				try_to_extend_reservation(my_rsv, sb,
+							*count - curr);
+		}
+
+		if ((my_rsv->rsv_start > group_last_block) ||
+				(my_rsv->rsv_end < group_first_block)) {
+			rsv_window_dump(&EXT2_SB(sb)->s_rsv_window_root, 1);
+			BUG();
+		}
+		ret = ext2_try_to_allocate(sb, group, bitmap_bh, grp_goal,
+					   &num, &my_rsv->rsv_window);
+		if (ret >= 0) {
+			my_rsv->rsv_alloc_hit += num;
+			*count = num;
+			break;				/* succeed */
+		}
+		num = *count;
+	}
+	return ret;
+}
+
+/**
+ * ext2_has_free_blocks()
+ * @sbi:		in-core super block structure.
+ *
+ * Check if filesystem has at least 1 free block available for allocation.
+ */
+static int ext2_has_free_blocks(struct ext2_sb_info *sbi)
+{
+	ext2_fsblk_t free_blocks, root_blocks;
+
+	free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
+	root_blocks = le32_to_cpu(sbi->s_es->s_r_blocks_count);
+	if (free_blocks < root_blocks + 1 && !capable(CAP_SYS_RESOURCE) &&
+		sbi->s_resuid != current->fsuid &&
+		(sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))) {
+		return 0;
+	}
+	return 1;
 }
 
 /*
- * ext2_new_block uses a goal block to assist allocation.  If the goal is
+ * ext2_new_blocks() -- core block(s) allocation function
+ * @inode:		file inode
+ * @goal:		given target block(filesystem wide)
+ * @count:		target number of blocks to allocate
+ * @errp:		error code
+ *
+ * ext2_new_blocks uses a goal block to assist allocation.  If the goal is
  * free, or there is a free block within 32 blocks of the goal, that block
  * is allocated.  Otherwise a forward search is made for a free block; within 
  * each block group the search first looks for an entire free byte in the block
  * bitmap, and then for any free bit if that fails.
  * This function also updates quota and i_blocks field.
  */
-int ext2_new_block(struct inode *inode, unsigned long goal,
-			u32 *prealloc_count, u32 *prealloc_block, int *err)
+ext2_fsblk_t ext2_new_blocks(struct inode *inode, ext2_fsblk_t goal,
+		    unsigned long *count, int *errp)
 {
 	struct buffer_head *bitmap_bh = NULL;
-	struct buffer_head *gdp_bh;	/* bh2 */
-	struct ext2_group_desc *desc;
-	int group_no;			/* i */
-	int ret_block;			/* j */
-	int group_idx;			/* k */
-	int target_block;		/* tmp */
-	int block = 0;
-	struct super_block *sb = inode->i_sb;
-	struct ext2_sb_info *sbi = EXT2_SB(sb);
-	struct ext2_super_block *es = sbi->s_es;
-	unsigned group_size = EXT2_BLOCKS_PER_GROUP(sb);
-	unsigned prealloc_goal = es->s_prealloc_blocks;
-	unsigned group_alloc = 0, es_alloc, dq_alloc;
-	int nr_scanned_groups;
-
-	if (!prealloc_goal--)
-		prealloc_goal = EXT2_DEFAULT_PREALLOC_BLOCKS - 1;
-	if (!prealloc_count || *prealloc_count)
-		prealloc_goal = 0;
-
-	if (DQUOT_ALLOC_BLOCK(inode, 1)) {
-		*err = -EDQUOT;
-		goto out;
+	struct buffer_head *gdp_bh;
+	int group_no;
+	int goal_group;
+	ext2_grpblk_t grp_target_blk;	/* blockgroup relative goal block */
+	ext2_grpblk_t grp_alloc_blk;	/* blockgroup-relative allocated block*/
+	ext2_fsblk_t ret_block;		/* filesyetem-wide allocated block */
+	int bgi;			/* blockgroup iteration index */
+	int performed_allocation = 0;
+	ext2_grpblk_t free_blocks;	/* number of free blocks in a group */
+	struct super_block *sb;
+	struct ext2_group_desc *gdp;
+	struct ext2_super_block *es;
+	struct ext2_sb_info *sbi;
+	struct ext2_reserve_window_node *my_rsv = NULL;
+	struct ext2_block_alloc_info *block_i;
+	unsigned short windowsz = 0;
+	unsigned long ngroups;
+	unsigned long num = *count;
+
+	*errp = -ENOSPC;
+	sb = inode->i_sb;
+	if (!sb) {
+		printk("ext2_new_blocks: nonexistent device");
+		return 0;
 	}
 
-	while (prealloc_goal && DQUOT_PREALLOC_BLOCK(inode, prealloc_goal))
-		prealloc_goal--;
+	/*
+	 * Check quota for allocation of this block.
+	 */
+	if (DQUOT_ALLOC_BLOCK(inode, num)) {
+		*errp = -EDQUOT;
+		return 0;
+	}
 
-	dq_alloc = prealloc_goal + 1;
-	es_alloc = reserve_blocks(sb, dq_alloc);
-	if (!es_alloc) {
-		*err = -ENOSPC;
-		goto out_dquot;
+	sbi = EXT2_SB(sb);
+	es = EXT2_SB(sb)->s_es;
+	ext2_debug("goal=%lu.\n", goal);
+	/*
+	 * Allocate a block from reservation only when
+	 * filesystem is mounted with reservation(default,-o reservation), and
+	 * it's a regular file, and
+	 * the desired window size is greater than 0 (One could use ioctl
+	 * command EXT2_IOC_SETRSVSZ to set the window size to 0 to turn off
+	 * reservation on that particular file)
+	 */
+	block_i = EXT2_I(inode)->i_block_alloc_info;
+	if (block_i) {
+		windowsz = block_i->rsv_window_node.rsv_goal_size;
+		if (windowsz > 0)
+			my_rsv = &block_i->rsv_window_node;
 	}
 
-	ext2_debug ("goal=%lu.\n", goal);
+	if (!ext2_has_free_blocks(sbi)) {
+		*errp = -ENOSPC;
+		goto out;
+	}
 
+	/*
+	 * First, test whether the goal block is free.
+	 */
 	if (goal < le32_to_cpu(es->s_first_data_block) ||
 	    goal >= le32_to_cpu(es->s_blocks_count))
 		goal = le32_to_cpu(es->s_first_data_block);
-	group_no = (goal - le32_to_cpu(es->s_first_data_block)) / group_size;
-	desc = ext2_get_group_desc (sb, group_no, &gdp_bh);
-	if (!desc) {
+	group_no = (goal - le32_to_cpu(es->s_first_data_block)) /
+			EXT2_BLOCKS_PER_GROUP(sb);
+	goal_group = group_no;
+retry_alloc:
+	gdp = ext2_get_group_desc(sb, group_no, &gdp_bh);
+	if (!gdp)
+		goto io_error;
+
+	free_blocks = le16_to_cpu(gdp->bg_free_blocks_count);
 	/*
-		 * gdp_bh may still be uninitialised.  But group_release_blocks
-		 * will not touch it because group_alloc is zero.
+	 * if there is not enough free blocks to make a new resevation
+	 * turn off reservation for this allocation
 	 */
-		goto io_error;
-	}
+	if (my_rsv && (free_blocks < windowsz)
+		&& (rsv_is_empty(&my_rsv->rsv_window)))
+		my_rsv = NULL;
 
-	group_alloc = group_reserve_blocks(sbi, group_no, desc,
-					gdp_bh, es_alloc);
-	if (group_alloc) {
-		ret_block = ((goal - le32_to_cpu(es->s_first_data_block)) %
-					group_size);
-		brelse(bitmap_bh);
+	if (free_blocks > 0) {
+		grp_target_blk = ((goal - le32_to_cpu(es->s_first_data_block)) %
+				EXT2_BLOCKS_PER_GROUP(sb));
 		bitmap_bh = read_block_bitmap(sb, group_no);
 		if (!bitmap_bh)
 			goto io_error;
-		
-		ext2_debug("goal is at %d:%d.\n", group_no, ret_block);
-
-		ret_block = grab_block(sb_bgl_lock(sbi, group_no),
-				bitmap_bh->b_data, group_size, ret_block);
-		if (ret_block >= 0)
-			goto got_block;
-		group_release_blocks(sb, group_no, desc, gdp_bh, group_alloc);
-		group_alloc = 0;
+		grp_alloc_blk = ext2_try_to_allocate_with_rsv(sb, group_no,
+					bitmap_bh, grp_target_blk,
+					my_rsv, &num);
+		if (grp_alloc_blk >= 0)
+			goto allocated;
 	}
 
-	ext2_debug ("Bit not found in block group %d.\n", group_no);
+	ngroups = EXT2_SB(sb)->s_groups_count;
+	smp_rmb();
 
 	/*
 	 * Now search the rest of the groups.  We assume that 
-	 * i and desc correctly point to the last group visited.
+	 * i and gdp correctly point to the last group visited.
 	 */
-	nr_scanned_groups = 0;
-retry:
-	for (group_idx = 0; !group_alloc &&
-			group_idx < sbi->s_groups_count; group_idx++) {
+	for (bgi = 0; bgi < ngroups; bgi++) {
 		group_no++;
-		if (group_no >= sbi->s_groups_count)
+		if (group_no >= ngroups)
 			group_no = 0;
-		desc = ext2_get_group_desc(sb, group_no, &gdp_bh);
-		if (!desc)
+		gdp = ext2_get_group_desc(sb, group_no, &gdp_bh);
+		if (!gdp)
 			goto io_error;
-		group_alloc = group_reserve_blocks(sbi, group_no, desc,
-						gdp_bh, es_alloc);
-	}
-	if (!group_alloc) {
-		*err = -ENOSPC;
-		goto out_release;
-	}
+
+		free_blocks = le16_to_cpu(gdp->bg_free_blocks_count);
+		/*
+		 * skip this group if the number of
+		 * free blocks is less than half of the reservation
+		 * window size.
+		 */
+		if (free_blocks <= (windowsz/2))
+			continue;
+
 		brelse(bitmap_bh);
 		bitmap_bh = read_block_bitmap(sb, group_no);
 		if (!bitmap_bh)
 			goto io_error;
-
-	ret_block = grab_block(sb_bgl_lock(sbi, group_no), bitmap_bh->b_data,
-				group_size, 0);
-	if (ret_block < 0) {
 		/*
-		 * If a free block counter is corrupted we can loop inifintely.
-		 * Detect that here.
+		 * try to allocate block(s) from this group, without a goal(-1).
 		 */
-		nr_scanned_groups++;
-		if (nr_scanned_groups > 2 * sbi->s_groups_count) {
-			ext2_error(sb, "ext2_new_block",
-				"corrupted free blocks counters");
-			goto io_error;
+		grp_alloc_blk = ext2_try_to_allocate_with_rsv(sb, group_no,
+					bitmap_bh, -1, my_rsv, &num);
+		if (grp_alloc_blk >= 0)
+			goto allocated;
 	}
 	/*
-		 * Someone else grabbed the last free block in this blockgroup
-		 * before us.  Retry the scan.
+	 * We may end up a bogus ealier ENOSPC error due to
+	 * filesystem is "full" of reservations, but
+	 * there maybe indeed free blocks avaliable on disk
+	 * In this case, we just forget about the reservations
+	 * just do block allocation as without reservations.
 	 */
-		group_release_blocks(sb, group_no, desc, gdp_bh, group_alloc);
-		group_alloc = 0;
-		goto retry;
+	if (my_rsv) {
+		my_rsv = NULL;
+		windowsz = 0;
+		group_no = goal_group;
+		goto retry_alloc;
 	}
+	/* No space left on the device */
+	*errp = -ENOSPC;
+	goto out;
+
+allocated:
 
-got_block:
 	ext2_debug("using block group %d(%d)\n",
-		group_no, desc->bg_free_blocks_count);
+			group_no, gdp->bg_free_blocks_count);
 
-	target_block = ret_block + group_no * group_size +
-			le32_to_cpu(es->s_first_data_block);
+	ret_block = grp_alloc_blk + ext2_group_first_block_no(sb, group_no);
 
-	if (target_block == le32_to_cpu(desc->bg_block_bitmap) ||
-	    target_block == le32_to_cpu(desc->bg_inode_bitmap) ||
-	    in_range(target_block, le32_to_cpu(desc->bg_inode_table),
-		      sbi->s_itb_per_group))
-		ext2_error (sb, "ext2_new_block",
+	if (in_range(le32_to_cpu(gdp->bg_block_bitmap), ret_block, num) ||
+	    in_range(le32_to_cpu(gdp->bg_inode_bitmap), ret_block, num) ||
+	    in_range(ret_block, le32_to_cpu(gdp->bg_inode_table),
+		      EXT2_SB(sb)->s_itb_per_group) ||
+	    in_range(ret_block + num - 1, le32_to_cpu(gdp->bg_inode_table),
+		      EXT2_SB(sb)->s_itb_per_group))
+		ext2_error(sb, "ext2_new_blocks",
 			    "Allocating block in system zone - "
-			    "block = %u", target_block);
+			    "blocks from "E2FSBLK", length %lu",
+			    ret_block, num);
+
+	performed_allocation = 1;
 
-	if (target_block >= le32_to_cpu(es->s_blocks_count)) {
-		ext2_error (sb, "ext2_new_block",
-			    "block(%d) >= blocks count(%d) - "
+	if (ret_block + num - 1 >= le32_to_cpu(es->s_blocks_count)) {
+		ext2_error(sb, "ext2_new_blocks",
+			    "block("E2FSBLK") >= blocks count(%d) - "
 			    "block_group = %d, es == %p ", ret_block,
 			le32_to_cpu(es->s_blocks_count), group_no, es);
-		goto io_error;
+		goto out;
 	}
-	block = target_block;
-
-	/* OK, we _had_ allocated something */
-	ext2_debug("found bit %d\n", ret_block);
 
-	dq_alloc--;
-	es_alloc--;
-	group_alloc--;
-
-	/*
-	 * Do block preallocation now if required.
-	 */
-	write_lock(&EXT2_I(inode)->i_meta_lock);
-	if (group_alloc && !*prealloc_count) {
-		unsigned n;
-
-		for (n = 0; n < group_alloc && ++ret_block < group_size; n++) {
-			if (ext2_set_bit_atomic(sb_bgl_lock(sbi, group_no),
-						ret_block,
-						(void*) bitmap_bh->b_data))
- 				break;
-		}
-		*prealloc_block = block + 1;
-		*prealloc_count = n;
-		es_alloc -= n;
-		dq_alloc -= n;
-		group_alloc -= n;
-	}
-	write_unlock(&EXT2_I(inode)->i_meta_lock);
+	group_adjust_blocks(sb, group_no, gdp, gdp_bh, -num);
+	percpu_counter_sub(&sbi->s_freeblocks_counter, num);
 
 	mark_buffer_dirty(bitmap_bh);
 	if (sb->s_flags & MS_SYNCHRONOUS)
 		sync_dirty_buffer(bitmap_bh);
 
