following glibc use the lowest rank 64bit integer type for ino_t etc.
this is eg. useful for printf format compatibility

The underflow exception is not raised correctly in some
cornercases (see previous fma commit), added comments
with examples for fmaf, fmal and non-x86 fma.

In fmaf store the result before returning so it has the
correct precision when FLT_EVAL_METHOD!=0

1) in downward rounding fma(1,1,-1) should be -0 but it was 0 with
gcc, the code was correct but gcc does not support FENV_ACCESS ON
so it used common subexpression elimination where it shouldn't have.
now volatile memory access is used as a barrier after fesetround.

2) in directed rounding modes there is no double rounding issue
so the complicated adjustments done for nearest rounding mode are
not needed. the only exception to this rule is raising the underflow
flag: assume "small" is an exactly representible subnormal value in
double precision and "verysmall" is a much smaller value so that
	(long double)(small plus verysmall) == small
then
	(double)(small plus verysmall)
raises underflow because the result is an inexact subnormal, but
	(double)(long double)(small plus verysmall)
does not because small is not a subnormal in long double precision
and it is exact in double precision.
now this problem is fixed by checking inexact using fenv when the
result is subnormal

* use unsigned arithmetics
* use unsigned to store arg reduction quotient (so n&3 is understood)
* remove z=0.0 variables, use literal 0
* raise underflow and inexact exceptions properly when x is small
* fix spurious underflow in tanl

patch by Strake. previously is was not feasible to duplicate this
functionality of the functions these were modeled on, since argv[0]
was not saved at program startup, but now that it's available it's
easy to use.

* use unsigned arithmetics on the representation
* store arg reduction quotient in unsigned (so n%2 would work like n&1)
* use different convention to pass the arg reduction bit to __tan
  (this argument used to be 1 for even and -1 for odd reduction
  which meant obscure bithacks, the new n&1 is cleaner)
* raise inexact and underflow flags correctly for small x
  (tanl(x) may still raise spurious underflow for small but normal x)
  (this exception raising code increases codesize a bit, similar fixes
  are needed in many other places, it may worth investigating at some
  point if the inexact and underflow flags are worth raising correctly
  as this is not strictly required by the standard)
* tanf manual reduction optimization is kept for now
* tanl code path is cleaned up to follow similar logic to tan and tanf

there was some question as to how many decimal places to use, since
one decimal place is always sufficient to identify the smallest
denormal uniquely. for now, I'm following the example in the C
standard which is consistent with the other min/max macros we already
had in place.

somehow I missed this when removing the corresponding
__STDC_LIMIT_MACROS and __STDC_CONSTANT_MACROS nonsense from stdint.h.
these were all attempts by the C committee to guess what the C++
committee would want, and the guesses turned out to be wrong.

support for these was recently added to sysmacros.h. note that the
syscall argument is a long, despite dev_t being 64-bit, so on 32-bit
archs the high bits will be lost. it appears the high bits are just
glibc silliness and not part of the kernel api, anyway, but it's nice
that we have them there for future expansion if needed.

When FLT_EVAL_METHOD!=0 (only i386 with x87 fp) the excess
precision of an expression must be removed in an assignment.
(gcc needs -fexcess-precision=standard or -std=c99 for this)

This is done by extra load/store instructions which adds code
bloat when lot of temporaries are used and it makes the result
less precise in many cases.
Using double_t and float_t avoids these issues on i386 and
it makes no difference on other archs.

For now only a few functions are modified where the excess
precision is clearly beneficial (mostly polynomial evaluations
with temporaries).

object size differences on i386, gcc-4.8:
             old   new
__cosdf.o    123    95
__cos.o      199   169
__sindf.o    131    95
__sin.o      225   203
__tandf.o    207   151
__tan.o      605   499
erff.o      1470  1416
erf.o       1703  1649
j0f.o       1779  1745
j0.o        2308  2274
j1f.o       1602  1568
j1.o        2286  2252
tgamma.o    1431  1424
math/*.o   64164 63635

__FLOAT_BITS and __DOUBLE_BITS macros used union compound literals,
now they are changed into static inline functions. A good C compiler
generates the same code for both and the later is C++ conformant.

since CLOCKS_PER_SEC is 1000000 (required by XSI) and the times
syscall reports values in 1/100 second units (Linux), the correct
scaling factor is 10000, not 100. note that only ancient kernels which
lack clock_gettime are affected.

all return values are valid, and on 32-bit systems, values that look
like errors can and will occur. since the only actual error this
function could return is EFAULT, and it is only returnable when the
application has invoked undefined behavior, simply ignore the
possibility that the return value is actually an error code.

there are several reasons for this change. one is getting rid of the
repetition of the syscall signature all over the place. another is
sharing the constant masks without costly GOT accesses in PIC.

the main motivation, however, is accurately representing whether we
want to block signals that might be handled by the application, or all
signals.

use __syscall rather than syscall when failure is not possible or not
to be considered.

they have already blocked signals before decrementing the thread
count, so the code being removed is unreachable in the case where the
thread is no longer counted.

this was simply a case of saving the state in the wrong place.

