Re: [PATCH net] net: qualcomm: rmnet: set skb->dev for deaggregated MAP commands

On Thu, Aug 20, 2026 at 2:49 AM Eric Dumazet <[email protected]> wrote:
>
> On Thu, Aug 20, 2026 at 10:11 AM Xiang Mei <[email protected]> wrote:
> >
> > rmnet_map_deaggregate() allocates each sub-frame with alloc_skb() and
> > leaves skb->dev NULL. __rmnet_map_ingress_handler() assigns
> > skb->dev = ep->egress_dev only on the data path, but a MAP command frame
> > is dispatched to rmnet_map_command() before that, so rmnet_map_send_ack()
> > runs netif_tx_lock(skb->dev) on a NULL device. An unprivileged user
> > reaches this by unsharing a user+net namespace, creating an rmnet link
> > over a tap device with INGRESS_DEAGGREGATION and INGRESS_MAP_COMMANDS,
> > and writing an aggregated frame carrying a flow-control command to the
> > tap fd.
> >
> > Assign port->dev before dispatch: the command was received on the real
> > device tracked by port->dev and the ACK must egress on that same device.
> > The non-deaggregated path already has skb->dev == real_dev.
> >
> >   BUG: KASAN: null-ptr-deref in _raw_spin_lock (kernel/locking/spinlock.c:158)
> >   Write of size 4 at addr 00000000000004b4 by task exploit/144
> >   Call Trace:
> >    _raw_spin_lock (kernel/locking/spinlock.c:158)
> >    netif_tx_lock (net/sched/sch_generic.c:497)
> >    rmnet_map_command (drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c:67)
> >    rmnet_rx_handler (drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c:125)
> >    __netif_receive_skb_core.constprop.0 (net/core/dev.c:6103)
> >    ...
> >    __netif_receive_skb_one_core (net/core/dev.c:6214)
> >    netif_receive_skb (net/core/dev.c:6474)
> >    tun_get_user (drivers/net/tun.c:1966)
> >    tun_chr_write_iter (drivers/net/tun.c:2012)
> >    vfs_write (fs/read_write.c:687)
> >    ksys_write (fs/read_write.c:739)
> >    do_syscall_64 (arch/x86/entry/syscall_64.c:94)
> >    entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
> >   Kernel panic - not syncing: Fatal exception in interrupt
> >
> > Fixes: 378e25357ac7 ("net: qualcomm: rmnet: Remove unnecessary device assignment")
> > Reported-by: [email protected]
> > Closes: https://lore.kernel.org/netdev/[email protected]/T/#u
> > Assisted-by: Claude:claude-opus-4-8
> > Signed-off-by: Xiang Mei <[email protected]>
> > ---
> >  drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c b/drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c
> > index add0f5ade2e6..3944925566fc 100644
> > --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c
> > +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_map_command.c
> > @@ -79,6 +79,8 @@ void rmnet_map_command(struct sk_buff *skb, struct rmnet_port *port)
> >         unsigned char command_name;
> >         unsigned char rc = 0;
> >
> > +       skb->dev = port->dev;
> > +
>
> This works, but the suggested patch is cleaner IMO
>
Thanks for your review!
> https://bugs.sh/b/4638111fe2a12980/patch.diff
Agreed, that is the better fix (fix the problem at producer end). v2
has been sent:
https://lore.kernel.org/netdev/[email protected]/T/#u

-Xiang