mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Li Chen <me@linux.beauty>
To: dm-devel@lists.linux.dev, linux-kernel@vger.kernel.org,
	Dongsheng Yang <dongsheng.yang@linux.dev>,
	Zheng Gu <cengku@gmail.com>
Subject: [PATCH 0/2] dm-pcache: fix metadata indexing and persistence issues
Date: Tue,  2 Dec 2025 20:11:56 +0800	[thread overview]
Message-ID: <20251202121158.111092-1-me@linux.beauty> (raw)

From: Li Chen <chenl311@chinatelecom.cn>

dm-pcache stores metadata (cache_info and segment_info) in 4K-aligned
slots on the cache device, using sequence numbers and CRC to identify
the latest valid copy.

However, the cache_info and segment_info paths were computing their
on-media index using sizeof(struct) instead of the 4K metadata stride.
As a result:

  * cache_info updates (including gc_percent set via a dmsetup message)
    were written to invalid offsets between metadata slots and failed
    to persist across table reloads or reboots.

  * segment_info indexing became desynchronized, so rotation to the
    "next" slot no longer matched the location returned by
    pcache_meta_find_latest().

The issue can be reproduced with:

  dmsetup create pcache_vdb --table \
    "0 $(blockdev --getsz /dev/vdb) pcache /dev/pmem0 /dev/vdb 4 \
     cache_mode writeback data_crc false"

  # Check default gc_percent (70)
  dmsetup status pcache_vdb

  # Change gc_percent to 10
  dmsetup message pcache_vdb 0 "gc_percent 10"

  # Verify change is active in memory
  dmsetup status pcache_vdb

  # Reboot the system...

  # Without patch (gc_percent reverts to 70):
  dmsetup status pcache_vdb

  # With patch (gc_percent persists as 10):
  dmsetup status pcache_vdb

This series fixes the issue by deriving the metadata slot index from
the pointer returned by pcache_meta_find_latest(), using the 4K stride
(CACHE_INFO_SIZE / SEG_INFO_SIZE). This ensures that updates to
cache_info and segment_info are written to valid slots and remain
consistent with the on-media layout.

Li Chen (2):
  dm pcache: fix cache info indexing
  dm pcache: fix segment info indexing

 drivers/md/dm-pcache/cache.c         | 13 ++++++++++---
 drivers/md/dm-pcache/cache_segment.c |  6 +++++-
 2 files changed, 15 insertions(+), 4 deletions(-)

-- 
2.51.0


             reply	other threads:[~2025-12-02 12:12 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-12-02 12:11 Li Chen [this message]
2025-12-02 12:11 ` [PATCH 1/2] dm pcache: fix cache info indexing Li Chen
2025-12-03  5:56   ` Dongsheng Yang
2025-12-03 23:38     ` Dongsheng Yang
2025-12-02 12:11 ` [PATCH 2/2] dm pcache: fix segment " Li Chen
2025-12-03  5:57   ` Dongsheng Yang
2025-12-03 23:38     ` Dongsheng Yang
2025-12-02 19:09 ` [PATCH 0/2] dm-pcache: fix metadata indexing and persistence issues Mikulas Patocka
2025-12-03  5:56 ` Dongsheng Yang
2025-12-03 23:38   ` Dongsheng Yang
2025-12-04  1:38     ` Dongsheng Yang
2025-12-05  9:12       ` Li Chen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20251202121158.111092-1-me@linux.beauty \
    --to=me@linux.beauty \
    --cc=cengku@gmail.com \
    --cc=dm-devel@lists.linux.dev \
    --cc=dongsheng.yang@linux.dev \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®