[Olsr-dev] OLSRv2 Segfault

[Henning Rogge]

The test cases are just testing the API without OLSRd2. They make sure
I don't mess up too much with the API (at least the parts with test
cases ^^).

You can run all of them by using "make test" in the build directory.

Henning Rogge

On Thu, Jan 23, 2014 at 1:46 PM, Vigneswaran R <(spam-protected)> wrote:
> On 01/23/2014 03:49 PM, Henning Rogge wrote:
>>
>> Hi,
>>
>> it took a while but I think I found the bug that created your segfault
>> and fixed it properly. I removed your hotfix and added a testcase for
>> the message that crashed the RFC5444 address compressor.
>>
>> (I also found a few places where I can still improve the compressor,
>> but I will do this later)
>>
>> Can you maybe verify that the new commit also works for you?
>
>
> Yes, it is working for me. I haven't got any segfault till now. However, I
> am not sure how to use the test cases to validate olsrd2. If I run
> test_nhdp_2, it says "10 tests ended: 10 successes, 0 fails" regardless of
> whether olsrd2 is running or not.
>
> Attached the logs produced by both the new code (olsrd2_success.log) and the
> old code which gave segfault (olsrd2_failure.log). Thanks.
>
> Regards,
> Vignesh
>
>>
>> Henning Rogge
>>
>> On Tue, Jan 21, 2014 at 12:50 PM, Vigneswaran R <(spam-protected)>
>> wrote:
>>>
>>> On 01/21/2014 05:13 PM, Henning Rogge wrote:
>>>>
>>>> On Tue, Jan 21, 2014 at 12:18 PM, Vigneswaran R <(spam-protected)>
>>>> wrote:
>>>>>
>>>>> PFA the log. I used the following command to start olsrd2 (after
>>>>> applying
>>>>> the patch and compiling).
>>>>>
>>>>>       ./olsrd2 --set interface[eth0].bindto=10.0.0.0/8 --set
>>>>> interface[eth1].bindto=10.0.0.0/8 --set
>>>>> interface[eth2].bindto=10.0.0.0/8
>>>>> --set log.debug=nhdp_w 2> olsrd2.log
>>>>
>>>> Good.
>>>>
>>>> At the end of the logfile the segfault happens, right?
>>>
>>>
>>> Yes.
>>>
>>>
>>>>> If the debug level needs to be increased, please suggest the suitable
>>>>> value.
>>>>
>>>> The "loglevel" is just a legacy part that will most likely be removed
>>>> soon. Its just a quick way to set log.debug and log.info.
>>>
>>>
>>> Ok.
>>>
>>> vignesh
>
>