From mboxrd@z Thu Jan 1 00:00:00 1970 From: Amos Kong Subject: Re: [PATCH] virtio_net: zero is an invald queue_pairs number Date: Mon, 21 Apr 2014 23:03:46 +0800 Message-ID: <20140421150346.GA2500@amosk.info> References: <1397799941-11371-1-git-send-email-akong@redhat.com> <5354BB94.4030908@redhat.com> <5354BE85.5030604@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, mst@redhat.com, virtualization@lists.linux-foundation.org To: Jason Wang Return-path: Content-Disposition: inline In-Reply-To: <5354BE85.5030604@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 List-Id: netdev.vger.kernel.org On Mon, Apr 21, 2014 at 02:45:25PM +0800, Jason Wang wrote: > On 04/21/2014 02:32 PM, Jason Wang wrote: > > On 04/18/2014 01:45 PM, Amos Kong wrote: > >> Execute "ethtool -L eth0 combined 0" in guest, if multiqueue > >> is enabled, virtnet_send_command() will return -EINVAL error, > >> there is a validation in QEMU. > >> > >> But if multiqueue is disabled, virtnet_set_queues() will just > >> return zero (success). We should return error for this situation. > >> > >> Signed-off-by: Amos Kong > >> --- > >> drivers/net/virtio_net.c | 2 +- > >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> > >> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c > >> index 7b68746..8a852b5 100644 > >> --- a/drivers/net/virtio_net.c > >> +++ b/drivers/net/virtio_net.c > >> @@ -1285,7 +1285,7 @@ static int virtnet_set_channels(struct net_device *dev, > >> if (channels->rx_count || channels->tx_count || channels->other_count) > >> return -EINVAL; > >> > >> - if (queue_pairs > vi->max_queue_pairs) > >> + if (queue_pairs > vi->max_queue_pairs || queue_pairs == 0) > >> return -EINVAL; > >> > >> get_online_cpus(); > > Acked-by: Jason Wang > > Looking at virtnet_set_queues(), it will return success even if no cvq > and no mq support. This is wrong. We need return error in this case. > We'd better fix it and check the its return value in virtnet_restore(). > > With this fix, there's no need for us to check queue_pairs against zero > like what this patch does. As we talked in IRC, the problem I tried to fix is actually caused by that we don't return error when MQ isn't support. I will post another patch to fix it. However, this fix is also useful, it will catch the error early. -- Amos.