https://bz.apache.org/bugzilla/show_bug.cgi?id=63516
--- Comment #3 from Joe Orton <[email protected]> --- Is this happening because mod_perl's register_hooks() is running before core.c's register_hooks? Can you break on mod_perl's register_hooks and print both of: saved_server_config_defines ap_server_config_defines at that point? When pconf is cleared reset_config_defines will set ap_server_config_defines to the value of saved_server_config_defines so I assume this is where mod_perl's MODPERL are ending up. The pool handling here looks confusing, ap_server_config_defines is initially allocated out of pcommands (essentially process-long pool) but then init_config_defines() will flip it with the saved copy and allocate a new pconf-lifetime copy. But then reset_config_defines() will flip it back again, which seems unsafe in the way you appear to see here, i.e. we can end up with a copy from a previous pconf. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
