Re: [PATCH v3] zram: Implement multi-page write-back

[Jens Axboe]

On 11/4/25 11:48 PM, Yuwen Chen wrote:
> For block devices, sequential write performance is significantly
> better than random write. Currently, zram's write-back function
> only supports single-page operations, which fails to leverage
> the sequential write advantage and leads to suboptimal performance.
> This patch implements multi-page batch write-back for zram to
> leverage sequential write performance of block devices.
> After applying this patch, a large number of pages being merged
> into batch write operations can be observed via the following test
> code, which effectively improves write-back performance.
> 
> mount -t debugfs none /sys/kernel/debug/
> echo "block:block_bio_frontmerge" >> /sys/kernel/debug/tracing/set_event
> echo "block:block_bio_backmerge" >> /sys/kernel/debug/tracing/set_event
> cat /sys/kernel/debug/tracing/trace_pipe &
> echo "page_indexes=1-10000" > /sys/block/zram0/writeback
> 
> Signed-off-by: Yuwen Chen <ywen.chen@xxxxxxxxxxx>
> Reviewed-by: Fengyu Lian <licayy@xxxxxxxxxxx>
> ---
> Changes in v3:
>   - Postpone the page allocation.
> Changes in v2:
>   - Rename some data structures.
>   - Fix an exception caused by accessing a null pointer.

Please either finish the patch before sending it out, or take your
time before posting again. Sending 3 versions in one day will just
make people ignore you.

This commit message is in dire need of some actual performance
results. This is a change for better performance, no? If so, you
should have some clear numbers in there describing where it's
better, and where it's worse (if appropriate).

-- 
Jens Axboe