Implement an up to date and cross-platform <elf.h> for userland use.

With this change:

- Elftoolchain's <sys/elfdefinitions.h> is used for the definitions
that make up the ELF gABI and its processor-specific ABI
supplements.

- Existing NetBSD-local definitions are brought in via
the kernel header <sys/exec_elf.h>, for continuity.

This changeset prepares the ground for migrating userland-only
definitions out of the kernel header, for better separation
of concerns.

This changeset also eases the development of cross-architecture
ELF tools since <elf.h> now offers ELF-related constants for all
of the architectures that NetBSD currently supports.

PR lib/59564

Add an implementation of the VIS C API for sparc64.

VIS is an extension of the SPARCv9 instruction set that adds packed
integer vector instructions that operate on 64-bit floating point
registers. It is primarily intended for accelerating common 2D
graphics and multimedia operations.

While GCC includes limited support for VIS via builtins, the builtins
are incomplete and incompatible with the C API described in
"VIS(tm) Instruction Set User's Manual" published by Sun Microsystems
in May 2001. This implementation intends to be 100% compatible with
the one described in the manual.

All sparc64 processors except the HAL SPARC64 (apparently not supported
by NetBSD) support the VIS instruction set. The paper documents
functions up to VIS 2. Post-Oracle CPUs support VIS 3 and VIS 4, but
I don't have access to C API documentation for those (presumably
the code is proprietary to SunPro on Solaris 11), and we can worry
about that when sun4v support is more mature.

The implementation includes some basic tests for packed arithmetic.

PR/58729: Taylor R Campbell: container_of is not documented or available
in userland

PR/58640: Riastradh: Provide sys/stdalign.h

uchar.h: New header file for C11 (and C++11) compliance.

Implementation of the new functions mbrtoc16, c16rtomb, mbrtoc32, and
c32rtomb to come later. Updates for C23 to come later.

PR lib/52374: <uchar.h> missing

branches: 1.146.10; 1.146.12;
Move the basic part of XDR to common/include/rpc and common/lib/libc/rpc.

No functional change intended.

Don't install YP headers when MKYP=no

branches: 1.144.4;
Add endian.h header. Simply includes sys/endian.h

<endian.h> has been accepted by the austin group.

branches: 1.143.6;
Import <stdalign.h>

This header conforms to the C11 standard
Reference: ISO/IEC 9899:201x 7.15 Alignment <stdalign.h>

According to ISO/IEC 9899:201x (draft) 7.15 Alignment <stdalign.h> defines
four macros:
- alignas which expands to _Alignas
- alignof which expands to _Alignof
- __alignas_is_defined and __alignof_is_defined which both expand to 1

The _Alignas declaration appears as one of the type specifiers to modify
the alignment requirement of the object being declared.

The _Alignof operator is used to query the alignment requirement of its
operand type.

ISO/IEC N3242=11-0012 (C++1x) and ISO/IEC N3797 (C++1y) both note a header
<cstdalign> which defines only __alignas_is_defined and shall not define
the alignas macro. It misses the alignof case as it's probably based on an
older C1x draft, which defined only alignas. Assume that this is a bug in
the standard and treat alignof the same way as alignas in C++11.

Allow to define alignas and alignof in C++ prior the C++11 standard. It
might be broken but a nonstandard C++ compiler might support C11-like
_Alignas and _Alignof. Note that it's fatal for g++(1) v.5.4.

Import <stdnoreturn.h>

This header conforms to the C11 standard
Reference: ISO/IEC 9899:201x 7.23 _Noreturn <stdnoreturn.h>

According to ISO/IEC 9899:201x (draft) <stdnoreturn.> defines the
noreturn macro which expands to _Noreturn.

The _Noreturn keyword appears in a function declaration and specifies
that the function does not return by reaching the end of the function
body.

Design choices:
- don't implicitly break C++ code including this header with #error
C++11 offers [[noreturn]] which conflicts with C11 _Noreturn

- don't check for __STDC_VERSION__, everybody is free to reuse it with
a nonstandard compiler not conforming to C11 but supporting _Noreturn

gcc(1) and clang(1) support _Noreturn in -std=c99 and older standards

this follows <stdbool.h> choice for not checking for C99

- follow <stdbool.h> and declare the __noreturn_is_defined guard

- use a standard header guard (_STDNORETURN_H_), similar to <stdbool.h>

Reviewed by <pgoyette> and <joerg>

branches: 1.141.2;
move stdbool.h to sys, reuire pool.h to include <sys/stdbool.h> for bool,
centralize definitions of bool, true, false, to <sys/stdbool.h>

Allow kernel code to access constant databases by moving cdbr(3) and the
required mi_vector_hash(3) into src/common.

branches: 1.139.6;
Add userland part of posix_spawn. Libc functions imported from FreeBSD.
Based on Charles Zhang's summer of code project.

Migrate one last leftover bit (used only by the kernel now) to
sys/ufs/ufs and remove the old quota headers and no-longer-used shared
code. Ok by releng.

Add <quota.h> with the new libquota API.

This is the user/application API and does not include the kernel
interface. Definitions shared with the kernel are found in
<sys/quota.h>.

Since I've been working on a local branch I have taken the liberty of
moving the latest-so-far version of <quota.h> to the front of the
patch queue. This loses (a small amount of) history but has the
advantage of avoiding unnecessary version bumps of libquota.so.

branches: 1.136.2;
Build and install ppath(3), property-list paths library.

Retire varargs.h support. Move machine/stdarg.h logic into MI
sys/stdarg.h and expect compiler to provide proper builtins, defaulting
to the GCC interface. lint still has a special fallback.
Reduce abuse of _BSD_VA_LIST_ by defining __va_list by default and
derive va_list as required by standards.

Add a new libquota library, which contains some blocks to build and/or
parse quota plists; as well as a getfsquota() function to retrieve quotas
for a single id from a single filesystem (whatever filesystem this is:
a local quota-enabled fs or NFS). This is build on functions getufsquota()
(for local filesystems with UFS-like quotas) and getnfsquota();
which are also available to userland programs.
move functions from quota2_subr.c to libquota or libprop as appropriate,
and ajust in-tree quota tools.
move some declarations from kernel headers to either sys/quota.h or
quota/quota.h as appropriate. ufs/ufs/quota.h still installed because
it's needed by other installed ufs headers.
ufs/ufs/quota1.h still installed as a quick&dirty way to get a code
using the old quotactl() to compile (just include ufs/ufs/quota1.h instead of
ufs/ufs/quota.h - old code won't compile without this change and this is
on purpose).
Discussed on tech-kern@ and tech-net@ (long thread, but not much about
libquota itself ...)

Add support for fenv.h interface for i386 and amd64.

Submitted by Stathis Kamperis as part of GSoC 2010 and ported from
FreeBSD.

Add the constant database reader (cdbr(3)) and writer ((cdbw(3)).
They implement a space efficent write-once database with fast access
path. Switch the services(5) database to use cdb. The size of the
database file decreases from 2.1MB disk space to 307KB. Access
performance is about the same if setservent(0) is used and about an
order of magnitude faster otherwise. services_mkdb defaults to the new
format, but can optionally create the old db(3) format as well for
statically linked legacy applications.

branches: 1.131.8; 1.131.16;
/usr/include/atomic.h, for Solaris source compat.

branches: 1.130.6; 1.130.8;
Implementation of POSIX message queues.

Reviewed by: <ad>, <tech-kern>

branches: 1.129.2;
Add C99 complex support, for double and float.
Most complex function implementations are from the "c9x-complex" library,
originating from the "cephes" math library, see
http://www.netlib.org/cephes/, from Stephen L. Moshier, incorporated and
redistributed with the NetBSD license by permission of the author.

