From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 468BFC61DD6 for ; Sat, 29 Aug 2026 06:26:13 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender:List-Subscribe:List-Help :List-Post:List-Archive:List-Unsubscribe:List-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:References:In-Reply-To:Message-ID:Date :Subject:CC:To:From:Reply-To:Content-ID:Content-Description:Resent-Date: Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Owner; bh=9NpBB7CEIjqjyGCb4SVckgVeutjU9azjoaZgEVa947o=; b=GN5x4NISwnu9VhtU8/iJtlhOr4 xRDfooJBb7BLMBFRxJZUQAD4+S6unFW8TP+TCNvV8lKJtklO0EMzBqopLuHQq6n29lLBZxrI80Aq6 DP5aL3sn8iYY91KVCYb5h83zhsQTSbAFMV1u3rCUk7G1klP2qKWoxCXqrrLJjNtm4q+P5eneZgLn2 EeldzBp6aMVDsWBsQBg0YKxT1riT6p3Wu0uzIMCux59KadGPIZ9Mrv9fbNctsStbc2fhWQWKH7tIb E/FXBSlqwMP9a6YjOkc86QfmEru7gzIrRS6pili9k7POUSPz7TJB/EUe7fa/9N2ZlJLX3cwPwS5yf nOXWOC+g==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.99.1 #2 (Red Hat Linux)) id 1x0CVu-00000006mdb-0nfl; Sat, 29 Aug 2026 06:26:02 +0000 Received: from canpmsgout09.his.huawei.com ([113.46.200.224]) by bombadil.infradead.org with esmtps (Exim 4.99.1 #2 (Red Hat Linux)) id 1x0CVd-00000006mPf-3yim; Sat, 29 Aug 2026 06:25:47 +0000 dkim-signature: v=1; a=rsa-sha256; d=huawei.com; s=dkim; c=relaxed/relaxed; q=dns/txt; h=From; bh=9NpBB7CEIjqjyGCb4SVckgVeutjU9azjoaZgEVa947o=; b=sLIQWxtrS8w5ERuoFDXmbubInkswYa662b4wJY61Zn7xeDFflLoibLq4MWzVHCe0KOiu45zz/ WsF0/wDXGz+6Irxls6bqJzT51YJ9VpzdmO28F/hdacMD9/6HUBU+B/lIjOaXLBS1rXND9jmd36W gmNkFosl91edLNaX6LjQcZk= Received: from mail.maildlp.com (unknown [172.19.163.127]) by canpmsgout09.his.huawei.com (SkyGuard) with ESMTPS id 4hX4jH4tCWz1cyPY; Sat, 29 Aug 2026 14:14:43 +0800 (CST) Received: from dggpemr100018.china.huawei.com (unknown [7.185.36.64]) by mail.maildlp.com (Postfix) with ESMTPS id 7539A40572; Sat, 29 Aug 2026 14:25:30 +0800 (CST) Received: from huawei.com (10.50.85.155) by dggpemr100018.china.huawei.com (7.185.36.64) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.36; Sat, 29 Aug 2026 14:25:29 +0800 From: zhouminqiang To: , , , , , CC: , , , , , , Subject: [PATCH v2 1/7] jffs2: wbuf: clear wbuf on recovery failure paths Date: Sat, 29 Aug 2026 14:16:51 +0800 Message-ID: <20260829061658.306854-2-zhouminqiang2@huawei.com> X-Mailer: git-send-email 2.52.0 In-Reply-To: <20260829061658.306854-1-zhouminqiang2@huawei.com> References: <20260829061658.306854-1-zhouminqiang2@huawei.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain X-Originating-IP: [10.50.85.155] X-ClientProxiedBy: kwepems100001.china.huawei.com (7.221.188.238) To dggpemr100018.china.huawei.com (7.185.36.64) X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.9.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20260828_232546_387481_283E032E X-CRM114-Status: GOOD ( 14.56 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org Write verification was introduced by commit a6bc432e296d ("[JFFS2] Add support for write-buffer verification.") to detect transport or program-time corruption. When verification fails, jffs2_wbuf_recover() attempts to recover the data: it first calls jffs2_block_refile() to mark the old eraseblock's remaining space as REF_OBSOLETE, then rewrites the old block's data along with the new data still in the wbuf to a new block. However, when the recovery write verification also fails, the function returns without clearing c->wbuf_len, leaving the wbuf still pointing to the refiled old block, which leads to two kinds of bugs. Both cases below are illustrated with a filesystem of erasesize=16KB and wbuf_pagesize=512B. Case 1: BUG_ON in jffs2_link_node_ref(): User: Two 1KB writes ref0: [A+0, A+1092) (ri:68B + data:1024B) ref1: [A+1092, A+2184) User: append 1KB write ... jffs2_write_dnode jffs2_flash_writev c->wbuf_ofs = A+2048, c->wbuf_len = 512 __jffs2_flush_wbuf mtd_write -> 0-to-1 bit flip jffs2_verify_write -> verify failed jffs2_wbuf_recover jffs2_block_refile c->nextblock = NULL jffs2_link_node_ref -> mark A remaining ref2: [A+2184, A+16384) space REF_OBSOLETE, jeb_A->free_size = 0 jffs2_reserve_space_gc jffs2_do_reserve_space jffs2_find_nextblock c->nextblock = B start = A+1092, end = A+2184 end - start >= c->wbuf_pagesize -> recover data in A mtd_write jffs2_verify_write -> verify also failed return -> c->wbuf_ofs = A+2048 c->wbuf_len = 512 retry jffs2_flash_writev if (SECTOR_ADDR(to) != SECTOR_ADDR(c->wbuf_ofs)) -> SECTOR(to) = B -> SECTOR(c->wbuf_ofs) = A __jffs2_flush_wbuf(c, PAD_NOACCOUNT) -> flush residual wbuf data wbuf_jeb = A -> c->wbuf_ofs still points to refiled old block mtd_write -> succeeds via NAND AND jffs2_verify_write -> passed if (pad) jffs2_link_node_ref ref_offset(ref) = c->wbuf_ofs + c->wbuf_len = A+2560 jeb_A->offset = A c->sector_size = 16384 jeb_A->free_size = 0 ref_offset(ref) != jeb_A->offset + c->sector_size - jeb_A->free_size -> BUG Case 2: deadlock in jffs2_flush_wbuf_pad(): User: 1KB write A_ref0: [A+0, A+1092) (ri:68B + data:1024B) User: append 1K write ... jffs2_write_dnode jffs2_flash_writev c->wbuf_ofs = A+1024, c->wbuf_len = 512 __jffs2_flush_wbuf mtd_write -> bit flip jffs2_verify_write -> verify failed jffs2_wbuf_recover jffs2_block_refile c->nextblock = NULL jffs2_link_node_ref -> mark A remaining A_ref1: [A+1092, A+16384) space as REF_OBSOLETE jffs2_reserve_space_gc jffs2_do_reserve_space jffs2_find_nextblock c->nextblock = B start = A, end = A+1092 end - start >= c->wbuf_pagesize -> recover data in A mtd_write jffs2_verify_write -> verify also failed jffs2_add_physical_node_ref -> mark written area in B_ref0: [B+0, B+1536) B as REF_OBSOLETE return -> c->wbuf_ofs = A+1024 c->wbuf_len = 512 retry jffs2_flash_writev down_write(&c->wbuf_sem) if (SECTOR_ADDR(to) != SECTOR_ADDR(c->wbuf_ofs)) -> SECTOR(to) = B -> SECTOR(c->wbuf_ofs) = A __jffs2_flush_wbuf(c, PAD_NOACCOUNT) -> flush residual wbuf data wbuf_jeb = A -> c->wbuf_ofs still points to refiled old block mtd_write -> bit flip jffs2_verify_write -> verify failed jffs2_wbuf_recover jffs2_block_refile c->nextblock != jeb -> jeb = A, nextblock = B jffs2_link_node_ref -> append zero-length ref A_ref2: [A+16384, A+16384) marked REF_OBSOLETE after the existing one end = jeb_A->last_node -> inflates to eraseblock tail start = A, end = A+16384 jffs2_reserve_space_gc minsize = end - start = 16384 jffs2_do_reserve_space jeb = c->nextblock -> points to B jeb_B->free_size = 16384 - 1536 minsize > jeb_B->free_size -> first recovery's OBSOLETE ref reduced free_size jffs2_wbuf_dirty(c) jffs2_flush_wbuf_pad(c) down_write(&c->wbuf_sem) -> already held, deadlock Fix this by setting c->wbuf_len = 0 when jffs2_verify_write fails in recovery path, and also in the jffs2_reserve_space_gc() and jffs2_prealloc_raw_node_refs() failure paths, ensuring subsequent flushes will not attempt writes on refiled blocks. Signed-off-by: zhouminqiang --- fs/jffs2/wbuf.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/fs/jffs2/wbuf.c b/fs/jffs2/wbuf.c index 3b7803c75d58..61e3dbd4cd7b 100644 --- a/fs/jffs2/wbuf.c +++ b/fs/jffs2/wbuf.c @@ -390,6 +390,7 @@ static void jffs2_wbuf_recover(struct jffs2_sb_info *c) if (ret) { pr_warn("Failed to allocate space for wbuf recovery. Data loss ensues.\n"); kfree(buf); + c->wbuf_len = 0; return; } @@ -400,6 +401,7 @@ static void jffs2_wbuf_recover(struct jffs2_sb_info *c) if (ret) { pr_warn("Failed to allocate node refs for wbuf recovery. Data loss ensues.\n"); kfree(buf); + c->wbuf_len = 0; return; } @@ -431,12 +433,13 @@ static void jffs2_wbuf_recover(struct jffs2_sb_info *c) if (ret || retlen != towrite || jffs2_verify_write(c, rewrite_buf, ofs)) { /* Argh. We tried. Really we did. */ - pr_crit("Recovery of wbuf failed due to a second write error\n"); + pr_crit("Recovery of wbuf failed due to a second write error. Data loss ensues.\n"); kfree(buf); if (retlen) jffs2_add_physical_node_ref(c, ofs | REF_OBSOLETE, ref_totlen(c, jeb, first_raw), NULL); + c->wbuf_len = 0; return; } pr_notice("Recovery of wbuf succeeded to %08x\n", ofs); -- 2.52.0