-	ext2_debug ("allocating block %d. ", block);
+	*errp = 0;
+	brelse(bitmap_bh);
+	DQUOT_FREE_BLOCK(inode, *count-num);
+	*count = num;
+	return ret_block;
 
-	*err = 0;
-out_release:
-	group_release_blocks(sb, group_no, desc, gdp_bh, group_alloc);
-	release_blocks(sb, es_alloc);
-out_dquot:
-	DQUOT_FREE_BLOCK(inode, dq_alloc);
+io_error:
+	*errp = -EIO;
 out:
+	/*
+	 * Undo the block allocation
+	 */
+	if (!performed_allocation)
+		DQUOT_FREE_BLOCK(inode, *count);
 	brelse(bitmap_bh);
-	return block;
+	return 0;
+}
 
-io_error:
-	*err = -EIO;
-	goto out_release;
+ext2_fsblk_t ext2_new_block(struct inode *inode, unsigned long goal, int *errp)
+{
+	unsigned long count = 1;
+
+	return ext2_new_blocks(inode, goal, &count, errp);
 }
 
 #ifdef EXT2FS_DEBUG

fs/ext2/ext2.h

@@ -33,22 +33,9 @@ struct ext2_inode_info {
 	 */
 	__u32	i_block_group;
 
-	/*
-	 * i_next_alloc_block is the logical (file-relative) number of the
-	 * most-recently-allocated block in this file.  Yes, it is misnamed.
-	 * We use this for detecting linearly ascending allocation requests.
-	 */
-	__u32	i_next_alloc_block;
+	/* block reservation info */
+	struct ext2_block_alloc_info *i_block_alloc_info;
 
-	/*
-	 * i_next_alloc_goal is the *physical* companion to i_next_alloc_block.
-	 * it the the physical block number of the block which was most-recently
-	 * allocated to this file.  This give us the goal (target) for the next
-	 * allocation when we detect linearly ascending requests.
-	 */
-	__u32	i_next_alloc_goal;
-	__u32	i_prealloc_block;
-	__u32	i_prealloc_count;
 	__u32	i_dir_start_lookup;
 #ifdef CONFIG_EXT2_FS_XATTR
 	/*
@@ -65,7 +52,16 @@ struct ext2_inode_info {
 	struct posix_acl	*i_default_acl;
 #endif
 	rwlock_t i_meta_lock;
+
+	/*
+	 * truncate_mutex is for serialising ext2_truncate() against
+	 * ext2_getblock().  It also protects the internals of the inode's
+	 * reservation data structures: ext2_reserve_window and
+	 * ext2_reserve_window_node.
+	 */
+	struct mutex truncate_mutex;
 	struct inode	vfs_inode;
+	struct list_head i_orphan;	/* unlinked but open inodes */
 };
 
 /*
@@ -91,8 +87,9 @@ static inline struct ext2_inode_info *EXT2_I(struct inode *inode)
 /* balloc.c */
 extern int ext2_bg_has_super(struct super_block *sb, int group);
 extern unsigned long ext2_bg_num_gdb(struct super_block *sb, int group);
-extern int ext2_new_block (struct inode *, unsigned long,
-			   __u32 *, __u32 *, int *);
+extern ext2_fsblk_t ext2_new_block(struct inode *, unsigned long, int *);
+extern ext2_fsblk_t ext2_new_blocks(struct inode *, unsigned long,
+				unsigned long *, int *);
 extern void ext2_free_blocks (struct inode *, unsigned long,
 			      unsigned long);
 extern unsigned long ext2_count_free_blocks (struct super_block *);
@@ -101,6 +98,10 @@ extern void ext2_check_blocks_bitmap (struct super_block *);
 extern struct ext2_group_desc * ext2_get_group_desc(struct super_block * sb,
 						    unsigned int block_group,
 						    struct buffer_head ** bh);
+extern void ext2_discard_reservation (struct inode *);
+extern int ext2_should_retry_alloc(struct super_block *sb, int *retries);
+extern void ext2_init_block_alloc_info(struct inode *);
+extern void ext2_rsv_window_add(struct super_block *sb, struct ext2_reserve_window_node *rsv);
 
 /* dir.c */
 extern int ext2_add_link (struct dentry *, struct inode *);
@@ -128,7 +129,6 @@ extern int ext2_write_inode (struct inode *, int);
 extern void ext2_put_inode (struct inode *);
 extern void ext2_delete_inode (struct inode *);
 extern int ext2_sync_inode (struct inode *);
-extern void ext2_discard_prealloc (struct inode *);
 extern int ext2_get_block(struct inode *, sector_t, struct buffer_head *, int);
 extern void ext2_truncate (struct inode *);
 extern int ext2_setattr (struct dentry *, struct iattr *);

fs/ext2/file.c

@@ -30,8 +30,11 @@
  */
 static int ext2_release_file (struct inode * inode, struct file * filp)
 {
-	if (filp->f_mode & FMODE_WRITE)
-		ext2_discard_prealloc (inode);
+	if (filp->f_mode & FMODE_WRITE) {
+		mutex_lock(&EXT2_I(inode)->truncate_mutex);
+		ext2_discard_reservation(inode);
+		mutex_unlock(&EXT2_I(inode)->truncate_mutex);
+	}
 	return 0;
 }
 

fs/ext2/ialloc.c

@@ -581,11 +581,8 @@ struct inode *ext2_new_inode(struct inode *dir, int mode)
 	ei->i_file_acl = 0;
 	ei->i_dir_acl = 0;
 	ei->i_dtime = 0;
+	ei->i_block_alloc_info = NULL;
 	ei->i_block_group = group;
-	ei->i_next_alloc_block = 0;
-	ei->i_next_alloc_goal = 0;
-	ei->i_prealloc_block = 0;
-	ei->i_prealloc_count = 0;
 	ei->i_dir_start_lookup = 0;
 	ei->i_state = EXT2_STATE_NEW;
 	ext2_set_inode_flags(inode);

fs/ext2/inode.c

@@ -53,19 +53,6 @@ static inline int ext2_inode_is_fast_symlink(struct inode *inode)
 		inode->i_blocks - ea_blocks == 0);
 }
 
-/*
- * Called at each iput().
- *
- * The inode may be "bad" if ext2_read_inode() saw an error from
- * ext2_get_inode(), so we need to check that to avoid freeing random disk
- * blocks.
- */
-void ext2_put_inode(struct inode *inode)
-{
-	if (!is_bad_inode(inode))
-		ext2_discard_prealloc(inode);
-}
-
 /*
  * Called at the last iput() if i_nlink is zero.
  */
@@ -89,61 +76,6 @@ void ext2_delete_inode (struct inode * inode)
 	clear_inode(inode);	/* We must guarantee clearing of inode... */
 }
 
-void ext2_discard_prealloc (struct inode * inode)
-{
-#ifdef EXT2_PREALLOCATE
-	struct ext2_inode_info *ei = EXT2_I(inode);
-	write_lock(&ei->i_meta_lock);
-	if (ei->i_prealloc_count) {
-		unsigned short total = ei->i_prealloc_count;
-		unsigned long block = ei->i_prealloc_block;
-		ei->i_prealloc_count = 0;
-		ei->i_prealloc_block = 0;
-		write_unlock(&ei->i_meta_lock);
-		ext2_free_blocks (inode, block, total);
-		return;
-	} else
-		write_unlock(&ei->i_meta_lock);
-#endif
-}
-
-static int ext2_alloc_block (struct inode * inode, unsigned long goal, int *err)
-{
-#ifdef EXT2FS_DEBUG
-	static unsigned long alloc_hits, alloc_attempts;
-#endif
-	unsigned long result;
-
-
-#ifdef EXT2_PREALLOCATE
-	struct ext2_inode_info *ei = EXT2_I(inode);
-	write_lock(&ei->i_meta_lock);
-	if (ei->i_prealloc_count &&
-	    (goal == ei->i_prealloc_block || goal + 1 == ei->i_prealloc_block))
-	{
-		result = ei->i_prealloc_block++;
-		ei->i_prealloc_count--;
-		write_unlock(&ei->i_meta_lock);
-		ext2_debug ("preallocation hit (%lu/%lu).\n",
-			    ++alloc_hits, ++alloc_attempts);
-	} else {
-		write_unlock(&ei->i_meta_lock);
-		ext2_discard_prealloc (inode);
-		ext2_debug ("preallocation miss (%lu/%lu).\n",
-			    alloc_hits, ++alloc_attempts);
-		if (S_ISREG(inode->i_mode))
-			result = ext2_new_block (inode, goal, 
-				 &ei->i_prealloc_count,
-				 &ei->i_prealloc_block, err);
-		else
-			result = ext2_new_block(inode, goal, NULL, NULL, err);
-	}
-#else
-	result = ext2_new_block (inode, goal, 0, 0, err);
-#endif
-	return result;
-}
-
 typedef struct {
 	__le32	*p;
 	__le32	key;
@@ -228,7 +160,8 @@ static int ext2_block_to_path(struct inode *inode,
 		ext2_warning (inode->i_sb, "ext2_block_to_path", "block > big");
 	}
 	if (boundary)
-		*boundary = (i_block & (ptrs - 1)) == (final - 1);
+		*boundary = final - 1 - (i_block & (ptrs - 1));
+
 	return n;
 }
 
@@ -355,39 +288,129 @@ static unsigned long ext2_find_near(struct inode *inode, Indirect *ind)
  *	@block:  block we want
  *	@chain:  chain of indirect blocks
  *	@partial: pointer to the last triple within a chain
- *	@goal:	place to store the result.
  *
- *	Normally this function find the prefered place for block allocation,
- *	stores it in *@goal and returns zero. If the branch had been changed
- *	under us we return -EAGAIN.
+ *	Returns preferred place for a block (the goal).
  */
 
 static inline int ext2_find_goal(struct inode *inode,
 				 long block,
 				 Indirect chain[4],
-				 Indirect *partial,
-				 unsigned long *goal)
+				 Indirect *partial)
 {
-	struct ext2_inode_info *ei = EXT2_I(inode);
-	write_lock(&ei->i_meta_lock);
-	if ((block == ei->i_next_alloc_block + 1) && ei->i_next_alloc_goal) {
-		ei->i_next_alloc_block++;
-		ei->i_next_alloc_goal++;
-	} 
-	if (verify_chain(chain, partial)) {
+	struct ext2_block_alloc_info *block_i;
+
+	block_i = EXT2_I(inode)->i_block_alloc_info;
+
 	/*
 	 * try the heuristic for sequential allocation,
 	 * failing that at least try to get decent locality.
 	 */
-		if (block == ei->i_next_alloc_block)
-			*goal = ei->i_next_alloc_goal;
-		if (!*goal)
-			*goal = ext2_find_near(inode, partial);
-		write_unlock(&ei->i_meta_lock);
-		return 0;
+	if (block_i && (block == block_i->last_alloc_logical_block + 1)
+		&& (block_i->last_alloc_physical_block != 0)) {
+		return block_i->last_alloc_physical_block + 1;
 	}
-	write_unlock(&ei->i_meta_lock);
-	return -EAGAIN;
+
+	return ext2_find_near(inode, partial);
+}
+
+/**
+ *	ext2_blks_to_allocate: Look up the block map and count the number
+ *	of direct blocks need to be allocated for the given branch.
+ *
+ * 	@branch: chain of indirect blocks
+ *	@k: number of blocks need for indirect blocks
+ *	@blks: number of data blocks to be mapped.
+ *	@blocks_to_boundary:  the offset in the indirect block
+ *
+ *	return the total number of blocks to be allocate, including the
+ *	direct and indirect blocks.
+ */
+static int
+ext2_blks_to_allocate(Indirect * branch, int k, unsigned long blks,
+		int blocks_to_boundary)
+{
+	unsigned long count = 0;
+
+	/*
+	 * Simple case, [t,d]Indirect block(s) has not allocated yet
+	 * then it's clear blocks on that path have not allocated
+	 */
+	if (k > 0) {
+		/* right now don't hanel cross boundary allocation */
+		if (blks < blocks_to_boundary + 1)
+			count += blks;
+		else
+			count += blocks_to_boundary + 1;
+		return count;
+	}
+
+	count++;
+	while (count < blks && count <= blocks_to_boundary
+		&& le32_to_cpu(*(branch[0].p + count)) == 0) {
+		count++;
+	}
+	return count;
+}
+
+/**
+ *	ext2_alloc_blocks: multiple allocate blocks needed for a branch
+ *	@indirect_blks: the number of blocks need to allocate for indirect
+ *			blocks
+ *
+ *	@new_blocks: on return it will store the new block numbers for
+ *	the indirect blocks(if needed) and the first direct block,
+ *	@blks:	on return it will store the total number of allocated
+ *		direct blocks
+ */
+static int ext2_alloc_blocks(struct inode *inode,
+			ext2_fsblk_t goal, int indirect_blks, int blks,
+			ext2_fsblk_t new_blocks[4], int *err)
+{
+	int target, i;
+	unsigned long count = 0;
+	int index = 0;
+	ext2_fsblk_t current_block = 0;
+	int ret = 0;
+
+	/*
+	 * Here we try to allocate the requested multiple blocks at once,
+	 * on a best-effort basis.
+	 * To build a branch, we should allocate blocks for
+	 * the indirect blocks(if not allocated yet), and at least
+	 * the first direct block of this branch.  That's the
+	 * minimum number of blocks need to allocate(required)
+	 */
+	target = blks + indirect_blks;
+
+	while (1) {
+		count = target;
+		/* allocating blocks for indirect blocks and direct blocks */
+		current_block = ext2_new_blocks(inode,goal,&count,err);
+		if (*err)
+			goto failed_out;
+
+		target -= count;
+		/* allocate blocks for indirect blocks */
+		while (index < indirect_blks && count) {
+			new_blocks[index++] = current_block++;
+			count--;
+		}
+
+		if (count > 0)
+			break;
+	}
+
+	/* save the new block number for the first direct block */
+	new_blocks[index] = current_block;
+
+	/* total number of blocks allocated for direct blocks */
+	ret = count;
+	*err = 0;
+	return ret;
+failed_out:
+	for (i = 0; i <index; i++)
+		ext2_free_blocks(inode, new_blocks[i], 1);
+	return ret;
 }
 
