Date: Sat, 13 Oct 2012 15:26:12 -0400
From: Christopher Faylor <cgf-use-the-mailinglist-please AT cygwin DOT com>
To: cygwin AT cygwin DOT com
Subject: Re: sshd fork failure
Message-ID: <20121013192612.GB10736@ednor.casa.cgf.cx>
Reply-To: cygwin AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
References: <503F9877 DOT 5070406 AT gmail DOT com> <20121013153936 DOT GB839 AT ednor DOT casa DOT cgf DOT cx> <5079926F DOT 1020503 AT gmail DOT com> <20121013165405 DOT GC9226 AT ednor DOT casa DOT cgf DOT cx> <5079B10D DOT 1030108 AT gmail DOT com> <20121013182449 DOT GA10736 AT ednor DOT casa DOT cgf DOT cx> <5079BF50 DOT 1020503 AT gmail DOT com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <5079BF50.1020503@gmail.com>
User-Agent: Mutt/1.5.20 (2009-06-14)
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Precedence: bulk
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

On Sat, Oct 13, 2012 at 09:21:52PM +0200, marco atzeri wrote:
>On 10/13/2012 8:24 PM, Christopher Faylor wrote:
>> On Sat, Oct 13, 2012 at 08:21:01PM +0200, marco atzeri wrote:
>>> On 10/13/2012 6:54 PM, Christopher Faylor wrote:
>>>> On Sat, Oct 13, 2012 at 06:10:23PM +0200, marco atzeri wrote:
>>>>> On 10/13/2012 5:39 PM, Christopher Faylor wrote:
>>>>>> On Thu, Aug 30, 2012 at 06:44:39PM +0200, marco atzeri wrote:
>>>>>>> Hi,
>>>>>>> tracing the sshd process on latest cygwin shapshot ,
>>>>>>> I see that the fork is failing due to exception C0000005
>>>>>>>
>>>>>>>        1       1 [main] sshd (4644)
>>>>>>> **********************************************
>>>>>>>     5921    5922 [main] sshd (4644) Program name:
>>>>>>> E:\cygwin\usr\sbin\sshd.exe (windows pid 4644)
>>>>>>>       30    5952 [main] sshd (4644) OS version:   Windows NT-6.1
>>>>>>>       25    5977 [main] sshd (4644)
>>>>>>> **********************************************
>>>>>>>      636    6613 [main] sshd (4644) sigprocmask: 0 = sigprocmask (0,
>>>>>>> 0x6123D428, 0x610FBB30)
>>>>>>>      122    6735 [main] sshd 4644 child_copy: cygheap - hp 0x20C low
>>>>>>> 0x612708D0, high 0x61277640, res 1
>>>>>>>     1368    8103 [main] sshd 4644 child_copy: done
>>>>>>> --- Process 4644, exception C0000005 at 61137277
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> $ addr2line.exe 61137277 -e /usr/bin/cygwin1.dll
>>>>>>> /netrel/src/cygwin-snapshot-20120827-1/newlib/libc/stdlib/mbtowc_r.c:323
>>>>>>>
>>>>>>> that should be:
>>>>>>>
>>>>>>>     ch = (state->__count == 1) ? t[i++] : state->__value.__wchb[1];
>>>>>>>
>>>>>>> inside __utf8_mbtowc .
>>>>>>> It looks a strange place for a fork failure, but it is used in
>>>>>>> winsup/cygwin/cygheap.cc so it could make sense.
>>>>>>>
>>>>>>> Any suggestion how to follow/debug the child portion of fork ?
>>>>>>
>>>>>> Is this still failing?  We're looking to release 1.7.17 soon.
>>>>>
>>>>> on  20121012 similar but in another location.
>>>>>
>>>>> --- Process 6280, exception C0000005 at 61137DC8
>>>>>     427  147466 [main] sshd 6280 exception::handle: In
>>>>> cygwin_except_handler exception 0xC0000005 at 0x61137DC8 sp 0x289270
>>>>>      67  147533 [main] sshd 6280 exception::handle: In
>>>>> cygwin_except_handler signal 11 at 0x61137DC8
>>>>>      38  147571 [main] sshd 6280 exception::handle: In
>>>>> cygwin_except_handler calling 0x0
>>>>>      37  147608 [main] sshd 6280 exception::handle: Exception:
>>>>> STATUS_ACCESS_VIOLATION
>>>>>
>>>>>    addr2line.exe 61137DC8 -e /usr/bin/cygwin1.dll
>>>>> /netrel/src/cygwin-snapshot-20121012-1/newlib/libc/stdio/findfp.c:296
>>>>
>>>>You seem to be the only person reporting this problem.  Everyone else
>>>>updates to a snapshot to cure ssh problems.  FWIW, that instruction
>>>>shouldn't even be executed.
>>>
>>>I have such issue only on sshd forking and for long time, so it is not
>>>relative to last snapshots.
>>
>>Corinna fixed an issue which could have manifested as ssh SIGSEGVing.
>>Given the fact that the 61137DC8 doesn't make any sense, are you sure
>>you're actually using a snapshot?  Could there be an old version of
>>cygwin1.dll sitting around somewhere?
>
>$ uname -a CYGWIN_NT-6.1-WOW64 MARCOATZERI 1.7.17s(0.262/5/3) 20121012 01:20:45 i686 Cygwin

That doesn't show anything other than the fact that uname is
sitting next to a snapshot cygwin1.dll.

>could be that the cygwin1.dbg of 20121012 is not correct ?

No.  This has nothing to do with cygwin1.dbg.  The address 61137DC8 is
in the middle of an instruction.

cgf

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple