On Sun, Nov 17, 2002 at 06:06:20PM -0800, Aaron Bannert wrote: > On Sat, Nov 16, 2002 at 06:05:05PM -0800, Brian Pane wrote: > > Intraprocess mutexes have the same problem. Because mutex->owner > > is of type apr_os_thread_t, there's no guarantee that it's small > > enough to be read or written atomically. > > apr_thread_mutex_t will have a problem if mutex->nested can not be > atomically loaded. I can't come up with a good way support nested > mutexes w/o introducing another mutex though, any ideas?
I take that back, you are correct, mutex->owner will be the problem. -aaron
