[RFC,v4,0/9] implement containerized syncfs for overlayfs

Message

Chengguang Xu Nov. 13, 2020, 6:55 a.m. UTC

Current syncfs(2) syscall on overlayfs just calls sync_filesystem()
on upper_sb to synchronize whole dirty inodes in upper filesystem
regardless of the overlay ownership of the inode. In the use case of
container, when multiple containers using the same underlying upper
filesystem, it has some shortcomings as below.

(1) Performance
Synchronization is probably heavy because it actually syncs unnecessary
inodes for target overlayfs.

(2) Interference
Unplanned synchronization will probably impact IO performance of
unrelated container processes on the other overlayfs.

This series try to implement containerized syncfs for overlayfs so that
only sync target dirty upper inodes which are belong to specific overlayfs
instance. By doing this, it is able to reduce cost of synchronization and
will not seriously impact IO performance of unrelated processes.

v1->v2:
- Mark overlayfs' inode dirty itself instead of adding notification
  mechanism to vfs inode.

v2->v3:
- Introduce overlayfs' extra syncfs wait list to wait target upper inodes
in ->sync_fs.

v3->v4:
- Using wait_sb_inodes() to wait syncing upper inodes.
- Mark overlay inode dirty only when having upper inode and  VM_SHARED
flag in ovl_mmap().
- Check upper i_state after checking upper mmap state
in ovl_write_inode.

Chengguang Xu (9):
  ovl: setup overlayfs' private bdi
  ovl: implement ->writepages operation
  ovl: implement overlayfs' ->evict_inode operation
  ovl: mark overlayfs' inode dirty on modification
  ovl: mark overlayfs' inode dirty on shared mmap
  ovl: implement overlayfs' ->write_inode operation
  ovl: cache dirty overlayfs' inode
  fs: export wait_sb_inodes()
  ovl: implement containerized syncfs for overlayfs

 fs/fs-writeback.c         |  3 +-
 fs/overlayfs/file.c       |  3 ++
 fs/overlayfs/inode.c      | 15 ++++++++++
 fs/overlayfs/overlayfs.h  |  4 +++
 fs/overlayfs/super.c      | 63 ++++++++++++++++++++++++++++++++++++---
 fs/overlayfs/util.c       | 14 +++++++++
 include/linux/writeback.h |  1 +
 7 files changed, 98 insertions(+), 5 deletions(-)

Comments

Chengguang Xu Dec. 4, 2020, 2:49 p.m. UTC | #1

---- 在 星期五, 2020-11-13 14:55:46 Chengguang Xu <[e-mail]> 撰写 ----
 > Current syncfs(2) syscall on overlayfs just calls sync_filesystem()
 > on upper_sb to synchronize whole dirty inodes in upper filesystem
 > regardless of the overlay ownership of the inode. In the use case of
 > container, when multiple containers using the same underlying upper
 > filesystem, it has some shortcomings as below.
 > 
 > (1) Performance
 > Synchronization is probably heavy because it actually syncs unnecessary
 > inodes for target overlayfs.
 > 
 > (2) Interference
 > Unplanned synchronization will probably impact IO performance of
 > unrelated container processes on the other overlayfs.
 > 
 > This series try to implement containerized syncfs for overlayfs so that
 > only sync target dirty upper inodes which are belong to specific overlayfs
 > instance. By doing this, it is able to reduce cost of synchronization and
 > will not seriously impact IO performance of unrelated processes.
 
Hi Miklos,

I think this version has addressed all previous issues and comments from Jack
and Amir.  Have you got time to review this patch series?

Thanks,
Chengguang

Miklos Szeredi Dec. 4, 2020, 3:02 p.m. UTC | #2

On Fri, Dec 4, 2020 at 3:50 PM Chengguang Xu <[e-mail]> wrote:
>
>  ---- 在 星期五, 2020-11-13 14:55:46 Chengguang Xu <[e-mail]> 撰写 ----
>  > Current syncfs(2) syscall on overlayfs just calls sync_filesystem()
>  > on upper_sb to synchronize whole dirty inodes in upper filesystem
>  > regardless of the overlay ownership of the inode. In the use case of
>  > container, when multiple containers using the same underlying upper
>  > filesystem, it has some shortcomings as below.
>  >
>  > (1) Performance
>  > Synchronization is probably heavy because it actually syncs unnecessary
>  > inodes for target overlayfs.
>  >
>  > (2) Interference
>  > Unplanned synchronization will probably impact IO performance of
>  > unrelated container processes on the other overlayfs.
>  >
>  > This series try to implement containerized syncfs for overlayfs so that
>  > only sync target dirty upper inodes which are belong to specific overlayfs
>  > instance. By doing this, it is able to reduce cost of synchronization and
>  > will not seriously impact IO performance of unrelated processes.
>
> Hi Miklos,
>
> I think this version has addressed all previous issues and comments from Jack
> and Amir.  Have you got time to review this patch series?

Hopefully yes.

I'm really keen to finish off the unprivileged overlay patches first.
Will test and post a new version shortly.

Thanks,
Miklos

Jan Kara Dec. 4, 2020, 4:40 p.m. UTC | #3

On Fri 04-12-20 22:49:13, Chengguang Xu wrote:
>  ---- 在 星期五, 2020-11-13 14:55:46 Chengguang Xu <[e-mail]> 撰写 ----
>  > Current syncfs(2) syscall on overlayfs just calls sync_filesystem()
>  > on upper_sb to synchronize whole dirty inodes in upper filesystem
>  > regardless of the overlay ownership of the inode. In the use case of
>  > container, when multiple containers using the same underlying upper
>  > filesystem, it has some shortcomings as below.
>  > 
>  > (1) Performance
>  > Synchronization is probably heavy because it actually syncs unnecessary
>  > inodes for target overlayfs.
>  > 
>  > (2) Interference
>  > Unplanned synchronization will probably impact IO performance of
>  > unrelated container processes on the other overlayfs.
>  > 
>  > This series try to implement containerized syncfs for overlayfs so that
>  > only sync target dirty upper inodes which are belong to specific overlayfs
>  > instance. By doing this, it is able to reduce cost of synchronization and
>  > will not seriously impact IO performance of unrelated processes.
>  
> Hi Miklos,
> 
> I think this version has addressed all previous issues and comments from Jack
> and Amir.  Have you got time to review this patch series?

Yes, the patches now look good to me. Feel free to add:

Acked-by: Jan Kara <[e-mail]>

								Honza