On Thu, Jul 19, 2001 at 09:57:24AM +0100, David Reid wrote: > - sms does seem to work, so your comment seems a little strange though very > Justin :)
Try hammering your SMS-enabled/thread-enabled httpd with requests. I get segfaults all over the place (I believe I posted the backtrace of one). Sander and I think the cleanup routines/mechanisms need to be reexamined. I've backed off using the SMS code for now until I have time to examine the cleanup code in detail. Sander has suggested a few places to look. > What problem are we trying to fix here? That's the one that never seems to > be answered... ISTR that we are basically saying that memory allocated > within a thread doesn't have to be locked as it's only available within a > single thread. Well, we have a lot of ways of doing that already... For a thread, I want the allocation parent to be the std SMS (i.e. malloc/free) - this makes the allocation of a thread to be independent of all other SMS (the locking moves into libc where it belongs). Others have asked that there be a separate concept for a "cleanup" parent. When a particular SMS is destroyed (even though it is unrelated hierarchically), the per-thread SMS needs to go away too even if its thread is still active (can't even try to kill the thread). And, you don't want to have a global allocation SMS that all SMS would use. That brings us back to the current pool code. -- justin
