libstdc++: Skip atomic instructions in shared_ptr when both counts are 1

libstdc++: Skip atomic instructions in shared_ptr when both counts are 1

This rewrites _Sp_counted_base::_M_release to skip the two atomic
instructions that decrement each of the use count and the weak count
when both are 1.

Benefits: Save the cost of the last atomic decrements of each of the use
count and the weak count in _Sp_counted_base. Atomic instructions are
significantly slower than regular loads and stores across major
architectures.

How current code works: _M_release() atomically decrements the use
count, checks if it was 1, if so calls _M_dispose(), atomically
decrements the weak count, checks if it was 1, and if so calls
_M_destroy().

How the proposed algorithm works: _M_release() loads both use count and
weak count together atomically (assuming suitable alignment, discussed
later), checks if the value corresponds to a 0x1 value in the individual
count members, and if so calls _M_dispose() and _M_destroy().
Otherwise, it follows the original algorithm.

Why it works: When the current thread executing _M_release() finds each
of the counts is equal to 1, then no other threads could possibly hold
use or weak references to this control block. That is, no other threads
could possibly access the counts or the protected object.

There are two crucial high-level issues that I'd like to point out first:
- Atomicity of access to the counts together
- Proper alignment of the counts together

The patch is intended to apply the proposed algorithm only to the case of
64-bit mode, 4-byte counts, and 8-byte aligned _Sp_counted_base.

** Atomicity **
- The proposed algorithm depends on the mutual atomicity among 8-byte
atomic operations and 4-byte atomic operations on each of the 4-byte halves
of the 8-byte aligned 8-byte block.
- The standard does not guarantee atomicity of 8-byte operations on a pair
of 8-byte aligned 4-byte objects.
- To my knowledge this works in practice on systems that guarantee native
implementation of 4-byte and 8-byte atomic operations.
- __atomic_always_lock_free is used to check for native atomic operations.

** Alignment **
- _Sp_counted_base is an internal base class, with a virtual destructor,
so it has a vptr at the beginning of the class, and will be aligned to
alignof(void*) i.e. 8 bytes.
- The first members of the class are the 4-byte use count and 4-byte
weak count, which will occupy 8 contiguous bytes immediately after the
vptr, i.e. they form an 8-byte aligned 8 byte range.

Other points:
- The proposed algorithm can interact correctly with the current algorithm.
That is, multiple threads using different versions of the code with and
without the patch operating on the same objects should always interact
correctly. The intent for the patch is to be ABI compatible with the
current implementation.
- The proposed patch involves a performance trade-off between saving the
costs of atomic instructions when the counts are both 1 vs adding the cost
of loading the 8-byte combined counts and comparison with {0x1, 0x1}.
- I noticed a big difference between the code generated by GCC vs LLVM. GCC
seems to generate noticeably more code and what seems to be redundant null
checks and branches.
- The patch has been in use (built using LLVM) in a large environment for
many months. The performance gains outweigh the losses (roughly 10 to 1)
across a large variety of workloads.

Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
Co-authored-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:

	* include/bits/c++config (_GLIBCXX_TSAN): Define macro
	indicating that TSan is in use.
	* include/bits/shared_ptr_base.h (_Sp_counted_base::_M_release):
	Replace definition in primary template with explicit
	specializations for _S_mutex and _S_atomic policies.
	(_Sp_counted_base<_S_mutex>::_M_release): New specialization.
	(_Sp_counted_base<_S_atomic>::_M_release): New specialization,
	using a single atomic load to access both reference counts at
	once.
	(_Sp_counted_base::_M_release_last_use): New member function.