[PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[longli]

From: Long Li <longli@microsoft.com>

When a VF is being exposed form the kernel, it should be marked as "slave"
before exposing to the user-mode. The VF is not usable without netvsc running
as master. The user-mode should never see a VF without the "slave" flag.

This commit moves the code of setting the slave flag to the time before VF is
exposed to user-mode.

Signed-off-by: Long Li <longli@microsoft.com>
---
 drivers/net/hyperv/netvsc_drv.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index ec77fb9dcf89..045777a4971d 100644
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -2206,9 +2206,6 @@ static int netvsc_vf_join(struct net_device *vf_netdev,
 		goto upper_link_failed;
 	}
 
-	/* set slave flag before open to prevent IPv6 addrconf */
-	vf_netdev->flags |= IFF_SLAVE;
-
 	schedule_delayed_work(&ndev_ctx->vf_takeover, VF_TAKEOVER_INT);
 
 	call_netdevice_notifiers(NETDEV_JOIN, vf_netdev);
@@ -2320,11 +2317,9 @@ static struct net_device *get_netvsc_byslot(const struct net_device *vf_netdev)
 	 */
 	list_for_each_entry(ndev_ctx, &netvsc_dev_list, list) {
 		ndev = hv_get_drvdata(ndev_ctx->device_ctx);
-		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr)) {
-			netdev_notice(vf_netdev,
-				      "falling back to mac addr based matching\n");
+		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr) ||
+		    ether_addr_equal(vf_netdev->dev_addr, ndev->perm_addr))
 			return ndev;
-		}
 	}
 
 	netdev_notice(vf_netdev,
@@ -2332,7 +2327,7 @@ static struct net_device *get_netvsc_byslot(const struct net_device *vf_netdev)
 	return NULL;
 }
 
-static int netvsc_register_vf(struct net_device *vf_netdev)
+static int netvsc_register_vf(struct net_device *vf_netdev, unsigned long event)
 {
 	struct net_device_context *net_device_ctx;
 	struct netvsc_device *netvsc_dev;
@@ -2347,6 +2342,12 @@ static int netvsc_register_vf(struct net_device *vf_netdev)
 	if (!ndev)
 		return NOTIFY_DONE;
 
+	if (event == NETDEV_POST_INIT) {
+		/* set slave flag before open to prevent IPv6 addrconf */
+		vf_netdev->flags |= IFF_SLAVE;
+		return NOTIFY_DONE;
+	}
+
 	net_device_ctx = netdev_priv(ndev);
 	netvsc_dev = rtnl_dereference(net_device_ctx->nvdev);
 	if (!netvsc_dev || rtnl_dereference(net_device_ctx->vf_netdev))
@@ -2753,8 +2754,9 @@ static int netvsc_netdev_event(struct notifier_block *this,
 		return NOTIFY_DONE;
 
 	switch (event) {
+	case NETDEV_POST_INIT:
 	case NETDEV_REGISTER:
-		return netvsc_register_vf(event_dev);
+		return netvsc_register_vf(event_dev, event);
 	case NETDEV_UNREGISTER:
 		return netvsc_unregister_vf(event_dev);
 	case NETDEV_UP:
-- 
2.34.1

Re: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[Stephen Hemminger]

On Wed, 25 Oct 2023 15:50:50 -0700
longli@linuxonhyperv.com wrote:

> 	list_for_each_entry(ndev_ctx, &netvsc_dev_list, list) {
>  		ndev = hv_get_drvdata(ndev_ctx->device_ctx);
> -		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr)) {
> -			netdev_notice(vf_netdev,
> -				      "falling back to mac addr based matching\n");
> +		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr) ||
> +		    ether_addr_equal(vf_netdev->dev_addr, ndev->perm_addr))

This part looks like unrelated change.
The VF mac address shouldn't change, but if it did don't look at i.

