Re: [mp2] Segfaults in mpxs_Apache2__RequestRec_rflush

Hi,

we use apache2-mpm-prefork, can threading issues still be an explanation 
for this kind of segfaults?
It seems rather complicated to backport version 2.0.8 to wheezy on my 
first try.

Regards
  Damian

Am 27.02.2015 09:07, schrieb Steve Hay:
> On 24 February 2015 at 16:53, Damian Lukowski<[email protected]>  wrote:
>> Hi,
>>
>> we are encountering occasional connection disruptions caused by segmentation
>> faults in xs/Apache2/RequestIO/Apache2__RequestIO.h Our application is a
>> PerlResponseHandler based on Catalyst::View::Mason on Debian Wheezy. The
>> segfault seems to always happen in mpxs_Apache2__RequestRec_rflush().
>> Sometimes we have NULL dereference, sometimes dangling pointers:
>>
>>> (gdb) bt
>>> #0  0x00007fd145345728 in mpxs_Apache2__RequestRec_rflush (my_perl=0x7fd1520b3950, items=1, mark=0x7fd1542ed2a0, sp=0x7fd1542ed298)
>>>      at /libapache2-mod-perl2-2.0.7/xs/Apache2/RequestIO/Apache2__RequestIO.h:177
> I would recommend trying mod_perl-2.0.8, and if you are able to test
> things in a development environment separate from your main production
> systems then the SVN trunk would definitely also be worth trying. The
> latter contains some significant changes in the area of threading,
> which I believe fix a number of possible SEGVs.
>