Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753916AbdGYTuG (ORCPT ); Tue, 25 Jul 2017 15:50:06 -0400 Received: from mail.linuxfoundation.org ([140.211.169.12]:36274 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753847AbdGYTYo (ORCPT ); Tue, 25 Jul 2017 15:24:44 -0400 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Marta Rybczynska , Sagi Grimberg , Christoph Hellwig Subject: [PATCH 4.12 119/196] nvme-rdma: remove race conditions from IB signalling Date: Tue, 25 Jul 2017 12:21:58 -0700 Message-Id: <20170725192052.132005185@linuxfoundation.org> X-Mailer: git-send-email 2.13.3 In-Reply-To: <20170725192046.422343510@linuxfoundation.org> References: <20170725192046.422343510@linuxfoundation.org> User-Agent: quilt/0.65 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2366 Lines: 72 4.12-stable review patch. If anyone has any objections, please let me know. ------------------ From: Marta Rybczynska commit 5e599d73c1c1816af07f94ddba879499aa39b43c upstream. This patch improves the way the RDMA IB signalling is done by using atomic operations for the signalling variable. This avoids race conditions on sig_count. The signalling interval changes slightly and is now the largest power of two not larger than queue depth / 2. ilog() usage idea by Bart Van Assche. Signed-off-by: Marta Rybczynska Reviewed-by: Sagi Grimberg Signed-off-by: Christoph Hellwig Signed-off-by: Greg Kroah-Hartman --- drivers/nvme/host/rdma.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) --- a/drivers/nvme/host/rdma.c +++ b/drivers/nvme/host/rdma.c @@ -88,7 +88,7 @@ enum nvme_rdma_queue_flags { struct nvme_rdma_queue { struct nvme_rdma_qe *rsp_ring; - u8 sig_count; + atomic_t sig_count; int queue_size; size_t cmnd_capsule_len; struct nvme_rdma_ctrl *ctrl; @@ -553,6 +553,7 @@ static int nvme_rdma_init_queue(struct n queue->cmnd_capsule_len = sizeof(struct nvme_command); queue->queue_size = queue_size; + atomic_set(&queue->sig_count, 0); queue->cm_id = rdma_create_id(&init_net, nvme_rdma_cm_handler, queue, RDMA_PS_TCP, IB_QPT_RC); @@ -1040,17 +1041,16 @@ static void nvme_rdma_send_done(struct i nvme_rdma_wr_error(cq, wc, "SEND"); } -static inline int nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue) +/* + * We want to signal completion at least every queue depth/2. This returns the + * largest power of two that is not above half of (queue size + 1) to optimize + * (avoid divisions). + */ +static inline bool nvme_rdma_queue_sig_limit(struct nvme_rdma_queue *queue) { - int sig_limit; + int limit = 1 << ilog2((queue->queue_size + 1) / 2); - /* - * We signal completion every queue depth/2 and also handle the - * degenerated case of a device with queue_depth=1, where we - * would need to signal every message. - */ - sig_limit = max(queue->queue_size / 2, 1); - return (++queue->sig_count % sig_limit) == 0; + return (atomic_inc_return(&queue->sig_count) & (limit - 1)) == 0; } static int nvme_rdma_post_send(struct nvme_rdma_queue *queue,