Error behaviour and other boundary conditions (branch cuts)
need to be looked at.

For namespace sanity, I've done the rename/weak alias procedure to
most of the exported functions which are also used internally.
Didn't do so for sin/cos(f) yet because assembler implementations use
them directly, and renaming functions shared between the main libm
and the machine specific "overlay" might raise binary compatibility
issues.

Move FORTIFY_SOURCE implementation from the somewhat ill-named "libssp"
(what other systems keep in libssp, we already have in libc) into libc
to match what other systems with FORTIFY_SOURCE do. Goodbye, libssp
dependency in libraries and executables. Discussed with christos and
mrg; Christos will merge the headers to get us the rest of the way to a
FORTIFY_SOURCE implementation that works as others' code expects.

Import of POSIX Asynchronous I/O.
Seems to be quite stable. Some work still left to do.

Please note, that syscalls are not yet MP-safe, because
of the file and vnode subsystems.

Reviewed by: <tech-kern>, <ad>

Sync with changes introduced by merging the newlock2 branch.

branches: 1.125.2; 1.125.4;
kill crypto/rmd160.h and crypto/sha2.h, and instead make symlinks to
/usr/include from /usr/include/sys. This makes all the one way hash
header handling identical.

remove support for building (with) GCC 2.95. also:
- always install <stdbool.h>
- don't generate a fake one for vax / gettext.

Install proplib include files.

retire HAVE_GCC3/HAVE_GCC4 and introduce HAVE_GCC that is set to 2, 3 or 4.

Add Todd Miller's ftw(3) and nftw(3) file-tree walking functionality,
from OpenBSD.

Bump libc minor to 136.

Add FreeBSD's strfmon(3). From Rui Paulo. No bump this time. Take advantage
of the getpwent_r bump.

Nuke threadlib.h. Library defensive threading is now provided through the
normal pthread.h interface.
Move stubs for libc back into a libc-internal header.

DCE 1.1 RPC compatible UUID routines for libc, adapted from FreeBSD.
Encode/decode to big/little endian binary blob routines adapted from
kernel.

PR 23470.

Add wordexp(3). The wordexp function performs shell-style word expansions.
This implementation is wrapper around the undocumented wordexp sh(1)
built-in command.

From FreeBSD.
Provided in PR lib/26123.
Approved by kleink@.

install new files.

Replace symbolic link "/usr/include/varargs.h" (pointing to
"/usr/include/machine/varargs.h") by a stub include file which will
emit an error if GCC 3.3 or newer is used and include "machine/varargs.h"
otherwise.

Based on a suggestion by Richard Earnshaw. This fixes PR toolchain/22888
by myself.

make some files symlinks from sys.

add randomid(3).

introduce an additional switch to enable building GCC3 instead of GCC2:
HAVE_GCC3. if this is set, we also set USE_TOOLS_TOOLCHAIN=no. change
the definition of the former to be restricted to whether tools/toolchain
is used or not.

install stdbool.h if gcc3

Citrus iconv support(4)
add iconv external I/F (iconv.h - iconv_open()/iconv()/iconv_close()) and
iconv(1) utility.
TODO:
- bump libc minor.
- add iconv data files.
- create manpages: iconv_open(3)/iconv(3)/iconv_close(3)/iconv(1)/
mkesdb(1)/mkcsmapper(1).

install mntopts.h

add support for p1003.1b semaphores. From FreeBSD

Expose libc's thread stubs, which get magically turned into real pthread
calls when the pthread library is linked in. Simplifies making other
libraries (such as X's) thread-safe.

Merge the nathanw_sa branch.

minor makefile delint

split /usr/include/sys/sha1.h and /usr/include/sha1.h completely.
future direction: nuke /usr/include/sys/sha1.h, it shouldn't be there as
we don't provide libkern to userland.

This mirrors the same change for md5.h made by itojun on 2000/12/11.

Split the notion of building Hesiod, Kerberos, S/key, and YP
infrastructure and using that infrastructure in programs.

* MKHESIOD, MKKERBEROS, MKSKEY, and MKYP control building
of the infratsructure (libraries, support programs, etc.)

* USE_HESIOD, USE_KERBEROS, USE_SKEY, and USE_YP control
building of support for using the corresponding API
in various libraries/programs that can use it.

As discussed on tech-toolchain.

A utmpx implementation.

Don't install <stdbool.h> for now. Lossage resulting from me testing
with a not-in-tree compiler.

ISO C / SUSv3 <stdbool.h>.

MKfoo=no -> NOfoo=

A <sched.h> file, which provides the clone(2) prototypes, in order
to be compatible with Linux.

* Move definitions of exact-width integer types from <machine/types.h>
to <sys/types.h> and <sys/stdint.h>.
* Add a new C99 <stdint.h> header, which provides integer types of
explicit width, related limits and integer constant macros.
* Extend <inttypes.h> to provide <stdint.h> definitions and format
macros for printf() and scanf().
* Add C99 strtoimax() and strtoumax() functions.
* Use the latter within scanf().
* Add C99 %j, %t and %z printf()/scanf() conversions for
intmax_t, pointer-type and size_t arguments.

Add md2 routines to libc.

branches: 1.95.2;
Don't handle ieeefp.h on vax.

have dummy wctype.h and isw*/tow*. part of libc minor 68 -> 69 crank.
equivalent to is*/to* at this moment. we'll put wchar_t code in the
near future.

split /usr/include/sys/md5.h and /usr/include/md5.h completely.
future direction: nuke /usr/include/sys/md5.h, it shouldn't be there as
we don't provide libkern to userland.

Pull in RIPEMD-160 functions from OpenBSD - this has the same API as our MD4,
MD5 and SHA1 stuff.

branches: 1.91.2;
Merge a bunch of things from crypto-us and crypto-intl into basesrc,
adding support for Heimdal/KTH Kerberos where easy to do so. Eliminate
bsd.crypto.mk.

There is still a bunch more work to do, but crypto is now more-or-less
fully merged into the base NetBSD distribution.

Use a seperate Makefile in the rpc/ subdir. Needed because it generates
a file (rpcb_prot.h).

Add missed include files.

Move rpcb_prot.x to here from libc/rpc.

Fix omission in previous.

New headers for the new RPC code (based on TI-RPC from Sun).

branches: 1.85.2;
* Move <wchar.h> from libcurses to include (for the missing functionality
is to be provided by libc).
* Define wint_t and size_t in <wchar.h> as well.

Add getopt.h and collect all the getopt related stuff here.
Make unistd.h include <getopt.h> to get the getopt stuff.
This adds support for getopt_long from Dieter Baron and Thomas Klausner.

add getifaddrs(3) from bsdi4, as mentioned in tech-net (in SIOCGIFCONF
related thread).
this will save future apps from complexity of SIOCGIFCONF.

getifaddrs(3) does not use SIOCGIFCONF internally, it uses sysctl.

Add necessary support routines for login.conf.

Move <null.h> to <sys/null.h> for the purpose of avoiding conflicts due to
std C headers including it, and a kernel module written in C++ would benfit
as well.

For the benefit of G++'s null pointer constant implementation, #define NULL
as __null with egcs 1.0 (GCC 2.90) and above. As several headers are affected
by this change, move the definition into a new header file, <null.h>, to ease
maintenance.

Add an XSH5 ulimit()/<ulimit.h> implementation.

Add an XSH5 fmtmsg() implementation.

optionally include CRYPTOPATH Makefile.frag files.

Use bsd.crypto.mk.

