[PATCH 1/4] f2fs: check node page contents all the time

[PATCH 1/4] f2fs: check node page contents all the time

[Jaegeuk Kim]

In get_node_page, if the page is up-to-date, we assumed that the page was not
reclaimed at all.
But, sometimes it was reported that its contents was missing.
So, just for sure, let's check its mapping and contents.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/node.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index ea22c32..1e354ff 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1036,11 +1036,11 @@ repeat:
 	err = read_node_page(page, READ_SYNC);
 	if (err < 0)
 		return ERR_PTR(err);
-	else if (err == LOCKED_PAGE)
-		goto got_it;
+	else if (err != LOCKED_PAGE)
+		lock_page(page);
 
-	lock_page(page);
 	if (unlikely(!PageUptodate(page) || nid != nid_of_node(page))) {
+		ClearPageUptodate(page);
 		f2fs_put_page(page, 1);
 		return ERR_PTR(-EIO);
 	}
@@ -1048,7 +1048,6 @@ repeat:
 		f2fs_put_page(page, 1);
 		goto repeat;
 	}
-got_it:
 	return page;
 }
 
-- 
2.1.1

[PATCH 2/4] f2fs: call set_buffer_new for get_block

[Jaegeuk Kim]

This patch fixes wrong handling of buffer_new flag in get_block.
If f2fs allocates new blocks and mapped buffer_head, it needs to set buffer_new
for the bh_result.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/data.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index d166557..26e2476 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -274,7 +274,7 @@ static int check_extent_cache(struct inode *inode, pgoff_t pgofs,
 		unsigned int blkbits = inode->i_sb->s_blocksize_bits;
 		size_t count;
 
-		clear_buffer_new(bh_result);
+		set_buffer_new(bh_result);
 		map_bh(bh_result, inode->i_sb,
 				start_blkaddr + pgofs - start_fofs);
 		count = end_fofs - pgofs + 1;
@@ -634,12 +634,14 @@ static int __get_data_block(struct inode *inode, sector_t iblock,
 		goto put_out;
 
 	if (dn.data_blkaddr != NULL_ADDR) {
+		set_buffer_new(bh_result);
 		map_bh(bh_result, inode->i_sb, dn.data_blkaddr);
 	} else if (create) {
 		err = __allocate_data_block(&dn);
 		if (err)
 			goto put_out;
 		allocated = true;
+		set_buffer_new(bh_result);
 		map_bh(bh_result, inode->i_sb, dn.data_blkaddr);
 	} else {
 		goto put_out;
-- 
2.1.1

[PATCH 3/4] f2fs: introduce macros to convert bytes and blocks in f2fs

[Jaegeuk Kim]

This patch adds two macros for transition between byte and block offsets.
Currently, f2fs only supports 4KB blocks, so use the default size for now.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/checkpoint.c    | 7 +++----
 fs/f2fs/file.c          | 3 +--
 fs/f2fs/segment.c       | 8 ++++----
 include/linux/f2fs_fs.h | 4 ++++
 4 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
index 31a715b..58d88df 100644
--- a/fs/f2fs/checkpoint.c
+++ b/fs/f2fs/checkpoint.c
@@ -981,15 +981,14 @@ static void do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 	/* write out checkpoint buffer at block 0 */
 	cp_page = grab_meta_page(sbi, start_blk++);
 	kaddr = page_address(cp_page);
-	memcpy(kaddr, ckpt, (1 << sbi->log_blocksize));
+	memcpy(kaddr, ckpt, F2FS_BLKSIZE);
 	set_page_dirty(cp_page);
 	f2fs_put_page(cp_page, 1);
 
 	for (i = 1; i < 1 + cp_payload_blks; i++) {
 		cp_page = grab_meta_page(sbi, start_blk++);
 		kaddr = page_address(cp_page);
-		memcpy(kaddr, (char *)ckpt + i * F2FS_BLKSIZE,
-				(1 << sbi->log_blocksize));
+		memcpy(kaddr, (char *)ckpt + i * F2FS_BLKSIZE, F2FS_BLKSIZE);
 		set_page_dirty(cp_page);
 		f2fs_put_page(cp_page, 1);
 	}
@@ -1009,7 +1008,7 @@ static void do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
 	/* writeout checkpoint block */
 	cp_page = grab_meta_page(sbi, start_blk);
 	kaddr = page_address(cp_page);
-	memcpy(kaddr, ckpt, (1 << sbi->log_blocksize));
+	memcpy(kaddr, ckpt, F2FS_BLKSIZE);
 	set_page_dirty(cp_page);
 	f2fs_put_page(cp_page, 1);
 
diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
index 7dc0ed8..790edb9 100644
--- a/fs/f2fs/file.c
+++ b/fs/f2fs/file.c
@@ -491,8 +491,7 @@ int truncate_blocks(struct inode *inode, u64 from, bool lock)
 
 	trace_f2fs_truncate_blocks_enter(inode, from);
 
-	free_from = (pgoff_t)
-		((from + blocksize - 1) >> (sbi->log_blocksize));
+	free_from = (pgoff_t)F2FS_BYTES_TO_BLK(from + blocksize - 1);
 
 	if (lock)
 		f2fs_lock_op(sbi);
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 9f278d1..877a272 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -1048,8 +1048,8 @@ static const struct segment_allocation default_salloc_ops = {
 
 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range)
 {
-	__u64 start = range->start >> sbi->log_blocksize;
-	__u64 end = start + (range->len >> sbi->log_blocksize) - 1;
+	__u64 start = F2FS_BYTES_TO_BLK(range->start);
+	__u64 end = start + F2FS_BYTES_TO_BLK(range->len) - 1;
 	unsigned int start_segno, end_segno;
 	struct cp_control cpc;
 
