6.1-stable review patch. If anyone has any objections, please let me know.
------------------
From: Jan Kara jack@suse.cz
commit 04e568a3b31cfbd545c04c8bfc35c20e5ccfce0f upstream.
Since we want to transition transaction commits to use ext4_writepages() for writing back ordered, add handling of page redirtying into ext4_bio_write_page(). Also move buffer dirty bit clearing into the same place other buffer state handling.
Reviewed-by: Ritesh Harjani (IBM) ritesh.list@gmail.com Signed-off-by: Jan Kara jack@suse.cz Link: https://lore.kernel.org/r/20221207112722.22220-1-jack@suse.cz Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- fs/ext4/page-io.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-)
--- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -484,6 +484,13 @@ int ext4_bio_write_page(struct ext4_io_s /* A hole? We can safely clear the dirty bit */ if (!buffer_mapped(bh)) clear_buffer_dirty(bh); + /* + * Keeping dirty some buffer we cannot write? Make + * sure to redirty the page. This happens e.g. when + * doing writeout for transaction commit. + */ + if (buffer_dirty(bh) && !PageDirty(page)) + redirty_page_for_writepage(wbc, page); if (io->io_bio) ext4_io_submit(io); continue; @@ -491,6 +498,7 @@ int ext4_bio_write_page(struct ext4_io_s if (buffer_new(bh)) clear_buffer_new(bh); set_buffer_async_write(bh); + clear_buffer_dirty(bh); nr_to_submit++; } while ((bh = bh->b_this_page) != head);
@@ -534,7 +542,10 @@ int ext4_bio_write_page(struct ext4_io_s printk_ratelimited(KERN_ERR "%s: ret = %d\n", __func__, ret); redirty_page_for_writepage(wbc, page); do { - clear_buffer_async_write(bh); + if (buffer_async_write(bh)) { + clear_buffer_async_write(bh); + set_buffer_dirty(bh); + } bh = bh->b_this_page; } while (bh != head); goto unlock; @@ -547,7 +558,6 @@ int ext4_bio_write_page(struct ext4_io_s continue; io_submit_add_bh(io, inode, page, bounce_page, bh); nr_submitted++; - clear_buffer_dirty(bh); } while ((bh = bh->b_this_page) != head);
unlock: