Re: [Bug-readline] SIGSEGV in _rl_dispatch_callback()

[Chet Ramey]

On 3/6/14 10:43 AM, Chet Ramey wrote:
> On 3/5/14 11:18 PM, Anatol Pomozov wrote:
>> Hi,
>>
>> It is a follow-up for a Linux Arch bug https://bugs.archlinux.org/task/39144
>>
>> After upgrading system to readline 6.3 we started seeing a crash that
>> can be reproduced using 'ipython' command-line tool. Here are the
>> steps:
>>
>> - run 'ipython'
>> - type 'import'
>> - press up, down, up
>> - SIGSEGV!!!
>>
>> Downgrading readline back to 6.2 fixes the issue.
> 
> I can't reproduce this with a simple callback test example.  I suspect it's
> something in how ipython configures readline that the example doesn't do.
> Where would I find that initialization code?

What I think would be really helpful is for someone to build readline-6.3
with debugging symbols (make CFLAGS=-g), install that version, run the
ipython commands to generate the core dump, and send that.

What might be even more helpful is for someone to start up ipython, run
gdb and attach to the core python interpreter, and watch _rl_kscxt.  That
should help identify when it is being set to NULL and then used without
going through the _rl_keyseq_cxt_alloc() interface.

Chet

-- 
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, ITS, CWRU    address@hidden    http://cnswww.cns.cwru.edu/~chet/