netdev - Re: [PATCH iproute2-next 2/3] ipmroute: don't complain about unicast routes

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] [day] [month] [year] [list]

Date:   Thu, 8 Mar 2018 10:12:34 -0700
From:   David Ahern <dsahern@...il.com>
To:     Stephen Hemminger <stephen@...workplumber.org>
Cc:     netdev@...r.kernel.org, Stephen Hemminger <sthemmin@...rosoft.com>
Subject: Re: [PATCH iproute2-next 2/3] ipmroute: don't complain about unicast
 routes

On 3/7/18 9:51 AM, Stephen Hemminger wrote:
> On Tue,  6 Mar 2018 17:03:54 -0800
> Stephen Hemminger <stephen@...workplumber.org> wrote:
> 
>> From: Stephen Hemminger <sthemmin@...rosoft.com>
>>
>> Every non-multicast route prints an error message.
>> Kernel doesn't filter out unicast routes, it is up to filter function
>> to do this.
>>
>> Signed-off-by: Stephen Hemminger <stephen@...workplumber.org>
> 
> I found the issue (in kernel) but not sure how to deal with it.
> If kernel is built without multicast routing configured !CONFIG_IP_MROUTE
> then the netlink request to return multicast routes will return all routes!
> 
> This is because in the kernel the way route dump works is that each
> address family registers a callback to dump routes for a specific address
> family.  If that address family is not registered then the fall back
> is to address family PF_UNSPEC which has a handler that dumps all
> routes.

If I ask the kernel for IP{6}MR address family and the kernel returns a
dump of all address families, that seems like a kernel bug to me. From a
short review it seems to be that way for a long time. I guess iproute2
has no choice but to adapt to the kernel design.

Are you going to resubmit this series?


> 
> Unfortunately, changing that behavior in kernel will certainly break
> some user. And there is no direct way to determine multicast routing
> is enabled in ip mroute code.
> 
> Maybe just change the message in ip mroute to do:
> 
> diff --git a/ip/ipmroute.c b/ip/ipmroute.c
> index aa5029b44f41..31b9bfe95596 100644
> --- a/ip/ipmroute.c
> +++ b/ip/ipmroute.c
> @@ -76,9 +76,8 @@ int print_mroute(const struct sockaddr_nl *who, struct nlmsghdr *n, void *arg)
>                 return -1;
>         }
>         if (r->rtm_type != RTN_MULTICAST) {
> -               fprintf(stderr, "Not a multicast route (type: %s)\n",
> -                       rtnl_rtntype_n2a(r->rtm_type, b1, sizeof(b1)));
> -               return 0;
> +               fprintf(stderr, "Multicast routing does not appear to be enabled\n");
> +               return -1;
>         }
>  
>         parse_rtattr(tb, RTA_MAX, RTM_RTA(r), len);
>