When querying info for an offloaded BPF map or program, bpf_map_offload_info_fill_ns() and bpf_prog_offload_info_fill_ns() obtain the network namespace with get_net(dev_net(offmap->netdev)). However, the associated netdev's netns may be racing with teardown during netns destruction. If the netns refcount has already reached 0, get_net() performs a refcount_t increment on 0, triggering: refcount_t: addition on 0; use-after-free. Although rtnl_lock and bpf_devs_lock ensure the netdev pointer remains valid, they cannot prevent the netns refcount from reaching zero. Fix this by using maybe_get_net() instead of get_net(). maybe_get_net() uses refcount_inc_not_zero() and returns NULL if the refcount is already zero, which causes ns_get_path_cb() to fail and the caller to return -ENOENT -- the correct behavior when the netns is being destroyed.
Showing the changes to kernel (1 of 1 file). Show all
Unified · Split
| @@ -435,9 +435,8 @@ static struct ns_common *bpf_prog_offload_info_fill_ns(void *private_data) · 434 unmodified lines | ||
| 435 | 435 | |
| 436 | 436 | if (aux->offload) { |
| 437 | 437 | args->info->ifindex = aux->offload->netdev->ifindex; |
| 438 | − net = dev_net(aux->offload->netdev); | |
| 439 | − get_net(net); | |
| 440 | − ns = &net->ns; | |
| 438 | + net = maybe_get_net(dev_net(aux->offload->netdev)); | |
| 439 | + ns = net ? &net->ns : NULL; | |
| 441 | 440 | } else { |
| 442 | 441 | args->info->ifindex = 0; |
| 443 | 442 | ns = NULL; |
| @@ -647,9 +646,8 @@ static struct ns_common *bpf_map_offload_info_fill_ns(void *private_data) · 203 unmodified lines | ||
| 647 | 646 | |
| 648 | 647 | if (args->offmap->netdev) { |
| 649 | 648 | args->info->ifindex = args->offmap->netdev->ifindex; |
| 650 | − net = dev_net(args->offmap->netdev); | |
| 651 | − get_net(net); | |
| 652 | − ns = &net->ns; | |
| 649 | + net = maybe_get_net(dev_net(args->offmap->netdev)); | |
| 650 | + ns = net ? &net->ns : NULL; | |
| 653 | 651 | } else { |
| 654 | 652 | args->info->ifindex = 0; |
| 655 | 653 | ns = NULL; |