In the Linux kernel, the following vulnerability has been resolved:
netlink: Fix wraparounds of sk->sk_rmem_alloc.
Netlink has this pattern in some places
if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf) atomic_add(skb->truesize, &sk->sk_rmem_alloc);
, which has the same problem fixed by commit 5a465a0da13e ("udp: Fix multiple wraparounds of sk->sk_rmem_alloc.").
For example, if we set INT_MAX to SO_RCVBUFFORCE, the condition is always false as the two operands are of int.
Then, a single socket can eat as many skb as possible until OOM happens, and we can see multiple wraparounds of sk->sk_rmem_alloc.
Let's fix it by using atomic_add_return() and comparing the two variables as unsigned int.
Before: [root@fedora ~]# ss -f netlink Recv-Q Send-Q Local Address:Port Peer Address:Port -1668710080 0 rtnl:nl_wraparound/293 *
After: [root@fedora ~]# ss -f netlink Recv-Q Send-Q Local Address:Port Peer Address:Port 2147483072 0 rtnl:nl_wraparound/290 * ^ `--- INT_MAX - 576
| Software | From | Fixed in |
|---|---|---|
| linux / linux_kernel | 2.6.13 | 5.4.296 |
| linux / linux_kernel | 5.5 | 5.10.240 |
| linux / linux_kernel | 5.11 | 5.15.189 |
| linux / linux_kernel | 5.16 | 6.1.146 |
| linux / linux_kernel | 6.2 | 6.6.99 |
| linux / linux_kernel | 6.7 | 6.12.39 |
| linux / linux_kernel | 6.13 | 6.15.7 |
| linux / linux_kernel | 2.6.12 | 2.6.12.x |
| linux / linux_kernel | 2.6.12-rc2 | 2.6.12-rc2.x |
| linux / linux_kernel | 2.6.12-rc3 | 2.6.12-rc3.x |
| linux / linux_kernel | 2.6.12-rc4 | 2.6.12-rc4.x |
| linux / linux_kernel | 2.6.12-rc5 | 2.6.12-rc5.x |
| linux / linux_kernel | 6.16-rc1 | 6.16-rc1.x |
| linux / linux_kernel | 6.16-rc2 | 6.16-rc2.x |
| linux / linux_kernel | 6.16-rc3 | 6.16-rc3.x |
| linux / linux_kernel | 6.16-rc4 | 6.16-rc4.x |
| linux / linux_kernel | 6.16-rc5 | 6.16-rc5.x |
| debian / debian_linux | 11.0 | 11.0.x |