From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jiucheng Xu Date: Mon, 07 Sep 2026 19:35:42 +0800 Subject: [PATCH] f2fs: force out-place update for all writes on compressed file MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 8bit Message-Id: <20260907-origin-dev-v1-1-85ae7ebc8c98@amlogic.com> X-B4-Tracking: v=1; b=H4sIAI2hnmoC/03MQQrCMBCF4auUWRvJjBqDq95DugjpmA7YRhIJS sndjcWFy//B+1bInIQzXLoVEhfJEpcWuOvAT24JrGRsDaTJaItWxSRBFjVyUZqtNwcmPJ4ctMM j8U1eG3YdWk+SnzG9N7vgd/0xZP6ZggqVJ6QzOkuode/mewzi9z7OMNRaP6PQm9qmAAAA X-Change-ID: 20260818-origin-dev-0e8c63e2145a To: Jaegeuk Kim , Chao Yu Cc: linux-f2fs-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org, stable@vger.kernel.org, jianxin.pan@amlogic.com, tao.zeng@amlogic.com, tuan.zhang@amlogic.com, Jiucheng Xu X-Mailer: b4 0.14.2 X-Developer-Signature: v=1; a=ed25519-sha256; t=1788780945; l=2497; i=jiucheng.xu@amlogic.com; s=20250821; h=from:subject:message-id; bh=YNrmwUC6hZBgGF3yjLDcDf3MEef/HpM3lXoE8cXM+oc=; b=2QKUyokgLLcsjxunR6k7wQ0huDmzz9QlttyBun5ZhJCS8hhFtf0fT5/2MtvQZv6XkjvH1SZt1 xqgZwn86I2+BJAHh+YZhlRgzQj05jHztIJ+/JhOUkSZjkvKltWETVXa X-Developer-Key: i=jiucheng.xu@amlogic.com; a=ed25519; pk=Q18IjkdWCCuncSplyu+dYqIrm+n42glvoLFJTQqpb2o= X-Endpoint-Received: by B4 Relay for jiucheng.xu@amlogic.com/20250821 with auth_id=498 List-Id: B4 Relay Submissions For compressed file, compressed write may fail and fall back to raw write. -Thread A - Thread B - f2fs_write_multi_pages - f2fs_down_write(&sbi->cp_rwsem); - f2fs_write_compressed_pages - ... - f2fs_trylock_op - ... - f2fs_down_read_trylock(&sbi->cp_rwsem); - ... - f2fs_write_raw_pages - ... Thread B acquires the lock first, which causes Thread A to fail lock acquisition and fall back to raw-data write. IPU_FORCE is enabled on small-capacity storage devices (< 16GB), so f2fs_write_raw_pages() overwrites the original compressed data in-place. The in-memory node has been updated with raw-data addresses, while the node metadata stored on eMMC still remains in compressed state. If a power-cut occurs before the node is flushed to disk, on-disk inconsistency arises: disk data is raw, yet metadata treats it as a compressed cluster, leading to decompression failure. To eliminate this risk completely, force out-place update for all write operations on compressed file. Signed-off-by: Jiucheng Xu --- - Link to v1: https://lore.kernel.org/r/20260826-origin-dev-v1-1-c21271a82100@amlogic.com --- fs/f2fs/data.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 6ae0eb37d20f96ed51a8124f59ba90dabf632fb5..aaeea561dcd30681b9b37f2f6f0493ed58ec2fe0 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -2973,10 +2973,12 @@ bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio) return true; if (f2fs_used_in_atomic_write(inode)) return true; - /* rewrite low ratio compress data w/ OPU mode to avoid fragmentation */ - if (f2fs_compressed_file(inode) && - F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER && - is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) + /* + * rewrite low ratio compress data w/ OPU mode to avoid fragmentation. + * If IO comes from compressed write path and fallback to raw write, + * force out‑place to prevent metadata‑data inconsistency. + */ + if (f2fs_compressed_file(inode)) return true; /* swap file is migrating in aligned write mode */ --- base-commit: 0a1703eba23707e3b2edfa2a2329352e7abc0ffa change-id: 20260818-origin-dev-0e8c63e2145a Best regards, -- Jiucheng Xu