From patchwork Fri Nov 17 17:22:47 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mikulas Patocka X-Patchwork-Id: 13459175 X-Patchwork-Delegate: snitzer@redhat.com Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.129.124]) (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 526EC37156 for ; Fri, 17 Nov 2023 17:22:52 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=redhat.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=redhat.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="YR6rlV24" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1700241771; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: in-reply-to:in-reply-to:references:references; bh=akSitNKLbWt9+Ota8avJH6qbNCjA5IDrL/Ubpvqmkf0=; b=YR6rlV24MPJRh0qeCqsZrIaGngCRx77NNK/9gKRQt7LrhsebvRu7DKVi64KEJzAoNiofej 8XC8ANf7LspOTaeVzw4Nz0nkeRr1R+UWFi6MN9241i0asdo56sametB7lfwCd16DbDEl0Q BBQmpwBga/AImWcQs5Yg55kBjiSkn74= Received: from mimecast-mx02.redhat.com (mimecast-mx02.redhat.com [66.187.233.88]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-659-xKXKar40PXmRiMAdffv2vQ-1; Fri, 17 Nov 2023 12:22:48 -0500 X-MC-Unique: xKXKar40PXmRiMAdffv2vQ-1 Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.rdu2.redhat.com [10.11.54.5]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id A242A8007B3; Fri, 17 Nov 2023 17:22:47 +0000 (UTC) Received: from file1-rdu.file-001.prod.rdu2.dc.redhat.com (unknown [10.11.5.21]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 9C85B5028; Fri, 17 Nov 2023 17:22:47 +0000 (UTC) Received: by file1-rdu.file-001.prod.rdu2.dc.redhat.com (Postfix, from userid 12668) id 964A630C1A8C; Fri, 17 Nov 2023 17:22:47 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by file1-rdu.file-001.prod.rdu2.dc.redhat.com (Postfix) with ESMTP id 954CB3FB76; Fri, 17 Nov 2023 18:22:47 +0100 (CET) Date: Fri, 17 Nov 2023 18:22:47 +0100 (CET) From: Mikulas Patocka To: Mike Snitzer cc: Christian Loehle , Harshit Mogalapalli , Heinz Mauelshagen , dm-devel@lists.linux.dev Subject: [PATCH 2/3] dm-delay: fix bugs in kthread mode In-Reply-To: <1b597cf-499b-a84c-fa5a-87a630987dff@redhat.com> Message-ID: References: <1b597cf-499b-a84c-fa5a-87a630987dff@redhat.com> Precedence: bulk X-Mailing-List: dm-devel@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.4.1 on 10.11.54.5 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com The patch 70bbeb29fab0 broke dm-delay seriously. It introduced the following bugs: * the function flush_worker_fn has no exit path - on unload, this function will just loop and consume 100% CPU without any progress * the wake-up mechanism in flush_worker_fn is racy - a wake up will be missed if the process adds entries to the delayed_bios list just before set_current_state(TASK_INTERRUPTIBLE) * flush_delayed_bios_fast submits a bio while holding a global mutex; this may deadlock if we have multiple stacked dm-delay devices and the underlying device attempts to acquire the mutex too * if the target constructor fails, it will call delay_dtr. delay_dtr would attempt to free dc->timer_lock without it being initialized by the constructor; if kthread allocation fails, delay_dtr would crash trying to dereference dc->worker because it would ontain error code This commit fixes these bugs. Fixes: 70bbeb29fab0 ("dm delay: for short delays, use kthread instead of timers and wq") Signed-off-by: Mikulas Patocka --- drivers/md/dm-delay.c | 61 ++++++++++++++++++++++++++++---------------------- 1 file changed, 35 insertions(+), 26 deletions(-) Index: linux-2.6/drivers/md/dm-delay.c =================================================================== --- linux-2.6.orig/drivers/md/dm-delay.c 2023-11-17 14:57:26.000000000 +0100 +++ linux-2.6/drivers/md/dm-delay.c 2023-11-17 14:58:44.000000000 +0100 @@ -73,9 +73,23 @@ static inline