Re: [net v3] drivers/net/wan/lapbether: Use needed_headroom instead of hard_header_len

From: Willem de Bruijn
Date: Mon Aug 03 2020 - 05:50:08 EST


On Sun, Aug 2, 2020 at 9:51 PM Xie He <xie.he.0141@xxxxxxxxx> wrote:
>
> In net/packet/af_packet.c, the function packet_snd first reserves a
> headroom of length (dev->hard_header_len + dev->needed_headroom).
> Then if the socket is a SOCK_DGRAM socket, it calls dev_hard_header,
> which calls dev->header_ops->create, to create the link layer header.
> If the socket is a SOCK_RAW socket, it "un-reserves" a headroom of
> length (dev->hard_header_len), and assumes the user to provide the
> appropriate link layer header.
>
> So according to the logic of af_packet.c, dev->hard_header_len should
> be the length of the header that would be created by
> dev->header_ops->create.
>
> However, this driver doesn't provide dev->header_ops, so logically
> dev->hard_header_len should be 0.
>
> So we should use dev->needed_headroom instead of dev->hard_header_len
> to request necessary headroom to be allocated.
>
> This change fixes kernel panic when this driver is used with AF_PACKET
> SOCK_RAW sockets. Call stack when panic:
>
> [ 168.399197] skbuff: skb_under_panic: text:ffffffff819d95fb len:20
> put:14 head:ffff8882704c0a00 data:ffff8882704c09fd tail:0x11 end:0xc0
> dev:veth0
> ...
> [ 168.399255] Call Trace:
> [ 168.399259] skb_push.cold+0x14/0x24
> [ 168.399262] eth_header+0x2b/0xc0
> [ 168.399267] lapbeth_data_transmit+0x9a/0xb0 [lapbether]
> [ 168.399275] lapb_data_transmit+0x22/0x2c [lapb]
> [ 168.399277] lapb_transmit_buffer+0x71/0xb0 [lapb]
> [ 168.399279] lapb_kick+0xe3/0x1c0 [lapb]
> [ 168.399281] lapb_data_request+0x76/0xc0 [lapb]
> [ 168.399283] lapbeth_xmit+0x56/0x90 [lapbether]
> [ 168.399286] dev_hard_start_xmit+0x91/0x1f0
> [ 168.399289] ? irq_init_percpu_irqstack+0xc0/0x100
> [ 168.399291] __dev_queue_xmit+0x721/0x8e0
> [ 168.399295] ? packet_parse_headers.isra.0+0xd2/0x110
> [ 168.399297] dev_queue_xmit+0x10/0x20
> [ 168.399298] packet_sendmsg+0xbf0/0x19b0
> ......
>
> Additional change:
> When sending, check skb->len to ensure the 1-byte pseudo header is
> present before reading it.
>
> Cc: Willem de Bruijn <willemdebruijn.kernel@xxxxxxxxx>
> Cc: Brian Norris <briannorris@xxxxxxxxxxxx>
> Signed-off-by: Xie He <xie.he.0141@xxxxxxxxx>

Overall, looks great. A few nits.

It's [PATCH net v3], not [net v3]

> diff --git a/drivers/net/wan/lapbether.c b/drivers/net/wan/lapbether.c
> index b2868433718f..8a3f7ba36f7e 100644
> --- a/drivers/net/wan/lapbether.c
> +++ b/drivers/net/wan/lapbether.c
> @@ -157,6 +157,9 @@ static netdev_tx_t lapbeth_xmit(struct sk_buff *skb,
> if (!netif_running(dev))
> goto drop;
>
> + if (skb->len < 1)
> + goto drop;
> +

Might be worth a comment along the lines of: /* upper layers pass a
control byte. must validate pf_packet input */

> switch (skb->data[0]) {
> case X25_IFACE_DATA:
> break;
> @@ -305,6 +308,7 @@ static void lapbeth_setup(struct net_device *dev)
> dev->netdev_ops = &lapbeth_netdev_ops;
> dev->needs_free_netdev = true;
> dev->type = ARPHRD_X25;
> + dev->hard_header_len = 0;

Technically not needed. The struct is allocated with kvzalloc, z for
__GFP_ZERO. Fine to leave if intended as self-describing comment, of
course.

> dev->mtu = 1000;
> dev->addr_len = 0;
> }
> @@ -331,7 +335,8 @@ static int lapbeth_new_device(struct net_device *dev)
> * then this driver prepends a length field of 2 bytes,
> * then the underlying Ethernet device prepends its own header.
> */
> - ndev->hard_header_len = -1 + 3 + 2 + dev->hard_header_len;
> + ndev->needed_headroom = -1 + 3 + 2 + dev->hard_header_len
> + + dev->needed_headroom;
>
> lapbeth = netdev_priv(ndev);
> lapbeth->axdev = ndev;
> --
> 2.25.1
>