Re: subversion 1.9.3-1 segfault

[David Rothenberger <daveroth@acm.org>]

Greg Chicares wrote:
> On 2016-02-13 18:41, David Rothenberger wrote:
>> On 2/9/2016 9:39 AM, Greg Chicares wrote:
>>> 'svn' segfaulted on a routine command:
>>>
>>> /lmi/mirror/lmi[0]$svn status --show-updates
>>>
>>> svn: E170013: Unable to connect to a repository at URL 'http://svn.savannah.nongnu.org/svn/lmi/lmi/trunk'
>>> svn: E000104: Error running context: Connection reset by peer
>>> svn update
>>> Updating '.':
>>> zsh: segmentation fault (core dumped)
>>>
>>> The command worked as expected when I reran it (after 'svn cleanup'):
>> [snip]
>>>
>>> It's working now, and I can't reproduce the problem, but I thought
>>> I should report it anyway. I've attached cygcheck -s -v -r output
>>> and copied 'svn.exe.stackdump' below.
>>>
>>> Exception: STATUS_ACCESS_VIOLATION at eip=00000000
>>> eax=200C61B8 ebx=200A8DE8 ecx=0022C39C edx=00000000 esi=00000000 edi=00000011
>>> ebp=200C01A0 esp=0022C35C program=C:\cygwin-lmi\bin\svn.exe, pid 1736, thread main
>>> cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
>>> Stack trace:
>>> Frame     Function  Args
>>> 200C01A0  00000000 (00000000, 00000000, 6C1D5580, 200A0E90)
>>> 200A8DE8  00000000 (00000002, 200A8DE8, 2009EDC0, 200AAEC8)
>>> 2009F5D8  00000000 (6C701180, 6C701160, 00000000, 2009F978)
>>> 2009EDC0  2009F628 (00000000, 20052794, 200AAF00, 00000000)
>>> 20052790  200C81B8 (2004D788, 200430F4, 00000000, 20052C08)
>>> 200430F0  2009EDC0 (2003EF00, 20038D6C, 20054CB0, 00000000)
>>> 20038D68  20052790 (00000000, 00000000, 20038E88, 00000000)
>>
>> I wasn't able to get any useful information out of the stack trace.
>> Running addr2line does not show any functions. I'm not sure if this is
>> because you don't have the subversion-debuginfo package installed. I
>> would suggest you install this in case it makes a difference and the
>> failure happens again.
>
> I've installed 'subversion-debuginfo' and rebooted. After working okay
> for a few days, 'svn' is now segfaulting again. A coworker who did a
> brand-new Cygwin installation on a new machine yesterday reports 'svn'
> failing for her too, also with a stackdump.
>
> Here's my failing command:
>
> /lmi/mirror/lmi[1]$svn update
>
> Updating '.':
>
> zsh: segmentation fault (core dumped)  svn update
>
>
> (The last line means that 'zsh' caught the segfault, not that zsh crashed.)
>
> Today's stackdump looks much the same to me as last week's:
>
> Exception: STATUS_ACCESS_VIOLATION at eip=00000000
> eax=200C6268 ebx=200A8E98 ecx=0022C39C edx=00000000 esi=00000000 edi=00000011
> ebp=200C0250 esp=0022C35C program=C:\cygwin-lmi\bin\svn.exe, pid 1164, thread main
> cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
> Stack trace:
> Frame     Function  Args
> 200C0250  00000000 (00000000, 00000000, 6C1D5580, 200A0F40)
> 200A8E98  00000000 (00000002, 200A8E98, 2009EE70, 200AAF78)
> 2009F688  00000000 (6C701180, 6C701160, 00000000, 2009FA28)
> 2009EE70  2009F6D8 (00000000, 20052794, 200AAFB0, 00000000)
> 20052790  200C8268 (2004D788, 200430F4, 00000000, 20052C08)
> 200430F0  2009EE70 (2003EF00, 20038D6C, 20054CB0, 00000000)
> 20038D68  20052790 (00000000, 00000000, 20038E88, 00000000)
> End of stack trace
>
> ...and I don't see any useful information from 'addr2line':
>
> $addr2line -e /usr/bin/svn 0x200c0250 0x2009ee70 0x2009f6d8
>
> ??:0
>
> ??:0
>
> ??:0
>
>
> Attached is fresh 'cygcheck' output.
>
> I'll try running 'svn' under 'gdb'. Is there anything else I can do to help?

That's the only other thing I can think of. If you can catch the 
segfault and get a backtrace, I may be able to track it down. Or if you 
can create a simple test case to reproduce the problem, I can use gdb. 
Honestly, though, it's been a long time since I've done C programming 
and used gdb in earnest, so it will be slow going for me to debug this.


-- 
David Rothenberger  ----  daveroth@acm.org


--
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