From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 6F3C13B2B2 for ; Wed, 20 Dec 2023 13:59:44 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="VAx8QlLM" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 7A60EC433CC; Wed, 20 Dec 2023 13:59:42 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1703080783; bh=HlIylD6IwGlkcvAcDhSn7XsD9KaF9Cn4cGLKY52iKZ4=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=VAx8QlLM9/SAVXDNxAz7Vm0DbGUGFuklAUi9oWrXB2siqcfYl9r8e7XOChcN4D/Qt xYOXSp7QxrttGDBWDLva4C/KR4QUc1uVnnB3taBjsdab4tRRz9aoM6eOn8KHhUSRik yoI1q/DaZ6Rv6lxN854d+RSSfV6XwtjZuKMsK5veySxq7mflzYcUwhBxKHMi4jqYmX yJHnbh4DQqEp3u57DtS2hcfwgTmo8WaZFLQPO2C1UE7qlbZBgy8xQKwrtfvailz4xi u6VpYHaTbn/olibbEvjwuTOfzOEJU+wehVd/hMyg+dYY6ZF/GPdoNGGiiF7L37IXvs kKXTwkHTljw3A== From: Chao Yu To: jaegeuk@kernel.org Cc: linux-f2fs-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org, Chao Yu Subject: [PATCH 2/6] f2fs: compress: fix to cover normal cluster write with cp_rwsem Date: Wed, 20 Dec 2023 21:59:30 +0800 Message-Id: <20231220135934.3471407-2-chao@kernel.org> X-Mailer: git-send-email 2.40.1 In-Reply-To: <20231220135934.3471407-1-chao@kernel.org> References: <20231220135934.3471407-1-chao@kernel.org> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit When we overwrite compressed cluster w/ normal cluster, we should not unlock cp_rwsem during f2fs_write_raw_pages(), otherwise data will be corrupted if partial blocks were persisted before CP & SPOR, due to cluster metadata wasn't updated atomically. Fixes: 4c8ff7095bef ("f2fs: support data compression") Signed-off-by: Chao Yu --- fs/f2fs/compress.c | 20 ++++++++++++++------ fs/f2fs/data.c | 3 ++- 2 files changed, 16 insertions(+), 7 deletions(-) diff --git a/fs/f2fs/compress.c b/fs/f2fs/compress.c index 1122db8cc0b0..a1426c3eadcc 100644 --- a/fs/f2fs/compress.c +++ b/fs/f2fs/compress.c @@ -1471,7 +1471,8 @@ static int f2fs_write_raw_pages(struct compress_ctx *cc, enum iostat_type io_type) { struct address_space *mapping = cc->inode->i_mapping; - int _submitted, compr_blocks, ret, i; + struct f2fs_sb_info *sbi = F2FS_M_SB(mapping); + int _submitted, compr_blocks, ret = 0, i; compr_blocks = f2fs_compressed_blocks(cc); @@ -1486,6 +1487,10 @@ static int f2fs_write_raw_pages(struct compress_ctx *cc, if (compr_blocks < 0) return compr_blocks; + /* overwrite compressed cluster w/ normal cluster */ + if (compr_blocks > 0) + f2fs_lock_op(sbi); + for (i = 0; i < cc->cluster_size; i++) { if (!cc->rpages[i]) continue; @@ -1518,26 +1523,29 @@ static int f2fs_write_raw_pages(struct compress_ctx *cc, unlock_page(cc->rpages[i]); ret = 0; } else if (ret == -EAGAIN) { + ret = 0; /* * for quota file, just redirty left pages to * avoid deadlock caused by cluster update race * from foreground operation. */ if (IS_NOQUOTA(cc->inode)) - return 0; - ret = 0; + goto out; f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT); goto retry_write; } - return ret; + goto out; } *submitted += _submitted; } - f2fs_balance_fs(F2FS_M_SB(mapping), true); +out: + if (compr_blocks > 0) + f2fs_unlock_op(sbi); - return 0; + f2fs_balance_fs(sbi, true); + return ret; } int f2fs_write_multi_pages(struct compress_ctx *cc, diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 6c72a6e86ba8..ceed5ac6c66b 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -2848,7 +2848,7 @@ int f2fs_write_single_data_page(struct page *page, int *submitted, .encrypted_page = NULL, .submitted = 0, .compr_blocks = compr_blocks, - .need_lock = LOCK_RETRY, + .need_lock = compr_blocks ? LOCK_DONE : LOCK_RETRY, .post_read = f2fs_post_read_required(inode) ? 1 : 0, .io_type = io_type, .io_wbc = wbc, @@ -2929,6 +2929,7 @@ int f2fs_write_single_data_page(struct page *page, int *submitted, if (err == -EAGAIN) { err = f2fs_do_write_data_page(&fio); if (err == -EAGAIN) { + f2fs_bug_on(sbi, compr_blocks); fio.need_lock = LOCK_REQ; err = f2fs_do_write_data_page(&fio); } -- 2.40.1