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=-8.2 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS, USER_AGENT_SANE_1 autolearn=unavailable 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 91165ECE59D for ; Wed, 16 Oct 2019 04:38:38 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 6373120873 for ; Wed, 16 Oct 2019 04:38:38 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S2389621AbfJPEid (ORCPT ); Wed, 16 Oct 2019 00:38:33 -0400 Received: from mx1.redhat.com ([209.132.183.28]:34654 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726502AbfJPEid (ORCPT ); Wed, 16 Oct 2019 00:38:33 -0400 Received: from smtp.corp.redhat.com (int-mx08.intmail.prod.int.phx2.redhat.com [10.5.11.23]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id ABFD53082B40; Wed, 16 Oct 2019 04:38:32 +0000 (UTC) Received: from [10.72.12.53] (ovpn-12-53.pek2.redhat.com [10.72.12.53]) by smtp.corp.redhat.com (Postfix) with ESMTP id 2228819C5B; Wed, 16 Oct 2019 04:38:19 +0000 (UTC) Subject: Re: [PATCH RFC v1 1/2] vhost: option to fetch descriptors through an independent struct To: "Michael S. Tsirkin" Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org, virtualization@lists.linux-foundation.org, netdev@vger.kernel.org References: <20191011134358.16912-1-mst@redhat.com> <20191011134358.16912-2-mst@redhat.com> <3b2a6309-9d21-7172-a581-9f0f1d5c1427@redhat.com> <20191012162445-mutt-send-email-mst@kernel.org> <20191014085806-mutt-send-email-mst@kernel.org> From: Jason Wang Message-ID: Date: Wed, 16 Oct 2019 12:38:15 +0800 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.8.0 MIME-Version: 1.0 In-Reply-To: <20191014085806-mutt-send-email-mst@kernel.org> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Content-Language: en-US X-Scanned-By: MIMEDefang 2.84 on 10.5.11.23 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.45]); Wed, 16 Oct 2019 04:38:32 +0000 (UTC) Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org On 2019/10/16 上午4:20, Michael S. Tsirkin wrote: > On Mon, Oct 14, 2019 at 09:43:25AM +0800, Jason Wang wrote: >> On 2019/10/13 上午4:27, Michael S. Tsirkin wrote: >>> On Sat, Oct 12, 2019 at 03:28:49PM +0800, Jason Wang wrote: >>>> On 2019/10/11 下午9:45, Michael S. Tsirkin wrote: >>>>> The idea is to support multiple ring formats by converting >>>>> to a format-independent array of descriptors. >>>>> >>>>> This costs extra cycles, but we gain in ability >>>>> to fetch a batch of descriptors in one go, which >>>>> is good for code cache locality. >>>>> >>>>> To simplify benchmarking, I kept the old code >>>>> around so one can switch back and forth by >>>>> writing into a module parameter. >>>>> This will go away in the final submission. >>>>> >>>>> This patch causes a minor performance degradation, >>>>> it's been kept as simple as possible for ease of review. >>>>> Next patch gets us back the performance by adding batching. >>>>> >>>>> Signed-off-by: Michael S. Tsirkin >>>>> --- >>>>> drivers/vhost/test.c | 17 ++- >>>>> drivers/vhost/vhost.c | 299 +++++++++++++++++++++++++++++++++++++++++- >>>>> drivers/vhost/vhost.h | 16 +++ >>>>> 3 files changed, 327 insertions(+), 5 deletions(-) >>>>> >>>>> diff --git a/drivers/vhost/test.c b/drivers/vhost/test.c >>>>> index 056308008288..39a018a7af2d 100644 >>>>> --- a/drivers/vhost/test.c >>>>> +++ b/drivers/vhost/test.c >>>>> @@ -18,6 +18,9 @@ >>>>> #include "test.h" >>>>> #include "vhost.h" >>>>> +static int newcode = 0; >>>>> +module_param(newcode, int, 0644); >>>>> + >>>>> /* Max number of bytes transferred before requeueing the job. >>>>> * Using this limit prevents one virtqueue from starving others. */ >>>>> #define VHOST_TEST_WEIGHT 0x80000 >>>>> @@ -58,10 +61,16 @@ static void handle_vq(struct vhost_test *n) >>>>> vhost_disable_notify(&n->dev, vq); >>>>> for (;;) { >>>>> - head = vhost_get_vq_desc(vq, vq->iov, >>>>> - ARRAY_SIZE(vq->iov), >>>>> - &out, &in, >>>>> - NULL, NULL); >>>>> + if (newcode) >>>>> + head = vhost_get_vq_desc_batch(vq, vq->iov, >>>>> + ARRAY_SIZE(vq->iov), >>>>> + &out, &in, >>>>> + NULL, NULL); >>>>> + else >>>>> + head = vhost_get_vq_desc(vq, vq->iov, >>>>> + ARRAY_SIZE(vq->iov), >>>>> + &out, &in, >>>>> + NULL, NULL); >>>>> /* On error, stop handling until the next kick. */ >>>>> if (unlikely(head < 0)) >>>>> break; >>>>> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c >>>>> index 36ca2cf419bf..36661d6cb51f 100644 >>>>> --- a/drivers/vhost/vhost.c >>>>> +++ b/drivers/vhost/vhost.c >>>>> @@ -301,6 +301,7 @@ static void vhost_vq_reset(struct vhost_dev *dev, >>>>> struct vhost_virtqueue *vq) >>>>> { >>>>> vq->num = 1; >>>>> + vq->ndescs = 0; >>>>> vq->desc = NULL; >>>>> vq->avail = NULL; >>>>> vq->used = NULL; >>>>> @@ -369,6 +370,9 @@ static int vhost_worker(void *data) >>>>> static void vhost_vq_free_iovecs(struct vhost_virtqueue *vq) >>>>> { >>>>> + kfree(vq->descs); >>>>> + vq->descs = NULL; >>>>> + vq->max_descs = 0; >>>>> kfree(vq->indirect); >>>>> vq->indirect = NULL; >>>>> kfree(vq->log); >>>>> @@ -385,6 +389,10 @@ static long vhost_dev_alloc_iovecs(struct vhost_dev *dev) >>>>> for (i = 0; i < dev->nvqs; ++i) { >>>>> vq = dev->vqs[i]; >>>>> + vq->max_descs = dev->iov_limit; >>>>> + vq->descs = kmalloc_array(vq->max_descs, >>>>> + sizeof(*vq->descs), >>>>> + GFP_KERNEL); >>>> Is iov_limit too much here? It can obviously increase the footprint. I guess >>>> the batching can only be done for descriptor without indirect or next set. >>>> Then we may batch 16 or 64. >>>> >>>> Thanks >>> Yes, next patch only batches up to 64. But we do need iov_limit because >>> guest can pass a long chain of scatter/gather. >>> We already have iovecs in a huge array so this does not look like >>> a big deal. If we ever teach the code to avoid the huge >>> iov arrays by handling huge s/g lists piece by piece, >>> we can make the desc array smaller at the same point. >>> >> Another possible issue, if we try to batch descriptor chain when we've >> already batched some descriptors, we may reach the limit then some of the >> descriptors might need re-read. >> >> Or we may need circular index (head, tail) in this case? >> >> Thanks > We never supported more than IOV_MAX descriptors. > And we don't batch more than iov_limit - IOV_MAX. Ok, but what happens when we've already batched 63 descriptors then come a 3 descriptor chain? And it looks to me we need forget the cached descriptor during set_vring_base() Thanks > > so buffer never overflows. >