From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:35762) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bAYKR-000394-45 for qemu-devel@nongnu.org; Wed, 08 Jun 2016 03:55:08 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bAYKM-0005r7-4x for qemu-devel@nongnu.org; Wed, 08 Jun 2016 03:55:06 -0400 References: <1465304408-16545-1-git-send-email-pl@kamp.de> <5757BDA0.8090504@redhat.com> <5757C593.3060209@kamp.de> From: Jason Wang Message-ID: <5757CF4D.9080806@redhat.com> Date: Wed, 8 Jun 2016 15:54:53 +0800 MIME-Version: 1.0 In-Reply-To: <5757C593.3060209@kamp.de> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [Qemu-stable] [PATCH V2] net: fix qemu_announce_self not emitting packets List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Lieven , qemu-devel@nongnu.org Cc: pbonzini@redhat.com, qemu-stable@nongnu.org, hongyang.yang@easystack.cn On 2016=E5=B9=B406=E6=9C=8808=E6=97=A5 15:13, Peter Lieven wrote: > Am 08.06.2016 um 08:39 schrieb Jason Wang: >> >> On 2016=E5=B9=B406=E6=9C=8807=E6=97=A5 21:00, Peter Lieven wrote: >>> commit fefe2a78 accidently dropped the code path for injecting >>> raw packets. This feature is needed for sending gratuitous ARPs >>> after an incoming migration has completed. The result is increased >>> network downtime for vservers where the network card is not virtio-ne= t >>> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature. >>> >>> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242 >>> Cc: qemu-stable@nongnu.org >>> Cc: hongyang.yang@easystack.cn >>> Signed-off-by: Peter Lieven >>> --- >>> v1->v2: assert that only raw packets with a plain buffer come in. [Pa= olo] >>> >>> net/net.c | 10 +++++++++- >>> 1 file changed, 9 insertions(+), 1 deletion(-) >>> >>> diff --git a/net/net.c b/net/net.c >>> index 5f3e5a9..5e1b5fa 100644 >>> --- a/net/net.c >>> +++ b/net/net.c >>> @@ -714,6 +714,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *= sender, >>> NetClientState *nc =3D opaque; >>> int ret; >>> + /* we currently only support sending raw packets via qemu_sen= d_packet_raw. >>> + * If we want generic raw iov support we need to implement somet= hing like >>> + * .receive_raw_iov in NetClientInfo first. */ >>> + assert(!(flags & QEMU_NET_PACKET_FLAG_RAW) || iovcnt =3D=3D 1); >>> + >>> if (nc->link_down) { >>> return iov_size(iov, iovcnt); >>> } >>> @@ -722,7 +727,10 @@ ssize_t qemu_deliver_packet_iov(NetClientState *= sender, >>> return 0; >>> } >>> - if (nc->info->receive_iov) { >>> + if (flags & QEMU_NET_PACKET_FLAG_RAW && nc->info->receive_raw) { >>> + /* this is required for qemu_announce_self() */ >>> + ret =3D nc->info->receive_raw(nc, iov[0].iov_base, iov[0].io= v_len); >>> + } else if (nc->info->receive_iov) { >>> ret =3D nc->info->receive_iov(nc, iov, iovcnt); >>> } else { >>> ret =3D nc_sendv_compat(nc, iov, iovcnt, flags); >> We still have raw packet support in nv_sendv_compat, can we use it? (a= nd then there's no need for the assert above). > Okay, so Version 1 of the patch is more appropiate. I would either use = V1 or revert fefe2a78. The issue is quite serious > for live migration. > > Peter > > V1 checks iovcnt and call receive_raw with hard coded first iov. Since=20 we have raw packet support in nc_sendv_compat(), I mean when we detect=20 QEMU_NET_PACKET_FLAG_RAW, call nc_sendv_compat() here.