@@ -1066,7 +1066,7 @@ int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range)
 	end_segno = (end >= MAX_BLKADDR(sbi)) ? MAIN_SEGS(sbi) - 1 :
 						GET_SEGNO(sbi, end);
 	cpc.reason = CP_DISCARD;
-	cpc.trim_minlen = range->minlen >> sbi->log_blocksize;
+	cpc.trim_minlen = F2FS_BYTES_TO_BLK(range->minlen);
 
 	/* do checkpoint to issue discard commands safely */
 	for (; start_segno <= end_segno; start_segno = cpc.trim_end + 1) {
@@ -1080,7 +1080,7 @@ int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range)
 		mutex_unlock(&sbi->gc_mutex);
 	}
 out:
-	range->len = cpc.trimmed << sbi->log_blocksize;
+	range->len = F2FS_BLK_TO_BYTES(cpc.trimmed);
 	return 0;
 }
 
diff --git a/include/linux/f2fs_fs.h b/include/linux/f2fs_fs.h
index 0980572..a23556c 100644
--- a/include/linux/f2fs_fs.h
+++ b/include/linux/f2fs_fs.h
@@ -19,12 +19,16 @@
 #define F2FS_MAX_LOG_SECTOR_SIZE	12	/* 12 bits for 4096 bytes */
 #define F2FS_LOG_SECTORS_PER_BLOCK	3	/* log number for sector/blk */
 #define F2FS_BLKSIZE			4096	/* support only 4KB block */
+#define F2FS_BLKSIZE_BITS		12	/* bits for F2FS_BLKSIZE */
 #define F2FS_MAX_EXTENSION		64	/* # of extension entries */
 #define F2FS_BLK_ALIGN(x)	(((x) + F2FS_BLKSIZE - 1) / F2FS_BLKSIZE)
 
 #define NULL_ADDR		((block_t)0)	/* used as block_t addresses */
 #define NEW_ADDR		((block_t)-1)	/* used as block_t addresses */
 
+#define F2FS_BYTES_TO_BLK(bytes)	((bytes) >> F2FS_BLKSIZE_BITS)
+#define F2FS_BLK_TO_BYTES(blk)		((blk) << F2FS_BLKSIZE_BITS)
+
 /* 0, 1(node nid), 2(meta nid) are reserved node id */
 #define F2FS_RESERVED_NODE_NUM		3
 
-- 
2.1.1

[PATCH 4/4] f2fs: allocate data blocks in advance for f2fs_direct_IO

[Jaegeuk Kim]

This patch adds preallocation for data blocks to prepare f2fs_direct_IO.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/data.c | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 54 insertions(+), 3 deletions(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 26e2476..985ed02 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -592,6 +592,56 @@ static int __allocate_data_block(struct dnode_of_data *dn)
 	return 0;
 }
 
+static void __allocate_data_blocks(struct inode *inode, loff_t offset,
+							size_t count)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct dnode_of_data dn;
+	u64 start = F2FS_BYTES_TO_BLK(offset);
+	u64 len = F2FS_BYTES_TO_BLK(count);
+	bool allocated;
+	u64 end_offset;
+
+	while (len) {
+		f2fs_balance_fs(sbi);
+		f2fs_lock_op(sbi);
+
+		/* When reading holes, we need its node page */
+		set_new_dnode(&dn, inode, NULL, NULL, 0);
+		if (get_dnode_of_data(&dn, start, ALLOC_NODE))
+			goto out;
+
+		allocated = false;
+		end_offset = ADDRS_PER_PAGE(dn.node_page, F2FS_I(inode));
+
+		while (dn.ofs_in_node < end_offset && len) {
+			if (dn.data_blkaddr == NULL_ADDR) {
+				if (__allocate_data_block(&dn))
+					goto sync_out;
+				allocated = true;
+			}
+			len--;
+			start++;
+			dn.ofs_in_node++;
+		}
+
+		if (allocated)
+			sync_inode_page(&dn);
+
+		f2fs_put_dnode(&dn);
+		f2fs_unlock_op(sbi);
+	}
+	return;
+
+sync_out:
+	if (allocated)
+		sync_inode_page(&dn);
+	f2fs_put_dnode(&dn);
+out:
+	f2fs_unlock_op(sbi);
+	return;
+}
+
 /*
  * get_data_block() now supported readahead/bmap/rw direct_IO with mapped bh.
  * If original data blocks are allocated, then give them to blockdev.
@@ -617,10 +667,8 @@ static int __get_data_block(struct inode *inode, sector_t iblock,
 	if (check_extent_cache(inode, pgofs, bh_result))
 		goto out;
 
-	if (create) {
-		f2fs_balance_fs(F2FS_I_SB(inode));
+	if (create)
 		f2fs_lock_op(F2FS_I_SB(inode));
-	}
 
 	/* When reading holes, we need its node page */
 	set_new_dnode(&dn, inode, NULL, NULL, 0);
@@ -1108,6 +1156,9 @@ static ssize_t f2fs_direct_IO(int rw, struct kiocb *iocb,
 
 	trace_f2fs_direct_IO_enter(inode, offset, count, rw);
 
+	if (rw & WRITE)
+		__allocate_data_blocks(inode, offset, count);
+
 	err = blockdev_direct_IO(rw, iocb, inode, iter, offset, get_data_block);
 	if (err < 0 && (rw & WRITE))
 		f2fs_write_failed(mapping, offset + count);
-- 
2.1.1