As writeback becomes more common there is another shortcoming
that needs to be addressed - compressed data writeback.  Currently
zram does uncompressed data writeback which is not optimal due to
potential CPU and battery wastage.  This series changes suboptimal
uncompressed writeback to a more optimal compressed data writeback.

v1 -> v2:
- made compressed writeback configurable via device attribute
- added missing batch_size documentation
- switched to guard() for init_lock
- more code tweaks and cleanups

Richard Chang (2):
  zram: introduce compressed data writeback
  zram: introduce writeback_compressed device attribute

Sergey Senozhatsky (5):
  zram: document writeback_batch_size
  zram: move bd_stat to writeback section
  zram: rename zram_free_page()
  zram: switch to guard() for init_lock
  zram: consolidate device-attr declarations

 Documentation/ABI/testing/sysfs-block-zram  |  14 +
 Documentation/admin-guide/blockdev/zram.rst |  24 +-
 drivers/block/zram/zram_drv.c               | 589 ++++++++++++--------
 drivers/block/zram/zram_drv.h               |   1 +
 4 files changed, 406 insertions(+), 222 deletions(-)

--
2.52.0.487.g5c8c507ade-goog

On Mon,  1 Dec 2025 18:47:47 +0900 Sergey Senozhatsky <senozhatsky@chromium.org> wrote:

> As writeback becomes more common there is another shortcoming
> that needs to be addressed - compressed data writeback.  Currently
> zram does uncompressed data writeback which is not optimal due to
> potential CPU and battery wastage.  This series changes suboptimal
> uncompressed writeback to a more optimal compressed data writeback.

I'll move this series and the "zram: use u32 for entry ac_time
tracking" series into mm-stable soon, but the review coverage is thin.

It would be good to get some more reviewer input on these zram/zsmalloc
changes.  We haven't heard from Minchan for a couple of months.  Is
there someone else who can step up and help out?

Thanks.