the previous few commits ended up leaving the thread count and signal
mask wrong for atexit handlers and stdio cleanup.

now that blocking signals prevents any application code from running
while the last thread is exiting, the cas logic is no longer needed to
prevent decrementing below zero.

the thread count (1+libc.threads_minus_1) must always be greater than
or equal to the number of threads which could have application code
running, even in an async-signal-safe sense. there is at least one
dangerous race condition if this invariant fails to hold: dlopen could
allocate too little TLS for existing threads, and a signal handler
running in the exiting thread could claim the allocated TLS for itself
(via __tls_get_addr), leaving too little for the other threads it was
allocated for and thereby causing out-of-bounds access.

there may be other situations where it's dangerous for the thread
count to be too low, particularly in the case where only one thread
should be left, in which case locking may be omitted. however, all
such code paths seem to arise from undefined behavior, since
async-signal-unsafe functions are not permitted to be called from a
signal handler that interrupts pthread_exit (which is itself
async-signal-unsafe).

this change may also simplify logic in __synccall and improve the
chances of making __synccall async-signal-safe.

for the duration of the vm-sharing clone used by posix_spawn, all
signals are blocked in the parent process, including
implementation-internal signals. since __synccall cannot do anything
until successfully signaling all threads, the fact that signals are
blocked automatically yields the necessary safety.

aside from debloating and general simplification, part of the
motivation for removing the explicit lock is to simplify the
synchronization logic of __synccall in hopes that it can be made
async-signal-safe, which is needed to make setuid and setgid, which
depend on __synccall, conform to the standard. whether this will be
possible remains to be seen.

C++11, the first C++ with stdint.h, requires the previously protected
macros to be exposed unconditionally by stdint.h. apparently these
checks were an early attempt by the C committee to guess what the C++
committee would want, and they guessed wrong.

this caused sigsetjmp not to save the signal mask but instead to
clobber it with whatever happened to be in the sigjmb_buf prior to the
call.

this allows /etc/ld-musl-$(ARCH).path to contain one path per line,
which is much more convenient for users than the :-delimited format,
which was a source of repeated and unnecessary confusion. for
simplicity, \n is also accepted in environment variables, though it
should probably not be used there.

at the same time, issues with overly long paths invoking UB or getting
truncated have been fixed. such issues should not have arisen with the
environment (which is size-limited) but could have been generated by a
path file larger than 2**31 bytes in length.

the getifaddrs interface seems to have been invented by glibc, and
they expose socket.h, so for us not to do so is just gratuitous
incompatibility with the interface we're mimicing.

the standard is clear that the old behavior is conforming: "In this
case, [EILSEQ] shall be stored in errno and the conversion state is
undefined."

however, the specification of mbrtowc has one peculiarity when the
source argument is a null pointer: in this case, it's required to
behave as mbrtowc(NULL, "", 1, ps). no motivation is provided for this
requirement, but the natural one that comes to mind is that the intent
is to reset the mbstate_t object. for stateful encodings, such
behavior is actually specified: "If the corresponding wide character
is the null wide character, the resulting state described shall be the
initial conversion state." but in the case of UTF-8 where the
mbstate_t object contains a partially-decoded character rather than a
shift state, a subsequent '\0' byte indicates that the previous
partial character is incomplete and thus an illegal sequence.

naturally, applications using their own mbstate_t object should clear
it themselves after an error, but the standard presently provides no
way to clear the builtin mbstate_t object used when the ps argument is
a null pointer. I suspect this issue may be addressed in the future by
specifying that a null source argument resets the state, as this seems
to have been the intent all along.

for what it's worth, this change also slightly reduces code size.

the interface contract for mbtowc admits a much faster implementation
than mbrtowc can achieve; wrapping mbrtowc with an extra call frame
only made the situation worse.

since the regex implementation uses mbtowc already, this change should
improve regex performance too. it may be possible to improve
performance in other places internally by switching from mbrtowc to
mbtowc.

this simple change, in my measurements, makes about a 7% performance
improvement. at first glance this change would seem like a
compiler-specific hack, since the modified code is not even used.
however, I suspect the reason is that I'm eliminating a second path
into the main body of the code, allowing the compiler more flexibility
to optimize the normal (hot) path into the main body. so even if it
weren't for the measurable (and quite notable) difference in
performance, I think the change makes sense.

SA and SB are used as the lowest and highest valid starter bytes, but
the value of SB was one-past the last valid starter. this caused
access past the end of the state table when the illegal byte '\xf5'
was encountered in a starter position. the error did not show up in
full-character decoding tests, since the bogus state read from just
past the table was unlikely to admit any continuation bytes as valid,
but would have shown up had we tested feeding '\xf5' to the
byte-at-a-time decoding in mbrtowc: it would cause the funtion to
wrongly return -2 rather than -1.

I may eventually go back and remove all references to SA and SB,
replacing them with the values; this would make the code more
transparent, I think. the original motivation for using macros was to
allow misguided users of the code to redefine them for the purpose of
enlarging the set of accepted sequences past the end of Unicode...

also include fallback code for broken kernels that don't support the
flags. as usual, the fallback has a race condition that can leak file
descriptors.