netdev - Re: [PATCH, net-next, v3, 0/2] net/ethtool: Introduce link_ksettings API for virtual network devices

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20191219195818.62501eea@hermes.lan>
Date:   Thu, 19 Dec 2019 19:58:18 -0800
From:   Stephen Hemminger <stephen@...workplumber.org>
To:     David Miller <davem@...emloft.net>
Cc:     cforno12@...ux.vnet.ibm.com, netdev@...r.kernel.org,
        mst@...hat.com, jasowang@...hat.com, haiyangz@...rosoft.com,
        sthemmin@...rosoft.com, sashal@...nel.org, tlfalcon@...ux.ibm.com
Subject: Re: [PATCH, net-next, v3, 0/2] net/ethtool: Introduce
 link_ksettings API for virtual network devices

On Thu, 19 Dec 2019 14:16:19 -0800 (PST)
David Miller <davem@...emloft.net> wrote:

> From: Stephen Hemminger <stephen@...workplumber.org>
> Date: Thu, 19 Dec 2019 13:11:56 -0800
> 
> > I don't think this makes sense for netvsc. The speed and duplex have no
> > meaning, why do you want to allow overriding it? If this is to try and make
> > some dashboard look good; then you aren't seeing the real speed which is
> > what only the host knows. Plus it does take into account the accelerated
> > networking path.  
> 
> Maybe that's the point, userspace has extraneous knowledge it might
> use to set it accurately.
> 
> This helps for bonding/team etc. as well.
> 
> I don't think there is any real harm in allowing to set this, and
> we've done this in the past I think.

The most widely used case with netvsc is using VF to get accelerated networking
in that case real speed and duplex value is reported by the VF device.

Maybe something like the following (COMPLETELY UNTESTED):


diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index eff8fef4f775..111847ca7e8c 100644
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -1083,11 +1083,14 @@ static int netvsc_get_link_ksettings(struct net_device *dev,
 				     struct ethtool_link_ksettings *cmd)
 {
 	struct net_device_context *ndc = netdev_priv(dev);
+	struct net_device *vf_netdev = rtnl_dereference(ndc->vf_netdev);
+
+	if (vf_netdev)
+		return __ethtool_get_link_ksettings(vf_netdev, cmd);
 
 	cmd->base.speed = ndc->speed;
 	cmd->base.duplex = ndc->duplex;
 	cmd->base.port = PORT_OTHER;
-
 	return 0;
 }
 
@@ -1095,8 +1098,17 @@ static int netvsc_set_link_ksettings(struct net_device *dev,
 				     const struct ethtool_link_ksettings *cmd)
 {
 	struct net_device_context *ndc = netdev_priv(dev);
+	struct net_device *vf_netdev = rtnl_dereference(ndc->vf_netdev);
 	u32 speed;
 
+	if (vf_netdev) {
+		if (!vf_netdev->ethtool_ops->set_link_ksettings)
+			return -EOPNOTSUPP;
+
+		return vf_netdev->ethtool_ops->set_link_ksettings(vf_netdev,
+								  cmd);
+	}
+
 	speed = cmd->base.speed;
 	if (!ethtool_validate_speed(speed) ||
 	    !ethtool_validate_duplex(cmd->base.duplex) ||
-- 
2.20.1