 /**
@@ -416,39 +439,49 @@ static inline int ext2_find_goal(struct inode *inode,
  */
 
 static int ext2_alloc_branch(struct inode *inode,
-			     int num,
-			     unsigned long goal,
-			     int *offsets,
-			     Indirect *branch)
+			int indirect_blks, int *blks, ext2_fsblk_t goal,
+			int *offsets, Indirect *branch)
 {
 	int blocksize = inode->i_sb->s_blocksize;
-	int n = 0;
-	int err;
-	int i;
-	int parent = ext2_alloc_block(inode, goal, &err);
-
-	branch[0].key = cpu_to_le32(parent);
-	if (parent) for (n = 1; n < num; n++) {
+	int i, n = 0;
+	int err = 0;
 	struct buffer_head *bh;
-		/* Allocate the next block */
-		int nr = ext2_alloc_block(inode, parent, &err);
-		if (!nr)
-			break;
-		branch[n].key = cpu_to_le32(nr);
+	int num;
+	ext2_fsblk_t new_blocks[4];
+	ext2_fsblk_t current_block;
+
+	num = ext2_alloc_blocks(inode, goal, indirect_blks,
+				*blks, new_blocks, &err);
+	if (err)
+		return err;
+
+	branch[0].key = cpu_to_le32(new_blocks[0]);
 	/*
-		 * Get buffer_head for parent block, zero it out and set 
-		 * the pointer to new one, then send parent to disk.
+	 * metadata blocks and data blocks are allocated.
 	 */
-		bh = sb_getblk(inode->i_sb, parent);
-		if (!bh) {
-			err = -EIO;
-			break;
-		}
+	for (n = 1; n <= indirect_blks;  n++) {
+		/*
+		 * Get buffer_head for parent block, zero it out
+		 * and set the pointer to new one, then send
+		 * parent to disk.
+		 */
+		bh = sb_getblk(inode->i_sb, new_blocks[n-1]);
+		branch[n].bh = bh;
 		lock_buffer(bh);
 		memset(bh->b_data, 0, blocksize);
-		branch[n].bh = bh;
 		branch[n].p = (__le32 *) bh->b_data + offsets[n];
+		branch[n].key = cpu_to_le32(new_blocks[n]);
 		*branch[n].p = branch[n].key;
+		if ( n == indirect_blks) {
+			current_block = new_blocks[n];
+			/*
+			 * End of chain, update the last new metablock of
+			 * the chain to point to the new allocated
+			 * data blocks numbers
+			 */
+			for (i=1; i < num; i++)
+				*(branch[n].p + i) = cpu_to_le32(++current_block);
+		}
 		set_buffer_uptodate(bh);
 		unlock_buffer(bh);
 		mark_buffer_dirty_inode(bh, inode);
@@ -458,16 +491,8 @@ static int ext2_alloc_branch(struct inode *inode,
 		 */
 		if (S_ISDIR(inode->i_mode) && IS_DIRSYNC(inode))
 			sync_dirty_buffer(bh);
-		parent = nr;
 	}
-	if (n == num)
-		return 0;
-
-	/* Allocation failed, free what we already allocated */
-	for (i = 1; i < n; i++)
-		bforget(branch[i].bh);
-	for (i = 0; i < n; i++)
-		ext2_free_blocks(inode, le32_to_cpu(branch[i].key), 1);
+	*blks = num;
 	return err;
 }
 
@@ -478,57 +503,56 @@ static int ext2_alloc_branch(struct inode *inode,
  * @chain: chain of indirect blocks (with a missing link - see
  *	ext2_alloc_branch)
  * @where: location of missing link
- *	@num:   number of blocks we are adding
+ * @num:   number of indirect blocks we are adding
+ * @blks:  number of direct blocks we are adding
  *
- *	This function verifies that chain (up to the missing link) had not
- *	changed, fills the missing link and does all housekeeping needed in
+ * This function fills the missing link and does all housekeeping needed in
  * inode (->i_blocks, etc.). In case of success we end up with the full
- *	chain to new block and return 0. Otherwise (== chain had been changed)
- *	we free the new blocks (forgetting their buffer_heads, indeed) and
- *	return -EAGAIN.
+ * chain to new block and return 0.
  */
-
-static inline int ext2_splice_branch(struct inode *inode,
-				     long block,
-				     Indirect chain[4],
-				     Indirect *where,
-				     int num)
+static void ext2_splice_branch(struct inode *inode,
+			long block, Indirect *where, int num, int blks)
 {
-	struct ext2_inode_info *ei = EXT2_I(inode);
 	int i;
+	struct ext2_block_alloc_info *block_i;
+	ext2_fsblk_t current_block;
 
-	/* Verify that place we are splicing to is still there and vacant */
-
-	write_lock(&ei->i_meta_lock);
-	if (!verify_chain(chain, where-1) || *where->p)
-		goto changed;
+	block_i = EXT2_I(inode)->i_block_alloc_info;
 
+	/* XXX LOCKING probably should have i_meta_lock ?*/
 	/* That's it */
 
 	*where->p = where->key;
-	ei->i_next_alloc_block = block;
-	ei->i_next_alloc_goal = le32_to_cpu(where[num-1].key);
 
-	write_unlock(&ei->i_meta_lock);
+	/*
+	 * Update the host buffer_head or inode to point to more just allocated
+	 * direct blocks blocks
+	 */
+	if (num == 0 && blks > 1) {
+		current_block = le32_to_cpu(where->key) + 1;
+		for (i = 1; i < blks; i++)
+			*(where->p + i ) = cpu_to_le32(current_block++);
+	}
 
-	/* We are done with atomic stuff, now do the rest of housekeeping */
+	/*
+	 * update the most recently allocated logical & physical block
+	 * in i_block_alloc_info, to assist find the proper goal block for next
+	 * allocation
+	 */
+	if (block_i) {
+		block_i->last_alloc_logical_block = block + blks - 1;
+		block_i->last_alloc_physical_block =
+				le32_to_cpu(where[num].key) + blks - 1;
+	}
 
-	inode->i_ctime = CURRENT_TIME_SEC;
+	/* We are done with atomic stuff, now do the rest of housekeeping */
 
 	/* had we spliced it onto indirect block? */
 	if (where->bh)
 		mark_buffer_dirty_inode(where->bh, inode);
 
+	inode->i_ctime = CURRENT_TIME_SEC;
 	mark_inode_dirty(inode);
-	return 0;
-
-changed:
-	write_unlock(&ei->i_meta_lock);
-	for (i = 1; i < num; i++)
-		bforget(where[i].bh);
-	for (i = 0; i < num; i++)
-		ext2_free_blocks(inode, le32_to_cpu(where[i].key), 1);
-	return -EAGAIN;
 }
 
