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 X-Spam-Level: X-Spam-Status: No, score=-6.1 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,HEADER_FROM_DIFFERENT_DOMAINS,MAILING_LIST_MULTI,SIGNED_OFF_BY, SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 626F6C43381 for ; Mon, 18 Feb 2019 14:43:51 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 3161121736 for ; Mon, 18 Feb 2019 14:43:51 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1550501031; bh=KddCNh8EehFoZsQAxas3bP9N4Ga/I5rT1UKN7Sms0cU=; h=From:To:Cc:Subject:Date:In-Reply-To:References:List-ID:From; b=euu3CND1z13ut2weXLlHU4taUItMXdPhbKAFrgT5rRL7gCDWE+qu7xoENUOQT55ux q0nUGUlJOE/wyBXkztXhH/Qpl3bbh5h+A5Q1SGreJ4AJ2QxYgu0dQeQ3FVc01+FaI0 0IymDYZehrbOlc3ILYaXkjj/OiLS+oSR1NegX4k8= Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1732887AbfBROnt (ORCPT ); Mon, 18 Feb 2019 09:43:49 -0500 Received: from mail.kernel.org ([198.145.29.99]:56034 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729707AbfBRNtA (ORCPT ); Mon, 18 Feb 2019 08:49:00 -0500 Received: from localhost (5356596B.cm-6-7b.dynamic.ziggo.nl [83.86.89.107]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id 78BC220449; Mon, 18 Feb 2019 13:48:58 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1550497739; bh=KddCNh8EehFoZsQAxas3bP9N4Ga/I5rT1UKN7Sms0cU=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=XMck4q4U3RraxgncucGofBBL7s5XPWpDwxX1Alkucc+G7DfiokfPg/zR5aS9V1/4f JLCr9b3WKDPWFwA09TI+XjyITXUQI47hQJGETRYGfa+2vZhJUP/YaXrZMKWOBXyTJE gArCczH3S0OP3TxH7VTVIhWj9rVypxth//PlFIa4= From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Nikos Tsironis , Joe Thornber , Mikulas Patocka , Mike Snitzer Subject: [PATCH 4.20 87/92] dm thin: fix bug where bio that overwrites thin block ignores FUA Date: Mon, 18 Feb 2019 14:43:30 +0100 Message-Id: <20190218133503.442692574@linuxfoundation.org> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20190218133454.668268457@linuxfoundation.org> References: <20190218133454.668268457@linuxfoundation.org> User-Agent: quilt/0.65 X-stable: review X-Patchwork-Hint: ignore MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 4.20-stable review patch. If anyone has any objections, please let me know. ------------------ From: Nikos Tsironis commit 4ae280b4ee3463fa57bbe6eede26b97daff8a0f1 upstream. When provisioning a new data block for a virtual block, either because the block was previously unallocated or because we are breaking sharing, if the whole block of data is being overwritten the bio that triggered the provisioning is issued immediately, skipping copying or zeroing of the data block. When this bio completes the new mapping is inserted in to the pool's metadata by process_prepared_mapping(), where the bio completion is signaled to the upper layers. This completion is signaled without first committing the metadata. If the bio in question has the REQ_FUA flag set and the system crashes right after its completion and before the next metadata commit, then the write is lost despite the REQ_FUA flag requiring that I/O completion for this request must only be signaled after the data has been committed to non-volatile storage. Fix this by deferring the completion of overwrite bios, with the REQ_FUA flag set, until after the metadata has been committed. Cc: stable@vger.kernel.org Signed-off-by: Nikos Tsironis Acked-by: Joe Thornber Acked-by: Mikulas Patocka Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman --- drivers/md/dm-thin.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 50 insertions(+), 5 deletions(-) --- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -257,6 +257,7 @@ struct pool { spinlock_t lock; struct bio_list deferred_flush_bios; + struct bio_list deferred_flush_completions; struct list_head prepared_mappings; struct list_head prepared_discards; struct list_head prepared_discards_pt2; @@ -956,6 +957,39 @@ static void process_prepared_mapping_fai mempool_free(m, &m->tc->pool->mapping_pool); } +static void complete_overwrite_bio(struct thin_c *tc, struct bio *bio) +{ + struct pool *pool = tc->pool; + unsigned long flags; + + /* + * If the bio has the REQ_FUA flag set we must commit the metadata + * before signaling its completion. + */ + if (!bio_triggers_commit(tc, bio)) { + bio_endio(bio); + return; + } + + /* + * Complete bio with an error if earlier I/O caused changes to the + * metadata that can't be committed, e.g, due to I/O errors on the + * metadata device. + */ + if (dm_thin_aborted_changes(tc->td)) { + bio_io_error(bio); + return; + } + + /* + * Batch together any bios that trigger commits and then issue a + * single commit for them in process_deferred_bios(). + */ + spin_lock_irqsave(&pool->lock, flags); + bio_list_add(&pool->deferred_flush_completions, bio); + spin_unlock_irqrestore(&pool->lock, flags); +} + static void process_prepared_mapping(struct dm_thin_new_mapping *m) { struct thin_c *tc = m->tc; @@ -988,7 +1022,7 @@ static void process_prepared_mapping(str */ if (bio) { inc_remap_and_issue_cell(tc, m->cell, m->data_block); - bio_endio(bio); + complete_overwrite_bio(tc, bio); } else { inc_all_io_entry(tc->pool, m->cell->holder); remap_and_issue(tc, m->cell->holder, m->data_block); @@ -2317,7 +2351,7 @@ static void process_deferred_bios(struct { unsigned long flags; struct bio *bio; - struct bio_list bios; + struct bio_list bios, bio_completions; struct thin_c *tc; tc = get_first_thin(pool); @@ -2328,26 +2362,36 @@ static void process_deferred_bios(struct } /* - * If there are any deferred flush bios, we must commit - * the metadata before issuing them. + * If there are any deferred flush bios, we must commit the metadata + * before issuing them or signaling their completion. */ bio_list_init(&bios); + bio_list_init(&bio_completions); + spin_lock_irqsave(&pool->lock, flags); bio_list_merge(&bios, &pool->deferred_flush_bios); bio_list_init(&pool->deferred_flush_bios); + + bio_list_merge(&bio_completions, &pool->deferred_flush_completions); + bio_list_init(&pool->deferred_flush_completions); spin_unlock_irqrestore(&pool->lock, flags); - if (bio_list_empty(&bios) && + if (bio_list_empty(&bios) && bio_list_empty(&bio_completions) && !(dm_pool_changed_this_transaction(pool->pmd) && need_commit_due_to_time(pool))) return; if (commit(pool)) { + bio_list_merge(&bios, &bio_completions); + while ((bio = bio_list_pop(&bios))) bio_io_error(bio); return; } pool->last_commit_jiffies = jiffies; + while ((bio = bio_list_pop(&bio_completions))) + bio_endio(bio); + while ((bio = bio_list_pop(&bios))) generic_make_request(bio); } @@ -2954,6 +2998,7 @@ static struct pool *pool_create(struct m INIT_DELAYED_WORK(&pool->no_space_timeout, do_no_space_timeout); spin_lock_init(&pool->lock); bio_list_init(&pool->deferred_flush_bios); + bio_list_init(&pool->deferred_flush_completions); INIT_LIST_HEAD(&pool->prepared_mappings); INIT_LIST_HEAD(&pool->prepared_discards); INIT_LIST_HEAD(&pool->prepared_discards_pt2);