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 5269ee1..d17d759 100644
--- a/fs/f2fs/checkpoint.c
+++ b/fs/f2fs/checkpoint.c
@@ -235,10 +235,15 @@
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 4309e77..95e05256 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -115,20 +115,18 @@
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(bvec, io->bio, i, 0) {
@@ -143,18 +141,34 @@
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;
@@ -163,6 +177,9 @@
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;
@@ -172,9 +189,24 @@
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.
@@ -1141,12 +1173,18 @@
inode_dec_dirty_pages(inode);
if (err)
ClearPageUptodate(page);
- unlock_page(page);
- f2fs_balance_fs(sbi, need_balance_fs);
- if (wbc->for_reclaim || unlikely(f2fs_cp_error(sbi))) {
- f2fs_submit_merged_bio(sbi, DATA, WRITE);
+
+ 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 (unlikely(f2fs_cp_error(sbi)))
+ f2fs_submit_merged_bio(sbi, DATA, WRITE);
+
return 0;
redirty_out:
@@ -1334,7 +1372,7 @@
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 67d0fbf..8e9fa4b 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1934,8 +1934,9 @@
/*
* 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 64f3fba..284bfee 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1337,8 +1337,13 @@
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 @@
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 ab3116e..cc41a95 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -313,7 +313,8 @@
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;
}
@@ -1508,8 +1509,7 @@
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