Fixed in r1601630.

On Tue, Jun 10, 2014 at 3:06 PM, Yann Ylavic <[email protected]> wrote:
> On Tue, Jun 10, 2014 at 3:02 PM, Plüm, Rüdiger, Vodafone Group
> <[email protected]> wrote:
>>
>>
>>> -----Original Message-----
>>> From: Takashi Sato [mailto:[email protected]]
>>> Sent: Dienstag, 10. Juni 2014 14:47
>>> To: [email protected]
>>> Subject: Re: svn commit: r1601291 - in /httpd/httpd/trunk: CHANGES
>>> docs/log-message-tags/next-number modules/proxy/proxy_util.c
>>>
>>> r1601291 causes SEGV.
>>>
>>> # Failed test 2 in t/ssl/proxy.t at line 56
>>> # Failed test 115 in t/ssl/proxy.t at line 56 fail #3
>>> [  error] oh gosh, server dumped core
>>> [  error] for stacktrace, run: gdb /home/st/c99fix/apache/bin/httpd
>>> -core /home/st/c99fix/httpd-test/t/core.3649
>>> [  error] oh nuts, server dumped core again
>>> [  error] for stacktrace, run: gdb /home/st/c99fix/apache/bin/httpd
>>> -core /home/st/c99fix/httpd-test/t/core.3647
>>> t/ssl/proxy.t .......................
>>> Failed 2/172 subtests
>>>
>>> The new function connection_shutdown is a pool cleanup func.
>>> connection_shutdown calls ap_shutdown_conn,
>>> but ap_shutdown_conn finally calls apr_pool_clear!
>>> (see attached strack trace)
>>
>> Hm, very bad. Actually we are trying to clean a subpool of the connection 
>> pool in setaside_remaining_output which
>> has been already destroyed at this point. So I guess we need to have to use 
>> apr_pool_pre_cleanup_register instead of
>> apr_pool_cleanup_register for the connection_shutdown.
>
> I'm testing this, will commit when sure it works.
>
> Regards,
> Yann.

Reply via email to