 /*
@@ -542,64 +566,99 @@ static inline int ext2_splice_branch(struct inode *inode,
  * That has a nice additional property: no special recovery from the failed
  * allocations is needed - we simply release blocks and do not touch anything
  * reachable from inode.
+ *
+ * `handle' can be NULL if create == 0.
+ *
+ * The BKL may not be held on entry here.  Be sure to take it early.
+ * return > 0, # of blocks mapped or allocated.
+ * return = 0, if plain lookup failed.
+ * return < 0, error case.
  */
-
-int ext2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create)
+static int ext2_get_blocks(struct inode *inode,
+			   sector_t iblock, unsigned long maxblocks,
+			   struct buffer_head *bh_result,
+			   int create)
 {
 	int err = -EIO;
 	int offsets[4];
 	Indirect chain[4];
 	Indirect *partial;
-	unsigned long goal;
-	int left;
-	int boundary = 0;
-	int depth = ext2_block_to_path(inode, iblock, offsets, &boundary);
+	ext2_fsblk_t goal;
+	int indirect_blks;
+	int blocks_to_boundary = 0;
+	int depth;
+	struct ext2_inode_info *ei = EXT2_I(inode);
+	int count = 0;
+	ext2_fsblk_t first_block = 0;
 
-	if (depth == 0)
-		goto out;
+	depth = ext2_block_to_path(inode,iblock,offsets,&blocks_to_boundary);
 
+	if (depth == 0)
+		return (err);
 reread:
 	partial = ext2_get_branch(inode, depth, offsets, chain, &err);
 
 	/* Simplest case - block found, no allocation needed */
 	if (!partial) {
-got_it:
-		map_bh(bh_result, inode->i_sb, le32_to_cpu(chain[depth-1].key));
-		if (boundary)
-			set_buffer_boundary(bh_result);
-		/* Clean up and exit */
-		partial = chain+depth-1; /* the whole chain */
-		goto cleanup;
-	}
+		first_block = le32_to_cpu(chain[depth - 1].key);
+		clear_buffer_new(bh_result); /* What's this do? */
+		count++;
+		/*map more blocks*/
+		while (count < maxblocks && count <= blocks_to_boundary) {
+			ext2_fsblk_t blk;
 
-	/* Next simple case - plain lookup or failed read of indirect block */
-	if (!create || err == -EIO) {
-cleanup:
-		while (partial > chain) {
-			brelse(partial->bh);
-			partial--;
+			if (!verify_chain(chain, partial)) {
+				/*
+				 * Indirect block might be removed by
+				 * truncate while we were reading it.
+				 * Handling of that case: forget what we've
+				 * got now, go to reread.
+				 */
+				count = 0;
+				goto changed;
 			}
-out:
-		return err;
+			blk = le32_to_cpu(*(chain[depth-1].p + count));
+			if (blk == first_block + count)
+				count++;
+			else
+				break;
 		}
+		goto got_it;
+	}
+
+	/* Next simple case - plain lookup or failed read of indirect block */
+	if (!create || err == -EIO)
+		goto cleanup;
+
+	mutex_lock(&ei->truncate_mutex);
 
 	/*
-	 * Indirect block might be removed by truncate while we were
-	 * reading it. Handling of that case (forget what we've got and
-	 * reread) is taken out of the main path.
+	 * Okay, we need to do block allocation.  Lazily initialize the block
+	 * allocation info here if necessary
 	*/
-	if (err == -EAGAIN)
-		goto changed;
+	if (S_ISREG(inode->i_mode) && (!ei->i_block_alloc_info))
+		ext2_init_block_alloc_info(inode);
 
-	goal = 0;
-	if (ext2_find_goal(inode, iblock, chain, partial, &goal) < 0)
-		goto changed;
+	goal = ext2_find_goal(inode, iblock, chain, partial);
 
-	left = (chain + depth) - partial;
-	err = ext2_alloc_branch(inode, left, goal,
-					offsets+(partial-chain), partial);
-	if (err)
+	/* the number of blocks need to allocate for [d,t]indirect blocks */
+	indirect_blks = (chain + depth) - partial - 1;
+	/*
+	 * Next look up the indirect map to count the totoal number of
+	 * direct blocks to allocate for this branch.
+	 */
+	count = ext2_blks_to_allocate(partial, indirect_blks,
+					maxblocks, blocks_to_boundary);
+	/*
+	 * XXX ???? Block out ext2_truncate while we alter the tree
+	 */
+	err = ext2_alloc_branch(inode, indirect_blks, &count, goal,
+				offsets + (partial - chain), partial);
+
+	if (err) {
+		mutex_unlock(&ei->truncate_mutex);
 		goto cleanup;
+	}
 
 	if (ext2_use_xip(inode->i_sb)) {
 		/*
@@ -607,16 +666,28 @@ int ext2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_
 		 */
 		err = ext2_clear_xip_target (inode,
 			le32_to_cpu(chain[depth-1].key));
-		if (err)
+		if (err) {
+			mutex_unlock(&ei->truncate_mutex);
 			goto cleanup;
 		}
+	}
 
-	if (ext2_splice_branch(inode, iblock, chain, partial, left) < 0)
-		goto changed;
-
+	ext2_splice_branch(inode, iblock, partial, indirect_blks, count);
+	mutex_unlock(&ei->truncate_mutex);
 	set_buffer_new(bh_result);
-	goto got_it;
-
+got_it:
+	map_bh(bh_result, inode->i_sb, le32_to_cpu(chain[depth-1].key));
+	if (count > blocks_to_boundary)
+		set_buffer_boundary(bh_result);
+	err = count;
+	/* Clean up and exit */
+	partial = chain + depth - 1;	/* the whole chain */
+cleanup:
+	while (partial > chain) {
+		brelse(partial->bh);
+		partial--;
+	}
+	return err;
 changed:
 	while (partial > chain) {
 		brelse(partial->bh);
@@ -625,6 +696,19 @@ int ext2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_
 	goto reread;
 }
 
+int ext2_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create)
+{
+	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;
+	int ret = ext2_get_blocks(inode, iblock, max_blocks,
+			      bh_result, create);
+	if (ret > 0) {
+		bh_result->b_size = (ret << inode->i_blkbits);
+		ret = 0;
+	}
+	return ret;
+
+}
+
 static int ext2_writepage(struct page *page, struct writeback_control *wbc)
 {
 	return block_write_full_page(page, ext2_get_block, wbc);
@@ -913,9 +997,10 @@ static void ext2_free_branches(struct inode *inode, __le32 *p, __le32 *q, int de
 		ext2_free_data(inode, p, q);
 }
 
-void ext2_truncate (struct inode * inode)
+void ext2_truncate(struct inode *inode)
 {
 	__le32 *i_data = EXT2_I(inode)->i_data;
+	struct ext2_inode_info *ei = EXT2_I(inode);
 	int addr_per_block = EXT2_ADDR_PER_BLOCK(inode->i_sb);
 	int offsets[4];
 	Indirect chain[4];
@@ -933,8 +1018,6 @@ void ext2_truncate (struct inode * inode)
 	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
 		return;
 
-	ext2_discard_prealloc(inode);
-
 	blocksize = inode->i_sb->s_blocksize;
 	iblock = (inode->i_size + blocksize-1)
 					>> EXT2_BLOCK_SIZE_BITS(inode->i_sb);
@@ -952,6 +1035,12 @@ void ext2_truncate (struct inode * inode)
 	if (n == 0)
 		return;
 
