netdev - Re: Kernel memory corruption in CIPSO labeled TCP packets processing.

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

Message-ID: <CAHC9VhQD=itT7jE22bo0UF24gE56y1z29fZec-7y1Zumz38FfQ@mail.gmail.com>
Date:   Fri, 25 Jan 2019 11:45:47 -0500
From:   Paul Moore <paul@...l-moore.com>
To:     Nazarov Sergey <s-nazarov@...dex.ru>
Cc:     "linux-security-module@...r.kernel.org" 
        <linux-security-module@...r.kernel.org>,
        "selinux@...r.kernel.org" <selinux@...r.kernel.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        Casey Schaufler <casey@...aufler-ca.com>
Subject: Re: Kernel memory corruption in CIPSO labeled TCP packets processing.

On Thu, Jan 24, 2019 at 9:46 AM Nazarov Sergey <s-nazarov@...dex.ru> wrote:
> 22.01.2019, 20:48, "Paul Moore" <paul@...l-moore.com>:
> >
> > Yes, exactly. If you don't pass the skb it shouldn't attempt to call
> > icmp_send() in case of error.
> >
> > --
> > paul moore
> > www.paul-moore.com
>
> You are right, sorry. We can do that without ip_options_compile modification.
> Simplified patch 2:
> ---
>  net/ipv4/cipso_ipv4.c |   18 ++++++++++++++++--
>  1 files changed, 16 insertions(+), 2 deletions(-)
>
> diff --git a/net/ipv4/cipso_ipv4.c b/net/ipv4/cipso_ipv4.c
> index 777fa3b..a4ed0a4 100644
> --- a/net/ipv4/cipso_ipv4.c
> +++ b/net/ipv4/cipso_ipv4.c
> @@ -1735,13 +1735,27 @@ int cipso_v4_validate(const struct sk_buff *skb, unsigned char **option)
>   */
>  void cipso_v4_error(struct sk_buff *skb, int error, u32 gateway)
>  {
> +       unsigned char optbuf[sizeof(struct ip_options) + 40];
> +       struct ip_options *opt = (struct ip_options *)optbuf;
> +
>         if (ip_hdr(skb)->protocol == IPPROTO_ICMP || error != -EACCES)
>                 return;
>
> +       /*
> +        * We might be called above the IP layer,
> +        * so we can not use icmp_send and IPCB here.
> +        */
> +
> +       memset(opt, 0, sizeof(struct ip_options));
> +       opt->optlen = ip_hdr(skb)->ihl*4 - sizeof(struct iphdr);

Hmm, I think the above calculation should take into account the actual
length of the IP options, and not just the max size (calculate it
based on iphdr->ihl).

Beyond that fix, I think it's time to put together a proper patchset
and post it to the lists for formal review/merging.

Thanks for your work on this.

> +       memcpy(opt->__data, (unsigned char *)&(ip_hdr(skb)[1]), opt->optlen);
> +       if (ip_options_compile(dev_net(skb->dev), opt, NULL))
> +               return;
> +
>         if (gateway)
> -               icmp_send(skb, ICMP_DEST_UNREACH, ICMP_NET_ANO, 0);
> +               __icmp_send(skb, ICMP_DEST_UNREACH, ICMP_NET_ANO, 0, opt);
>         else
> -               icmp_send(skb, ICMP_DEST_UNREACH, ICMP_HOST_ANO, 0);
> +               __icmp_send(skb, ICMP_DEST_UNREACH, ICMP_HOST_ANO, 0, opt);
>  }

-- 
paul moore
www.paul-moore.com