Move the link.h files for the different shlib formats via repository copy,
and rename them to match the executable format. Introduce a new link.h
file which automatically picks the correct header based on the target of
the toolchain.

convert from NOxxx= to MKxxx=no.
include <bsd.own.mk> if testing a MKxxx variable.

merge nsswitch into main branch

Separate the userland and libsa errno declarations from <sys/errno.h> into
their canonical places.

merge lite-2

Add <libgen.h> header, from XPG4.2.

branches: 1.69.2;
Move md5.h to <sys/md5.h> so that it may be included by the kernel.
A symlink is used to preserve <md5.h> compatibility.

Install files via INCS, and GC the bogus permission fixing hack.

Install md4.h and md5.h

A stringlist implementation (previously ``private'' to libc netgroup code,
but christos ok-ed making public as it's useful)

link.h no longer liver here

install dlfcn.h on the Alpha, since the Alpha (with the ELF shared library
tools) provides the dl*() interfaces. the only remaining special-case header
is the a.out link.h, which should really be installed by the dynamic linker.

branches: 1.63.2;
Use ${INSTALL}.

install sys headers from sys/Makefile.

Make a symlink for poll.h.

install rmt.h

Added new header files for functions in libutil.
Added macro for master passwd lock file.
From Greg Hudson <ghudson@MIT.EDU>.

don't install dlfcn.h or link.h on the Alpha, because they make no
sense for that architecture. (It doesn't have dynamic linking or
shared libraries.)

Use 'includes' instead of 'install' targets to install headers

Add cpio.h, tar.h, and re_comp.h headers for POSIX & XPG4 conformance

Add dlfcn.h and alphabetize.

Follow symlinks when copying .h files. From Greg Hudson.

New header file <ieeefp.h>, which provides a API for getting and setting
IEEE rounding mode, exception enable flags, and exception sticky flags.

For Normative Addendum 1.
We have this cr*p because no one would sa "no" to the danes.

fix tipo

Use and .

Add netgroup.h

new RCS ID format.

make <stdarg.h> a symlink, and clean up ports' stdarg.h and varargs.h files.

branches: 1.46.2;
Simplify copying of sys include files.

get permissions right on installed directories

Use pax instead of two tar's to install ${MACHINE} and ${MACHINE_ARCH}
header files. It's faster, because the file data doesn't have to be
passed through a pipe; and it's cleaner, because pax's rich set of
features allows us to copy the files without changing directories, etc.

branches: 1.43.2;
Fix the install for real. Note this uses pax(1).

add bm header, from Bostic

Added langinfo.h header for nl_langinfo() function

make install work for copies, but hack it to work for symlinks

Added nl_types.h for message catalog functions

oops

update all but ctype.h, dumprestore.h, time.h to 4.4-Lite versions.
USL copyright additions on those. Kill varargs.h, because it can simply
be a link to the machine-dependent version.

user-land signal include file

don't use MACHINE_ARCH if it's dir doesn't exist...

clean things up some

don't hide install -d, and make owner/group correct

regexp.h should still be here, for things needing the compat lib.

fix SYS_INCLUDE definition, and explicitly include bsd.own.mk.
minor cleanup, and add support for creating and/or symlinking
/usr/include/${MACHINE_ARCH} as necessary.

add 'dev' to the list of LDIRS

Install regex.h header.

Added X/Open 1170's <search.h> header --- it only declares bsearch, insque,
and remque at this time. The remainder of the functions will be added as
time allows.

Moved <link.h> here from usr.bin/ld as it is needed by other modules
and utilities (eg. crt0, libc, nm).

install the 2 includes in ./rpcsvc, too.

move rpcsvc files and include generation into src/lib/librpcsvc

Use `:', rather than `.' to separate user and group in chown commands
in case we ever decide to remove the support for the (obsolete) `.'
separator.

Oops. `copies' was getting done before making /usr/include.

There would have to be just one Makefile in the entire tree which referenced
_PROGSUBDIR. GRRRRRR.

Add RCS identifiers.

Allow SYS_INCLUDES={copies,symlinks} choice in bsd.own.mk.

Don't forget to install fnmatch.h.

install incluide files mode 644

install /usr/include/malloc.h!

change for 'arch'

ftp put doesn't cleanly truncate and overwrite a file.

rpc directory installs now.

Set permissions and ownership all at once. (No more directories mode 0777)

Don't forget to install err.h.

changed sys.386bsd to sys

Install mpool.h.

added netccitt, netiso, and netns support, from (currently
alpha) patch 126

make sure directory to install stuff exists before installing

now that we have subdirs, need to do them on install, too...

updated for addition of rpcsvc

install of include subdirs from sys.386bsd now done relative to current
dir, rather than from /sys.

marked dirs netimp, netiso, and netns as "missing" from LDIRS

added "vm" to list of subdirs of /usr/include, and made a minor
cleanup so /usr/include/machine is handled right w/different types
of installs (links vs. copies).

made install copy, rather than symlink includes

branches: 1.1.1;
Initial revision

Repair more fallout from the exec_aout.h change.

No guarantee that these are the "right" things to do, but they do allow
building of port-amd64 again.

Modify the licences of code written by Theo De Raadt from a 4-clause
to a 2-clause licence (retaining UCB clauses (1) and (2)), per PR
22409 from Joel Baker, approved by Theo de Raadt, and ratified by
myself - the only discrepancy being the handling of the original
clause 3 in src/usr.sbin/yppoll/yppoll.c.

Remove Berkeley's advertising clause 3.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

merge lite-2

new RCS ID format.

macros in exec_aout.h now

document the magic number gibble somewhat

Make NMAGIC work again.

Add N_PAGSIZ(), N_TRELOFF(), and N_DRELOFF(), and N_BSSADDR(), to match
SunOS interface.

Moved __LDPGSZ and `struct relocation_info' to <machine/exec.h>.

Added fields to struct relocation_info to accommodate PIC code and
run-time relocations. I've usurped all formerly reserved bits in the process.

sun3 has 8k pages

addition of the pc532 architecture

Add RCS identifiers.

convert magic numbers to network byte order, and attendent changes

Add N_RELOFF().

Add QMAGIC support.

Add support for QMAGIC.

branches: 1.1.1;
Initial revision

fix reversed test from Tobias Nygren.

merge christos-time_t

branches: 1.5.8;
Adjust and thus unify my license.

branches: 1.4.6; 1.4.8;
Unify the license: All rights reserved.
No functional change.

branches: 1.3.4;
Fix the C99 restrict keyword with array defintion in prototype problem.

http://mail-index.netbsd.org/tech-userlevel/2007/06/17/0004.html
Closes PR/36451.

- Add linkage for C++
- KNF

Patch from <ad>, thanks!

Import of POSIX Asynchronous I/O.
Seems to be quite stable. Some work still left to do.

Please note, that syscalls are not yet MP-safe, because
of the file and vnode subsystems.

Reviewed by: <tech-kern>, <ad>

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

new RCS ID format.

update all but ctype.h, dumprestore.h, time.h to 4.4-Lite versions.
USL copyright additions on those. Kill varargs.h, because it can simply
be a link to the machine-dependent version.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

assert.h: Sprinkle __predict_true.

Assertions are always supposed to be true, so this should be a good
prediction.

This matches what we have done for KASSERT in the kernel since the
year 2000.

PR lib/59231: assert.h: missing branch prediction

branches: 1.26.2;
assert.h: remove custom macro definitions for lint mode

In lint mode, 'assert(cond)' expanded to nothing. This broke code like
sqlite that assumes 'assert(cond)' expands to an expression that can be
used as operand of the comma operator, as in 'return assert(i > 0), i'.

Remove the static_assert() fallback for pre-C11 and pre-C++11

C++ without real static_assert() can be incompatible with the C fallback
as presented in openjdk.

A pre-C11 compiler can be picky on the implementation.

