f2fs: introduce f2fs_submit_merged_bio_cond

f2fs use single bio buffer per type data (META/NODE/DATA) for caching
writes locating in continuous block address as many as possible, after
submitting, these writes may be still cached in bio buffer, so we have
to flush cached writes in bio buffer by calling f2fs_submit_merged_bio.

Unfortunately, in the scenario of high concurrency, bio buffer could be
flushed by someone else before we submit it as below reasons:
a) there is no space in bio buffer.
b) add a request of different type (SYNC, ASYNC).
c) add a discontinuous block address.

For this condition, f2fs_submit_merged_bio will be devastating, because
it could break the following merging of writes in bio buffer, split one
big bio into two smaller one.

This patch introduces f2fs_submit_merged_bio_cond which can do a
conditional submitting with bio buffer, before submitting it will judge
whether:
 - page in DATA type bio buffer is matching with specified page;
 - page in DATA type bio buffer is belong to specified inode;
 - page in NODE type bio buffer is belong to specified inode;
If there is no eligible page in bio buffer, we will skip submitting step,
result in gaining more chance to merge consecutive block IOs in bio cache.

Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
index 96d606d93f2fdee9bd6bbb5d26b5711f37c3dab5..9c0d721598e5d566b025a6f8ca97d0c6531f06d0 100644 (file)
--- a/fs/f2fs/checkpoint.c
+++ b/fs/f2fs/checkpoint.c
@@ -234,10 +234,15 @@ static int f2fs_write_meta_page(struct page *page,
 
        write_meta_page(sbi, page);
        dec_page_count(sbi, F2FS_DIRTY_META);
+
+       if (wbc->for_reclaim)
+               f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, META, WRITE);
+
        unlock_page(page);
 
-       if (wbc->for_reclaim || unlikely(f2fs_cp_error(sbi)))
+       if (unlikely(f2fs_cp_error(sbi)))
                f2fs_submit_merged_bio(sbi, META, WRITE);
+
        return 0;
 
 redirty_out:

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index e0b09c33fc0f91b748e574dfdcf3164d9888f7d7..9ae43a725d54bd1fab61d95cd4355b083684912b 100644 (file)
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -114,20 +114,18 @@ static void __submit_merged_bio(struct f2fs_bio_info *io)
        io->bio = NULL;
 }
 
-bool is_merged_page(struct f2fs_sb_info *sbi, struct page *page,
-                                                       enum page_type type)
+static bool __has_merged_page(struct f2fs_bio_info *io, struct inode *inode,
+                                               struct page *page, nid_t ino)
 {
-       enum page_type btype = PAGE_TYPE_OF_BIO(type);
-       struct f2fs_bio_info *io = &sbi->write_io[btype];
        struct bio_vec *bvec;
        struct page *target;
        int i;
 
-       down_read(&io->io_rwsem);
-       if (!io->bio) {
-               up_read(&io->io_rwsem);
+       if (!io->bio)
                return false;
-       }
+
+       if (!inode && !page && !ino)
+               return true;
 
        bio_for_each_segment_all(bvec, io->bio, i) {
 
@@ -142,18 +140,34 @@ bool is_merged_page(struct f2fs_sb_info *sbi, struct page *page,
                        target = ctx->w.control_page;
                }
 
-               if (page == target) {
-                       up_read(&io->io_rwsem);
+               if (inode && inode == target->mapping->host)
+                       return true;
+               if (page && page == target)
+                       return true;
+               if (ino && ino == ino_of_node(target))
                        return true;
-               }
        }
 
-       up_read(&io->io_rwsem);
        return false;
 }
 
-void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
-                               enum page_type type, int rw)
+static bool has_merged_page(struct f2fs_sb_info *sbi, struct inode *inode,
+                                               struct page *page, nid_t ino,
+                                               enum page_type type)
+{
+       enum page_type btype = PAGE_TYPE_OF_BIO(type);
+       struct f2fs_bio_info *io = &sbi->write_io[btype];
+       bool ret;
+
+       down_read(&io->io_rwsem);
+       ret = __has_merged_page(io, inode, page, ino);
+       up_read(&io->io_rwsem);
+       return ret;
+}
+
+static void __f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
+                               struct inode *inode, struct page *page,
+                               nid_t ino, enum page_type type, int rw)
 {
        enum page_type btype = PAGE_TYPE_OF_BIO(type);
        struct f2fs_bio_info *io;
@@ -162,6 +176,9 @@ void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
 
        down_write(&io->io_rwsem);
 
+       if (!__has_merged_page(io, inode, page, ino))
+               goto out;
+
        /* change META to META_FLUSH in the checkpoint procedure */
        if (type >= META_FLUSH) {
                io->fio.type = META_FLUSH;
@@ -171,9 +188,24 @@ void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
                        io->fio.rw = WRITE_FLUSH_FUA | REQ_META | REQ_PRIO;
        }
        __submit_merged_bio(io);
+out:
        up_write(&io->io_rwsem);
 }
 
