Re: [PATCH net v3] net: introduce CAN specific pointer in the struct net_device

From: Oleksij Rempel
Date: Tue Feb 23 2021 - 01:53:21 EST


Hi Jakub,

On Mon, Feb 22, 2021 at 05:30:12PM -0800, Jakub Kicinski wrote:
> On Mon, 22 Feb 2021 16:02:51 +0100 Oleksij Rempel wrote:
> > Since 20dd3850bcf8 ("can: Speed up CAN frame receiption by using
> > ml_priv") the CAN framework uses per device specific data in the AF_CAN
> > protocol. For this purpose the struct net_device->ml_priv is used. Later
> > the ml_priv usage in CAN was extended for other users, one of them being
> > CAN_J1939.
> >
> > Later in the kernel ml_priv was converted to an union, used by other
> > drivers. E.g. the tun driver started storing it's stats pointer.
> >
> > Since tun devices can claim to be a CAN device, CAN specific protocols
> > will wrongly interpret this pointer, which will cause system crashes.
> > Mostly this issue is visible in the CAN_J1939 stack.
> >
> > To fix this issue, we request a dedicated CAN pointer within the
> > net_device struct.
> >
> > Reported-by: syzbot+5138c4dd15a0401bec7b@xxxxxxxxxxxxxxxxxxxxxxxxx
> > Fixes: 20dd3850bcf8 ("can: Speed up CAN frame receiption by using ml_priv")
> > Fixes: ffd956eef69b ("can: introduce CAN midlayer private and allocate it automatically")
> > Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol")
> > Fixes: 497a5757ce4e ("tun: switch to net core provided statistics counters")
> > Signed-off-by: Oleksij Rempel <o.rempel@xxxxxxxxxxxxxx>
>
> > diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> > index ddf4cfc12615..6e25c6f0f190 100644
> > --- a/include/linux/netdevice.h
> > +++ b/include/linux/netdevice.h
> > @@ -1584,6 +1584,16 @@ enum netdev_priv_flags {
> > #define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
> > #define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
> >
> > +/**
> > + * enum netdev_ml_priv_type - &struct net_device ml_priv_type
> > + *
> > + * This enum specifies the type of the struct net_device::ml_priv pointer.
> > + */
>
> kdoc (scripts/kernel-doc -none include/linux/netdevice.h) is not happy
> about the fact enum values are not defined. Perhaps they will be
> sufficiently self-explanatory to not bother documenting?
>
> Maybe just:
>
> /* Specifies the type of the struct net_device::ml_priv pointer */
>
> ?

sounds good, done.

> > +enum netdev_ml_priv_type {
> > + ML_PRIV_NONE,
> > + ML_PRIV_CAN,
> > +};
> > +
> > /**
> > * struct net_device - The DEVICE structure.
> > *
> > @@ -1779,6 +1789,7 @@ enum netdev_priv_flags {
> > * @nd_net: Network namespace this network device is inside
> > *
> > * @ml_priv: Mid-layer private
> > + @ml_priv_type: Mid-layer private type
>
> missing '*' at the start of the line

done

> > * @lstats: Loopback statistics
> > * @tstats: Tunnel statistics
> > * @dstats: Dummy statistics
> > @@ -2094,8 +2105,10 @@ struct net_device {
> > possible_net_t nd_net;
> >
> > /* mid-layer private */
> > + void *ml_priv;
> > + enum netdev_ml_priv_type ml_priv_type;
> > +
> > union {
> > - void *ml_priv;
> > struct pcpu_lstats __percpu *lstats;
> > struct pcpu_sw_netstats __percpu *tstats;
> > struct pcpu_dstats __percpu *dstats;
> > @@ -2286,6 +2299,29 @@ static inline void netdev_reset_rx_headroom(struct net_device *dev)
> > netdev_set_rx_headroom(dev, -1);
> > }
> >
> > +static inline void *netdev_get_ml_priv(struct net_device *dev,
> > + enum netdev_ml_priv_type type)
> > +{
> > + if (dev->ml_priv_type != type)
> > + return NULL;
> > +
> > + return dev->ml_priv;
> > +}
> > +
> > +static inline void netdev_set_ml_priv(struct net_device *dev,
> > + void *ml_priv,
> > + enum netdev_ml_priv_type type)
> > +{
> > + WARN_ONCE(dev->ml_priv_type && dev->ml_priv_type != type,
> > + "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
> > + dev->ml_priv_type, type);
> > + WARN_ONCE(!dev->ml_priv_type && dev->ml_priv,
> > + "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
>
> nit: do we need the _ONCE() this helper should be used on control path
> and relatively rarely, no?

I have no strong opinion right now. Changed to WARN()

> > + dev->ml_priv = ml_priv;
> > + dev->ml_priv_type = type;
> > +}
> > +
> > /*
> > * Net namespace inlines
> > */
>
> > @@ -454,6 +455,7 @@ static int j1939_sk_bind(struct socket *sock, struct sockaddr *uaddr, int len)
> > j1939_local_ecu_put(priv, jsk->addr.src_name, jsk->addr.sa);
> > } else {
> > struct net_device *ndev;
> > + struct can_ml_priv *can_ml;
>
> nit: rev xmas treei

done

>
> >
> > ndev = dev_get_by_index(net, addr->can_ifindex);
> > if (!ndev) {
>

Thank you,
Regards,
Oleksij
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |