ok deraadt

- MATH_ERRNO, MATH_ERREXCEPTION and math_errhandling
- Optional FP_FAST_FMA{,F,L} macros if fma() executes as fast or faster
than (x * y) + z;  which in practice is achievable if gcc implements
__FP_FAST_FMA{,F,L}
Reported by John Marino @ DragonFlyBSD.

The man page was already correct.

prototype to match other prototypes in the file.  OK guenther@ deraadt@

Add timezone and daylight symbols for XSI compatibility.

ok millert@

that should never have been installed.

What's worse, the tzfile.h that gets installed is over 20 years old
and doesn't match the real tzfile.h in libc/time.  This makes the
tree safe for /usr/include/tzfile.h removal.  The TM_YEAR_BASE
define has been moved to time.h temporarily until its usage is
replaced by 1900 in the tree.  Actual removal of tzfile.h is pending
a ports build.  Based on a diff from deraadt@

No actual change, but makes it easier to reuse the code elsewhere.
Suggested by Andre Smagin

builds will have a bogus /usr/include/openssl/openssl as found by naddy. ok jca@

ok jca sthen

OK tedu@ kettenis@

for a couple decades.  Keep the OSIOCGIFCONF ioctl to support COMPAT_LINUX
but move the rest of the Linux-specific ioctl() handling into linux_socket.c
This lets struct osockaddr finally move from sys/socket.h to protocols/talkd.h
ok krw@ deraadt@ mpi@

but safer not to wraparound by accident.
ok deraadt millert miod

superflous '*' after '/*' and adding blank after terminating '$'.
Also eases parsing of the lines by simple awk scripts.
Aesthetic approval from tedu@.

include <sys/param.h> (to operate in legacy mode) or if it wishes to operate
in the POSIX world use <limits.h>
ok guenther millert doug naddy

of the functions prototyped here exist in libc, making it useless.

not be in scope, so hardcode as 256.  Other options have more downside.

This hides the unavoidably cruddy sys/param.h namespace in some programs.
Little impact, because many programs are cruddy and still get it via
other includes.
ports testing thanks to sthen and naddy
ok guenther millert

instead, #define to __builtin_alloca. if that doesn't work, you're hosed.
from an old discussion about the perils of alloca. (don't use alloca!)

when it's really old. but no actual ports fallout in espie's build.
ok deraadt

rewrite the function to be simpler as well. the compiler can unroll the
loop for us if necessary.
ok schwarze

to returning strong random by default, source from arc4random(3).
Parameters to the seeding functions are ignored, and the subsystems remain
in strong random mode.  If you wish the standardized deterministic mode,
call srand_deterministic(), srandom_determistic(), srand48_deterministic(),
seed48_deterministic() or lcong48_deterministic() instead.
The re-entrant functions rand_r(), erand48(), nrand48(), jrand48() are
unaffected by this change and remain in deterministic mode (for now).
Verified as a good roadmap forward by auditing 8800 pieces of software.
Roughly 60 pieces of software will need adaptation to request the
deterministic mode.
Violates POSIX and C89, which violate best practice in this century.
ok guenther tedu millert

add siphash
delete setkey, encrypt, cfree

pwd_gensalt origins, but a string argument works equally work and is more
friendly to consumers beyond local user accounts.
ok deraadt

login_cap_t lives and i don't want to forward declare it in unistd.h

distinguish between LibreSSL (the project) and libressl (the library).
Discussed with many.

(Both NetBSD and FreeBSD provide these prototypes)
ok millert@

version with zero args, so it should only be visible if __BSD_VISIBLE
(and not also for __XPG_VISIBLE).  Contrawise, readlink() has been
part of base POSIX since 1995, so move to proper #if area for that.
Move crypt_checkpass() to the pure-BSD section of the file.
setpgrp() issue noted by Matti Karnaattu (mkarnaattu (at) gmail.com)
ok millert@

discussion, help and ok guenther@

Markus Mueller.

indirectly prodded by krw@

when creating them: mkostemp(), mkostemps(), the 'e' mode letter for
fopen(), freopen(), fdopen(), and popen().  The close-on-exec flag will
be cleared by the action created by posix_spawn_file_actions_adddup2().
Also, add support for the C11 'x' mode letter for fopen() and freopen(),
setting O_EXCL when possibly creating files.
Note: this requires kernel support for pipe2() and dup3()!
ok millert@

when creating them: pipe2(), dup3(), accept4(), MSG_CMSG_CLOEXEC,
SOCK_CLOEXEC.  Includes SOCK_NONBLOCK support.
ok matthew@

diff from frantisek holop (minusf (at) obiit.org)

align with POSIX and other systems.
Pointed out by Elliott Hughes on tech
ok deraadt

required for libc++
ok guenther afresh1

* you can #include <sys/endian.h> instead of <machine/endian.h>,
and ditto <endian.h>  (fixes code that pulls in <sys/endian.h> first)
* those will always export the symbols that POSIX specified for
<endian.h>, including the new {be,le}{16,32,64}toh() set.  c.f.
http://austingroupbugs.net/view.php?id=162
if __BSD_VISIBLE then you also get the symbols that our <machine/endian.h>
currently exports (ntohs, NTOHS, dlg's bemtoh*, etc)
* when doing POSIX compiles (not __BSD_VISIBLE), then <netinet/in.h> and
<arpa/inet.h> will *stop* exporting the extra symbols like BYTE_ORDER
and betoh*
ok deraadt@