As a _NETBSD_SOURCE extension define static_assert for pre-c11/c++99

Proposed on source-changes-d.

Limit static_assert visibility to C11.

The existing definition caused issues as GCC only provides _Static_assert
when building C11 code.
This follows the C standard: static_assert available since C11.

Fixes https://rt.perl.org/Public/Bug/Display.html?id=134023

branches: 1.22.6; 1.22.14;
Add static_assert macro definition in <assert.h>

This declaration conforms to the C11 standard
Reference: ISO/IEC 9899:201x 7.2 Diagnostics <assert.h>

_Static_assert performs compile-time assertion checking.

According to ISO/IEC 9899:201x (draft) 7.2 Diagnostics <assert.h> defines
the static_assert macro which expands to _Static_assert. It's not
conditionalized by NDEBUG like the assert macro.

According to ISO/IEC N3242=11-0012 (C++1x) the <cassert> header shall
define only the assert macro, but not static_assert as it's already part
of the C++11 language.

Allow to define static_assert in C++ prior the C++11 standard. It might be
broken but a nonstandard C++ compiler might support C11-like _Static_assert
feature. Note that it's fatal for g++ 5.4, but it works for clang++ 3.8.1.

Approved by <joerg>.

branches: 1.21.24;
__assert and __assert13 are dead

branches: 1.20.2;
include <sys/null.h> because _DIAGASSERT(foo == NULL) is too common and some
include it.

Define away __assert_function__ if __lint__.

branches: 1.18.26;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Tidy up the namespace: lint -> __lint__.

Remove _assert() again. As has been noted by Ben Harris, this had been
removed already in rev. 1.4 and was subsequently reinstated by the
Lite-2 import. Nothing uses it, and it pollutes user namespace.

branches: 1.15.2;
Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

EBENISANIDIOT: Remove a spurious '#endif' that crept in under the radar.

Namespace cleanup: _assert is in the application's namespace, and thus should
only be exposed if _NETBSD_SOURCE is defined. To make life less complicated,
define _assert() in terms of assert(), since they always seem to have identical
definitions.

_assert() was removed back in 1993, but got reinstated with the 4.4Lite
import. Maybe it should go again. Nothing in the tree uses it.

Employ __static_cast(); suggested by Dave Sainty in PR lib/11766.

branches: 1.11.2;
avoid multiple declarations.

C99: Print the name of the function enclosing the assertion, if possible.

#ifdef __STDC__ -> #if __STDC__

if _DIAGNOSTIC is defined, define _DIAGASSERT() to call __diagassert(),
otherwise it's a no-op.
prototype __diagassert().

Define assert and _assert macros for lint when NDEBUG is present as nothing.

new RCS ID format.

update all but ctype.h, dumprestore.h, time.h to 4.4-Lite versions.
USL copyright additions on those. Kill varargs.h, because it can simply
be a link to the machine-dependent version.

Use _ASSERT_H_ guard macro so that <sys/cdefs.h> is only included once and
the __assert() helper function is declared once if <assert.h> is included
multiple times.
Don't support the _assert() macro, nothing uses it and it pollutes the
user's namespace.

Add RCS identifiers.

Make assert macro ansi complient.

branches: 1.1.1;
Initial revision

branches: 1.1.2; 1.1.6; 1.1.12;
/usr/include/atomic.h, for Solaris source compat.

include/bitstring.h: remove CONSTCOND comment from do-while-0 loop

Lint doesn't need this comment anymore, as there is nothing surprising
or unusual about this idiom.

sed -Ei 's,} *while *\(/\* *CONSTCOND *\*/ *0\),} while (0),' *.h

more casting fixes

appease gcc-5 |= #$@#%

add some const

keep value as an int.

update to size_t

lint friendly

branches: 1.8.44;
remove obsolete "register" declarations

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

merge lite-2

Make macros evaluate to non-compound statements (Jarle Greipsland; PR#3614)

new RCS ID format.

Add RCS identifiers.

fixed, thanks to Mike Murphy <mrm@optigfx.com>, with minor hacks by cgd

branches: 1.1.1;
Initial revision

Needs stddef.h for size_t.

u_char, u_int, u_long -> unsigned char, unsigned, unsigned long

This makes these headers work without _NETBSD_SOURCE.

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

new RCS ID format.

rcs id's

add bm header, from Bostic

Allow kernel code to access constant databases by moving cdbr(3) and the
required mi_vector_hash(3) into src/common.

Add cdbr_open_mem.

branches: 1.1.2; 1.1.8; 1.1.14;
Add the constant database reader (cdbr(3)) and writer ((cdbw(3)).
They implement a space efficent write-once database with fast access
path. Switch the services(5) database to use cdb. The size of the
database file decreases from 2.1MB disk space to 307KB. Access
performance is about the same if setservent(0) is used and about an
order of magnitude faster otherwise. services_mkdb defaults to the new
format, but can optionally create the old db(3) format as well for
statically linked legacy applications.

cdbw(3): Make cdbw_output descr parameter type less confusing.

This is a string of _up to_ 16 bytes, used with strncpy(..., 16).
Specifying `const char descr[16]', while formally equivalent to
`const char *descr' in standard C, now provokes the ire of gcc when
the caller does not provide a buffer of at least 16 bytes.

Prototype for cdbw_stable_seeder.

branches: 1.1.2; 1.1.8; 1.1.10;
Add the constant database reader (cdbr(3)) and writer ((cdbw(3)).
They implement a space efficent write-once database with fast access
path. Switch the services(5) database to use cdb. The size of the
database file decreases from 2.1MB disk space to 307KB. Access
performance is about the same if setservent(0) is used and about an
order of magnitude faster otherwise. services_mkdb defaults to the new
format, but can optionally create the old db(3) format as well for
statically linked legacy applications.

Add missing long double prototypes.

branches: 1.3.6; 1.3.12;
Commit SoC long double support from Stathis Kamperis

branches: 1.2.2;
add a "public domain" note (there is nothing creative justifying a
long license)

Add C99 complex support, for double and float.
Most complex function implementations are from the "c9x-complex" library,
originating from the "cephes" math library, see
http://www.netlib.org/cephes/, from Stephen L. Moshier, incorporated and
redistributed with the NetBSD license by permission of the author.

Error behaviour and other boundary conditions (branch cuts)
need to be looked at.

For namespace sanity, I've done the rename/weak alias procedure to
most of the exported functions which are also used internally.
Didn't do so for sin/cos(f) yet because assembler implementations use
them directly, and renaming functions shared between the main libm
and the machine specific "overlay" might raise binary compatibility
issues.

branches: 1.3.10;
Remove clause 3 and 4 from TNF licenses

branches: 1.2.46;
Fix files using old TNF copyright notice

Add cpio.h, tar.h, and re_comp.h headers for POSIX & XPG4 conformance

Fix visibility of various C99 features, esp. for C++11.

branches: 1.34.28;
Include at least ctype_bits.h, even when not providing the macros.

Do not define the ctype(3) interface via macros for C++.