bool delay_is_fast(struct return !!dc->worker; } +static void flush_bios(struct bio *bio) +{ + struct bio *n; + + while (bio) { + n = bio->bi_next; + bio->bi_next = NULL; + dm_submit_bio_remap(bio, NULL); + bio = n; + } +} + static void flush_delayed_bios_fast(struct delay_c *dc, bool flush_all) { struct dm_delay_info *delayed, *next; + struct bio_list flush_bio_list; + bio_list_init(&flush_bio_list); mutex_lock(&delayed_bios_lock); list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) { @@ -83,47 +97,42 @@ static void flush_delayed_bios_fast(stru struct bio *bio = dm_bio_from_per_bio_data(delayed, sizeof(struct dm_delay_info)); list_del(&delayed->list); - dm_submit_bio_remap(bio, NULL); + bio_list_add(&flush_bio_list, bio); delayed->class->ops--; } } mutex_unlock(&delayed_bios_lock); + + flush_bios(bio_list_get(&flush_bio_list)); } static int flush_worker_fn(void *data) { struct delay_c *dc = data; - while (1) { + while (!kthread_should_stop()) { flush_delayed_bios_fast(dc, false); + mutex_lock(&delayed_bios_lock); if (unlikely(list_empty(&dc->delayed_bios))) { set_current_state(TASK_INTERRUPTIBLE); + mutex_unlock(&delayed_bios_lock); schedule(); - } else + } else { + mutex_unlock(&delayed_bios_lock); cond_resched(); + } } return 0; } -static void flush_bios(struct bio *bio) -{ - struct bio *n; - - while (bio) { - n = bio->bi_next; - bio->bi_next = NULL; - dm_submit_bio_remap(bio, NULL); - bio = n; - } -} - -static struct bio *flush_delayed_bios(struct delay_c *dc, bool flush_all) +static void flush_delayed_bios(struct delay_c *dc, bool flush_all) { struct dm_delay_info *delayed, *next; unsigned long next_expires = 0; unsigned long start_timer = 0; - struct bio_list flush_bios = { }; + struct bio_list flush_bio_list; + bio_list_init(&flush_bio_list); mutex_lock(&delayed_bios_lock); list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) { @@ -131,7 +140,7 @@ static struct bio *flush_delayed_bios(st struct bio *bio = dm_bio_from_per_bio_data(delayed, sizeof(struct dm_delay_info)); list_del(&delayed->list); - bio_list_add(&flush_bios, bio); + bio_list_add(&flush_bio_list, bio); delayed->class->ops--; continue; } @@ -147,7 +156,7 @@ static struct bio *flush_delayed_bios(st if (start_timer) queue_timeout(dc, next_expires); - return bio_list_get(&flush_bios); + flush_bios(bio_list_get(&flush_bio_list)); } static void flush_expired_bios(struct work_struct *work) @@ -158,7 +167,7 @@ static void flush_expired_bios(struct wo if (delay_is_fast(dc)) flush_delayed_bios_fast(dc, false); else - flush_bios(flush_delayed_bios(dc, false)); + flush_delayed_bios(dc, false); } static void delay_dtr(struct dm_target *ti) @@ -177,8 +186,7 @@ static void delay_dtr(struct dm_target * if (dc->worker) kthread_stop(dc->worker); - if (!delay_is_fast(dc)) - mutex_destroy(&dc->timer_lock); + mutex_destroy(&dc->timer_lock); kfree(dc); } @@ -236,6 +244,7 @@ static int delay_ctr(struct dm_target *t ti->private = dc; INIT_LIST_HEAD(&dc->delayed_bios); + mutex_init(&dc->timer_lock); dc->may_delay = true; dc->argc = argc; @@ -282,12 +291,12 @@ out: "dm-delay-flush-worker"); if (IS_ERR(dc->worker)) { ret = PTR_ERR(dc->worker); + dc->worker = NULL; goto bad; } } else { timer_setup(&dc->delay_timer, handle_delayed_timer, 0); INIT_WORK(&dc->flush_expired_bios, flush_expired_bios); - mutex_init(&dc->timer_lock); dc->kdelayd_wq = alloc_workqueue("kdelayd", WQ_MEM_RECLAIM, 0); if (!dc->kdelayd_wq) { ret = -EINVAL; @@ -345,11 +354,11 @@ static void delay_presuspend(struct dm_t dc->may_delay = false; mutex_unlock(&delayed_bios_lock); - if (delay_is_fast(dc)) + if (delay_is_fast(dc)) { flush_delayed_bios_fast(dc, true); - else { + } else { del_timer_sync(&dc->delay_timer); - flush_bios(flush_delayed_bios(dc, true)); + flush_delayed_bios(dc, true); } }