Alexander Bergolth wrote:
When I'm using the request-object in a subroutine without having passed
it to the sub as a parameter, the apache child processes die with a segmentation fault when being reused.
Well, it probably shouldn't segfault, but you can't do that. You are creating a closure which will permanently keep a copy of the very first $r that existed in that process, but $r is really an apache structure that gets freed after each request. If you need a copy of $r but don't want to pass it, and you are running in pre-fork MPM, you can use the Apache->request() call.
I've looked at trying to avoid the segfault. I think we can't much do about it. Any object with C guts can be trapped in the closure and contain a pointer to invalid data on subsequent accesses. You can't check whether the pointer to the C object is invalid or not. Since its members all look valid and the memory it used to occupy has been freed already, it may or may not be overtaken by someone else, so there is no reliable test.
Let's say we have added some signature to each object that marks it as destroyed on DESTROY. Then we could check that validity flag in every method attempting to use that object. But it'll add CPU and memory overhead which normally is not needed, since the closure created by Alexander is a bug in his program and not a normal behavior.
I'm quite sure that you will have the same problem with mp1, if you run that closure script on it.
__________________________________________________________________ Stas Bekman JAm_pH ------> Just Another mod_perl Hacker http://stason.org/ mod_perl Guide ---> http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
-- Reporting bugs: http://perl.apache.org/bugs/ Mail list info: http://perl.apache.org/maillist/modperl.html
