6.11-stable review patch. If anyone has any objections, please let me know.
------------------
From: Chao Yu chao@kernel.org
[ Upstream commit 26e6f59d0bbaac76fa3413462d780bd2b5f9f653 ]
Jinsu Lee reported a performance regression issue, after commit 5c8764f8679e ("f2fs: fix to force buffered IO on inline_data inode"), we forced direct write to use buffered IO on inline_data inode, it will cause performace regression due to memory copy and data flush.
It's fine to not force direct write to use buffered IO, as it can convert inline inode before committing direct write IO.
Fixes: 5c8764f8679e ("f2fs: fix to force buffered IO on inline_data inode") Reported-by: Jinsu Lee jinsu1.lee@samsung.com Closes: https://lore.kernel.org/linux-f2fs-devel/af03dd2c-e361-4f80-b2fd-39440766cf6... Signed-off-by: Chao Yu chao@kernel.org Signed-off-by: Jaegeuk Kim jaegeuk@kernel.org Signed-off-by: Sasha Levin sashal@kernel.org --- fs/f2fs/file.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 611c0aac721a3..b4047d8092890 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -868,7 +868,11 @@ static bool f2fs_force_buffered_io(struct inode *inode, int rw) return true; if (f2fs_compressed_file(inode)) return true; - if (f2fs_has_inline_data(inode)) + /* + * only force direct read to use buffered IO, for direct write, + * it expects inline data conversion before committing IO. + */ + if (f2fs_has_inline_data(inode) && rw == READ) return true;
/* disallow direct IO if any of devices has unaligned blksize */