From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 4C48BA0524; Fri, 31 Jan 2020 19:38:32 +0100 (CET) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id CAE241C0DC; Fri, 31 Jan 2020 19:38:31 +0100 (CET) Received: from us-smtp-1.mimecast.com (us-smtp-delivery-1.mimecast.com [205.139.110.120]) by dpdk.org (Postfix) with ESMTP id 884211C0C8 for ; Fri, 31 Jan 2020 19:38:30 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1580495910; 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: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references:autocrypt:autocrypt; bh=e0JK2JCtkBgpMK7VpTBVJ8FjebDgnT4WwtOEf83/f0g=; b=dGiO7jWZfnvJGpN9Go1lZrWFP6jVh3ZcL9teOMLEN1ypSqEvuCpBKdtoTbfCD+JRzUV9t0 h9qF+zNpEx0DgkSDCWrwpW1lU2WEnXEMSng+2XxXy6jTBU3f0CsQC1UHKxxUwO8KGhe399 8Dm/q8sxC6/uDg2xs+W1zjw+9Pg1ygY= Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-54-TYNCboQFPnyoXnUAwB1QUg-1; Fri, 31 Jan 2020 13:38:26 -0500 Received: from smtp.corp.redhat.com (int-mx04.intmail.prod.int.phx2.redhat.com [10.5.11.14]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id 0B7C9189F760; Fri, 31 Jan 2020 18:38:25 +0000 (UTC) Received: from [10.33.36.12] (unknown [10.33.36.12]) by smtp.corp.redhat.com (Postfix) with ESMTP id 61CC25D9E5; Fri, 31 Jan 2020 18:38:19 +0000 (UTC) To: =?UTF-8?Q?Eugenio_P=c3=a9rez?= , dev@dpdk.org, yong.liu@intel.com Cc: Maxime Coquelin , Adrian Moreno Zapata , Jason Wang , "Michael S. Tsirkin" References: <20200129193310.9157-1-eperezma@redhat.com> From: Kevin Traynor Autocrypt: addr=ktraynor@redhat.com; keydata= mQINBF2J2awBEADUEPNhgNI+nJNgiTAUcw4YIgVXEoHlsNPyyzG1BEXkWXALy0Y3fNTiw6+r ltWDkF9jzL9kfkecgQ67itGfk1OaBXgSGKuw1PUpxAwX2Bi76LAR6M5OsyGM9TSVVQwARalz hMwRBIZPzPc7or6Pw7jAOJ8SQGJ1Zlp1YJCjrvpe87V1tH/LY8Wnxn/EuoseFmWILAQZAtYS tGjcrAgYn3SPMLR1B0BP5bTBY06vWQjiufH8drenfDnMJAzuBdG1mqjnTqCjULZ3Hunv4xqZ aMnkvL/K5Tj1c12Oe4930EE53LrXIBUltRg5mBudSWHnC7twjH0082HH9f963Z/2UI63SFIT iUvRvAzJYytgy7XnWLQ0+goZBADKYfolOuC0H8VgCaux8u8KFF28Dy+N6TV2KI58jTlyg1Zu l7QwykZpnOkJFiy37Gfbu3YEOzO72cP/S7/A+zvuqkxi63jyEkd+FY99vLt/HN2MUZwRmKDw UPbLkmrs8WU01/POVsqDcfvz7vu2St8hqqTiSIdQGS2zyTKB2/DvPSM3jws3udkIYSuhn+X4 QBiV6lkVZ7DSE6a065gnAauAql+b32Eymy+xnG5jCt1tR+0Cp2VZYCR9OU2gmomUKBDoX/He pSgED01CqYPNjN+TddirwmQX7ep4DtXc8FWvv2g/pq9WZFQk2QARAQABtCNLZXZpbiBUcmF5 bm9yIDxrdHJheW5vckByZWRoYXQuY29tPokCTgQTAQgAOBYhBAoiOaH51tHF7VYtEI9CINER a+yJBQJdidmsAhsDBQsJCAcCBhUKCQgLAgQWAgMBAh4BAheAAAoJEI9CINERa+yJoxIP/3VF 2TIgW4ckxhRFCvFu/606bnvCPie88ake4uWVWMAWwcMc4fKEltRWRCpkSVOwgqoMHnyHxK5r kOKzx2CLJMX5TgTMfKzPuaBDHngHLUzl2DStpBzrod0cVg5TShdmmfjY61uxRJKz+DlSkwgJ riADdVF5PPosQXTkKSGf2ombpTGpx/pue9ocjnr3x4SDpRLlnooM6Jf/3Y3Ib4jX6HPEyWuY b+owIIk9y2nRRGPQ6jbqAhsrXd9V+77UL0QuGWloMuKMZFbNg8hbu7X5aFijAbfxj4YUgojS ba7gfGZQan8h32A9KGQWrmsCBc3j2GqEPsX0r05X7cn7WL6IOPgQJ5EiQ7PlazQYVLrvZg9B n0GKK0k6895mLG0ZZ5v/qajOPF52etSmvFD1WUPb4OqaHqGA9ZtMpaKFRt7Y6rpXqKNU1xzW F5KjbTPtTb9WF3An8dciVv+AYUI7totkZYkWvQtgss8lfaX3NKUvXLVxqK0z3dQyr7rF/tYz PneTKypSksjCgaEBLSrsRmM5zKfe7tSNF/fDntfIq/029Jtcw29TcWEP57peNu6TtejewQD9 sTI+oqiXvW2D5l7LNUDYG8eMJp2oT7I0ZSBRvwcbmjH0DtN/bXCCFfCvk8Yic68F3tV1ctix wQARVKDBhT30uCxycRWojCYqTgNJJS71uQINBF2J2awBEADP57PR2IpSYBeNSrsAjeIcsahE N4SQP2C4s50S8QEWAUhqMRI7WNv5cfeef0nDvcl1IUA6oz5SokbcsbMa+mRgaNF4N5KikWTO LPYxq2YVJoXwJ+tKmNzyOLFUIfFJ4NBJZple5dTfWzD00Dbb19Mri1hy1mWMqNTPGBee1+hw Qcp6n3mmGECvajs8G5A7NyXbwL8ihN7HX9D01ucD62b4G03yKe2g/hvKgcdUVmhCldJlF27I 2fSR9tDxH9pZqRODY4rjbFZEey/vWKXqjE+DQ8AtMSEaDfFe5D+i4Aw6erWQ3Wr+DwZt1/7G dIAElGA/q90T1ENVwJX9y7fsQssawKYYdDqURHCl5JuDXI+VXUypExipUUT5SPycMmbLsx0D iKEqPPDQWKxkIDVKqj2+EhamSuJznZUwBLJKn0h4zrIWiXWUy07lRwtVuhaDXhF3GfW+5W/x wAg7Qg3w00ASsb/XTHBIhMnenKDfS7ihtQA8SacwX8ySdxb+15XPyiplM979qBQ0mhnilulm MIJzEf/JxoYR5huuj4f1PFqqrsP06Dl+YGB7dQZp3IKggS5c3/TAynARRg9N89UsDXNtp7X0 tgIPFF5k6fnHE0J5O64GYHeTqN/1aE6dAEOV9WrGzQAJxU9ipikb8jKAWXzLewRIKGmoPcRZ WdB0NmIjmQARAQABiQI2BBgBCAAgFiEECiI5ofnW0cXtVi0Qj0Ig0RFr7IkFAl2J2awCGwwA CgkQj0Ig0RFr7IkkORAAl/NbX93WK5MEoRw7/DaPTo/Lo6Pj1XMeSqGyACigHK/452UDvlEH NjNJMzYYrNIjMtEmN9VVCfjT38CSca7mpGQVwchc0mC7QSPAETLCS+UacVf/Kwxz5FfkEUUw UT7A+uyVOIgW3d9ldlRzkHA2czonSSgTQU+i2g6DM4ha+BuQb4byAXH6HQHt/Zh1J64z0ohH v6iGsCzCY/sMWF8+LEGSnzMGRCLiiwSF0vJBHbzWK68fANaF4gBV0Z/+6tQRFN7YMhj/INmk qgvHj1ZzHFNtirjMGPRxoZs51YoLQM/aBPxKrnmXThx1ufH+0L6sGmFTugiDt0XSEkC5reH7 a+VhQ1VTFFQrClA8NmDSPzFeuhru4ryaaDHO+uEB16cNHxHrQtlP/2hts2JM5lwkZRWJ5A57 h8eDEIK5be47T85NVHfuTaboNRmgg1HygVejhGUtt69u/0MVRg/roUTa0FyEbNsvz4qAecyW yWzMcVrcGJDQLC9JLKEpoyUF6gdTKaiDL2Vao4+XRIA3Y57b6MO35a3HuzAv7+i5Z0mnDEJO XxXqTOmKYpMIGexzM/PtuA0712sT1abG9tAJ17ao/B7cqMW5IkKkalemFbWfI2unns4Papvo tk9igVqyp6EJDU98z5TJioCVojwK2laDaoIjTJk9YYv3iwCsqPd5feU= Message-ID: <1bdb5d16-2a39-1a19-9c47-69b8cb4607a1@redhat.com> Date: Fri, 31 Jan 2020 18:38:18 +0000 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.2.2 MIME-Version: 1.0 In-Reply-To: <20200129193310.9157-1-eperezma@redhat.com> Content-Language: en-US X-Scanned-By: MIMEDefang 2.79 on 10.5.11.14 X-MC-Unique: TYNCboQFPnyoXnUAwB1QUg-1 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [dpdk-dev] [PATCH] vhost: flush shadow tx if there is no more packets X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Hi Eugenio, On 29/01/2020 19:33, Eugenio