Date: Mon, 11 Aug 2008 11:31:09 +0200 From: Jordi Moles Blanco <jordi@cdmon.com> Cc: freebsd-questions@freebsd.org, Patrick Mahan <mahan@mahan.org> Subject: Re: problems with a C script, exiting with signal 10 Message-ID: <48A006DD.4030300@cdmon.com> In-Reply-To: <489BF19D.1010804@cdmon.com> References: <489ACAB8.7000503@cdmon.com> <489B2689.8070209@mahan.org> <489BF19D.1010804@cdmon.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,
i've been trying to debug what you suggested, but no luck so far :(
The thing is that i checked out all the calls to arrays, space handling
and so on, and i couldn't find anything wrong.
After that, i ended up trying the "hard" way, which is to keep a file
/tmp/debug.log where the script writes everything that it does. So...
the problem was that even in those cases when postfix logged a "signal
10" error, the logs showed that the C script got to the end of the file,
it executed every single line, it doesn't get stuck manipulating arrays
or anything like that.
any idea?
Thanks.
En/na Jordi Moles Blanco ha escrit:
> Hi,
>
> thanks for the reply, i will have a close look at what you suggested.
> The thing is that, yes, i work with arrays, pointers, mallocs and so
> on. I'll try to make sure everything is initiliazed properly before
> being used.
>
> Thanks for the advice.
>
>
>
>
> En/na Patrick Mahan ha escrit:
>>
>>
>> Jordi Moles Blanco presented these words - circa 8/7/08 3:13 AM->
>>> Hi,
>>>
>>> I've got this home-made script, written in C, on a Freebsd 7.0
>>> server with different versions of postfix: 2.3,2,4 and 2.5
>>>
>>> The problem is that, while most of the time it works like a charm,
>>> sometimes it crashes and bounces the message. It's not really a big
>>> deal, cause the sender gets notified that their mail wasn't
>>> delivered and hopefully, they will resend it. However, the problem
>>> is that I've tried to debug my script but found nothing wrong at
>>> all, cause it only fails from time to time, let's say... once for
>>> each 2000 messages that postfix receives, and it appears to do so in
>>> a random way.
>>>
>>> As i said... postfix can fail to deliver a message to one particular
>>> mailbox, but if then you resend the very same message to the very
>>> same mailbox, it will be delivered.
>>>
>>> The error is reported in both "maillog" and "messages", like this:
>>>
>>>
>>> ******/var/log/maillog********
>>> Aug 7 01:55:19 mail01 postfix/pipe[27534]: 3E1A0143709:
>>> to=<EMAIL_ACCOUNT>, relay=quota_postfix, delay=0.23,
>>> delays=0.11/0/0/0.11, dsn=5.3.0, status=bounced (Command died with
>>> signal 10: "/usr/local/etc/postfix/quota_postfix")
>>>
>>>
>>> *****/var/log/messages*******
>>> Aug 7 01:55:19 mail01 kernel: pid 29535 (quota_postfix), uid 125:
>>> exited on signal 10
>>>
>>
>> Well signal 10 is SIGBUS which is indicative of (generally) a bad
>> address,
>> non-aligned memory address (on platforms it matters) or a hardware
>> error.
>> I would look for places you are dereferencing a pointer without perhaps
>> first validating it.
>>
>> Given that it rarely occurs, I might suspect that you are allocating
>> some
>> memory, but failing to completely initialize (malloc() doesn't zero out
>> memory) it or assuming it is already initialize.
>>
>> Good luck,
>>
>> Patrick
>>>
>>> Here you have some extra information about the script itself and the
>>> master.cf
>>>
>>>
>>> *****/usr/local/etc/postfix/quota_postfix***
>>>
>>> # ls -la /usr/local/etc/postfix/quota_postfix
>>> -rwsr-xr-x 1 postfix postfix 20048 Aug 4 10:18
>>> /usr/local/etc/postfix/quota_postfix
>>>
>>> It's got de suid flag cause it performs a "du" command and other
>>> file operations which need permissions, although i've tried with
>>> other groups of permissions and it eventually crashes anyway with
>>> "signal 10"
>>>
>>> ******master.cf*********
>>>
>>> .........
>>>
>>> # spamfilter
>>> spamfilter unix - n n - 20 pipe
>>> flags=R user=filter argv=/home/antispam.pl "localhost:10027"
>>> "antispam" "${sender}" "${recipient}" "/usr/local/bin/spamc"
>>>
>>> # from spamfilter to smtpd:10026
>>> localhost:10027 inet n - n - 100
>>> smtpd -o content_filter=quota_postfix
>>>
>>>
>>> # quota_postfix
>>> quota_postfix unix - n n - 20 pipe
>>> flags=R user=filter argv=/usr/local/etc/postfix/quota_postfix
>>> "localhost" "10028" "${sender}" "${recipient}" "${domain}"
>>>
>>> # from quota_postfix to smtpd:10028
>>> localhost:10028 inet n - n - 100
>>> smtpd -o content_filter=
>>>
>>> ................
>>>
>>> So far, any program which crashed would leave a ".core" file in
>>> /usr/crash, but this one is not doing the same, so... i can't
>>> actually debug from the core file either.
>>> Sysctl in my FreeBSD server is ok, but i guess that postfix, somehow
>>> is preventing this filter from generating a core file. Is that
>>> possible? Or am i completely wrong?
>>>
>>> How could I, at least, generate the .core file?
>>>
>>> Thanks.
>>>
>>>
>>> _______________________________________________
>>> freebsd-questions@freebsd.org mailing list
>>> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
>>> To unsubscribe, send any mail to
>>> "freebsd-questions-unsubscribe@freebsd.org"
>>>
>>>
>
> _______________________________________________
> freebsd-questions@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to
> "freebsd-questions-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?48A006DD.4030300>