With advice from jca@
OK jca@ millert@

Based on discussion here https://github.com/libressl-portable/openbsd/pull/82
Suggested fix from jsing@

don't need to use sys/param.h at all, guess which one i believe is
greater namespace polution
ok otto

ok guenther kettenis

- Start with a full page of struct region_info's
- Save an mprotect in the init code: allocate 3 pages with none and
make the middle page r/w instead of a r/w allocation and two calls to make the
guard pages none

junked
- freezero(): only clear requested size

came also up with this diff. Simple, no bias and benchmarks show the extra
random calls disappear in te measurement noise.

but at that time ffs function calls were generated instead of the
compiler inlining the code.  Now that ffs is marked protected in
libc this is handled better.  Thanks to kshe who prompted me to
look at this again.

dance, mark it protected.  This works better for both gcc and clang: gcc
blocks overriding of internal calls, while clang permits inlining again.
ok otto@

some cleanup; tested by sthen@ on a ports build

relevant. from Mark Karpilovskij.
ok millert

cleanups.  From Jan Kokemueller.  OK deraadt@

(w/ _dlctl reference) into static executables.  It's all Mark's code so
put his preferred copyright on it.
ok kettenis@

to control which symbols are exported from the shared library.
ok guenther@, deraadt@, jca@

_libc___cxa_thread_atexit_impl reference on gcc architectures that breaks
the build.

interface is also made available as __cxa_thread_atexit_impl to satisfy the
needs of GNU libstdc++.
ok guenther@, millert@

to builtins like mem{set,cpy,move} and __stack_smash_handler.  So, when
building with clang, instead mark those as protected visibility to get rid
of the PLT relocations.  We can't take the address of them then, but that's
ok: it's a build-time error not a run-time error.
ok kettenis@

with "warning: " since 2003, so the messages themselves need not
contain the prefix anymore.
From Scott Cheloha
ok jca, deraadt

but not all.  From Peter J. Philipp.

which results in an internal double free when internal functions are not
in use.
ok otto@

OK deraadt@ jca@ jmc@

ok tb@ tedu@

Complete in the sense that all POSIX *locale(3) and *_l(3) functions
are included, but in OpenBSD, we of course only really care about
LC_CTYPE and we only support ASCII and UTF-8.
With important help from kettenis@, guenther@, and jca@.
Repeated testing in ports bulk builds by naddy@.
Additional testing by jca@, sebastia@, dcoppa@, and others.
OK kettenis@ dcoppa@, and guenther@ on an earlier version.
Riding guenther@'s libc/librthread major bump.

POSIX 2001 instead of POSIX.1c suggested by millert@ and jmc@, ok jmc@

messages, to avoid pulling in piles of other machinery unnecessarily
problem observed by schwarze@
ok deraadt@ millert@

ok tedu@

found with regress/usr.bin/mandoc/db/dbm_dump;
OK jmc@

definite value in the size == 0 case

is not initialized. Problem spotted by Carlin Bingham; ok phessler@ tedu@

prefix if the character following it is a valid hex char.  The C99
standard is clear that given the string "0xy" zero should be returned
and endptr set to point to the "x".  OK deraadt@ espie@