| Message ID | 20200214133404.30643-4-Jason@zx2c4.com |
|---|---|
| State | Changes Requested |
| Delegated to: | David Miller |
| Headers | show |
| Series | wireguard fixes for 5.6-rc2 | expand |
On 2/14/20 5:34 AM, Jason A. Donenfeld wrote: > It turns out there's an easy way to get packets queued up while still > having an MTU of zero, and that's via persistent keep alive. This commit > makes sure that in whatever condition, we don't wind up dividing by > zero. Note that an MTU of zero for a wireguard interface is something > quasi-valid, so I don't think the correct fix is to limit it via > min_mtu. This can be reproduced easily with: > > ip link add wg0 type wireguard > ip link add wg1 type wireguard > ip link set wg0 up mtu 0 > ip link set wg1 up > wg set wg0 private-key <(wg genkey) > wg set wg1 listen-port 1 private-key <(wg genkey) peer $(wg show wg0 public-key) > wg set wg0 peer $(wg show wg1 public-key) persistent-keepalive 1 endpoint 127.0.0.1:1 > > However, while min_mtu=0 seems fine, it makes sense to restrict the > max_mtu. This commit also restricts the maximum MTU to the greatest > number for which rounding up to the padding multiple won't overflow a > signed integer. Packets this large were always rejected anyway > eventually, due to checks deeper in, but it seems more sound not to even > let the administrator configure something that won't work anyway. > If mtu is set to 0, the device must not send any payload. > > diff --git a/drivers/net/wireguard/send.c b/drivers/net/wireguard/send.c > index c13260563446..ae77474dadeb 100644 > --- a/drivers/net/wireguard/send.c > +++ b/drivers/net/wireguard/send.c > @@ -148,7 +148,8 @@ static unsigned int calculate_skb_padding(struct sk_buff *skb) > * wouldn't want the final subtraction to overflow in the case of the > * padded_size being clamped. > */ > - unsigned int last_unit = skb->len % PACKET_CB(skb)->mtu; > + unsigned int last_unit = PACKET_CB(skb)->mtu ? > + skb->len % PACKET_CB(skb)->mtu : skb->len; > unsigned int padded_size = ALIGN(last_unit, MESSAGE_PADDING_MULTIPLE); > > if (padded_size > PACKET_CB(skb)->mtu) > Are you sure this works ? Last statement : return padded_size - last_unit; will return a a ' negative number'
On Fri, Feb 14, 2020 at 4:18 PM Eric Dumazet <eric.dumazet@gmail.com> wrote: > > However, while min_mtu=0 seems fine, it makes sense to restrict the > > max_mtu. This commit also restricts the maximum MTU to the greatest > > number for which rounding up to the padding multiple won't overflow a > > signed integer. Packets this large were always rejected anyway > > eventually, due to checks deeper in, but it seems more sound not to even > > let the administrator configure something that won't work anyway. > > > If mtu is set to 0, the device must not send any payload. Yes, but there's still internal keepalive messages. > > Are you sure this works ? > > Last statement : > > return padded_size - last_unit; > > will return a a ' negative number' Woah nelly I don't know how I missed this. Thanks! Will send a v2 of this patchset.
diff --git a/drivers/net/wireguard/device.c b/drivers/net/wireguard/device.c index 43db442b1373..cdc96968b0f4 100644 --- a/drivers/net/wireguard/device.c +++ b/drivers/net/wireguard/device.c @@ -258,6 +258,8 @@ static void wg_setup(struct net_device *dev) enum { WG_NETDEV_FEATURES = NETIF_F_HW_CSUM | NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_GSO | NETIF_F_GSO_SOFTWARE | NETIF_F_HIGHDMA }; + const int overhead = MESSAGE_MINIMUM_LENGTH + sizeof(struct udphdr) + + max(sizeof(struct ipv6hdr), sizeof(struct iphdr)); dev->netdev_ops = &netdev_ops; dev->hard_header_len = 0; @@ -271,9 +273,8 @@ static void wg_setup(struct net_device *dev) dev->features |= WG_NETDEV_FEATURES; dev->hw_features |= WG_NETDEV_FEATURES; dev->hw_enc_features |= WG_NETDEV_FEATURES; - dev->mtu = ETH_DATA_LEN - MESSAGE_MINIMUM_LENGTH - - sizeof(struct udphdr) - - max(sizeof(struct ipv6hdr), sizeof(struct iphdr)); + dev->mtu = ETH_DATA_LEN - overhead; + dev->max_mtu = round_down(INT_MAX, MESSAGE_PADDING_MULTIPLE) - overhead; SET_NETDEV_DEVTYPE(dev, &device_type); diff --git a/drivers/net/wireguard/send.c b/drivers/net/wireguard/send.c index c13260563446..ae77474dadeb 100644 --- a/drivers/net/wireguard/send.c +++ b/drivers/net/wireguard/send.c @@ -148,7 +148,8 @@ static unsigned int calculate_skb_padding(struct sk_buff *skb) * wouldn't want the final subtraction to overflow in the case of the * padded_size being clamped. */ - unsigned int last_unit = skb->len % PACKET_CB(skb)->mtu; + unsigned int last_unit = PACKET_CB(skb)->mtu ? + skb->len % PACKET_CB(skb)->mtu : skb->len; unsigned int padded_size = ALIGN(last_unit, MESSAGE_PADDING_MULTIPLE); if (padded_size > PACKET_CB(skb)->mtu)
It turns out there's an easy way to get packets queued up while still having an MTU of zero, and that's via persistent keep alive. This commit makes sure that in whatever condition, we don't wind up dividing by zero. Note that an MTU of zero for a wireguard interface is something quasi-valid, so I don't think the correct fix is to limit it via min_mtu. This can be reproduced easily with: ip link add wg0 type wireguard ip link add wg1 type wireguard ip link set wg0 up mtu 0 ip link set wg1 up wg set wg0 private-key <(wg genkey) wg set wg1 listen-port 1 private-key <(wg genkey) peer $(wg show wg0 public-key) wg set wg0 peer $(wg show wg1 public-key) persistent-keepalive 1 endpoint 127.0.0.1:1 However, while min_mtu=0 seems fine, it makes sense to restrict the max_mtu. This commit also restricts the maximum MTU to the greatest number for which rounding up to the padding multiple won't overflow a signed integer. Packets this large were always rejected anyway eventually, due to checks deeper in, but it seems more sound not to even let the administrator configure something that won't work anyway. Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> --- drivers/net/wireguard/device.c | 7 ++++--- drivers/net/wireguard/send.c | 3 ++- 2 files changed, 6 insertions(+), 4 deletions(-)