ext4: enable "punch hole" functionality

This patch adds new routines: "ext4_punch_hole" "ext4_ext_punch_hole"
and "ext4_ext_check_cache"

fallocate has been modified to call ext4_punch_hole when the punch hole
flag is passed.  At the moment, we only support punching holes in
extents, so this routine is pretty much a wrapper for the ext4_ext_punch_hole
routine.

The ext4_ext_punch_hole routine first completes all outstanding writes
with the associated pages, and then releases them.  The unblock
aligned data is zeroed, and all blocks in between are punched out.

The ext4_ext_check_cache routine is very similar to ext4_ext_in_cache
except it accepts a ext4_ext_cache parameter instead of a ext4_extent
parameter.  This routine is used by ext4_ext_punch_hole to check and
see if a block in a hole that has been cached.  The ext4_ext_cache
parameter is necessary because the members ext4_extent structure are
not large enough to hold a 32 bit value.  The existing
ext4_ext_in_cache routine has become a wrapper to this new function.

[ext4 punch hole patch series 5/5 v7] 

Signed-off-by: Allison Henderson <achender@us.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reviewed-by: Mingming Cao <cmm@us.ibm.com>

3 files changed, 253 insertions, 11 deletions

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index c48f9db5b96b..7bc7c724805d 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1816,6 +1816,7 @@ extern int ext4_change_inode_journal_flag(struct inode *, int);
 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
 extern int ext4_can_truncate(struct inode *inode);
 extern void ext4_truncate(struct inode *);
+extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
 extern void ext4_set_inode_flags(struct inode *);
 extern void ext4_get_inode_flags(struct ext4_inode_info *);
@@ -2157,6 +2158,8 @@ extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 			       struct ext4_map_blocks *map, int flags);
 extern void ext4_ext_truncate(struct inode *);
+extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
+				loff_t length);
 extern void ext4_ext_init(struct super_block *);
 extern void ext4_ext_release(struct super_block *);
 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 1d456b2ac377..88ff3a74787b 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -2044,12 +2044,23 @@ ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
 }
 
 /*
+ * ext4_ext_in_cache()
+ * Checks to see if the given block is in the cache.
+ * If it is, the cached extent is stored in the given
+ * cache extent pointer.  If the cached extent is a hole,
+ * this routine should be used instead of
+ * ext4_ext_in_cache if the calling function needs to
+ * know the size of the hole.
+ *
+ * @inode: The files inode
+ * @block: The block to look for in the cache
+ * @ex:    Pointer where the cached extent will be stored
+ *         if it contains block
+ *
  * Return 0 if cache is invalid; 1 if the cache is valid
  */
-static int
-ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
-			struct ext4_extent *ex)
-{
+static int ext4_ext_check_cache(struct inode *inode, ext4_lblk_t block,
+	struct ext4_ext_cache *ex){
 	struct ext4_ext_cache *cex;
 	struct ext4_sb_info *sbi;
 	int ret = 0;
@@ -2066,9 +2077,7 @@ ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
 		goto errout;
 
 	if (in_range(block, cex->ec_block, cex->ec_len)) {
-		ex->ee_block = cpu_to_le32(cex->ec_block);
-		ext4_ext_store_pblock(ex, cex->ec_start);
-		ex->ee_len = cpu_to_le16(cex->ec_len);
+		memcpy(ex, cex, sizeof(struct ext4_ext_cache));
 		ext_debug("%u cached by %u:%u:%llu\n",
 				block,
 				cex->ec_block, cex->ec_len, cex->ec_start);
@@ -2084,6 +2093,37 @@ errout:
 }
 
