News: 0001551404

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Proposed Persistent Cache For Block Devices "PCACHE" Ported To DM Framework

([Linux Storage] 6 Hours Ago PCACHE)


An initial patch series sent back out in April [1]proposed PCACHE as a persistent memory cache for block devices . PCACHE was born out of the CXL block device driver and brings some benefits over the likes of BCache and dm-writecache.

PCACHE was proposed for offering lower write latency than the existing Linux block cache "Bcache" code, greater concurrency using persistent memory DAX, much greater performance, read cache support unlike dm-writecache, and greater integrity characteristics.

Posted today to the Linux kernel mailing list were the [2]RFC v2 patch series for PCACHE... Or rather now, dm-pcache.

The PCACHE code has been ported over to the Device Mapper (DM) framework and thus now taking on the name dm-pcache. This port allows for this new persistent cache to be exposed as a regular DM target.

Main features

- 16 MiB pmem segments, log-structured allocation.

- Multi-subtree RB-tree index for high parallelism.

- Optional per-entry *CRC32* on cached data.

- Background *write-back* worker and watermark-driven *GC*.

- Crash-safe replay: key-sets are scanned from *key_tail* on start-up.

Current limitations

- Only *write-back* mode implemented.

- Only FIFO cache invalidate; other (LRU, ARC...) planned.

Those wanting to learn more about dm-pcache as a proposed persistent cache for Linux block devices can check out the new [3]patches for all the details.



[1] https://www.phoronix.com/news/Linux-PCACHE-RFC

[2] https://lore.kernel.org/lkml/20250605142306.1930831-1-dongsheng.yang@linux.dev/

[3] https://lore.kernel.org/lkml/20250605142306.1930831-1-dongsheng.yang@linux.dev/



Kjell

spyke

pWe00Iri3e7Z9lHOX2Qx

Anon'ym'

Volta

caligula

Anon'ym'

... Fortunately, the responsibility for providing evidence is on the part of
the person making the claim, not the critic. It is not the responsibility
of UFO skeptics to prove that a UFO has never existed, nor is it the
responsibility of paranormal-health-claims skeptics to prove that crystals
or colored lights never healed anyone. The skeptic's role is to point out
claims that are not adequately supported by acceptable evidence and to
provide plausible alternative explanations that are more in keeping with
the accepted body of scientific evidence. ...
-- Thomas L. Creed, The Skeptical Inquirer, Vol. XII No. 2, pg. 215