Re: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[Stephen Hemminger]

On Wed, 25 Oct 2023 15:50:50 -0700
longli@linuxonhyperv.com wrote:

> @@ -2347,6 +2342,12 @@ static int netvsc_register_vf(struct net_device *vf_netdev)
>  	if (!ndev)
>  		return NOTIFY_DONE;
>  
> +	if (event == NETDEV_POST_INIT) {
> +		/* set slave flag before open to prevent IPv6 addrconf */
> +		vf_netdev->flags |= IFF_SLAVE;
> +		return NOTIFY_DONE;
> +	}
> +
>  	net_device_ctx = netdev_priv(ndev);
>  	netvsc_dev = rtnl_dereference(net_device_ctx->nvdev);
>  	if (!netvsc_dev || rtnl_dereference(net_device_ctx->vf_netdev))
> @@ -2753,8 +2754,9 @@ static int netvsc_netdev_event(struct notifier_block *this,
>  		return NOTIFY_DONE;
>  
>  	switch (event) {
> +	case NETDEV_POST_INIT:
>  	case NETDEV_REGISTER:
> -		return netvsc_register_vf(event_dev);
> +		return netvsc_register_vf(event_dev, event);

Although correct, this is an awkward way to write this.
There are two events which call register_vf() but the post init
one short circuits and doesn't really register the VF.

The code is clearer if flag is set in switch statement.

@@ -2206,9 +2206,6 @@ static int netvsc_vf_join(struct net_device *vf_netdev,
 		goto upper_link_failed;
 	}
 
-	/* set slave flag before open to prevent IPv6 addrconf */
-	vf_netdev->flags |= IFF_SLAVE;
-
 	schedule_delayed_work(&ndev_ctx->vf_takeover, VF_TAKEOVER_INT);
 
 	call_netdevice_notifiers(NETDEV_JOIN, vf_netdev);
@@ -2753,6 +2750,10 @@ static int netvsc_netdev_event(struct notifier_block *this,
 		return NOTIFY_DONE;
 
 	switch (event) {
+	case NETDEV_POST_INIT:
+		/* set slave flag before open to prevent IPv6 addrconf */
+		vf_netdev->flags |= IFF_SLAVE;
+		return NOTIFY_DONE;
 	case NETDEV_REGISTER:
 		return netvsc_register_vf(event_dev);
 	case NETDEV_UNREGISTER:

RE: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[Long Li]

> Subject: Re: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode
> 
> On Wed, 25 Oct 2023 15:50:50 -0700
> longli@linuxonhyperv.com wrote:
> 
> > 	list_for_each_entry(ndev_ctx, &netvsc_dev_list, list) {
> >  		ndev = hv_get_drvdata(ndev_ctx->device_ctx);
> > -		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr))
> {
> > -			netdev_notice(vf_netdev,
> > -				      "falling back to mac addr based
> matching\n");
> > +		if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr)
> ||
> > +		    ether_addr_equal(vf_netdev->dev_addr, ndev->perm_addr))
> 
> This part looks like unrelated change.
> The VF mac address shouldn't change, but if it did don't look at i.

At the time of NETDEV_POST_INIT, the perm_addr is not assigned yet (last it was copied from dev_addr when device is registered).

RE: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode

[Long Li]

> Subject: Re: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode
> 
> On Wed, 25 Oct 2023 15:50:50 -0700
> longli@linuxonhyperv.com wrote:
> 
> > @@ -2347,6 +2342,12 @@ static int netvsc_register_vf(struct net_device
> *vf_netdev)
> >  	if (!ndev)
> >  		return NOTIFY_DONE;
> >
> > +	if (event == NETDEV_POST_INIT) {
> > +		/* set slave flag before open to prevent IPv6 addrconf */
> > +		vf_netdev->flags |= IFF_SLAVE;
> > +		return NOTIFY_DONE;
> > +	}
> > +
> >  	net_device_ctx = netdev_priv(ndev);
> >  	netvsc_dev = rtnl_dereference(net_device_ctx->nvdev);
> >  	if (!netvsc_dev || rtnl_dereference(net_device_ctx->vf_netdev))
> > @@ -2753,8 +2754,9 @@ static int netvsc_netdev_event(struct notifier_block
> *this,
> >  		return NOTIFY_DONE;
> >
> >  	switch (event) {
> > +	case NETDEV_POST_INIT:
> >  	case NETDEV_REGISTER:
> > -		return netvsc_register_vf(event_dev);
> > +		return netvsc_register_vf(event_dev, event);
> 
> Although correct, this is an awkward way to write this.
> There are two events which call register_vf() but the post init one short circuits
> and doesn't really register the VF.
> 
> The code is clearer if flag is set in switch statement.
> 

I will add a dedicated function to handle NETDEV_POST_INIT.

Thanks,
Long