 /*
+ * ext4_ext_in_cache()
+ * Checks to see if the given block is in the cache.
+ * If it is, the cached extent is stored in the given
+ * extent pointer.
+ *
+ * @inode: The files inode
+ * @block: The block to look for in the cache
+ * @ex:    Pointer where the cached extent will be stored
+ *         if it contains block
+ *
+ * Return 0 if cache is invalid; 1 if the cache is valid
+ */
+static int
+ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
+			struct ext4_extent *ex)
+{
+	struct ext4_ext_cache cex;
+	int ret = 0;
+
+	if (ext4_ext_check_cache(inode, block, &cex)) {
+		ex->ee_block = cpu_to_le32(cex.ec_block);
+		ext4_ext_store_pblock(ex, cex.ec_start);
+		ex->ee_len = cpu_to_le16(cex.ec_len);
+		ret = 1;
+	}
+
+	return ret;
+}
+
+
+/*
  * ext4_ext_rm_idx:
  * removes index from the index block.
  * It's used in truncate case only, thus all requests are for
@@ -3724,10 +3764,6 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 	struct ext4_map_blocks map;
 	unsigned int credits, blkbits = inode->i_blkbits;
 
-	/* We only support the FALLOC_FL_KEEP_SIZE mode */
-	if (mode & ~FALLOC_FL_KEEP_SIZE)
-		return -EOPNOTSUPP;
-
 	/*
 	 * currently supporting (pre)allocate mode for extent-based
 	 * files _only_
@@ -3735,6 +3771,13 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
 		return -EOPNOTSUPP;
 
+	/* Return error if mode is not supported */
+	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
+		return -EOPNOTSUPP;
+
+	if (mode & FALLOC_FL_PUNCH_HOLE)
+		return ext4_punch_hole(file, offset, len);
+
 	trace_ext4_fallocate_enter(inode, offset, len, mode);
 	map.m_lblk = offset >> blkbits;
 	/*
@@ -4100,6 +4143,177 @@ static int ext4_xattr_fiemap(struct inode *inode,
 	return (error < 0 ? error : 0);
 }
 
