Fatal error from Cygwin emacs-w32 every day or so

[Ken Brown]

On 4/21/2014 1:25 PM, KARR, DAVID wrote:
>> -----Original Message-----
>> Of Ken Brown
>> Sent: Monday, April 21, 2014 4:16 AM
>> Subject: Re: Fatal error from Cygwin emacs-w32 every day or so
>>
>> On 4/21/2014 1:14 AM, KARR, DAVID wrote:
>>>> -----Original Message-----
>>>> Of KARR, DAVID
>>>> Sent: Saturday, April 19, 2014 6:19 PM
>>>> Subject: RE: Fatal error from Cygwin emacs-w32 every day or so
>>>>
>>>>> -----Original Message-----
>>>>> Of Ken Brown
>>>>> Sent: Saturday, April 19, 2014 4:38 PM
>>>>> Subject: Re: Fatal error from Cygwin emacs-w32 every day or so
>>>>>
>>>>> On 4/15/2014 12:13 PM, Ken Brown wrote:
>>>>>> On 4/15/2014 12:12 AM, Eli Zaretskii wrote:
>>>>>>>> Date: Mon, 14 Apr 2014 21:08:15 -0700
>>>>>>>> From: Ken Brown
>>>>>>>>
>>>>>>>>> I just saw it die, and this is the bt I get:
>>>>>>>>>
>>>>>>>>> Program received signal SIGSEGV, Segmentation fault.
>>>>>>>>> 0x0000000100551354 in wait_reading_process_output (
>>>>>>>>>         time_limit=time_limit@entry=0, nsecs=nsecs@entry=0,
>>>>>>>>>         read_kbd=read_kbd@entry=-1,
>> do_display=do_display@entry=true,
>>>>>>>>>         wait_for_cell=wait_for_cell@entry=4304630834,
>>>>>>>>>         wait_proc=wait_proc@entry=0x0,
>>>>>>>>> just_wait_proc=just_wait_proc@entry=0)
>>>>>>>>>        @/usr/src/debug/emacs-24.3-7/src/process.c:4677
>>>>>>>>> 4677                      if (wait_proc->gnutls_p /* Check for
>>>> valid
>>>>>>>>> process.  */
>>>>>>>>
>>>>>>>> This backtrace doesn't make sense.  If you look at the source code,
>>>>>>>> you'll see that if wait_proc is NULL on entry to
>>>>>>>> wait_reading_process_output, then line 4677 is never reached.  I'm
>>>> not
>>>>>>>> sure what would cause a bogus backtrace like this.  BLODA?
>>>>>>>> Optimization?
>>>>>>>>
>>>>>>>> In any case, I suggest that you wait a week until I have a chance
>> to
>>>>>>>> build a pretest of 24.4 for you to try.  I'll build it without
>>>>>>>> optimization to make debugging easier.
>>>>>>>
>>>>>>> This is a known problem with GnuTLS support, it was solved in the
>>>>>>> Emacs repository last November (bzr revision 114956, if someone
>> wants
>>>>>>> the diffs), and surely should be solved in the upcoming Emacs 24.4.
>>>>>>
>>>>>> Thanks, Eli!  In that case I'll make a new release of emacs-24.3 with
>>>>>> that patch applied, to see if it resolves the issue for the OP.
>>>>>
>>>>> I've rebuilt emacs-24.3 with the gnutls fix.  David and Achim (and
>>>>> anyone else who's been experiencing these crashes), please try the
>>>>> following binary and let me know if it solves the problem:
>>>>>
>>>>>       http://sanibeltranquility.com/cygwin/emacs-w32.exe.xz
>>>>>
>>>>> You might have to do "chmod +x emacs-w32.exe" after uncompressing it.
>>>>>
>>>>> If it fixes the problem, I'll issue a new release of emacs-24.3.  If
>>>>> not, I'll build a pretest of emacs-24.4 for you to try.
>>>>
>>>> Ok.  It's running.  As it only fails after random intervals, I'll have
>> to
>>>> keep it running for a while.  I'll give it a day or so.
>>>
>>> I came back to my computer and saw two dialogs, one saying that my VPN
>> connection had died (that happens from time to time), and the other the
>> Emacs failure dialog.  I started this executable from gdb, but it's saying
>> that it doesn't have symbols, so the stacktrace probably isn't helpful,
>> but this is what I saw:
>>
>> Right.  I just supplied a stripped binary for a quick test.  It's not
>> suitable for debugging.
>>
>>> I've started it again, but not in gdb this time.
>>
>> If it crashes again, I'll build a pretest of 24.4 without optimization
>> and issue it as a test release.
>
> It just failed again while sitting here in my office (no relation to VPN/network issues).

Could you try one more thing if you haven't already:  Start emacs with 
the -Q option, in case there's something in your initialization files 
that triggers the problem.

Ken


--
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