From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Wang Subject: Re: [PATCH V2 1/2] vhost_net: stop polling socket during rx processing Date: Tue, 31 May 2016 11:14:10 +0800 Message-ID: <574D0182.9020701@redhat.com> References: <1464590874-39539-1-git-send-email-jasowang@redhat.com> <1464590874-39539-2-git-send-email-jasowang@redhat.com> <20160530184211-mutt-send-email-mst@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8"; Format="flowed" Content-Transfer-Encoding: base64 Return-path: In-Reply-To: <20160530184211-mutt-send-email-mst@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: "Michael S. Tsirkin" Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, kvm@vger.kernel.org, virtualization@lists.linux-foundation.org List-Id: virtualization@lists.linuxfoundation.org CgpPbiAyMDE25bm0MDXmnIgzMOaXpSAyMzo0NywgTWljaGFlbCBTLiBUc2lya2luIHdyb3RlOgo+ IE9uIE1vbiwgTWF5IDMwLCAyMDE2IGF0IDAyOjQ3OjUzQU0gLTA0MDAsIEphc29uIFdhbmcgd3Jv dGU6Cj4+IFdlIGRvbid0IHN0b3AgcnggcG9sbGluZyBzb2NrZXQgZHVyaW5nIHJ4IHByb2Nlc3Np bmcsIHRoaXMgd2lsbCBsZWFkCj4+IHVubmVjZXNzYXJ5IHdha2V1cHMgZnJvbSB1bmRlciBsYXll ciBuZXQgZGV2aWNlcyAoRS5nCj4+IHNvY2tfZGVmX3JlYWRhYmxlKCkgZm9ybSB0dW4pLiBSeCB3 aWxsIGJlIHNsb3dlZCBkb3duIGluIHRoaXMKPj4gd2F5LiBUaGlzIHBhdGNoIGF2b2lkcyB0aGlz IGJ5IHN0b3AgcG9sbGluZyBzb2NrZXQgZHVyaW5nIHJ4Cj4+IHByb2Nlc3NpbmcuIEEgc21hbGwg ZHJhd2JhY2sgaXMgdGhhdCB0aGlzIGludHJvZHVjZXMgc29tZSBvdmVyaGVhZHMgaW4KPj4gbGln aHQgbG9hZCBjYXNlIGJlY2F1c2Ugb2YgdGhlIGV4dHJhIHN0YXJ0L3N0b3AgcG9sbGluZywgYnV0 IHNpbmdsZQo+PiBuZXRwZXJmIFRDUF9SUiBkb2VzIG5vdCBub3RpY2UgYW55IGNoYW5nZS4gSW4g YSBzdXBlciBoZWF2eSBsb2FkIGNhc2UsCj4+IGUuZyB1c2luZyBwa3RnZW4gdG8gaW5qZWN0IHBh Y2tldCB0byBndWVzdCwgd2UgZ2V0IGFib3V0IH44LjglCj4+IGltcHJvdmVtZW50IG9uIHBwczoK Pj4KPj4gYmVmb3JlOiB+MTI0MDAwMCBwa3Qvcwo+PiBhZnRlcjogIH4xMzUwMDAwIHBrdC9zCj4+ Cj4+IFNpZ25lZC1vZmYtYnk6IEphc29uIFdhbmcgPGphc293YW5nQHJlZGhhdC5jb20+Cj4+IC0t LQo+PiAgIGRyaXZlcnMvdmhvc3QvbmV0LmMgfCA1NiArKysrKysrKysrKysrKysrKysrKysrKysr KystLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQo+PiAgIDEgZmlsZSBjaGFuZ2VkLCAyOSBpbnNl cnRpb25zKCspLCAyNyBkZWxldGlvbnMoLSkKPj4KPj4gZGlmZiAtLWdpdCBhL2RyaXZlcnMvdmhv c3QvbmV0LmMgYi9kcml2ZXJzL3Zob3N0L25ldC5jCj4+IGluZGV4IDEwZmY0OTQuLmU5MTYwM2Ig MTAwNjQ0Cj4+IC0tLSBhL2RyaXZlcnMvdmhvc3QvbmV0LmMKPj4gKysrIGIvZHJpdmVycy92aG9z dC9uZXQuYwo+PiBAQCAtMzAxLDYgKzMwMSwzMiBAQCBzdGF0aWMgYm9vbCB2aG9zdF9jYW5fYnVz eV9wb2xsKHN0cnVjdCB2aG9zdF9kZXYgKmRldiwKPj4gICAJICAgICAgICF2aG9zdF9oYXNfd29y ayhkZXYpOwo+PiAgIH0KPj4gICAKPj4gK3N0YXRpYyB2b2lkIHZob3N0X25ldF9kaXNhYmxlX3Zx KHN0cnVjdCB2aG9zdF9uZXQgKm4sCj4+ICsJCQkJIHN0cnVjdCB2aG9zdF92aXJ0cXVldWUgKnZx KQo+PiArewo+PiArCXN0cnVjdCB2aG9zdF9uZXRfdmlydHF1ZXVlICpudnEgPQo+PiArCQljb250 YWluZXJfb2YodnEsIHN0cnVjdCB2aG9zdF9uZXRfdmlydHF1ZXVlLCB2cSk7Cj4+ICsJc3RydWN0 IHZob3N0X3BvbGwgKnBvbGwgPSBuLT5wb2xsICsgKG52cSAtIG4tPnZxcyk7Cj4+ICsJaWYgKCF2 cS0+cHJpdmF0ZV9kYXRhKQo+PiArCQlyZXR1cm47Cj4+ICsJdmhvc3RfcG9sbF9zdG9wKHBvbGwp Owo+PiArfQo+PiArCj4+ICtzdGF0aWMgaW50IHZob3N0X25ldF9lbmFibGVfdnEoc3RydWN0IHZo b3N0X25ldCAqbiwKPj4gKwkJCQlzdHJ1Y3Qgdmhvc3RfdmlydHF1ZXVlICp2cSkKPj4gK3sKPj4g KwlzdHJ1Y3Qgdmhvc3RfbmV0X3ZpcnRxdWV1ZSAqbnZxID0KPj4gKwkJY29udGFpbmVyX29mKHZx LCBzdHJ1Y3Qgdmhvc3RfbmV0X3ZpcnRxdWV1ZSwgdnEpOwo+PiArCXN0cnVjdCB2aG9zdF9wb2xs ICpwb2xsID0gbi0+cG9sbCArIChudnEgLSBuLT52cXMpOwo+PiArCXN0cnVjdCBzb2NrZXQgKnNv Y2s7Cj4+ICsKPj4gKwlzb2NrID0gdnEtPnByaXZhdGVfZGF0YTsKPj4gKwlpZiAoIXNvY2spCj4+ ICsJCXJldHVybiAwOwo+PiArCj4+ICsJcmV0dXJuIHZob3N0X3BvbGxfc3RhcnQocG9sbCwgc29j ay0+ZmlsZSk7Cj4+ICt9Cj4+ICsKPj4gICBzdGF0aWMgaW50IHZob3N0X25ldF90eF9nZXRfdnFf ZGVzYyhzdHJ1Y3Qgdmhvc3RfbmV0ICpuZXQsCj4+ICAgCQkJCSAgICBzdHJ1Y3Qgdmhvc3Rfdmly dHF1ZXVlICp2cSwKPj4gICAJCQkJICAgIHN0cnVjdCBpb3ZlYyBpb3ZbXSwgdW5zaWduZWQgaW50 IGlvdl9zaXplLAo+IEJUVyB3ZSBtaWdodCB3YW50IHRvIHJlbmFtZSB0aGVzZSBmdW5jdGlvbnMs IG5hbWUgbm8gbG9uZ2VyCj4gcmVmbGVjdHMgZnVuY3Rpb24gLi4uCgpEbyB5b3UgbWVhbiBhZGRp bmcgc29tZXRoaW5nIHJlZmxlY3QgYnVzeSBwb2xsaW5nIGluIHRoZSBuYW1lPyBUaGVuIHRoZSAK bmFtZSBtYXkgYmUgdG9vIGxvbmcgb3IgaGF2ZSBzdWdnZXN0aW9uIG9uIHRoZSBuYW1lPwoKPgo+ Cj4+IEBAIC02MjcsNiArNjUzLDcgQEAgc3RhdGljIHZvaWQgaGFuZGxlX3J4KHN0cnVjdCB2aG9z dF9uZXQgKm5ldCkKPj4gICAJaWYgKCFzb2NrKQo+PiAgIAkJZ290byBvdXQ7Cj4+ICAgCXZob3N0 X2Rpc2FibGVfbm90aWZ5KCZuZXQtPmRldiwgdnEpOwo+PiArCXZob3N0X25ldF9kaXNhYmxlX3Zx KG5ldCwgdnEpOwo+PiAgIAo+PiAgIAl2aG9zdF9obGVuID0gbnZxLT52aG9zdF9obGVuOwo+PiAg IAlzb2NrX2hsZW4gPSBudnEtPnNvY2tfaGxlbjsKPj4gQEAgLTcxNSw5ICs3NDIsMTAgQEAgc3Rh dGljIHZvaWQgaGFuZGxlX3J4KHN0cnVjdCB2aG9zdF9uZXQgKm5ldCkKPj4gICAJCXRvdGFsX2xl biArPSB2aG9zdF9sZW47Cj4+ICAgCQlpZiAodW5saWtlbHkodG90YWxfbGVuID49IFZIT1NUX05F VF9XRUlHSFQpKSB7Cj4+ICAgCQkJdmhvc3RfcG9sbF9xdWV1ZSgmdnEtPnBvbGwpOwo+PiAtCQkJ YnJlYWs7Cj4+ICsJCQlnb3RvIG91dDsKPj4gICAJCX0KPj4gICAJfQo+PiArCXZob3N0X25ldF9l bmFibGVfdnEobmV0LCB2cSk7Cj4gT0sgc28gaWYgc29jayBpcyByZWFkYWJsZSBidXQgUlggVlEg aXMgZW1wdHksIHRoaXMgd2lsbAo+IGltbWVkaWF0ZWx5IHNjaGVkdWxlIGFub3RoZXIgcm91bmQg b2YgaGFuZGxlX3J4IGFuZCBzbyBhZAo+IGluZmluaXR1bSwKPgo+IExvb2tzIGxpa2UgYSBidWcu CgpZZXMgaXQgaXMsIHdpbGwgY2hhbmdlIHRoZSBhYm92ZSBoZWFkY291bnQgY2hlY2sgdG86Cgog ICAgICAgICAgICAgICAgIC8qIE9LLCBub3cgd2UgbmVlZCB0byBrbm93IGFib3V0IGFkZGVkIGRl c2NyaXB0b3JzLiAqLwogICAgICAgICAgICAgICAgIGlmICghaGVhZGNvdW50KSB7CiAgICAgICAg ICAgICAgICAgICAgICAgICBpZiAodW5saWtlbHkodmhvc3RfZW5hYmxlX25vdGlmeSgmbmV0LT5k ZXYsIHZxKSkpIHsKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgLyogVGhleSBoYXZl IHNsaXBwZWQgb25lIGluIGFzIHdlIHdlcmUKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICogZG9pbmcgdGhhdDogY2hlY2sgYWdhaW4uICovCnZob3N0X2Rpc2FibGVfbm90aWZ5KCZu ZXQtPmRldiwgdnEpOwogICAgICAgICAgICAgICAgIGNvbnRpbnVlOwogICAgICAgICAgICAgICAg ICAgICAgICAgfQogICAgICAgICAgICAgICAgICAgICAgICAgLyogTm90aGluZyBuZXc/ICBXYWl0 IGZvciBldmVudGZkIHRvIHRlbGwgdXMKICAgICAgICAgICAgICAgICAgICAgICAgICAqIHRoZXkg cmVmaWxsZWQuICovCiAgICAgICAgICAgICAgICAgICAgICAgICBnb3RvIG91dDsKICAgICAgICAg ICAgICAgICB9CgoKPgo+Cj4+ICAgb3V0Ogo+PiAgIAltdXRleF91bmxvY2soJnZxLT5tdXRleCk7 Cj4+ICAgfQo+PiBAQCAtNzk2LDMyICs4MjQsNiBAQCBzdGF0aWMgaW50IHZob3N0X25ldF9vcGVu KHN0cnVjdCBpbm9kZSAqaW5vZGUsIHN0cnVjdCBmaWxlICpmKQo+PiAgIAlyZXR1cm4gMDsKPj4g ICB9Cj4+ICAgCj4+IC1zdGF0aWMgdm9pZCB2aG9zdF9uZXRfZGlzYWJsZV92cShzdHJ1Y3Qgdmhv c3RfbmV0ICpuLAo+PiAtCQkJCSBzdHJ1Y3Qgdmhvc3RfdmlydHF1ZXVlICp2cSkKPj4gLXsKPj4g LQlzdHJ1Y3Qgdmhvc3RfbmV0X3ZpcnRxdWV1ZSAqbnZxID0KPj4gLQkJY29udGFpbmVyX29mKHZx LCBzdHJ1Y3Qgdmhvc3RfbmV0X3ZpcnRxdWV1ZSwgdnEpOwo+PiAtCXN0cnVjdCB2aG9zdF9wb2xs ICpwb2xsID0gbi0+cG9sbCArIChudnEgLSBuLT52cXMpOwo+PiAtCWlmICghdnEtPnByaXZhdGVf ZGF0YSkKPj4gLQkJcmV0dXJuOwo+PiAtCXZob3N0X3BvbGxfc3RvcChwb2xsKTsKPj4gLX0KPj4g LQo+PiAtc3RhdGljIGludCB2aG9zdF9uZXRfZW5hYmxlX3ZxKHN0cnVjdCB2aG9zdF9uZXQgKm4s Cj4+IC0JCQkJc3RydWN0IHZob3N0X3ZpcnRxdWV1ZSAqdnEpCj4+IC17Cj4+IC0Jc3RydWN0IHZo b3N0X25ldF92aXJ0cXVldWUgKm52cSA9Cj4+IC0JCWNvbnRhaW5lcl9vZih2cSwgc3RydWN0IHZo b3N0X25ldF92aXJ0cXVldWUsIHZxKTsKPj4gLQlzdHJ1Y3Qgdmhvc3RfcG9sbCAqcG9sbCA9IG4t PnBvbGwgKyAobnZxIC0gbi0+dnFzKTsKPj4gLQlzdHJ1Y3Qgc29ja2V0ICpzb2NrOwo+PiAtCj4+ IC0Jc29jayA9IHZxLT5wcml2YXRlX2RhdGE7Cj4+IC0JaWYgKCFzb2NrKQo+PiAtCQlyZXR1cm4g MDsKPj4gLQo+PiAtCXJldHVybiB2aG9zdF9wb2xsX3N0YXJ0KHBvbGwsIHNvY2stPmZpbGUpOwo+ PiAtfQo+PiAtCj4+ICAgc3RhdGljIHN0cnVjdCBzb2NrZXQgKnZob3N0X25ldF9zdG9wX3ZxKHN0 cnVjdCB2aG9zdF9uZXQgKm4sCj4+ICAgCQkJCQlzdHJ1Y3Qgdmhvc3RfdmlydHF1ZXVlICp2cSkK Pj4gICB7Cj4+IC0tIAo+PiAxLjguMy4xCj4gLS0KPiBUbyB1bnN1YnNjcmliZSBmcm9tIHRoaXMg bGlzdDogc2VuZCB0aGUgbGluZSAidW5zdWJzY3JpYmUga3ZtIiBpbgo+IHRoZSBib2R5IG9mIGEg bWVzc2FnZSB0byBtYWpvcmRvbW9Admdlci5rZXJuZWwub3JnCj4gTW9yZSBtYWpvcmRvbW8gaW5m byBhdCAgaHR0cDovL3ZnZXIua2VybmVsLm9yZy9tYWpvcmRvbW8taW5mby5odG1sCgpfX19fX19f X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fXwpWaXJ0dWFsaXphdGlvbiBt YWlsaW5nIGxpc3QKVmlydHVhbGl6YXRpb25AbGlzdHMubGludXgtZm91bmRhdGlvbi5vcmcKaHR0 cHM6Ly9saXN0cy5saW51eGZvdW5kYXRpb24ub3JnL21haWxtYW4vbGlzdGluZm8vdmlydHVhbGl6 YXRpb24= From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161940AbcEaDOW (ORCPT ); Mon, 30 May 2016 23:14:22 -0400 Received: from mx1.redhat.com ([209.132.183.28]:35104 