Re: ssh coredump if gssapi-with-mic preferred, but no kinit done (yet)?

On 2 Jul 2007, at 15:17, Ian FREISLICH wrote:

> Ian FREISLICH wrote:
>> Doug Rabson wrote:
>>> On 1 Jul 2007, at 19:38, David Wolfskill wrote:
>>>
>>>> I'm willing to try some things on the machine to help diagnose the
>>>> problem(s) under CURRENT, but there are some things I'd like to do
>>>> with it (that will take a few hours,such as a "make release") that
>>>> I'd much rather do while it's running STABLE.  And Sundays are
>>>> about the
>>>> only days I can really count on being able to do anything with  
>>>> CURRENT
>>>> on the machine.  (I'm in US/PAcific time zone.)
>>>>
>>>> So:  clues?  Would building a debugging version of ssh be a  
>>>> reasonable
>>>> step?
>>>
>>> I may have broken this with my last commit to the gss-api code. I'll
>>> try to take a look at this as soon as I fix my FreeBSD scratch  
>>> box (I
>>> managed to fry its motherboard yesterday). In the meantime, could  
>>> you
>>> get a stack-trace so I can at least see where its crashing. If you
>>> can build libgssapi with debugging information that would probably
>>> help a lot.
>>
>> It's quite a mangled stack trace.  I saw a similar trace in the
>> recent nmh core dump problem - not sure if you followed that thread.
>
> Not sure why the list truncated the stack trace.  Hopefully this
> makes it through.
>
> 0  0x280eec71 in _gss_oid_equal () from /usr/lib/libgssapi.so.8
> (gdb) bt
> 0  0x280eec71 in _gss_oid_equal () from /usr/lib/libgssapi.so.8
> 1  0x280eeb78 in _gss_find_mech_switch () from /usr/lib/libgssapi.so.8
> 2  0x280ecffd in gss_display_status () from /usr/lib/libgssapi.so.8

It looks like I did break it. I should be able to repro and fix this  
as soon as I put my FreeBSD box back together (probably tomorrow). In  
the meantime, you can avoid this by rolling src/lib/libgssapi back to  
about last Friday or so.