netdev - Re: Memory leaks in IPv6 ndisc on v4.19.56

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]

Message-ID: <e430b2cd-e358-f328-5445-b5f681118ce3@linuxlounge.net>
Date:   Tue, 2 Jul 2019 15:54:35 +0200
From:   Martin Weinelt <martin@...uxlounge.net>
To:     "David S. Miller" <davem@...emloft.net>,
        Alexey Kuznetsov <kuznet@....inr.ac.ru>,
        Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
        netdev@...r.kernel.org
Subject: Re: Memory leaks in IPv6 ndisc on v4.19.56

Hi again,

another backtrace just came up.

Best,
  Martin


$ ./scripts/faddr2line /usr/lib/debug/lib/modules/4.19.56/vmlinux build_skb+0x11/0x80
build_skb+0x11/0x80:
build_skb at net/core/skbuff.c:314

$ ./scripts/faddr2line /usr/lib/debug/lib/modules/4.19.56/kernel/drivers/net/tun.ko tun_build_skb.isra.56+0x191/0x4d0
tun_build_skb.isra.56+0x191/0x4d0:
tun_build_skb at /home/hexa/git/linux-stable/drivers/net/tun.c:1687


On 7/2/19 3:05 PM, Martin Weinelt wrote:
> Hi everyone,
> 
> I've been experiencing memory leaks on the v4.19 series. I've started
> seeing them on Debian with v4.19.16 and I can reproduce them on v4.19.56
> using Debians kernel config. I was unable to reproduce this on 
> v5.2.0-rc6/rc7.
> 
>   [ 1899.380321] kmemleak: 1138 new suspected memory leaks (see /sys/kernel/debug/kmemleak)
> 
> On the machines in question we're running routers for a mesh networking
> setup based on the batman-adv kmod. Our setup consists of KVM guests 
> running Debian, with each router having 18 bridges with the following
> master/slave relationship:
> 
>   bridge -> batman-adv -> {L2 tunnel, virtio net device}
> 
> I've attached the output of kmemleak and I've looked up the top-most
> function offsets below:
> 
> Best,
>   Martin
> 
> 
> $ ./faddr2line /usr/lib/debug/lib/modules/4.19.56/vmlinux ndisc_recv_ns+0x356/0x5f0
> ndisc_recv_ns+0x356/0x5f0:
> __neigh_lookup at include/net/neighbour.h:513
> (inlined by) ndisc_recv_ns at net/ipv6/ndisc.c:916
> 
> $ ./faddr2line /usr/lib/debug/lib/modules/4.19.56/vmlinux ndisc_router_discovery+0x4ab/0xae0
> ndisc_router_discovery+0x4ab/0xae0:
> __neigh_lookup at include/net/neighbour.h:513
> (inlined by) ndisc_router_discovery at net/ipv6/ndisc.c:1387
> 
> $ ./faddr2line /usr/lib/debug/lib/modules/4.19.56/vmlinux ndisc_recv_rs+0x173/0x1b0
> ndisc_recv_rs+0x173/0x1b0:
> ndisc_recv_rs at net/ipv6/ndisc.c:1095
> 
> $ ./faddr2line /usr/lib/debug/lib/modules/4.19.56/vmlinux ip6_finish_output2+0x211/0x570
> ip6_finish_output2+0x211/0x570:
> ip6_finish_output2 at net/ipv6/ip6_output.c:117
> 


View attachment "kmemleak2.log" of type "text/x-log" (1644 bytes)