another abrt problem

Jiri Moskovcak jmoskovc at redhat.com
Mon Mar 4 12:40:00 UTC 2013


On 03/04/2013 01:37 PM, Neal Becker wrote:
> Jiri Moskovcak wrote:
>
>> On 03/04/2013 01:28 PM, Neal Becker wrote:
>>> Jiri Moskovcak wrote:
>>>
>>>> On 03/04/2013 01:17 PM, Neal Becker wrote:
>>>>> Is it possible to use abrt as intended and still be able to get core dumps
>>>>> from
>>>>> non-fedora binaries (e.g., my own code)?  Right now, all core dumps are
>>>>> redirected to abrt.
>>>>>
>>>>> We need a way to be able to use gdb to debug core dumps.
>>>>>
>>>>> I know we can turn off abrt entirely
>>>>> /proc/sys/kernel/core_pattern
>>>>>
>>>>> but that's not acceptable!
>>>>>
>>>>> We need a way that allows abrt to be used for fedora packages, while at the
>>>>> same time allowing capturing core dumps to run gdb on for non-fedora
>>>>> packages.
>>>>>
>>>>> If there is a way in the current abrt design, it is not sufficiently
>>>>> obvious/discoverable.
>>>>>
>>>>>
>>>>
>>>> Hi,
>>>> I'm pretty sure there is a way to achieve this with the current abrt
>>>> design, can you please describe how do you imagine this should work?
>>>>
>>>> Thank you,
>>>> Jirka
>>>
>>> One possibility would be to simply add an dialog to abrt-gui to save the core
>>> dump to some user-specified location.  It could offer to run gdb, but that's
>>> not necessary.
>>>
>>> I suppose it would also be good to have a non-gui option as well, but I don't
>>> know what that would be.
>>>
>>
>> You have two options:
>>
>> 1) in /etc/abrt/abrt-action-save-package-data.conf you can change the
>> option ProcessUnpackaged = no to yes
>>
>> 2) you can just set ulimit -c unlimited and abrt will create the core in
>> the CWD in format core.<PID> as it is by default without abrt and then
>> you can pass it to gdb
>>
>> Regards,
>> Jirka
>>
>
> If 2), will abrt still operate as normal on fedora package core files?  We don't
> want to permanently disable abrt just to enable normal use of gdb on our own
> code.
>

- yes, the only problem is, that it will create 2 cores (one in /var/ 
one in CWD) for every crash

--Jirka


More information about the devel mailing list