Add isalpha_l and friends.

branches: 1.31.6; 1.31.12;
more split ctype.h -> sys/ctype_inline.h, sys/ctype_bits.h

1. hide _CTYPE_PRIVATE section in ctype.h, move them to private header ctype_local.h.
2. do not use _CTYPE_NUM_CHARS macro to read data from LC_CTYPE(old BSDCTYPE style) database.
because 1<<CHAR_BIT is MD, so i added MI macro _CTYPE_CACHE_SIZE(1<<8).
3. remove _NB_CACHED_RUNE macro, use _CTYPE_CACHE_SIZE instead.

branches: 1.29.38;
u_intN_t -> uintN_t

Replaced the macro name "notyet" with a 0. There's always the possibility
that a strictly conforming C program uses that name itself, so it had to
be removed. Also put the long comment above the #if 0, because it describes
the #if, not the isblank() macro. Added two blank lines to make the grouping
clear. Approved by kleink.

branches: 1.27.2;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Remove the (int) cast from the array subscript of the ctype #defines.
The cast just stops gcc detecting places where the functions are
passed a 'char' variable - and will index off the array for byte values
0x80 to 0xff.
All (I hope) of the places in the netbsd sources where this was a
potential problem have had an (unsigned char) cast applied.
Christos said he'll fix any fallout I've missed.

branches: 1.25.4;
_ISO_C99_SOURCE -> _ISOC99_SOURCE

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Header file cleanup:
* Sort includes.
* isblank() is in C99, POSIX 2001 and XSH6. Update ifdefs accordingly.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

If _CTYPE_PRIVATE, need to include <machine/limits.h> for CHAR_BIT.

Replace (sizeof (char) << 3) with a less silly-looking CHAR_BIT;
from Peter Seebach in PR standards/9333.

branches: 1.19.2;
have comment about strange use of _B

Remove 'extern' from function declarations.

branches: 1.17.8;
Reorganize name space protection.

cast the argument of ctype macros to (int).

branches: 1.15.2;
Add support for localized character sets (a.k.a. LC_CTYPE).

Thanks go to Matthias Scheler <tron@lyssa.owl.de> for contributing his initial
work in PR/3592, and to Christos Zoulas for refining it!

new RCS ID format.

update all but ctype.h, dumprestore.h, time.h to 4.4-Lite versions.
USL copyright additions on those. Kill varargs.h, because it can simply
be a link to the machine-dependent version.

#ifdef out the isblank macro. Programs that use it will still work, as
they will link in the isblank function.

Added isblank() used by 4.4 and POSIX utilities like tr.

ANSI_SOURCE --> _ANSI_SOURCE

use "_to*_tab_" rather than "_to*_table_" -- table is defined as the former

_ctype_, _tolower_tab_, and _toupper_tab_ are now pointers to the tables.
The tables have been renamed to _C_ctype_, _C_tolower_, and _C_toupper_
as they are tables for the C locale. When switching to a new locale, the
pointers will be set to point to tables specific to the new locale.

zapped extra ; at end of two #define lines

Declare translation tables for toupper and tolower. To be replaced by
pointers to the tables to the current locale.
Reintroduce toupper and tolower macros that use the translation tables.

Use const qualifier with _ctype_ table. Smart compilers can then store it
in the text segment. When we implement locales, the isctype macros/functions
will reference this table (or a locale specific table) through a pointer, but
for right now, it continues to reference the _ctype_ table directly.

Rename tolower & toupper macros to _tolower and _toupper.
Standard C requires tolower to return a character that is !isupper unchanged
which was not being done with the macro. The function version does the
right thing, so the loss of the macro is no great deal.

I didn't eliminate the macros entirely, since X/Open's XPG3 requires _tolower
and _toupper with the same semantics. But, like isascii/toascii, they are
removed from the namespace if either ANSI_SOURCE or _POSIX_SOURCE is defined.

Declare function versions of is* and to* functions -- discovered by
P.J. Plauger's ANSI Library regression test.
Don't export isascii & toascii in strict ansi mode.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

sync with mit/kerberos db

branches: 1.26.8;
add new internal functions

branches: 1.25.2;
use __type_fit() instead of locally brewed macro

branches: 1.24.2; 1.24.6;
Convert db from UCB types to POSIX types. Prefer uint8_t for u_char as
it is mostly about the wire format.

Patched libc.so is bit-identical.

OK matt@

branches: 1.23.12;
Add macros for overflow detection

u_intN_t -> uintN_t

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Remove variable names from prototypes.

do not define types like int32_t or u_int32_t locally.

branches: 1.17.8;
Modify the P_{16,32}_SWAP() macros to avoid unaligned 16 and 32 bit
memory accesses. This fixes a problem seen on the Shark when trying
to use the db(3) functions to access a BIG_ENDIAN btree.

For some reason, the db(3) regression tests fail to spot this problem
on the Shark...

branches: 1.16.2; 1.16.4;
delint.

delint

Change the file mode argument to dbopen() and the db backends from int to
mode_t.

new RCS ID format.

branches: 1.12.2;
clean up import

update to v1.72 of new DB code

forgot the friggin' rcsid last time around.

upgrade new DB code to version 1.7

Add RCS identifiers.

update db code to v1.6 from vangogh, and put it in "db" rather than "DB"

new vangogh db (v1.5)

Merged Charles's enhancements to SWAP macros into new db.h

Took out an include line. Libc should compile now.

Updated header files for libdb.

after 0.2.2 "stable" patches applied

branches: 1.1.1;
Initial revision

scandir(3) and alphasort(3) are part of POSIX-2008, so don't require
_NETBSD_SOURCE to use them.

branches: 1.37.2;
Make fdopendir visible according to POSIX.1 spec.

scandir/alphasort take "const struct dirent **" not "const void *" in
modern unix. since we claim to be 'IEEE Std 1003.1-2008', make it so.

branches: 1.35.14;
comments

branches: 1.34.6;
fix rewinddir on nfs. fix PR/42879 (and probably PR/40229.)

branches: 1.33.2;
version alphasort.

branches: 1.32.2;
merge christos-time_t

Add fdopendir(), from Solaris/Linux.

branches: 1.30.14;
tell about getdents, not getdirentries

branches: 1.29.10;
rename dd_rewind to dd_internal so that it can be re-used to implement
telldir/seekdir/rewinddir differently. While there change the prototype
to telldir to take a non-const DIR, per:
http://www.opengroup.org/onlinepubs/009695399/basedefs/dirent.h.html

PR/33124: Murray Armfield: readdir_r is missing a restrict qualifier.
http://www.opengroup.org/onlinepubs/000095399/basedefs/dirent.h.html

remove functions that we should not be exposing.

expose unlocked versions of readdir and telldir.

Fix opendir30 screwup and sort the compat defs for clarity.

compat code reorg

64 bit inode changes

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

add mutex locking and dirent_r.

Fix things so that _NETBSD_SOURCE correctly exposes the dirfd() and
the insides of struct _dirdesc. Also make the existence of struct
_dirdesc visible to POSIX applications since this is less ugly than
pretending that DIR is a void *.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

make scandir(3)'s 3rd argument take a function that takes a *const*
struct dirent *, rather than non-const. this makes scandir(3) the
same as the scandir implementations in libiberty and glibc, and the
select function has no need to modify the dirent.

Move <null.h> to <sys/null.h> for the purpose of avoiding conflicts due to
std C headers including it, and a kernel module written in C++ would benfit
as well.

For the benefit of G++'s null pointer constant implementation, #define NULL
as __null with egcs 1.0 (GCC 2.90) and above. As several headers are affected
by this change, move the definition into a new header file, <null.h>, to ease
maintenance.

Reorganize name space protection.

oops. Checked in wrong dirent.h version. Add missing getdents prototype
to this one.

Correct comment: getdirentries() -> getdents()

Added missing __seekdir prototype

Changed to unconditionally include <sys/types.h> as required by X/Open XPG4.2.

KERNEL -> _KERNEL

Sync with CSRG.

new RCS ID format.

branches: 1.6.2;
Update to 4.4-Lite version.

Struct dirent definition moved to <sys/dirent.h>.

Expand "u_long" & "u_short" to "unsigned long" and "unsigned short".
Some programs (like NSCA httpd) include <dirent.h> without first including
<sys/types.h> so the u_foo types aren't declared. I don't think POSIX.1
allows us to include <sys/types.h> in <dirent.h> either.

