f2fs: avoid issuing small bios due to several dirty node pages

If some small bios of dirty node pages are supposed to be issued during the
sequential data writes, there-in well-produced consecutive data bios are able
to be split by the small node bios, resulting in performance degradation.
So, let's collect a number of dirty node pages until reaching a threshold.
And, by default, I set the threshold as 2MB, a segment size.

This improves sequential write performance on i5, 512GB SSD (830 w/ SATA2) as
follows.
Before: 231 MB/s -> After: 255 MB/s

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com>

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index f177c018745c6e8e495bdf50d53f2fa64245c806..9bda63c9c166a886c1585cc34de06dfef3f59cc2 100644 (file)
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1124,6 +1124,12 @@ static int f2fs_write_node_page(struct page *page,
        return 0;
 }
 
+/*
+ * It is very important to gather dirty pages and write at once, so that we can
+ * submit a big bio without interfering other data writes.
+ * Be default, 512 pages (2MB), a segment size, is quite reasonable.
+ */
+#define COLLECT_DIRTY_NODES    512
 static int f2fs_write_node_pages(struct address_space *mapping,
                            struct writeback_control *wbc)
 {
@@ -1131,17 +1137,16 @@ static int f2fs_write_node_pages(struct address_space *mapping,
        struct block_device *bdev = sbi->sb->s_bdev;
        long nr_to_write = wbc->nr_to_write;
 
-       if (wbc->for_kupdate)
-               return 0;
-
-       if (get_pages(sbi, F2FS_DIRTY_NODES) == 0)
-               return 0;
-
+       /* First check balancing cached NAT entries */
        if (try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK)) {
                write_checkpoint(sbi, false, false);
                return 0;
        }
 
+       /* collect a number of dirty node pages and write together */
+       if (get_pages(sbi, F2FS_DIRTY_NODES) < COLLECT_DIRTY_NODES)
+               return 0;
+
        /* if mounting is failed, skip writing node pages */
        wbc->nr_to_write = bio_get_nr_vecs(bdev);
        sync_node_pages(sbi, 0, wbc);