[PD] pd-0.46-7: segfault with jack-1.9.10

[Xavier Miller]

Hi!

All those 4 commits from git are OK, and also the latest version, but I
compiled them by hand (autogen.sh ; configure --enable-jack ; make ;
make install) and not using the Gentoo compilation tools (emerge which
does almost the same).

And compiling 0.46-7 by hand also works.

It seems that the problem resides on what I did in the Gentoo
compilation script (ebuild), not in the source

So the problem is not at pd's side.

Thank you anyway for the support.

Kind regards,
Xavier Miller.

Le 17/02/16 17:58, Miller Puckette a écrit :
> Actually, rather than figure out git bisect (warning: it opens a new branch
> and after you're done you have to manually get rid of it and get back to
> 'master') - since there are only 4 commits in question you could simply check
> out each of the 4 and see which is the first one that gives you the crash.
> 
> Thanks, by the way, for being willng to help figure this out!
> 
> Miller
> 
> On Wed, Feb 17, 2016 at 09:28:36AM +0000, Xavier Miller wrote:
>>
>> IOhannes m zmoelnig <zmoelnig at iem.at> a écrit :
>>
>>> On 2016-02-17 09:11, Xavier Miller wrote:
>>>> I have diffed the jack related source between 0.45-5 and 0.46-0.
>>>>
>>>> I see added test on null pointers, but I have questions about a for loop
>>>> mixed with this test, or the fact that some actions are still done where
>>>> there are null buffers.
>>>>
>>>> I will check every block later, and find the problem.
>>>
>>>
>>> there are at least 4 commits that have touched s_audio_jack.c since 0.45-5.
>>> - 5ff6366a5ad030ccd83730a14dcadadce943c352
>>> - ab33f9b23ecd2014ec89d43541b9e4d3c080a424
>>> - 73362cc6a57d89337815023f1ffc4ee0adc1de55
>>> - ffd1a20bc8972c84e164d2266960b5d35275e907
>> OK
>>
>>>
>>> as a first try i would test each of them and see which hs introduced the
>>> problem (and what was the purpose of the patch).
>>>
>>> since Pd crashes, it should be easy to create an automated test whether
>>> the problem is there or not - so you can let "git bisect" do all the
>>> work of finding the problematic commit.
>>
>> I like the word "simple" ;) Will still take a look on how to run git bisect
>>
>>>
>>> gfmasdr
>>> IOhannes
>>
>> Xavier
>>
>> _______________________________________________
>> Pd-list at lists.iem.at mailing list
>> UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list