+	/*
+	 * From here we block out all ext2_get_block() callers who want to
+	 * modify the block allocation tree.
+	 */
+	mutex_lock(&ei->truncate_mutex);
+
 	if (n == 1) {
 		ext2_free_data(inode, i_data+offsets[0],
 					i_data + EXT2_NDIR_BLOCKS);
@@ -1004,6 +1093,10 @@ void ext2_truncate (struct inode * inode)
 		case EXT2_TIND_BLOCK:
 			;
 	}
+
+	ext2_discard_reservation(inode);
+
+	mutex_unlock(&ei->truncate_mutex);
 	inode->i_mtime = inode->i_ctime = CURRENT_TIME_SEC;
 	if (inode_needs_sync(inode)) {
 		sync_mapping_buffers(inode->i_mapping);
@@ -1104,6 +1197,8 @@ void ext2_read_inode (struct inode * inode)
 	ei->i_acl = EXT2_ACL_NOT_CACHED;
 	ei->i_default_acl = EXT2_ACL_NOT_CACHED;
 #endif
+	ei->i_block_alloc_info = NULL;
+
 	if (IS_ERR(raw_inode))
  		goto bad_inode;
 
@@ -1145,9 +1240,6 @@ void ext2_read_inode (struct inode * inode)
 	ei->i_dtime = 0;
 	inode->i_generation = le32_to_cpu(raw_inode->i_generation);
 	ei->i_state = 0;
-	ei->i_next_alloc_block = 0;
-	ei->i_next_alloc_goal = 0;
-	ei->i_prealloc_count = 0;
 	ei->i_block_group = (ino - 1) / EXT2_INODES_PER_GROUP(inode->i_sb);
 	ei->i_dir_start_lookup = 0;
 

fs/ext2/ioctl.c

@@ -22,6 +22,7 @@ int ext2_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 {
 	struct ext2_inode_info *ei = EXT2_I(inode);
 	unsigned int flags;
+	unsigned short rsv_window_size;
 
 	ext2_debug ("cmd = %u, arg = %lu\n", cmd, arg);
 
@@ -83,6 +84,50 @@ int ext2_ioctl (struct inode * inode, struct file * filp, unsigned int cmd,
 		inode->i_ctime = CURRENT_TIME_SEC;
 		mark_inode_dirty(inode);
 		return 0;
+	case EXT2_IOC_GETRSVSZ:
+		if (test_opt(inode->i_sb, RESERVATION)
+			&& S_ISREG(inode->i_mode)
+			&& ei->i_block_alloc_info) {
+			rsv_window_size = ei->i_block_alloc_info->rsv_window_node.rsv_goal_size;
+			return put_user(rsv_window_size, (int __user *)arg);
+		}
+		return -ENOTTY;
+	case EXT2_IOC_SETRSVSZ: {
+
+		if (!test_opt(inode->i_sb, RESERVATION) ||!S_ISREG(inode->i_mode))
+			return -ENOTTY;
+
+		if (IS_RDONLY(inode))
+			return -EROFS;
+
+		if ((current->fsuid != inode->i_uid) && !capable(CAP_FOWNER))
+			return -EACCES;
+
+		if (get_user(rsv_window_size, (int __user *)arg))
+			return -EFAULT;
+
+		if (rsv_window_size > EXT2_MAX_RESERVE_BLOCKS)
+			rsv_window_size = EXT2_MAX_RESERVE_BLOCKS;
+
+		/*
+		 * need to allocate reservation structure for this inode
+		 * before set the window size
+		 */
+		/*
+		 * XXX What lock should protect the rsv_goal_size?
+		 * Accessed in ext2_get_block only.  ext3 uses i_truncate.
+		 */
+		mutex_lock(&ei->truncate_mutex);
+		if (!ei->i_block_alloc_info)
+			ext2_init_block_alloc_info(inode);
+
+		if (ei->i_block_alloc_info){
+			struct ext2_reserve_window_node *rsv = &ei->i_block_alloc_info->rsv_window_node;
+			rsv->rsv_goal_size = rsv_window_size;
+		}
+		mutex_unlock(&ei->truncate_mutex);
+		return 0;
+	}
 	default:
 		return -ENOTTY;
 	}

fs/ext2/super.c

@@ -149,6 +149,7 @@ static struct inode *ext2_alloc_inode(struct super_block *sb)
 	ei->i_acl = EXT2_ACL_NOT_CACHED;
 	ei->i_default_acl = EXT2_ACL_NOT_CACHED;
 #endif
+	ei->i_block_alloc_info = NULL;
 	ei->vfs_inode.i_version = 1;
 	return &ei->vfs_inode;
 }
@@ -166,6 +167,7 @@ static void init_once(struct kmem_cache * cachep, void *foo)
 #ifdef CONFIG_EXT2_FS_XATTR
 	init_rwsem(&ei->xattr_sem);
 #endif
+	mutex_init(&ei->truncate_mutex);
 	inode_init_once(&ei->vfs_inode);
 }
 
@@ -188,6 +190,7 @@ static void destroy_inodecache(void)
 
 static void ext2_clear_inode(struct inode *inode)
 {
+	struct ext2_block_alloc_info *rsv = EXT2_I(inode)->i_block_alloc_info;
 #ifdef CONFIG_EXT2_FS_POSIX_ACL
 	struct ext2_inode_info *ei = EXT2_I(inode);
 
@@ -200,6 +203,10 @@ static void ext2_clear_inode(struct inode *inode)
 		ei->i_default_acl = EXT2_ACL_NOT_CACHED;
 	}
 #endif
+	ext2_discard_reservation(inode);
+	EXT2_I(inode)->i_block_alloc_info = NULL;
+	if (unlikely(rsv))
+		kfree(rsv);
 }
 
 static int ext2_show_options(struct seq_file *seq, struct vfsmount *vfs)
@@ -291,7 +298,6 @@ static const struct super_operations ext2_sops = {
 	.destroy_inode	= ext2_destroy_inode,
 	.read_inode	= ext2_read_inode,
 	.write_inode	= ext2_write_inode,
-	.put_inode	= ext2_put_inode,
 	.delete_inode	= ext2_delete_inode,
 	.put_super	= ext2_put_super,
 	.write_super	= ext2_write_super,
@@ -379,7 +385,7 @@ enum {
 	Opt_err_ro, Opt_nouid32, Opt_nocheck, Opt_debug,
 	Opt_oldalloc, Opt_orlov, Opt_nobh, Opt_user_xattr, Opt_nouser_xattr,
 	Opt_acl, Opt_noacl, Opt_xip, Opt_ignore, Opt_err, Opt_quota,
-	Opt_usrquota, Opt_grpquota
+	Opt_usrquota, Opt_grpquota, Opt_reservation, Opt_noreservation
 };
 
 static match_table_t tokens = {
@@ -411,6 +417,8 @@ static match_table_t tokens = {
 	{Opt_ignore, "noquota"},
 	{Opt_quota, "quota"},
 	{Opt_usrquota, "usrquota"},
+	{Opt_reservation, "reservation"},
+	{Opt_noreservation, "noreservation"},
 	{Opt_err, NULL}
 };
 
@@ -543,6 +551,14 @@ static int parse_options (char * options,
 			break;
 #endif
 
+		case Opt_reservation:
+			set_opt(sbi->s_mount_opt, RESERVATION);
+			printk("reservations ON\n");
+			break;
+		case Opt_noreservation:
+			clear_opt(sbi->s_mount_opt, RESERVATION);
+			printk("reservations OFF\n");
+			break;
 		case Opt_ignore:
 			break;
 		default:
@@ -784,6 +800,8 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
 	sbi->s_resuid = le16_to_cpu(es->s_def_resuid);
 	sbi->s_resgid = le16_to_cpu(es->s_def_resgid);
 	
+	set_opt(sbi->s_mount_opt, RESERVATION);
+
 	if (!parse_options ((char *) data, sbi))
 		goto failed_mount;
 
