This patch removes check_prefree_segments initially designed to enhance the
performance by narrowing the range of LBA usage across the whole block device.
When allocating a new segment, previous f2fs tries to find proper prefree
segments, and then, if finds a segment, it reuses the segment for further
data or node block allocation.
However, I found that this was totally wrong approach since the prefree segments
have several data or node blocks that will be used by the roll-forward mechanism
operated after sudden-power-off.
Let's assume the following scenario.
/* write 8MB with fsync */
for (i = 0; i < 2048; i++) {
offset = i * 4096;
write(fd, offset, 4KB);
fsync(fd);
}
In this case, naive segment allocation sequence will be like:
data segment: x, x+1, x+2, x+3
node segment: y, y+1, y+2, y+3.
But, if we can reuse prefree segments, the sequence can be like:
data segment: x, x+1, y, y+1
node segment: y, y+1, y+2, y+3.
Because, y, y+1, and y+2 became prefree segments one by one, and those are
reused by data allocation.
After conducting this workload, we should consider how to recover the latest
inode with its data.
If we reuse the prefree segments such as y or y+1, we lost the old node blocks
so that f2fs even cannot start roll-forward recovery.
Therefore, I suggest that we should remove reusing prefree segments.
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/segment.c | 56 +------------------------------------------------------
1 file changed, 1 insertion(+), 55 deletions(-)
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 3ac4d29..a86d125 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -309,56 +309,6 @@ static void write_sum_page(struct f2fs_sb_info *sbi,
f2fs_put_page(page, 1);
}
-static unsigned int check_prefree_segments(struct f2fs_sb_info *sbi, int type)
-{
- struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
- unsigned long *prefree_segmap = dirty_i->dirty_segmap[PRE];
- unsigned int segno;
- unsigned int ofs = 0;
-
- /*
- * If there is not enough reserved sections,
- * we should not reuse prefree segments.
- */
- if (has_not_enough_free_secs(sbi, 0))
- return NULL_SEGNO;
-
- /*
- * NODE page should not reuse prefree segment,
- * since those information is used for SPOR.
- */
- if (IS_NODESEG(type))
- return NULL_SEGNO;
-next:
- segno = find_next_bit(prefree_segmap, TOTAL_SEGS(sbi), ofs);
- ofs += sbi->segs_per_sec;
-
- if (segno < TOTAL_SEGS(sbi)) {
- int i;
-
- /* skip intermediate segments in a section */
- if (segno % sbi->segs_per_sec)
- goto next;
-
- /* skip if the section is currently used */
- if (sec_usage_check(sbi, GET_SECNO(sbi, segno)))
- goto next;
-
- /* skip if whole section is not prefree */
- for (i = 1; i < sbi->segs_per_sec; i++)
- if (!test_bit(segno + i, prefree_segmap))
- goto next;
-
- /* skip if whole section was not free at the last checkpoint */
- for (i = 0; i < sbi->segs_per_sec; i++)
- if (get_seg_entry(sbi, segno + i)->ckpt_valid_blocks)
- goto next;
-
- return segno;
- }
- return NULL_SEGNO;
-}
-
static int is_next_segment_free(struct f2fs_sb_info *sbi, int type)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
@@ -597,11 +547,7 @@ static void allocate_segment_by_default(struct f2fs_sb_info *sbi,
goto out;
}
- curseg->next_segno = check_prefree_segments(sbi, type);
-
- if (curseg->next_segno != NULL_SEGNO)
- change_curseg(sbi, type, false);
- else if (type == CURSEG_WARM_NODE)
+ if (type == CURSEG_WARM_NODE)
new_curseg(sbi, type, false);
else if (curseg->alloc_type == LFS && is_next_segment_free(sbi, type))
new_curseg(sbi, type, false);
--
1.8.3.1.437.g0dbd812
If user requests many data writes and fsync together, the last updated i_size
should be stored to the inode block consistently.
But, previous write_end just marks the inode as dirty and doesn't update its
metadata into its inode block.
After that, fsync just writes the inode block with newly updated data index
excluding inode metadata updates.
So, this patch introduces write_end in which updates inode block too when the
i_size is changed.
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/data.c | 23 ++++++++++++++++++++++-
1 file changed, 22 insertions(+), 1 deletion(-)
diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 6d4a743..e88f46f 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -701,6 +701,27 @@ err:
return err;
}
+static int f2fs_write_end(struct file *file,
+ struct address_space *mapping,
+ loff_t pos, unsigned len, unsigned copied,
+ struct page *page, void *fsdata)
+{
+ struct inode *inode = page->mapping->host;
+
+ SetPageUptodate(page);
+ set_page_dirty(page);
+
+ if (pos + copied > i_size_read(inode)) {
+ i_size_write(inode, pos + copied);
+ mark_inode_dirty(inode);
+ update_inode_page(inode);
+ }
+
+ unlock_page(page);
+ page_cache_release(page);
+ return copied;
+}
+
static ssize_t f2fs_direct_IO(int rw, struct kiocb *iocb,
const struct iovec *iov, loff_t offset, unsigned long nr_segs)
{
@@ -757,7 +778,7 @@ const struct address_space_operations f2fs_dblock_aops = {
.writepage = f2fs_write_data_page,
.writepages = f2fs_write_data_pages,
.write_begin = f2fs_write_begin,
- .write_end = nobh_write_end,
+ .write_end = f2fs_write_end,
.set_page_dirty = f2fs_set_data_page_dirty,
.invalidatepage = f2fs_invalidate_data_page,
.releasepage = f2fs_release_data_page,
--
1.8.3.1.437.g0dbd812