> On 31 Jan 2002, Jeff Trawick wrote: > > | Dale Ghent <[EMAIL PROTECTED]> writes: > | > | > | > Yeah, I see that, too. I have to add '-lthread -lpthread' to the > | > | > src/Makefile manually. > | > | What is the actual symptom you see if you don't do that? > > I don't know, i've always been adding to LIBS in the Makefile since I > started using HPSA in apache. > > But we seem to only be using pthread_mutex_init(), > pthread_mutex_lock/unlock(), pthread_mutexattr_init(), and > pthread_mutexattr_setpshared(). > > If httpd is linked sans thread libs, then calling these functions should > always return true, but not actually acomplish anything intended.
:-OOO Is that the emoticon for suprise? Astonishment? You mean on Solaris, it is possible to: 1. Sucessfully compile and link a program that requires pthread libraries but the linker doesn;t complain if the pthread library is not present? 2. Said program will run w/o the pthread library linked in and return bogus stuff on pthread calls, rather than segfaulting outright? If this is true, I am completely speechless... Or I must have been kidnapped while I was sleeping and taken to an alternate universe; Do apples fall from trees into the sky here? Does it rain donuts? (Ummm... donuts....) :-) Bill > > Maybe this is the main source of heartache for people who try using HPSA > on Solaris? > > | If a plug-in uses threads and httpd is bound (?) to some dummy thread > | routines in libc, doesn't that screw up the plug-in or confuse libpthread? > | > | Emperical results show that as long as httpd doesn't bind to dummy > | thread routines in libc the plug-in's use of threads is fine. > > That meshes with what I observe with 'lsof' on a HPSA-enabled httpd-1.3.23 > with PHP-4.1.1 - both of which are linked against the thread libs. > > Never tried it with dummies in the parent and real fcns linked into the > plug-in. I don't think I want to :) > > /dale >