+/*
+ * ext4_ext_punch_hole
+ *
+ * Punches a hole of "length" bytes in a file starting
+ * at byte "offset"
+ *
+ * @inode:  The inode of the file to punch a hole in
+ * @offset: The starting byte offset of the hole
+ * @length: The length of the hole
+ *
+ * Returns the number of blocks removed or negative on err
+ */
+int ext4_ext_punch_hole(struct file *file, loff_t offset, loff_t length)
+{
+	struct inode *inode = file->f_path.dentry->d_inode;
+	struct super_block *sb = inode->i_sb;
+	struct ext4_ext_cache cache_ex;
+	ext4_lblk_t first_block, last_block, num_blocks, iblock, max_blocks;
+	struct address_space *mapping = inode->i_mapping;
+	struct ext4_map_blocks map;
+	handle_t *handle;
+	loff_t first_block_offset, last_block_offset, block_len;
+	loff_t first_page, last_page, first_page_offset, last_page_offset;
+	int ret, credits, blocks_released, err = 0;
+
+	first_block = (offset + sb->s_blocksize - 1) >>
+		EXT4_BLOCK_SIZE_BITS(sb);
+	last_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);
+
+	first_block_offset = first_block << EXT4_BLOCK_SIZE_BITS(sb);
+	last_block_offset = last_block << EXT4_BLOCK_SIZE_BITS(sb);
+
+	first_page = (offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	last_page = (offset + length) >> PAGE_CACHE_SHIFT;
+
+	first_page_offset = first_page << PAGE_CACHE_SHIFT;
+	last_page_offset = last_page << PAGE_CACHE_SHIFT;
+
+	/*
+	 * Write out all dirty pages to avoid race conditions
+	 * Then release them.
+	 */
+	if (mapping->nrpages && mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
+		err = filemap_write_and_wait_range(mapping,
+			first_page_offset == 0 ? 0 : first_page_offset-1,
+			last_page_offset);
+
+			if (err)
+				return err;
+	}
+
+	/* Now release the pages */
+	if (last_page_offset > first_page_offset) {
+		truncate_inode_pages_range(mapping, first_page_offset,
+					   last_page_offset-1);
+	}
+
+	/* finish any pending end_io work */
+	ext4_flush_completed_IO(inode);
+
+	credits = ext4_writepage_trans_blocks(inode);
+	handle = ext4_journal_start(inode, credits);
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
+
+	err = ext4_orphan_add(handle, inode);
+	if (err)
+		goto out;
+
+	/*
+	 * Now we need to zero out the un block aligned data.
+	 * If the file is smaller than a block, just
+	 * zero out the middle
+	 */
+	if (first_block > last_block)
+		ext4_block_zero_page_range(handle, mapping, offset, length);
+	else {
+		/* zero out the head of the hole before the first block */
+		block_len  = first_block_offset - offset;
+		if (block_len > 0)
+			ext4_block_zero_page_range(handle, mapping,
+						   offset, block_len);
+
+		/* zero out the tail of the hole after the last block */
+		block_len = offset + length - last_block_offset;
+		if (block_len > 0) {
+			ext4_block_zero_page_range(handle, mapping,
+					last_block_offset, block_len);
+		}
+	}
+
+	/* If there are no blocks to remove, return now */
+	if (first_block >= last_block)
+		goto out;
+
+	down_write(&EXT4_I(inode)->i_data_sem);
+	ext4_ext_invalidate_cache(inode);
+	ext4_discard_preallocations(inode);
+
+	/*
+	 * Loop over all the blocks and identify blocks
+	 * that need to be punched out
+	 */
+	iblock = first_block;
+	blocks_released = 0;
+	while (iblock < last_block) {
+		max_blocks = last_block - iblock;
+		num_blocks = 1;
+		memset(&map, 0, sizeof(map));
+		map.m_lblk = iblock;
+		map.m_len = max_blocks;
+		ret = ext4_ext_map_blocks(handle, inode, &map,
+			EXT4_GET_BLOCKS_PUNCH_OUT_EXT);
+
+		if (ret > 0) {
+			blocks_released += ret;
+			num_blocks = ret;
+		} else if (ret == 0) {
+			/*
+			 * If map blocks could not find the block,
+			 * then it is in a hole.  If the hole was
+			 * not already cached, then map blocks should
+			 * put it in the cache.  So we can get the hole
+			 * out of the cache
+			 */
+			memset(&cache_ex, 0, sizeof(cache_ex));
+			if ((ext4_ext_check_cache(inode, iblock, &cache_ex)) &&
+				!cache_ex.ec_start) {
+
+				/* The hole is cached */
+				num_blocks = cache_ex.ec_block +
+				cache_ex.ec_len - iblock;
+
+			} else {
+				/* The block could not be identified */
+				err = -EIO;
+				break;
+			}
+		} else {
+			/* Map blocks error */
+			err = ret;
+			break;
+		}
+
+		if (num_blocks == 0) {
+			/* This condition should never happen */
+			ext_debug("Block lookup failed");
+			err = -EIO;
+			break;
+		}
+
+		iblock += num_blocks;
+	}
+
+	if (blocks_released > 0) {
+		ext4_ext_invalidate_cache(inode);
+		ext4_discard_preallocations(inode);
+	}
+
+	if (IS_SYNC(inode))
+		ext4_handle_sync(handle);
+
+	up_write(&EXT4_I(inode)->i_data_sem);
+
+out:
+	ext4_orphan_del(handle, inode);
+	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
+	ext4_mark_inode_dirty(handle, inode);
+	ext4_journal_stop(handle);
+	return err;
+}
 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		__u64 start, __u64 len)
 {
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index fbe11e676624..65576c02c6f5 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -4419,6 +4419,31 @@ int ext4_can_truncate(struct inode *inode)
 }
 
 /*
+ * ext4_punch_hole: punches a hole in a file by releaseing the blocks
+ * associated with the given offset and length
+ *
+ * @inode:  File inode
+ * @offset: The offset where the hole will begin
+ * @len:    The length of the hole
+ *
+ * Returns: 0 on sucess or negative on failure
+ */
+
+int ext4_punch_hole(struct file *file, loff_t offset, loff_t length)
+{
+	struct inode *inode = file->f_path.dentry->d_inode;
+	if (!S_ISREG(inode->i_mode))
+		return -ENOTSUPP;
+
+	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
+		/* TODO: Add support for non extent hole punching */
+		return -ENOTSUPP;
+	}
+
+	return ext4_ext_punch_hole(file, offset, length);
+}
+
+/*
  * ext4_truncate()
  *
  * We block out ext4_get_block() block instantiations across the entire