[Swan] Libreswan 3.27 segfault

Andrew Cagney andrew.cagney at gmail.com
Fri Jan 11 14:43:26 UTC 2019


One thing.  The value of p1c from the assignment:

        const struct connection *p1c = p1st->st_connection;

looks bogus.  Given the value came from *p1st, I'd suspect the passed
in state's either been free()d or corrupted :-(

On Fri, 11 Jan 2019 at 04:57, csszep <csszep at gmail.com> wrote:
>
> Hi!
>
> Still crashing with Libreswan master from 10 jan.
>
> I updating the github issue #169 with new gdb backtrace.
>
> The RHEL bugzilla enry is not accessible with regular RH account.
>
> Thx Csszep
>
> csszep <csszep at gmail.com> ezt írta (időpont: 2018. dec. 4., K, 9:23):
>>
>> Hi Paul!
>>
>> Thx for the Answer. I will try and report. Unfortunately the crash now happens ony once or twice a week....
>>
>> Paul Wouters <paul at nohats.ca> ezt írta (időpont: 2018. dec. 3., H, 15:40):
>>>
>>> On Thu, 29 Nov 2018, csszep wrote:
>>>
>>> > I have a longstanding problem w libreswan. See github issue #169
>>> >
>>> > Can anyone help identify the problem?
>>> >
>>> > The crash happened daily (SA delete? rekey?), and after 4-5 crashes it works again.
>>> >
>>> > The last few messages, before every  crash:
>>> >
>>> >
>>> > 2018-11-28T10:43:15+01:00 firewall1 pluto[16834]: "customer2" #701: received Delete SA(0xb6ca75dc) payload: deleting IPSEC State #702
>>> > 2018-11-28T10:43:15+01:00 firewall1 pluto[16834]: "customer2" #702: deleting other state #702 (STATE_QUICK_R2) and sending notification
>>> > 2018-11-28T10:43:15+01:00 firewall1 pluto[16834]: "customer2" #702: ESP traffic information: in=1MB out=248KB
>>> > 2018-11-28T10:43:15+01:00 firewall1 pluto[16834]: "customer2 #701: deleting state (STATE_MAIN_R3) and sending notification
>>> > 2018-11-28T10:40:23+01:00 firewall1 kernel: traps: pluto[16834] general protection ip:7f71e05e212b sp:7ffcd12c9180 error:0 in pluto[7f71e0587000+154000]
>>> >
>>> > The connection "customer2" is not the same in every crash, but maybe? all connections that causes the crash come from F5/BIG-IP peer....
>>>
>>> Can you try git master? I think this issue is fixed there. This is when
>>> there is a Delete plus an additional notify payload.
>>>
>>> A different backport of the same bug is applied for RHEL via https://bugzilla.redhat.com/show_bug.cgi?id=1630355
>>>
>>> Paul
>
> _______________________________________________
> Swan mailing list
> Swan at lists.libreswan.org
> https://lists.libreswan.org/mailman/listinfo/swan


More information about the Swan mailing list