[libc] Rename pthread_impl.h to threads_impl.h

Baby step towards separating pthreads out of libc

Change-Id: Ibb1ba6656c796218909f4a3451db9bb66b99fb80

[zx] Magenta -> Zircon

The Great Renaming is here!

Change-Id: I3229bdeb2a3d0e40fb4db6fec8ca7d971fbffb94

[musl] Deduplicate the requeue-ing lock abstraction that condvars use

Change-Id: I95319f957f8190420e1c3ddc2a67f25facaf6ece

[musl][condvars] Name some constants in internal condvar locks

Change-Id: Ieb12cb0d5af5d59c1add1cb80bd990b8e4da8550

[musl][pthread] Remove pthread cancellation machinery

Nothing currently in Fuchsia uses this, nor am I aware of any 2nd or
3rd party code that is likely to. I do not believe that, if someone
were to attempt to truly implement pthread_cancel etc. on Magenta,
that the removal of this code and comments would be a meaningful
impediment.

Change-Id: Ib8435a68d0b053ba3f1d0590640ef54c23e59840

[musl][condvar] Remove unneeded void* casts in condvar code

Change-Id: I2f2bab298a0b7509cb129419b9a5b6b362953882

[musl][atomics] Use C11 atomics in public types

Change-Id: Ie684384c127d881fbeccb1290adce622d368d136

Revert "[musl][atomics] Use C11 atomics in public types"

This reverts commits f2373edae22bc921a41286318243b3985102b510 and
c97bbfb0755f81841769a12a12ba5fbe86144a6b

This breaks libc++ builds:

FAILED: src/CMakeFiles/unwind_objects.dir/libunwind.cpp.o
/b/swarm_slave/w/irzvTVgH/kitchen-workdir/buildtools/toolchain/clang+llvm-x86_64-linux/bin/clang++ -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -I/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libunwind/include -I/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include -stdlib=libc++ -fvisibility-inlines-hidden -Werror=date-time -std=c++11 -fcolor-diagnostics -target x86_64-fuchsia --sysroot=/b/swarm_slave/w/irzvTVgH/kitchen-workdir/out/sysroot/x86_64-fuchsia -std=c++11 -g -Werror=return-type -W -Wall -Wchar-subscripts -Wconversion -Wmismatched-tags -Wmissing-braces -Wnewline-eof -Wno-unused-function -Wshadow -Wshorten-64-to-32 -Wsign-compare -Wsign-conversion -Wstrict-aliasing=2 -Wstrict-overflow=4 -Wunused-parameter -Wunused-variable -Wwrite-strings -Wundef -Wno-error -pedantic -D_DEBUG -D_LIBUNWIND_IS_NATIVE_ONLY -fPIC -fno-exceptions -funwind-tables -I/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include -stdlib=libc++ -fvisibility-inlines-hidden -Werror=date-time -std=c++11 -fcolor-diagnostics -target x86_64-fuchsia --sysroot=/b/swarm_slave/w/irzvTVgH/kitchen-workdir/out/sysroot/x86_64-fuchsia -std=c++11 -fstrict-aliasing -fno-rtti -MD -MT src/CMakeFiles/unwind_objects.dir/libunwind.cpp.o -MF src/CMakeFiles/unwind_objects.dir/libunwind.cpp.o.d -o src/CMakeFiles/unwind_objects.dir/libunwind.cpp.o -c /b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libunwind/src/libunwind.cpp
In file included from /b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libunwind/src/libunwind.cpp:19:
In file included from /b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/algorithm:640:
In file included from /b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/memory:643:
/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/atomic:1713:17: error: address argument to atomic operation must be a pointer to _Atomic type ('volatile bool *' invalid)
{return __c11_atomic_exchange(&__a_, true, __m);}
^ ~~~~~
/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/atomic:1716:17: error: address argument to atomic operation must be a pointer to _Atomic type ('bool *' invalid)
{return __c11_atomic_exchange(&__a_, true, __m);}
^ ~~~~~
/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/atomic:1719:10: error: address argument to atomic operation must be a pointer to _Atomic type ('volatile bool *' invalid)
{__c11_atomic_store(&__a_, false, __m);}
^ ~~~~~
/b/swarm_slave/w/irzvTVgH/kitchen-workdir/third_party/llvm/runtimes/libcxx/include/atomic:1722:10: error: address argument to atomic operation must be a pointer to _Atomic type ('bool *' invalid)
{__c11_atomic_store(&__a_, false, __m);}

Change-Id: I2fba5de2b1abc16f6e2ab3ae11f4b6ee1e70f135

[musl][atomics] Use C11 atomics in public types

For now, use a shim around compare and swap. Followup patches will
rewrite them directly into the C11 style, but for now minimize noise.