Add RCS identifiers.

Removed bugs and memory leaks from the directory functions.

branches: 1.1.1;
Initial revision

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

constify setdisktab()

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Add '-f disktab' option to disklabel, via lib/6623 from
Greg A. Woods <woods@most.weird.com>. (Very) slightly tweaked disklabel.8

Make many of the new consts dependent on a new `__AUDIT__' flag, to avoid
breaking third party software. *sigh*

const poisoning.

merge lite-2

new RCS ID format.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

Expose dladdr(3) to POSIX.1.2024 applications.

branches: 1.25.18;
Implement __cxa_thread_atexit and __cxa_thread_atexit_impl. This
functions are used for destructors of thread_local objects.

If a pending destructor exists, prevent unloading of shared objects.
Introduce __dl_cxa_refcount interface for this purpose. When the last
reference is gone and the object has been dlclose'd before, the
unloading is finalized.

Ideally, __cxa_thread_atexit_impl wouldn't exist, but libstdc++ insists
on providing __cxa_thread_atexit as direct wrapper without further
patching.

branches: 1.24.24;
Add _dlauxinfo helper function to provide access to the ELF auxilary
vector. This can normally be found via __ps_strings, but libc is
initialised too early when linked dynamically and doesn't have access to
it yet, so provide an alternative mechanism via ld.elf_so.
Bump libc minor.

branches: 1.23.2;
PR/45015: ld.elf_so: support ELF symbol versioning
Applied latest patch.

Add support for DF_1_BIND_NOW, DF_1_NODELETE and DF_1_NOOPEN marked
objects, and the RTLD_NODELETE and RTLD_NOLOAD flags to dlopen(3).

Mark libpthread as DF_1_NOOPEN and use it to test the functionality.

Somewhat taken from FreeBSD.

Fixes PR 42029.

OK from christos and joerg.

Typo in comment.

Add Solarisa-like dlinfo() interface to the ELF dynamic linker.
Implement RTLD_DI_LINKMAP which returns a pointer to the linkmap
chain at the given object. Other Solaris queries are currently
unimplemented.

branches: 1.19.8; 1.19.10; 1.19.16;
Remove clause 3 and 4 from TNF licenses

branches: 1.18.18;
Add restrict qualifiers to dlsym() (XSI) and dladdr() arguments.

branches: 1.17.2;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Remove reference to dlinfo - it doesn't exist in NetBSD.

Add RTLD_SELF, RTLD_NEXT, RTLD_DEFAULT, from FreeBSD

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

branches: 1.13.6;
Remove 'extern' from function declarations.

branches: 1.12.2;
Const'ify dli_saddr in struct _dl_info.

change first argument of dladdr to const.

Declare dlerror()'s return value __aconst.

Assign my copyright to the TNF.

Change ld.so default behavior to RTLD_LOCAL, and make RTLD_LOCAL into a
new flag bit. Keep RTLD_GLOBAL as an actual flag bit so it won't be
re-used, or may be used for something later. Mask the dlopen() mode bits
in ld.elf_so so a comparison against RTLD_NOW works. (RTLD_{LOCAL,GLOBAL}
needs to be implemented for ld.elf_so.)

Add dladdr() and `Dl_info' structure; interface cloned from Solaris 2.

Per XSH98, define the RTLD_LOCAL flag. In this implementation it's a NOP,
since non-local searches have to be explicitly enabled via RTLD_GLOBAL.

Protect the dlctl() prototype and the DL_* macros against _XOPEN_SOURCE;
they are neither specified by the XPG nor part of the name space reserved for
implementation-defined use.

Define RTLD_GLOBAL.

Add `const' to dlopen() & dlsym() protos, per PR#2453.
Add RTLD_NOW.

{BEGIN,END}DECLS

Put dlopen() et.al. declarations in separate header file for compatibility.

Implement an up to date and cross-platform <elf.h> for userland use.

With this change:

- Elftoolchain's <sys/elfdefinitions.h> is used for the definitions
that make up the ELF gABI and its processor-specific ABI
supplements.

- Existing NetBSD-local definitions are brought in via
the kernel header <sys/exec_elf.h>, for continuity.

This changeset prepares the ground for migrating userland-only
definitions out of the kernel header, for better separation
of concerns.

This changeset also eases the development of cross-architecture
ELF tools since <elf.h> now offers ELF-related constants for all
of the architectures that NetBSD currently supports.

PR lib/59564

branches: 1.1.2;
Add endian.h header. Simply includes sys/endian.h

<endian.h> has been accepted by the austin group.

err.h: dead once is dead enough

Per joerg's advice keep the __dead in front position as it is where
_Noreturn would go.

Add the {v,}{warn,err}c function flavors from FreeBSD

branches: 1.16.2; 1.16.8;
Retire varargs.h support. Move machine/stdarg.h logic into MI
sys/stdarg.h and expect compiler to provide proper builtins, defaulting
to the GCC interface. lint still has a special fallback.
Reduce abuse of _BSD_VA_LIST_ by defining __va_list by default and
derive va_list as required by standards.

Consistently use __printflike and __scanflike.

branches: 1.14.44;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

__Namespace__ __protection__.

new RCS ID format.

better attributes.

_VA_LIST_ -> _BSD_VA_LIST_

Change __attribute -> __attribute__. The former isn't turned off by
sys/cdefs.h when not using Gcc, the latter is.

Use GCC's format attribute (for bug detection with -Wformat).

trim _BSD off of _BSD_VA_LIST, so that we don't have to change all
standard headers (yet).

update to latest version

Change volatile qualifier to __dead, for when cc != gcc.

Add RCS identifiers.

s/__dead/volatile/

branches: 1.1.1;
oops

Add a comment noting that the definition of __errno is now replicated in
signal.h. Per request of joerg@

signal.h is not supposed to define errno, so use ___errno instead.
Adjust errno.h in a compatible fashion to avoid -Wredundant-decls issues.
I have run a build to check the tree for places where errno.h is missing,
and fixed the cases I found, but there might be more in MD code.

branches: 1.9.10; 1.9.16;
Remove leading __ from __(const|inline|signed|volatile) -- it is obsolete.

compat code reorg

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

protect against multiple declarations.

Remove 'extern' from function declarations.

branches: 1.2.2;
Do the pthread errno hack. That is, *always* do:
#define errno (*__errno())
This is necessary to make non-threaded libraries (e.g. libX11) continue to
work correctly when linked with -lpthread. glibc also does this.

branches: 1.1.2;
Separate the userland and libsa errno declarations from <sys/errno.h> into
their canonical places.

fexcept_t + fenv_t should be handled the same way for softfloat

provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

Add fenv for ia64, imported from FreeBSD, compile tested only.

branches: 1.23.2;
add sh

add alpha

branches: 1.21.2;
fix inverted condition...

Add missing eol-escape

We need at least 68010 or coldfire for fenv.

add m68k, default types.

rearrange...

Add __HAVE_FENV

Add fenv.h for mips

Powerpc now has fenv.h

fenv(3) support for hppa

RISC-V support <fenv.h>

or1k support <fenv.h>

branches: 1.10.2;
Changes to existing files to enable building AARCH64 userland.
evbarm64-el
This is clang only. While gcc4.8 supports aarch64, no netbsd support has
been written for aarch64 with gcc4.8.

Enable fenv support for arm.

Systematically include sys/featuretest.h when _NETBSD_SOURCE is used.
Some are redundant, but make verification with grep much easier.

branches: 1.7.2;
typo in comment

branches: 1.6.4;
Add fenv support for sparc. Mostly copied from sparc64 and share with it.