P=C3=A9rez wrote: > The current implementation of vhost_net in packed vring tries to fill > the shadow vector before send any actual changes to the guest. While > this can be beneficial for the throughput, it conflicts with some > bufferfloats methods like the linux kernel napi, that stops > transmitting packets if there are too much bytes/buffers in the > driver. >=20 > To solve it, we flush the shadow packets at the end of > virtio_dev_tx_packed if we have starved the vring, i.e., the next > buffer is not available for the device. >=20 > Since this last check can be expensive because of the atomic, we only > check it if we have not obtained the expected (count) packets. If it > happens to obtain "count" packets and there is no more available > packets the caller needs to keep call virtio_dev_tx_packed again. >=20 It seems to be fixing an issue and should be considered for stable branches? You can add the tags needed in the commit message here: Fixes: Cc: stable@dpdk.org > Signed-off-by: Eugenio P=C3=A9rez > --- > lib/librte_vhost/virtio_net.c | 27 ++++++++++++++++++++++++++- > 1 file changed, 26 insertions(+), 1 deletion(-) >=20 > diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.= c > index 21c311732..ac2842b2d 100644 > --- a/lib/librte_vhost/virtio_net.c > +++ b/lib/librte_vhost/virtio_net.c > @@ -2133,6 +2133,20 @@ virtio_dev_tx_packed_zmbuf(struct virtio_net *dev, > =09return pkt_idx; > } > =20 > +static __rte_always_inline bool > +next_desc_is_avail(const struct vhost_virtqueue *vq) > +{ > +=09bool wrap_counter =3D vq->avail_wrap_counter; > +=09uint16_t next_used_idx =3D vq->last_used_idx + 1; > + > +=09if (next_used_idx >=3D vq->size) { > +=09=09next_used_idx -=3D vq->size; > +=09=09wrap_counter ^=3D 1; > +=09} > + > +=09return desc_is_avail(&vq->desc_packed[next_used_idx], wrap_counter); > +} > + > static __rte_noinline uint16_t > virtio_dev_tx_packed(struct virtio_net *dev, > =09=09 struct vhost_virtqueue *vq, > @@ -2165,9 +2179,20 @@ virtio_dev_tx_packed(struct virtio_net *dev, > =20 > =09} while (remained); > =20 > -=09if (vq->shadow_used_idx) > +=09if (vq->shadow_used_idx) { > =09=09do_data_copy_dequeue(vq); > =20 > +=09=09if (remained && !next_desc_is_avail(vq)) { > +=09=09=09/* > +=09=09=09 * The guest may be waiting to TX some buffers to > +=09=09=09 * enqueue more to avoid bufferfloat, so we try to > +=09=09=09 * reduce latency here. > +=09=09=09 */ > +=09=09=09vhost_flush_dequeue_shadow_packed(dev, vq); > +=09=09=09vhost_vring_call_packed(dev, vq); > +=09=09} > +=09} > + > =09return pkt_idx; > } > =20 >=20