2024-03-24 22:21:56

by NeilBrown

Subject: [PATCH] ceph: redirty page before returning AOP_WRITEPAGE_ACTIVATE

The page has been marked clean before writepage is called. If we don't
redirty it before postponing the write, it might never get written.

Fixes: 503d4fa6ee28 ("ceph: remove reliance on bdi congestion")
Signed-off-by: NeilBrown <[email protected]>
---
fs/ceph/addr.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 1340d77124ae..ee9caf7916fb 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -795,8 +795,10 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
ihold(inode);

if (wbc->sync_mode == WB_SYNC_NONE &&
- ceph_inode_to_fs_client(inode)->write_congested)
+ ceph_inode_to_fs_client(inode)->write_congested) {
+ redirty_page_for_writepage(wbc, page);
return AOP_WRITEPAGE_ACTIVATE;
+ }

wait_on_page_fscache(page);

--
2.44.0

2024-03-25 08:47:43

by Xiubo Li

Subject: Re: [PATCH] ceph: redirty page before returning AOP_WRITEPAGE_ACTIVATE

On 3/25/24 06:21, NeilBrown wrote:
> The page has been marked clean before writepage is called. If we don't
> redirty it before postponing the write, it might never get written.
>
> Fixes: 503d4fa6ee28 ("ceph: remove reliance on bdi congestion")
> Signed-off-by: NeilBrown <[email protected]>
> ---
> fs/ceph/addr.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
> index 1340d77124ae..ee9caf7916fb 100644
> --- a/fs/ceph/addr.c
> +++ b/fs/ceph/addr.c
> @@ -795,8 +795,10 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
> ihold(inode);
>
> if (wbc->sync_mode == WB_SYNC_NONE &&
> - ceph_inode_to_fs_client(inode)->write_congested)
> + ceph_inode_to_fs_client(inode)->write_congested) {
> + redirty_page_for_writepage(wbc, page);
> return AOP_WRITEPAGE_ACTIVATE;
> + }
>
> wait_on_page_fscache(page);
>

Good catch!

Applied to the testing branch to run the tests.

Thanks NeilBrown

- Xiubo

2024-03-25 09:30:02

by Jeffrey Layton

Subject: Re: [PATCH] ceph: redirty page before returning AOP_WRITEPAGE_ACTIVATE

On Mon, 2024-03-25 at 09:21 +1100, NeilBrown wrote:
> The page has been marked clean before writepage is called. If we don't
> redirty it before postponing the write, it might never get written.
>
> Fixes: 503d4fa6ee28 ("ceph: remove reliance on bdi congestion")
> Signed-off-by: NeilBrown <[email protected]>
> ---
> fs/ceph/addr.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
> index 1340d77124ae..ee9caf7916fb 100644
> --- a/fs/ceph/addr.c
> +++ b/fs/ceph/addr.c
> @@ -795,8 +795,10 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
> ihold(inode);
>
> if (wbc->sync_mode == WB_SYNC_NONE &&
> - ceph_inode_to_fs_client(inode)->write_congested)
> + ceph_inode_to_fs_client(inode)->write_congested) {
> + redirty_page_for_writepage(wbc, page);
> return AOP_WRITEPAGE_ACTIVATE;
> + }
>
> wait_on_page_fscache(page);
>

Reviewed-by: Jeff Layton <[email protected]>