Prefer <elf.h> to the non portable <sys/exec_elf.h>.

ok jca@, deraadt@

Move mutex, condvar, and thread-specific data routes, pthread_once, and
pthread_exit from libpthread to libc, along with low-level bits to
support them. Major bump to both libc and libpthread.

Requested by libressl team. Ports testing by naddy@
ok kettenis@

disable post fork checks for now, too much turbulence in the air

bad things can (and will) happen if a threaded program calls fork() and
then strays off the path to exec(). one common manifestation of this
problem occurs in pthread_join(), so we can add a little check there.
first person to hit this in real life gets to change the error message.

Get rid of ticket support, replace "struct _spinlock" with "_atomic_lock_t".

ok tedu@

Less lock contention by using more pools for mult-threaded programs.
tested by many (thanks!) ok tedu, guenther@

Use a Thread Information Block in both single and multi-threaded programs.
This stores errno, the cancelation flags, and related bits for each thread
and is allocated by ld.so or libc.a. This is an ABI break from 5.9-stable!

Make libpthread dlopen'able by moving the cancelation wrappers into libc
and doing locking and fork/errno handling via callbacks that libpthread
registers when it first initializes. 'errno' *must* be declared via
<errno.h> now!

Clean up libpthread's symbol exports like libc.

On powerpc, offset the TIB/TCB/TLS data from the register per the ELF spec.

Testing by various, particularly sthen@ and patrick@
ok kettenis@

Eliminate the need to explicitly invoke syscalls via their _thread_sys_*
aliases by using a macro REDIRECT_SYSCALL() to map the symbols. Apply
that to getthrid(), sysctl(), and issetugid() as well.

ok mpi@ beck@

Replace the malloc spinlock with a mutex. This lock is held over system calls
which run for many cycles and may even sleep. This leads to other threads
spinning for a long time waiting on the lock. Using a mutex means those
threads go to sleep and get woken up when the lock is released, which results
in a lot less CPU usage. More work is needed to improve the performance of
threaded code that suffers from malloc lock contention, but this diff makes
ports like Firefox significantly more usable.

Tested by many.

ok mpi@, guenther@, tedu@, jca@

ld.so no longer needs or uses a bind lock, so stop setting it. This
eliminates a chunk of complexity from the libpthread init and the fork
wrapper, as it was the bind lock that needed prebinding before use.

Instead of testing for __ELF__ and/or vax, leave out the bits for interfacing
with ld.so locking whenever building NOPIC

pointless use of __ELF__ noted by brad@
ok miod@

In the child after fork, the dl lock has to be forced as its inner spinlock
may have been grabbed by another thread in the parent before the fork

problem report from dcoppa@, ok kettenis@

Make pthread_atfork() track the DSO that called it like atexit() does,
unregistering callbacks if the DSO is unloaded. Move the callback
handling from libpthread to libc, though libpthread still overrides the
inner call to handle locking and thread-library reinitialization.
Major version bump for both libc and libpthread.

verification that this fixes various ports ajacoutot@
asm assistance miod@; ok millert@ deraadt@

Don't try to reuse _initial_thread in the fork() wrapper, as the
thread's existing handle must continue to be valid and it didn't
fully 'change' the thread handle anyway. For pthread_main_np(),
use a new flag, THREAD_ORIGINAL, to indicate that the flagged thread
is the original thread for *this* process.

Fixes some ConsoleKit failures according to aja@

nmask and omask only used on ELF. Noted by todd@

The bind lock must only be grabbed with all signals masked.

backtrace of example case from sthen@
ok kettenis@ millert@

cleanup and consolidate the spinlock_lock (what a name!) code.
it's now atomic_lock to better reflect its usage, and librthread now
features a new spinlock that's really a ticket lock.
thrlseep can handle both types of lock via a flag in the clock arg.
(temp back compat hack)
remove some old stuff that's accumulated along the way and no longer used.
some feedback from dlg, who is concerned with all things ticket lock.
(you need to boot a new kernel before installing librthread)

We want to check that the dynamic linker is available at run-time, so
we should actually check for _DYNAMIC at run-time rather than checking
for __PIC__ at compile time, since the two are actually independent.

Problem and solution identified by guenther; minor tweaks by me.
ok guenther

Test for __PIC__ instead of PIC in preparation for eliminating -DPIC
from bsd.lib.mk for C source files.

ok deraadt, pascal

Move <machine/spinlock.h> into rthread.h; strip out unnecessary #includes

Change threxit() to take a pointer to a pid_t to zero out from the
kernel so that librthread can detect when a thread is completely
done with its stack without need a kqueue. The dying thread moves
itself to a GC list, other threads scan the GC list on pthread_create()
and pthread_join() and free the stack and handle once the thread's
thread id is zeroed.

"get it in" deraadt@, tedu@, cheers by others

After forking, the child is single threaded, so tell libc that. This
is needed to avoid deadlocks in popen() on FILE locking.

ok kurt@

- Add fork/vfork wrapper functions to reset state in the child process.
- Make an effort to protect important libc and ld.so critical areas during
the fork(2)/vfork(2) sys call.
- Add pthread_atfork(3) implementation based on Daniel Eischen's code.

Original diff by Philip Guenther <guenther at gmail.com> with some
additions and refinements by me. Positive test report from brad@ with
many kde apps. fork(2) and pthread_atfork(3) pthread regresses pass.
okay tedu@, kettenis@, marc@