netdev - Race: data race between rawv6_send_hdrinc() and __dev_set

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Mon, 30 Nov 2020 17:35:52 +0000
From:   "Gong, Sishuai" <sishuai@...due.edu>
To:     "davem@...emloft.net" <davem@...emloft.net>,
        "kuznet@....inr.ac.ru" <kuznet@....inr.ac.ru>,
        "yoshfuji@...ux-ipv6.org" <yoshfuji@...ux-ipv6.org>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Race: data race between rawv6_send_hdrinc() and __dev_set_mtu() 

Hi,

We found a data race in linux kernel 5.3.11 that we are able to reproduce in x86 under specific interleavings. Currently, we are not sure about the consequence of this race so we would like to confirm with the community if this can be a harmful bug.

------------------------------------------
Writer site

/tmp/tmp.B7zb7od2zE-5.3.11/extract/linux-5.3.11/net/core/dev.c:7665
       7658  int __dev_set_mtu(struct net_device *dev, int new_mtu)
       7659  {
       7660      const struct net_device_ops *ops = dev->netdev_ops;
       7661
       7662      if (ops->ndo_change_mtu)
       7663          return ops->ndo_change_mtu(dev, new_mtu);
       7664
 ==>   7665      dev->mtu = new_mtu;
       7666      return 0;


------------------------------------------
Reader site

 /tmp/tmp.B7zb7od2zE-5.3.11/extract/linux-5.3.11/net/ipv6/raw.c:632
        618  static int rawv6_send_hdrinc(struct sock *sk, struct msghdr *msg, int length,
        619              struct flowi6 *fl6, struct dst_entry **dstp,
        620              unsigned int flags, const struct sockcm_cookie *sockc)
        621  {
        622      struct ipv6_pinfo *np = inet6_sk(sk);
        623      struct net *net = sock_net(sk);
        624      struct ipv6hdr *iph;
        625      struct sk_buff *skb;
        626      int err;
        627      struct rt6_info *rt = (struct rt6_info *)*dstp;
        628      int hlen = LL_RESERVED_SPACE(rt->dst.dev);
        629      int tlen = rt->dst.dev->needed_tailroom;
        630
        631      if (length > rt->dst.dev->mtu) {
 ==>    632          ipv6_local_error(sk, EMSGSIZE, fl6, rt->dst.dev->mtu);
        633          return -EMSGSIZE;
        634      }
        635      if (length < sizeof(struct ipv6hdr))
        636          return -EINVAL;
        637      if (flags&MSG_PROBE)
        638          goto out;
        639
        640      skb = sock_alloc_send_skb(sk,
        641                    length + hlen + tlen + 15,
        642                    flags & MSG_DONTWAIT, &err);
        643      if (!skb)
        644          goto error;
        645      skb_reserve(skb, hlen);
        646
        647      skb->protocol = htons(ETH_P_IPV6);
        648      skb->priority = sk->sk_priority;
        649      skb->mark = sk->sk_mark;
        650      skb->tstamp = sockc->transmit_time;
        651
        652      skb_put(skb, length);


------------------------------------------
Writer calling trace

- ksys_ioctl
-- do_vfs_ioctl 
--- vfs_ioctl
---- dev_ioctl
----- dev_ifsioc
------ dev_set_mtu
------- dev_set_mtu_ext
-------- __dev_set_mtu

------------------------------------------
Reader calling trace

- __sys_sendto
-- sock_sendmsg
--- inet_sendmsg
---- rawv6_sendmsg



Thanks,
Sishuai