+void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi, enum page_type type,
+                                                                       int rw)
+{
+       __f2fs_submit_merged_bio(sbi, NULL, NULL, 0, type, rw);
+}
+
+void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *sbi,
+                               struct inode *inode, struct page *page,
+                               nid_t ino, enum page_type type, int rw)
+{
+       if (has_merged_page(sbi, inode, page, ino, type))
+               __f2fs_submit_merged_bio(sbi, inode, page, ino, type, rw);
+}
+
 /*
  * Fill the locked page with data located in the block address.
  * Return unlocked page.
@@ -1140,12 +1172,18 @@ out:
        inode_dec_dirty_pages(inode);
        if (err)
                ClearPageUptodate(page);
+
+       if (wbc->for_reclaim) {
+               f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, DATA, WRITE);
+               remove_dirty_inode(inode);
+       }
+
        unlock_page(page);
        f2fs_balance_fs(sbi, need_balance_fs);
-       if (wbc->for_reclaim || unlikely(f2fs_cp_error(sbi))) {
+
+       if (unlikely(f2fs_cp_error(sbi)))
                f2fs_submit_merged_bio(sbi, DATA, WRITE);
-               remove_dirty_inode(inode);
-       }
+
        return 0;
 
 redirty_out:
@@ -1333,7 +1371,7 @@ static int f2fs_write_data_pages(struct address_space *mapping,
                locked = true;
        }
        ret = f2fs_write_cache_pages(mapping, wbc, __f2fs_writepage, mapping);
-       f2fs_submit_merged_bio(sbi, DATA, WRITE);
+       f2fs_submit_merged_bio_cond(sbi, inode, NULL, 0, DATA, WRITE);
        if (locked)
                mutex_unlock(&sbi->writepages);
 

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 469a1f49a3cdd3a3cc407717b43034057f79ebb3..5f98236f4d580f18a92da4e77666d56d84b16bc9 100644 (file)
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1897,8 +1897,9 @@ void destroy_checkpoint_caches(void);
 /*
  * data.c
  */
-bool is_merged_page(struct f2fs_sb_info *, struct page *, enum page_type);
 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
+void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *, struct inode *,
+                               struct page *, nid_t, enum page_type, int);
 int f2fs_submit_page_bio(struct f2fs_io_info *);
 void f2fs_submit_page_mbio(struct f2fs_io_info *);
 void set_data_blkaddr(struct dnode_of_data *);

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 511c0e7c3ae182e53d37437cd307b2594f82545f..e08b8d5e8ca2b8a1bdb9002a1bb82c21fba0fcdd 100644 (file)
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1337,8 +1337,13 @@ continue_unlock:
                goto next_step;
        }
 
-       if (wrote)
-               f2fs_submit_merged_bio(sbi, NODE, WRITE);
+       if (wrote) {
+               if (ino)
+                       f2fs_submit_merged_bio_cond(sbi, NULL, NULL,
+                                                       ino, NODE, WRITE);
+               else
+                       f2fs_submit_merged_bio(sbi, NODE, WRITE);
+       }
        return nwritten;
 }
 
@@ -1433,9 +1438,13 @@ static int f2fs_write_node_page(struct page *page,
        set_node_addr(sbi, &ni, fio.blk_addr, is_fsync_dnode(page));
        dec_page_count(sbi, F2FS_DIRTY_NODES);
        up_read(&sbi->node_write);
+
+       if (wbc->for_reclaim)
+               f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, NODE, WRITE);
+
        unlock_page(page);
 
-       if (wbc->for_reclaim || unlikely(f2fs_cp_error(sbi)))
+       if (unlikely(f2fs_cp_error(sbi)))
                f2fs_submit_merged_bio(sbi, NODE, WRITE);
 
        return 0;

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 7cd07d54be4e0c1647df8860b27c5a458a2bd86f..57a5f7bb275ae21b0935e1d8f89e83624bf06117 100644 (file)
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -254,7 +254,8 @@ int commit_inmem_pages(struct inode *inode, bool abort)
        if (!abort) {
                f2fs_unlock_op(sbi);
                if (submit_bio)
-                       f2fs_submit_merged_bio(sbi, DATA, WRITE);
+                       f2fs_submit_merged_bio_cond(sbi, inode, NULL, 0,
+                                                               DATA, WRITE);
        }
        return err;
 }
@@ -1422,8 +1423,7 @@ void f2fs_wait_on_page_writeback(struct page *page,
        if (PageWriteback(page)) {
                struct f2fs_sb_info *sbi = F2FS_P_SB(page);
 
-               if (is_merged_page(sbi, page, type))
-                       f2fs_submit_merged_bio(sbi, type, WRITE);
+               f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, type, WRITE);
                if (ordered)
                        wait_on_page_writeback(page);
                else