Remove unused code that's confusing when using cscope/opengrok.

branches: 1.11.8;
Remove clause 3 and 4 from TNF licenses

branches: 1.10.4;
- Remove libpthread's atomic ops.
- Remove the old spinlock-based mutex and rwlock implementations.
- Use the atomic ops from libc.

Mutexes:

- Play scrooge again and chop more cycles off acquire/release.
- Spin while the lock holder is running on another CPU (adaptive mutexes).
- Do non-atomic release.

Threadreg:

- Add the necessary hooks to use a thread register.
- Add the code for i386, using %gs.
- Leave i386 code disabled until xen and COMPAT_NETBSD32 have the changes.

For PR bin/37347:

- Override __libc_thr_init() instead of using our own constructor.
- Add pthread__getenv() and use instead of getenv(). This is used before
we are up and running and unfortunatley getenv() takes locks.

Other changes:

- Cache the spinlock vectors in pthread__st. Internal spinlock operations
now take 1 function call instead of 3 (i386).
- Use pthread__self() internally, not pthread_self().
- Use __attribute__ ((visibility("hidden"))) in some places.
- Kill PTHREAD_MAIN_DEBUG.

Fix a dodgy bit of assembly.

- Don't take the mutex's spinlock (ptr_interlock) in pthread_cond_wait().
Instead, make the deferred wakeup list a per-thread array and pass down
the lwpid_t's that way.

- In pthread_cond_wait(), take the mutex before dealing with early wakeup.
In this way there should never be contention on the CV's spinlock if
the app follows POSIX rules (there should only be contention on the
user-provided mutex).

- Add a port of the kernel's rwlocks. The rwlock's spinlock is only taken if
there is contention. This is enabled where atomic ops are available. Right
now that is only i386 and amd64 because I don't have other hardware to
test with. It's trivial to add stubs for other architectures as long as
they have compare-and-swap. When we have proper atomic ops the old rwlock
code can be removed.

- Add a new mutex implementation that's similar to the kernel's mutexes, but
uses compare-and-swap to maintain the waiters list, so no spinlocks are
involved. Same caveats apply as for the rwlocks.

Add: pthread__atomic_cas_ptr, pthread__atomic_swap_ptr, pthread__membar_full
This is a stopgap until the thorpej-atomic branch is complete.

branches: 1.4.14; 1.4.18;
Fix thread context switching to take the stack ABI into account.
From Wolfgang Solfrank.

branches: 1.3.2;
Restore %rax correctly during a full context restore (oops).

netbsd.org->NetBSD.org

libpthread support for x86_64.

Remove unused code that's confusing when using cscope/opengrok.

branches: 1.6.8;
Remove clause 3 and 4 from TNF licenses

branches: 1.5.12;
Remove the PTHREAD_SA option. If M:N threads is reimplemented it's
better off done with a seperate library.

Remove possible race condition in upcall recycling.

Adapt to structure name changes.

Remove PT_SLEEPUC and add PT_TRAPUC.

libpthread support for x86_64.

libpthread: New pthread__smt_wait to put CPU in low power for spin.

This is now distinct from pthread__smt_pause, which is for spin lock
backoff with no paired wakeup.

On Arm, there is a single-bit event register per CPU, and there are two
instructions to manage it:

- wfe, wait for event -- if event register is clear, enter low power
mode and wait until event register is set; then exit low power mode
and clear event register

- sev, signal event -- sets event register on all CPUs (other
circumstances like interrupts also set the event register and cause
wfe to wake)

These can be used to reduce the power consumption of spinning for a
lock, but only if they are actually paired -- if there's no sev, wfe
might hang indefinitely. Currently only pthread_spin(3) actually
pairs them; the other lock primitives (internal lock, mutex, rwlock)
do not -- they have spin lock backoff loops, but no corresponding
wakeup to cancel a wfe.