@@ -965,6 +983,21 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
 	get_random_bytes(&sbi->s_next_generation, sizeof(u32));
 	spin_lock_init(&sbi->s_next_gen_lock);
 
+	/* per fileystem reservation list head & lock */
+	spin_lock_init(&sbi->s_rsv_window_lock);
+	sbi->s_rsv_window_root = RB_ROOT;
+	/*
+	 * Add a single, static dummy reservation to the start of the
+	 * reservation window list --- it gives us a placeholder for
+	 * append-at-start-of-list which makes the allocation logic
+	 * _much_ simpler.
+	 */
+	sbi->s_rsv_window_head.rsv_start = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+	sbi->s_rsv_window_head.rsv_end = EXT2_RESERVE_WINDOW_NOT_ALLOCATED;
+	sbi->s_rsv_window_head.rsv_alloc_hit = 0;
+	sbi->s_rsv_window_head.rsv_goal_size = 0;
+	ext2_rsv_window_add(sb, &sbi->s_rsv_window_head);
+
 	err = percpu_counter_init(&sbi->s_freeblocks_counter,
 				ext2_count_free_blocks(sb));
 	if (!err) {
@@ -1260,7 +1293,7 @@ static ssize_t ext2_quota_read(struct super_block *sb, int type, char *data,
 
 		tmp_bh.b_state = 0;
 		err = ext2_get_block(inode, blk, &tmp_bh, 0);
-		if (err)
+		if (err < 0)
 			return err;
 		if (!buffer_mapped(&tmp_bh))	/* A hole? */
 			memset(data, 0, tocopy);
@@ -1299,7 +1332,7 @@ static ssize_t ext2_quota_write(struct super_block *sb, int type,
 
 		tmp_bh.b_state = 0;
 		err = ext2_get_block(inode, blk, &tmp_bh, 1);
-		if (err)
+		if (err < 0)
 			goto out;
 		if (offset || tocopy != EXT2_BLOCK_SIZE(sb))
 			bh = sb_bread(sb, tmp_bh.b_blocknr);

fs/ext2/xattr.c

@@ -664,8 +664,7 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 						           s_first_data_block) +
 				   EXT2_I(inode)->i_block_group *
 				   EXT2_BLOCKS_PER_GROUP(sb);
-			int block = ext2_new_block(inode, goal,
-						   NULL, NULL, &error);
+			int block = ext2_new_block(inode, goal, &error);
 			if (error)
 				goto cleanup;
 			ea_idebug(inode, "creating block %d", block);

include/linux/ext2_fs.h

@@ -29,11 +29,12 @@
 #undef EXT2FS_DEBUG
 
 /*
- * Define EXT2_PREALLOCATE to preallocate data blocks for expanding files
+ * Define EXT2_RESERVATION to reserve data blocks for expanding files
  */
-#define EXT2_PREALLOCATE
-#define EXT2_DEFAULT_PREALLOC_BLOCKS	8
-
+#define EXT2_DEFAULT_RESERVE_BLOCKS     8
+/*max window size: 1024(direct blocks) + 3([t,d]indirect blocks) */
+#define EXT2_MAX_RESERVE_BLOCKS         1027
+#define EXT2_RESERVE_WINDOW_NOT_ALLOCATED 0
 /*
  * The second extended file system version
  */
@@ -200,6 +201,8 @@ struct ext2_group_desc
 #define	EXT2_IOC_SETFLAGS		FS_IOC_SETFLAGS
 #define	EXT2_IOC_GETVERSION		FS_IOC_GETVERSION
 #define	EXT2_IOC_SETVERSION		FS_IOC_SETVERSION
+#define	EXT2_IOC_GETRSVSZ		_IOR('f', 5, long)
+#define	EXT2_IOC_SETRSVSZ		_IOW('f', 6, long)
 
 /*
  * ioctl commands in 32 bit emulation
@@ -319,6 +322,7 @@ struct ext2_inode {
 #define EXT2_MOUNT_XIP			0x010000  /* Execute in place */
 #define EXT2_MOUNT_USRQUOTA		0x020000  /* user quota */
 #define EXT2_MOUNT_GRPQUOTA		0x040000  /* group quota */
+#define EXT2_MOUNT_RESERVATION		0x080000  /* Preallocation */
 
 
 #define clear_opt(o, opt)		o &= ~EXT2_MOUNT_##opt
@@ -558,4 +562,11 @@ enum {
 #define EXT2_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT2_DIR_ROUND) & \
 					 ~EXT2_DIR_ROUND)
 
+static inline ext2_fsblk_t
+ext2_group_first_block_no(struct super_block *sb, unsigned long group_no)
+{
+	return group_no * (ext2_fsblk_t)EXT2_BLOCKS_PER_GROUP(sb) +
+		le32_to_cpu(EXT2_SB(sb)->s_es->s_first_data_block);
+}
+
 #endif	/* _LINUX_EXT2_FS_H */

include/linux/ext2_fs_sb.h

@@ -18,6 +18,52 @@
 
 #include <linux/blockgroup_lock.h>
 #include <linux/percpu_counter.h>
+#include <linux/rbtree.h>
+
+/* XXX Here for now... not interested in restructing headers JUST now */
+
+/* data type for block offset of block group */
+typedef int ext2_grpblk_t;
+
+/* data type for filesystem-wide blocks number */
+typedef unsigned long ext2_fsblk_t;
+
+#define E2FSBLK "%lu"
+
+struct ext2_reserve_window {
+	ext2_fsblk_t		_rsv_start;	/* First byte reserved */
+	ext2_fsblk_t		_rsv_end;	/* Last byte reserved or 0 */
+};
+
+struct ext2_reserve_window_node {
+	struct rb_node	 	rsv_node;
+	__u32			rsv_goal_size;
+	__u32			rsv_alloc_hit;
+	struct ext2_reserve_window	rsv_window;
+};
+
+struct ext2_block_alloc_info {
+	/* information about reservation window */
+	struct ext2_reserve_window_node	rsv_window_node;
+	/*
+	 * was i_next_alloc_block in ext2_inode_info
+	 * is the logical (file-relative) number of the
+	 * most-recently-allocated block in this file.
+	 * We use this for detecting linearly ascending allocation requests.
+	 */
+	__u32			last_alloc_logical_block;
+	/*
+	 * Was i_next_alloc_goal in ext2_inode_info
+	 * is the *physical* companion to i_next_alloc_block.
+	 * it the the physical block number of the block which was most-recentl
+	 * allocated to this file.  This give us the goal (target) for the next
+	 * allocation when we detect linearly ascending requests.
+	 */
+	ext2_fsblk_t		last_alloc_physical_block;
+};
+
+#define rsv_start rsv_window._rsv_start
+#define rsv_end rsv_window._rsv_end
 
 /*
  * second extended-fs super-block data in memory
@@ -56,6 +102,10 @@ struct ext2_sb_info {
 	struct percpu_counter s_freeinodes_counter;
 	struct percpu_counter s_dirs_counter;
 	struct blockgroup_lock s_blockgroup_lock;
+	/* root of the per fs reservation window tree */
+	spinlock_t s_rsv_window_lock;
+	struct rb_root s_rsv_window_root;
+	struct ext2_reserve_window_node s_rsv_window_head;
 };
 
 #endif	/* _LINUX_EXT2_FS_SB */