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932469AbcEaDOU (ORCPT ); Mon, 30 May 2016 23:14:20 -0400 Subject: Re: [PATCH V2 1/2] vhost_net: stop polling socket during rx processing To: "Michael S. Tsirkin" References: <1464590874-39539-1-git-send-email-jasowang@redhat.com> <1464590874-39539-2-git-send-email-jasowang@redhat.com> <20160530184211-mutt-send-email-mst@redhat.com> Cc: kvm@vger.kernel.org, virtualization@lists.linux-foundation.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org From: Jason Wang Message-ID: <574D0182.9020701@redhat.com> Date: Tue, 31 May 2016 11:14:10 +0800 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.8.0 MIME-Version: 1.0 In-Reply-To: <20160530184211-mutt-send-email-mst@redhat.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.31]); Tue, 31 May 2016 03:14:19 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2016年05月30日 23:47, Michael S. Tsirkin wrote: > On Mon, May 30, 2016 at 02:47:53AM -0400, Jason Wang wrote: >> We don't stop rx polling socket during rx processing, this will lead >> unnecessary wakeups from under layer net devices (E.g >> sock_def_readable() form tun). Rx will be slowed down in this >> way. This patch avoids this by stop polling socket during rx >> processing. A small drawback is that this introduces some overheads in >> light load case because of the extra start/stop polling, but single >> netperf TCP_RR does not notice any change. In a super heavy load case, >> e.g using pktgen to inject packet to guest, we get about ~8.8% >> improvement on pps: >> >> before: ~1240000 pkt/s >> after: ~1350000 pkt/s >> >> Signed-off-by: Jason Wang >> --- >> drivers/vhost/net.c | 56 +++++++++++++++++++++++++++-------------------------- >> 1 file changed, 29 insertions(+), 27 deletions(-) >> >> diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c >> index 10ff494..e91603b 100644 >> --- a/drivers/vhost/net.c >> +++ b/drivers/vhost/net.c >> @@ -301,6 +301,32 @@ static bool vhost_can_busy_poll(struct vhost_dev *dev, >> !vhost_has_work(dev); >> } >> >> +static void vhost_net_disable_vq(struct vhost_net *n, >> + struct vhost_virtqueue *vq) >> +{ >> + struct vhost_net_virtqueue *nvq = >> + container_of(vq, struct vhost_net_virtqueue, vq); >> + struct vhost_poll *poll = n->poll + (nvq - n->vqs); >> + if (!vq->private_data) >> + return; >> + vhost_poll_stop(poll); >> +} >> + >> +static int vhost_net_enable_vq(struct vhost_net *n, >> + struct vhost_virtqueue *vq) >> +{ >> + struct vhost_net_virtqueue *nvq = >> + container_of(vq, struct vhost_net_virtqueue, vq); >> + struct vhost_poll *poll = n->poll + (nvq - n->vqs); >> + struct socket *sock; >> + >> + sock = vq->private_data; >> + if (!sock) >> + return 0; >> + >> + return vhost_poll_start(poll, sock->file); >> +} >> + >> static int vhost_net_tx_get_vq_desc(struct vhost_net *net, >> struct vhost_virtqueue *vq, >> struct iovec iov[], unsigned int iov_size, > BTW we might want to rename these functions, name no longer > reflects function ... Do you mean adding something reflect busy polling in the name? Then the name may be too long or have suggestion on the name? > > >> @@ -627,6 +653,7 @@ static void handle_rx(struct vhost_net *net) >> if (!sock) >> goto out; >> vhost_disable_notify(&net->dev, vq); >> + vhost_net_disable_vq(net, vq); >> >> vhost_hlen = nvq->vhost_hlen; >> sock_hlen = nvq->sock_hlen; >> @@ -715,9 +742,10 @@ static void handle_rx(struct vhost_net *net) >> total_len += vhost_len; >> if (unlikely(total_len >= VHOST_NET_WEIGHT)) { >> vhost_poll_queue(&vq->poll); >> - break; >> + goto out; >> } >> } >> + vhost_net_enable_vq(net, vq); > OK so if sock is readable but RX VQ is empty, this will > immediately schedule another round of handle_rx and so ad > infinitum, > > Looks like a bug. Yes it is, will change the above headcount check to: /* OK, now we need to know about added descriptors. */ if (!headcount) { if (unlikely(vhost_enable_notify(&net->dev, vq))) { /* They have slipped one in as we were * doing that: check again. */ vhost_disable_notify(&net->dev, vq); continue; } /* Nothing new? Wait for eventfd to tell us * they refilled. */ goto out; } > > >> out: >> mutex_unlock(&vq->mutex); >> } >> @@ -796,32 +824,6 @@ static int vhost_net_open(struct inode *inode, struct file *f) >> return 0; >> } >> >> -static void vhost_net_disable_vq(struct vhost_net *n, >> - struct vhost_virtqueue *vq) >> -{ >> - struct vhost_net_virtqueue *nvq = >> - container_of(vq, struct vhost_net_virtqueue, vq); >> - struct vhost_poll *poll = n->poll + (nvq - n->vqs); >> - if (!vq->private_data) >> - return; >> - vhost_poll_stop(poll); >> -} >> - >> -static int vhost_net_enable_vq(struct vhost_net *n, >> - struct vhost_virtqueue *vq) >> -{ >> - struct vhost_net_virtqueue *nvq = >> - container_of(vq, struct vhost_net_virtqueue, vq); >> - struct vhost_poll *poll = n->poll + (nvq - n->vqs); >> - struct socket *sock; >> - >> - sock = vq->private_data; >> - if (!sock) >> - return 0; >> - >> - return vhost_poll_start(poll, sock->file); >> -} >> - >> static struct socket *vhost_net_stop_vq(struct vhost_net *n, >> struct vhost_virtqueue *vq) >> { >> -- >> 1.8.3.1 > -- > To unsubscribe from this list: send the line "unsubscribe kvm" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html