It may be worthwhile to teach the other lock primitives to pair
wfe/sev, but that requires some performance measurement to verify
it's actually worthwhile. So for now, we just make sure not to use
wfe when there's no sev, and keep everything else the same -- this
should fix severe performance degredation in libpthread on Arm
without hurting anything else.

No change in the generated code on amd64 and i386. No change in the
generated code for pthread_spin.c on arm and aarch64 -- changes only
the generated code for pthread_lock.c, pthread_mutex.c, and
pthread_rwlock.c, as intended.

PR port-arm/57437

XXX pullup-10

branches: 1.12.36; 1.12.46; 1.12.54;
make pthread__sp unsigned long.

branches: 1.11.2;
Remove unused code that's confusing when using cscope/opengrok.

branches: 1.10.8;
Remove clause 3 and 4 from TNF licenses

branches: 1.9.2;
Cheat and add inlines for _atomic_cas_ptr() to work around gcc emitting
unneeded PIC stuff in mutex_lock() and mutex_unlock(), when a thread
register is used.

- Remove libpthread's atomic ops.
- Remove the old spinlock-based mutex and rwlock implementations.
- Use the atomic ops from libc.

Mutexes:

- Play scrooge again and chop more cycles off acquire/release.
- Spin while the lock holder is running on another CPU (adaptive mutexes).
- Do non-atomic release.

Threadreg:

- Add the necessary hooks to use a thread register.
- Add the code for i386, using %gs.
- Leave i386 code disabled until xen and COMPAT_NETBSD32 have the changes.

Resurrect the function pointers for lock operations and allow each
architecture to provide asm versions of the RAS operations.

We do this because relying on the compiler to get the RAS right is not
sensible. (It gets alpha wrong and hppa is suboptimal)

Provide asm RAS ops for hppa.

(A slightly different version) reviewed by Andrew Doran.

Add: pthread__atomic_cas_ptr, pthread__atomic_swap_ptr, pthread__membar_full
This is a stopgap until the thorpej-atomic branch is complete.

branches: 1.4.8; 1.4.12;
Remove leading __ from __(const|inline|signed|volatile) -- it is obsolete.

netbsd.org->NetBSD.org

Catch up with changed struct reg (same as gregs in mcontext now).

libpthread support for x86_64.

Remove the PTHREAD_SA option. If M:N threads is reimplemented it's
better off done with a seperate library.

I was a bit over zealous with my last change so revert the
locked_return_point change.

Loading the instruction at locked_return_point as a return value didn't
work so well.

Thanks to uwe, cube, and dsl.

A couple of fixes to make libpthread really shared, i.e. not have text re-
locations:

- Don't declare pthread__switch_away global
- Do the PIC dance for pthread__switch_return_point and
pthread__locked_switch. Ideally these (and other) symbols would
be hidden.

Thanks to uwe@, dyoung@ and elad@ for help.

XXX sh3 is still to be done.
XXX vax does strange things.

branches: 1.10.2;
s/the the/the/ (only in sources that aren't regularly imported from
elsewhere).

Make register usage more consistent, also in comparison with the i386
version, for easier maintenance.

Correct and simplify computing the return point; just use PC-relative
addressing.

Make sure the stack stays aligned.

Remove possible race condition in upcall recycling.

netbsd.org->NetBSD.org

Adapt to pt_trapuc: change STACK_SWITCH to check for a value in pt_trapuc
and use it preferentially to a value in pt_uc, clearing it once on the new
stack. Move stores into pt_uc back to before the stack switch; storing
after the stack switch opened a one-instruction race condition where an upcall
that had just started a chain could be preempted again, and would bomb when
restarted due to its pt_uc not yet having been updated. Now that pt_trapuc
is what the upcall code writes to, it is safe to store to pt_uc before
switching stacks.

Remove obsolete pt_sleepuc code.

Apply Nathan's switch-away fix.

Continue at the plain switch return point in pthread__switch, not the
locked one, in the !PIC case. From Tor Egge via Havard Eidnes.

libpthread support for x86_64.