From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754061AbbALSJj (ORCPT ); Mon, 12 Jan 2015 13:09:39 -0500 Received: from youngberry.canonical.com ([91.189.89.112]:57818 "EHLO youngberry.canonical.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753909AbbALSGw (ORCPT ); Mon, 12 Jan 2015 13:06:52 -0500 From: Luis Henriques To: linux-kernel@vger.kernel.org, stable@vger.kernel.org, kernel-team@lists.ubuntu.com Cc: Bart Van Assche , Christoph Hellwig , Robert Elliott , Ming Lei , Alexander Gordeev , Jens Axboe , Luis Henriques Subject: [PATCH 3.16.y-ckt 103/216] blk-mq: Fix a race between bt_clear_tag() and bt_get() Date: Mon, 12 Jan 2015 18:03:40 +0000 Message-Id: <1421085933-32536-104-git-send-email-luis.henriques@canonical.com> X-Mailer: git-send-email 2.1.4 In-Reply-To: <1421085933-32536-1-git-send-email-luis.henriques@canonical.com> References: <1421085933-32536-1-git-send-email-luis.henriques@canonical.com> X-Extended-Stable: 3.16 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 3.16.7-ckt4 -stable review patch. If anyone has any objections, please let me know. ------------------ From: Bart Van Assche commit c38d185d4af12e8be63ca4b6745d99449c450f12 upstream. What we need is the following two guarantees: * Any thread that observes the effect of the test_and_set_bit() by __bt_get_word() also observes the preceding addition of 'current' to the appropriate wait list. This is guaranteed by the semantics of the spin_unlock() operation performed by prepare_and_wait(). Hence the conversion of test_and_set_bit_lock() into test_and_set_bit(). * The wait lists are examined by bt_clear() after the tag bit has been cleared. clear_bit_unlock() guarantees that any thread that observes that the bit has been cleared also observes the store operations preceding clear_bit_unlock(). However, clear_bit_unlock() does not prevent that the wait lists are examined before that the tag bit is cleared. Hence the addition of a memory barrier between clear_bit() and the wait list examination. Signed-off-by: Bart Van Assche Cc: Christoph Hellwig Cc: Robert Elliott Cc: Ming Lei Cc: Alexander Gordeev Signed-off-by: Jens Axboe Signed-off-by: Luis Henriques --- block/blk-mq-tag.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/block/blk-mq-tag.c b/block/blk-mq-tag.c index a1bc33974fc4..1caaa7c4605d 100644 --- a/block/blk-mq-tag.c +++ b/block/blk-mq-tag.c @@ -158,7 +158,7 @@ restart: return -1; } last_tag = tag + 1; - } while (test_and_set_bit_lock(tag, &bm->word)); + } while (test_and_set_bit(tag, &bm->word)); return tag; } @@ -342,11 +342,10 @@ static void bt_clear_tag(struct blk_mq_bitmap_tags *bt, unsigned int tag) struct bt_wait_state *bs; int wait_cnt; - /* - * The unlock memory barrier need to order access to req in free - * path and clearing tag bit - */ - clear_bit_unlock(TAG_TO_BIT(bt, tag), &bt->map[index].word); + clear_bit(TAG_TO_BIT(bt, tag), &bt->map[index].word); + + /* Ensure that the wait list checks occur after clear_bit(). */ + smp_mb(); bs = bt_wake_ptr(bt); if (!bs) -- 2.1.4