Actually test for sparc v9 architecture, doesn't matter if we are building
32bit or 64bit (fixes 32bit compat library build on sparc64 - yes, they
differ from native sparc libs!)

The support of <fenv.h> is actually for __sparc64__, not for __sparc__
(which is also defined on sparc64). This fixes a build problem for
plain sparc.

fenv support for sparc64 from Stathis Kamperis

branches: 1.2.2; 1.2.4;
Missing defined.

Add support for fenv.h interface for i386 and amd64.

Submitted by Stathis Kamperis as part of GSoC 2010 and ported from
FreeBSD.

fix triple s typos in comments.

branches: 1.3.10;
Remove clause 3 and 4 from TNF licenses

branches: 1.2.24;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Add an XSH5 fmtmsg() implementation.

new error.

branches: 1.12.50;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

implement FNM_LEADING_DIR; matches Linux and other *BSDs; approved thorpej

Make FNM_CASEFOLD !_XOPEN_SOURCE, too.

branches: 1.7.2;
Implement FNM_CASEFOLD, for matching the pattern in a case-insensitive
way. Flag name taken from glibc.

branches: 1.6.10;
Add a FNM_NOSYS return code. This condition will never happen (since fnmatch()
is implemented), but this preprocessor symbol is required by XPG4.2.

new RCS ID format.

Remove #ifdefs introduced in last change -- the <fnmatch.h> header is not
specified by 1003.1, so any program that includes it is automatically not
POSIX.1 compliant.

update to latest version; don't proto fnmatch() unless _POSIX_SOURCE
not defined.

Add RCS identifiers.

branches: 1.1.1;
Update fnmatch to be more posix complient (from keith bostic)

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

add a new fstab type "dp" for the user-specified dump device.
change swapctl -A to see this and add it via swapctl(2). also
add a new swapctl -D <device> to change the dump device on the
fly.

Slight rearrangement.

Make many of the new consts dependent on a new `__AUDIT__' flag, to avoid
breaking third party software. *sigh*

const poisoning.

merge lite-2

fs_passno comment dump->fsck, fixes 3150 from Andrew Wheadon

new RCS ID format.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

Version fts one more time! This is the most versioned API ever... Bump
fts_level from short to int to avoid a potential DoS.
From Maksymilian Arciemowicz

merge christos-time_t

branches: 1.17.12;
Use "unsigned foo" instead of "u_foo".
Be consistent with tabs.

branches: 1.16.10; 1.16.16;
fts_number is a quad; bump version.

PR/33163: Auster Vl.: fts: cant allocate memory | filename too long
Change the ftp_pathlen and fts_namelen to u_int from u_short so that
pathnames > 32K work.

compat code reorg

64 bit inode changes

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Make fts_pathlen and fts_nitems unsigned.

branches: 1.9.12;
complete prototype for fts_compar.

Fix __RENAME quoting lossage (don't pass it a quoted string). From
Chris Demetriou <cgd@pa.dec.com>.

New hacks to make libc work painlessly without bumping the major number:
use type func(arg1s) asm("emitted_name") gcc mechanism.
Suggested by Bill Sommerfeld.

branches: 1.6.2;
Hacks to enable libc to work without bumping the major.

Sync with CSRG.

new RCS ID format.

version number update

New version to match new fts.c.

branches: 1.1.1;
Initial revision

Add Todd Miller's ftw(3) and nftw(3) file-tree walking functionality,
from OpenBSD.

Bump libc minor to 136.

branches: 1.11.10;
Remove clause 3 and 4 from TNF licenses

branches: 1.10.6;
revert the reversion of the previous change. Kill _NETBSD_SOURCE again.

Revert the other part of the getopt_long changes to fix non-NetBSD builds;
viz -r1.22 -r1.23 src/lib/libc/stdlib/getopt_long.c

remove featuretest.h and NETBSD_SOURCE conditional ifdefs. This is not
a standard header and does not have namespace issues.

branches: 1.7.16;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Completely rework how tools/compat is done. Purge all uses/references to
_NETBSD_SOURCE as this makes cross building from older/newer versions of
NetBSD harder, not easier (and also makes the resulting tools 'different')

Wrap all required code with the inclusion of nbtool_config.h, attempt to
only use POSIX code in all places (or when reasonable test w. configure and
provide definitions: ala u_int, etc).

Reviewed by lukem. Tested on FreeBSD 4.9, Redhat Linux ES3, NetBSD 1.6.2 x86
NetBSD current (x86 and amd64) and Solaris 9.

Fixes PR's: PR#17762 PR#25944

branches: 1.5.2;
Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

Pull in <sys/cdefs.h>, since we're using __BEGIN_DECLS/__END_DECLS.

Include <unistd.h> to get the rest of the getopt goo.

backout previous. the definition of struct option caused too many namespace
collisions. Instead, define getopt_long in getopt.h and getopt in unistd.h
where it belongs.

Add getopt.h and collect all the getopt related stuff here.
Make unistd.h include <getopt.h> to get the getopt stuff.
This adds support for getopt_long from Dieter Baron and Thomas Klausner.

Add GLOB_TILDE_CHECK (from GNU)

branches: 1.26.46;
add globstar.

branches: 1.25.2;
add glob_pattern_p for glibc because cvs needs it and it is a huge mess
to compile in the gnu glob code.

branches: 1.24.6; 1.24.8;
PR/39534: Henning Petersen: Typo = instead of / to close comment.

branches: 1.23.4;
GLOB_NO_DOTDIRS patch from mouse@
Fixes re-definition of GLOB_PERIOD.

branches: 1.22.4;
From Richard M Kreuter, add GLOB_PERIOD.

branches: 1.21.10;
PR/33123: Murray Armfield: standards compliance & glob.h
Certain fields in glob.h need to be size_t; fix this and version glob(3).
http://www.opengroup.org/onlinepubs/000095399/basedefs/glob.h.html

compat code reorg

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Completely rework how tools/compat is done. Purge all uses/references to
_NETBSD_SOURCE as this makes cross building from older/newer versions of
NetBSD harder, not easier (and also makes the resulting tools 'different')

Wrap all required code with the inclusion of nbtool_config.h, attempt to
only use POSIX code in all places (or when reasonable test w. configure and
provide definitions: ala u_int, etc).

Reviewed by lukem. Tested on FreeBSD 4.9, Redhat Linux ES3, NetBSD 1.6.2 x86
NetBSD current (x86 and amd64) and Solaris 9.

Fixes PR's: PR#17762 PR#25944

branches: 1.17.2;
Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

Partially back out previous: don't apply restrict qualifier to pointer
to function.

Sprinkle some __restrict into <glob.h>.

fix comment.

Add GLOB_LIMIT

branches: 1.11.2; 1.11.10; 1.11.12;
Typo: POSXI -> POSIX.
Re-add GLOB_QUOTE and GLOB_ABEND bracketed by standards ifs for source
compatibility.

* Remove GLOB_QUOTE, add GLOB_NOESCAPE. Per POSIX, backslash quoting of
special characters is the default behaviour, which may be disabled by
setting GLOB_NOESCAPE.
* Add GLOB_NOMATCH, which will be returned if no matching pathnames have been
found and GLOB_NOCHECK was not set.
* Add GLOB_NOSYS for completeness; it will never happen.

Replace the old GLOB_ABEND constant with (the standardized) GLOB_ABORTED.

Fix __RENAME quoting lossage (don't pass it a quoted string). From
Chris Demetriou <cgd@pa.dec.com>.

New hacks to make libc work painlessly without bumping the major number:
use type func(arg1s) asm("emitted_name") gcc mechanism.
Suggested by Bill Sommerfeld.

branches: 1.6.2;
Hacks to enable libc to work without bumping the major.

new RCS ID format.

update to latest version

New versions to go with new fts() et al.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

kill _NETBSD_TOOLS.

Introduce _NETBSD_TOOLS for symbols the are needed in tools build

branches: 1.22.10;
Make namespace protection feature tests for reentrant functions more
accurate (and actually less restrictive).

Add getpwent_r, getgrent_r

branches: 1.20.2;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Overhaul the use of nsdispatch(3) by public APIs so that the back-end
methods use va_list in a manner that is directly related to the public API.
This makes it much easier to write dynamic nsswitch backends for getgrent(3).
Per my proposal on tech-userlevel.

Implement getgrgid_r() and getgrnam_r() APIs per the POSIX 1003.1, 2004 Ed.
These aren't fully reentrant or threadsafe yet, because the compat stuff
currently uses non-reentrant data sources (getnetgrent(3), getgrent(3)),
and there is probably some locking to be improved in the backends.
This will be fixed in the near future.
We also need to add _SC_GETGR_R_SIZE_MAX to sysconf(3).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22270, verified by myself.

Add a new feature-test macro, _NETBSD_SOURCE. If this is defined
by the application, all NetBSD interfaces are made visible, even
if some other feature-test macro (like _POSIX_C_SOURCE) is defined.
<sys/featuretest.h> defined _NETBSD_SOURCE if none of _ANSI_SOURCE,
_POSIX_C_SOURCE and _XOPEN_SOURCE is defined, so as to preserve
existing behaviour.

This has two major advantages:
+ Programs that require non-POSIX facilities but define _POSIX_C_SOURCE
can trivially be overruled by putting -D_NETBSD_SOURCE in their CFLAGS.
+ It makes most of the #ifs simpler, in that they're all now ORs of the
various macros, rather than having checks for (!defined(_ANSI_SOURCE) ||
!defined(_POSIX_C_SOURCE) || !defined(_XOPEN_SOURCE)) all over the place.

I've tried not to change the semantics of the headers in any case where
_NETBSD_SOURCE wasn't defined, but there were some places where the
current semantics were clearly mad, and retaining them was harder than
correcting them. In particular, I've mostly normalised things so that
_ANSI_SOURCE gets you the smallest set of stuff, then _POSIX_C_SOURCE,
_XOPEN_SOURCE and _NETBSD_SOURCE in that order.

Tested by building for vax, encouraged by thorpej, and uncontested in
tech-userlevel for a week.

public prototypes should use __P(); fix entries for pwcache_{user,group}db()

- Implement pwcache_userdb(3), which changes the routines that
user_from_uid(3) and uid_from_user(3) use to lookup user information.
- Implement pwcache_groupdb(3), which changes the routines that
group_from_gid(3) and gid_from_group(3) use to lookup group information.
- Ensure that private functions in pwcache.c are declared static
- Use strlcpy(3) instead of strncpy(3)

branches: 1.14.12;
Add gid_from_group() and uid_from_user().

Mark more functions returning internal buffers for const auditing.

Make many of the new consts dependent on a new `__AUDIT__' flag, to avoid
breaking third party software. *sigh*

const poisoning.

pid/gid is not int but pid_t/gid_t (u_int32_t (unsigned!)).

Reorganize name space protection.

POSIX.1 requires programs to include <sys/types.h> before including
<grp.h> or <utime.h> while XPG4.2 makes <sys/types.h> optional. So
I've included <sys/types.h> to provide the XPG4.2 behavior.

provide group_from_gid() proto if not _POSIX_SOURCE and not _XOPEN_SOURCE

new RCS ID format.

add back jtc's _XOPEN_SOURCE changes, return type of setgrent back to void

update all but ctype.h, dumprestore.h, time.h to 4.4-Lite versions.
USL copyright additions on those. Kill varargs.h, because it can simply
be a link to the machine-dependent version.

X/Open 1170 defines the return type of setgrent() to be void -- a trival
change since our setgrent() allways succeeded.
Removed prototype for obsolete function setgrfile(), added prototype for
setgroupent().
Added _XOPEN_SOURCE to conditionals in case someone wants to compile
something in an strict X/Open environment.

Add RCS identifiers.

branches: 1.1.1;
Initial revision

branches: 1.5.10;
Remove clause 3 and 4 from TNF licenses

branches: 1.4.24;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

merge in stuff from hesiod3.0.2 distribution

merge nsswitch into main branch

branches: 1.1.2; 1.1.4;
file hesiod.h was initially added on branch nsswitch.

Switch the iconv(3) prototype to the POSIX conformant variation

Remove const from the 2nd argument.

const char ** and char ** are incompatible types and it was a cost to keep
the technically incompatible form for a more purist variation. NetBSD was
almost the last alive OS to still keep the const argument (known leftovers:
Minix and Illumos).

Keep the const form for the internal purposes inside citrus and rump.

Address the build breakage fallout in the same change.

There are no ABI changes.

Change accepted by core@.

branches: 1.6.96;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

revert the last all changes related to iconv(3).

make sure that the iconv(3) follows the POSIX specification;
change the 2nd argument from "const char ** restrict" to "char ** restrict".

Sprinkle some restrict, repair a minor nit in iconv(3)'s synopsis.

change iconv_t definision : void * -> struct __tag_iconv_t *.
suggested by yamt.

Citrus iconv support(4)
add iconv external I/F (iconv.h - iconv_open()/iconv()/iconv_close()) and
iconv(1) utility.
TODO:
- bump libc minor.
- add iconv data files.
- create manpages: iconv_open(3)/iconv(3)/iconv_close(3)/iconv(1)/
mkesdb(1)/mkcsmapper(1).

finish the ieeefp.h vs fp_prec_t hack from the previous change and
don't try to define a type when "fp_prec" isn't defined.

this lets my non-x86 (and i presume non-m68k) builds pass again.

add fpgetprec/fpsetprec

PR/32108: Greg Oster: fpgetmask() linking failure with g++ on macppc
Missing __{BEGIN,END}_DECLS.

branches: 1.6.2;
de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

branches: 1.5.10;
Remove 'extern' from function declarations.

branches: 1.4.8;
RCS Id Police.

Move fp_except and fp_rnd definitions to machine/include.

Correct fp_rnd for i386. Add FP_X_* for m68k.

New header file <ieeefp.h>, which provides a API for getting and setting
IEEE rounding mode, exception enable flags, and exception sticky flags.

Add ifa_addrflags to ifaddrs (forgot to commit this file, thanks Ryo!)

branches: 1.7.2;
Apparently a recent round of vandalism by the C++ standards committee
includes disallowing the standard and historic type name "unsigned". Add
a gratuitous "int" to recent changes.

C++ is really a blight on the world.

u_char, u_int, u_long -> unsigned char, unsigned, unsigned long

This makes these headers work without _NETBSD_SOURCE.

de-__P -- the hack is long since useless. Discussed with christos,
matt, kleink, others. Approved by christos.

Remove 'extern' from function declarations.

branches: 1.3.2;
Minor style nit.

add freeifaddrs(), which reclaims region allocated by getifaddrs().
in sync with kame and bsdi.

add getifaddrs(3) from bsdi4, as mentioned in tech-net (in SIOCGIFCONF
related thread).
this will save future apps from complexity of SIOCGIFCONF.

getifaddrs(3) does not use SIOCGIFCONF internally, it uses sysctl.

strtoi and strtou additions

branches: 1.10.6;
Include sys/featuretest.h before using _NETBSD_SOURCE.

Add strtol_l and friends. Switch _citrus_bcs_strtol to use plain
strtol_l unless in tools mode. Add note to retire the BCS code on the
next libc major bump.

Add support for wcstoimax_l and friends.