Change-Id: Idf3626fea6318b8508d6e65f2634a71e93cf866c

[runtime] Make mxr_mutex more efficient: change from 2-state lock to 3-state

This is a standard 3-state futex-based locking algorithm. See the
comment.

This requires updating the C11 cnd_* (condvar) implementation to
match.

This change means that mxr_mutex_unlock() won't do a futex_wake()
syscall in the uncontended case (when there are no waiting threads).

Also, mxr_mutex_unlock() will now wake only one thread at a time.
Previously, it would wake all waiting threads (though that was
unnecessary, even with a 2-state lock).

This also affects the C11 mtx_* implementation, which uses mxr_mutex.

This is a modified version of e4ae71336ad8b56b53aa5cb385374610b284c6a6
with MG-487 fixed: cnd_timedwait() must do
mxr_mutex_lock_with_waiter() unconditionally to ensure that threads
waiting on the mutex will get woken later.

MG-361

Change-Id: I94733d3a26c05852d34ce5c6571511c3e3e55456

[musl] Add explanatory comments to pthread_cond_timedwait() and cnd_timedwait()

The algorithm used by Musl's condvar implementation is quite complex
and subtle, and the implementation is cryptic with little explanation.

This change adds some explanation of how the algorithm works.

cnd_timedwait() has a duplicate implementation, so I've copied the
comments there too, changing the pthread_*() references to refer to
mxr_mutex_*() and cnd_*() instead.

MG-361

Change-Id: I9023435f6a629a398ccfcf3729b7a9d865ed0a53

Revert "[runtime] Make mxr_mutex more efficient: change from 2-state lock to 3-state"

This reverts commit e4ae71336ad8b56b53aa5cb385374610b284c6a6.

That change introduced a bug in the interaction between the mutex
implementation and the condvar implementation
(third_party/ulib/musl/src/thread/cnd_timedwait.c) which could lead to
missed mutex_unlock wakeups. That could cause c11-condvar-test to
hang.

MG-487 #done

Change-Id: I1519733e66330da27707ebb6b8a8c111f290ca73

[runtime] Make mxr_mutex more efficient: change from 2-state lock to 3-state

This is a standard 3-state futex-based locking algorithm. See the
comment.

This requires updating the C11 cnd_* (condvar) implementation to
match.

This change means that mxr_mutex_unlock() won't do a futex_wake()
syscall in the uncontended case (when there are no waiting threads).

Also, mxr_mutex_unlock() will now wake only one thread at a time.
Previously, it would wake all waiting threads (though that was
unnecessary, even with a 2-state lock).

This also affects the C11 mtx_* implementation, which uses mxr_mutex.

MG-361

Change-Id: I6397bc0b2d4fc6c153d5eb22238e02fc5c8b5d33

[musl] Reenable missing-field-initializers for clang

Change-Id: I375e3a15d156a5c8eb1de4af0e380133ec33595d

[musl] Tweak some types used in futex operations

This means moving a few types and operations to stdatomic.h types.

Change-Id: Ied45a4c49628dbfee55e708d581443bedafec742

[musl][format] Clang format musl

Change-Id: I9d0d2d586a6587dbedda055a3b3664296c7eb89a

[ulib][magenta][runtime][musl] Give system calls _mx_* names, use them in musl

The mx_* name space is not kosher for the standard C library to use
for its own purpose. A conforming standard C program can define its
own functions called mx_* and this must not break the C library.

Give all the system call entry points in the vDSO names starting
with _mx_, with the old mx_ names as aliases. Use only the _mx_
names in runtime and musl.

Change-Id: Id01b57b87c0bb91733e9492159401ff019943822

[musl][cnd_t] Fix cnd_timedwait

The original implementation of cnd_t relied on mtx_t having the same
layout as pthread_mutex_t. However, we have changed mtx_t to be
considerably smaller and simpler, based on mxr_mutex_t. Accordingly,
this patch essentially recreates the cnd_timedwait implementation, with
the internal locking changed. It is essentially the same as the
pthread_cond_timedwait implementation, with some dead code removed due
to the simpler mutex implementation.

Change-Id: If56f3be4547d16a567dfa9cda0954430c126a52c

[magenta] Initial commit