Hi Am 16.08.2013 um 13:23 schrieb Martyn Russell <mar...@lanedo.com>:
> On 14/08/13 15:36, Ralph Böhme wrote:
>>
>> Am 14.08.2013 um 14:33 schrieb Ralph Böhme <r...@netafp.com>:
>>> Looks like we're using malloc() in a signal handler. That must be
>>> l.469 in tracker-extract-pdf.c:
>>>
>>> g_debug ("Parent: Zombie %d reaped", child_pid); ... I'll run
>>> another test with the log statements removed and report back.
>>
>> success, that fixes the issue.
>>
>> <https://github.com/slowfranklin/tracker/commit/88ae53a78b48e39f4fe39b03217d629733d746c0>
>>
>> The change to increase the watchdog timeout in tracker-extract from
>> 10 to (afair) 120 seconds was in place too though, but from looking
>> at the tracker logs the whole process of indexing ~1100 files, a
>> subset of that are the large PDFs I mentioned, only took 58 seconds
>> so I assume fixing the signal handler is all that's needed for now.
>
> Seriously? Removing the debug statement is fixing the issue for you?
yes, it fixes one issue and since adding the fix to my code I haven't run into
any other issue yet.
This one is definitely a bug, you're not supposed to use non async-signal-safe
stuff in a signal handler. As g_debug() calls malloc(), which is not
async-signal-safe, we MUST remove the log statement from the signal handler.
-r
--
Ralph Böhme <r...@netafp.com>
Netatalk Developer | Support | Services
Curslacker Deich 254, 21039 Hamburg, Germany
http://www.netafp.com/
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ tracker-list mailing list tracker-list@gnome.org https://mail.gnome.org/mailman/listinfo/tracker-list
