`nm`: Generate a stable symbol table (Jan-Benedict Glaw)

Release artifacts may contain symbol lists. Binutils's `nm` sorts them by
symbol name, locale based. Inserting a plain `sort` (with `LC_ALL=C`) here
sorts them by address, and (with several symbols pointing to the same address)
sorting by name is also stable (think of embedded '_'.)

Allow individual MDSETTARGETS/MDSET_SUFFIXES to define optional dependencies
between themselves.
Use that to make netbsd-RPI_INSTALL.img depend on netbsd-RPI_INSTALL.bin,
resolving the parallel build race.

Back out previous, it causes build fallout like:
release ===> etc/dreamcast/cdroms/installcd
shle--netbsdelf-install: .WAIT: stat: No such file or directory

Various suffixes for the same kernel image may depend on each
other, so .WAIT after creating one before moving on to the next.
Case at hand: evbarm/installation/instkernel:
netbsd-RPI_INSTALL.bin needs to be complete before netbsd-RPI_INSTALL.img
can be created.
Better solutions welcome...

branches: 1.43.2;
- clean the .tmp kernel targets as well

branches: 1.42.4;
Make MDSET_NOGZIP behave as documented also if MDSET_SUFFIX is set

MDSET_NOGZIP is documented as intended to apply for all kernels, even if
MDSET_NOGZIP.filename is not set.

From Harold Gutch in PR toolchain/52836

Add the ability to specify more sections to strip.

branches: 1.40.6;
Use ${TOOL_GZIP_N} instead of ${TOOL_GZIP} -n, and similarly
use ${TOOL_GZIP_N} -9 instead of ${TOOL_GZIP} -9n, etc.

This will pass -nT instead of just -n if the gzip program is
actually pigz.

branches: 1.39.2;
Use ${TOOL_GZIP} instead of just gzip in all Makefiles
outside */dist/* subdirectories.

When USE_PIGZGZIP=yes, bsd.own.mk sets TOOL_GZIP=${TOOL_PIGZ},
so there's no need to test USE_PIGZGZIP in these Makefiles.

Save space and strip .eh_frame sections for the ramdisks as well.

Fix the ALL_KERNELS default value assignment, so that the double
quotes don't become a part of the value, causing all the targets
to be skipped by default.

build.sh lets you build releases with a limited set of kernels by
using the ALL_KERNELS environment variable to override the arch default.

Fix this mechanism for evbarm by limiting the generated boot images
accordingly.

branches: 1.35.6; 1.35.12;
Document MDSET_NOIMAGE.

Consistently use gzip with the -n flag so dates are not embedded.

Not tested on all platforms.

branches: 1.33.4; 1.33.8;
Implement, and document, an ugly hack which allows cross-building of
NetBSD on Darwin - Mac OS X Leopard - to complete successfully

hack cross-building hack on Darwin
mdate 20 July 2008
who agc
file src/distrib/common/Makefile.mdset : 1.33
descr
Darwin has problems with getopt() when used in mdsetimage -v,
due to the difference between BSD and libiberty() getopt
implementations, more fully described in
http://mail-index.netbsd.org/current-users/2008/06/27/msg003242.html
and the subsequent thread. For just now, we just have an ugly
workaround not to call mdsetimage with the -v argument on Darwin
kcah

Change MACHINE to RELEASEMACHINEDIR

branches: 1.31.2;
Put the GENERIC kernel onto i386 installation CDs and load a kernel module
from the boot loader to provide the miniroot.

branches: 1.30.8; 1.30.10;
Fix bug which causes cats builds to not produce an a.out install kernels,
which are needed to boot from Cyclone firmware.

Bug reported on port-cats mailing list:
http://mail-index.netbsd.org/port-cats/2007/09/08/0000.html

branches: 1.29.4;
After ${_FILENAME} got assigned its value when it's "-", we can
no longer use it as a copy of ${_F} in the subsequent test for
additional targets under the MDSET_SUFFIXES test.

This brings back builds of srec & bin files for evbarm and img
files for evbppc.

branches: 1.28.2; 1.28.4;
Properly set ${_FILENAME} if the _F item of MDSETTARGETS is '-'.
From Jachym Holecek on tech-toolchain.

Support MDSET_POST; apply the operation to every image.

Implement MDSET_NOGZIP & MDSET_NOGZIP.${FILENAME}, to prevent the gzip
of the final image. This is useful if that image has already been
munged & gzipped by MDSET_POST* or MDSET_SUFFIXES*.

Remove the .comment and .ident sections when stripping kernels.
Saves about 4k from all the install floppies

MAKEVERBOSE support

Run mdsetimage on a temporary copy of the kernel, then we retry if/when
it fails.

Add MDSET_NOSTRIP and MDSET_NOSYMBOLS; global versions of the "per ${FILE}"
equivalents that currently exist.

Rename a large chunk of the make(1) variables which refer to a
program/tool from "FOO" to "TOOL_FOO". The new variables are:
TOOL_ASN1_COMPILE TOOL_CAP_MKDB TOOL_CAT TOOL_CKSUM TOOL_COMPILE_ET
TOOL_CONFIG TOOL_CRUNCHGEN TOOL_CTAGS TOOL_DB TOOL_EQN TOOL_FGEN
TOOL_GENCAT TOOL_GROFF TOOL_HEXDUMP TOOL_INDXBIB TOOL_INSTALLBOOT
TOOL_INSTALL_INFO TOOL_M4 TOOL_MAKEFS TOOL_MAKEINFO TOOL_MAKEWHATIS
TOOL_MDSETIMAGE TOOL_MENUC TOOL_MKCSMAPPER TOOL_MKESDB
TOOL_MKLOCALE TOOL_MKMAGIC TOOL_MKTEMP TOOL_MSGC TOOL_MTREE
TOOL_PAX TOOL_PIC TOOL_PREPMKBOOTIMAGE TOOL_PWD_MKDB TOOL_REFER
TOOL_ROFF_ASCII TOOL_ROFF_DVI TOOL_ROFF_HTML TOOL_ROFF_PS
TOOL_ROFF_RAW TOOL_RPCGEN TOOL_SOELIM TOOL_SUNLABEL TOOL_TBL
TOOL_UUDECODE TOOL_VGRIND TOOL_ZIC

For each, provide default in <bsd.sys.mk> of the form:
TOOL_FOO?= foo
and for the ${USETOOLS}=="yes" case in <bsd.own.mk>, provide override:
TOOL_FOO= ${TOOLDIR}/bin/${_TOOL_PREFIX}foo

Document all of these in bsd.README.

This cleans up a chunk of potential (and actual) namespace collision
within our build infrastructure, as well as improves consistency in
the share/mk documentation and provision of appropriate defaults for
each of these variables.

add multiple-include protection.
clean up some comments.

Augment the `KERNEL' field of MDSETIMAGES; if it contains a `/'
character, use that as the path to the kernel to use, rather than
using the default of "${KERNOBJDIR}/${KERNEL}/netbsd".
(Using this feature means that FILENAME == "-" isn't useful. such is life...)

Install release files under "${RELEASEDIR}/${MACHINE}/..." instead of
"${RELEASEDIR}/...".

${RELEASEDIR} is never cleaned , and ${RELEASEDIR}/${MACHINE} is only
cleaned if UPDATE is not defined.

Consistently use "netbsd-XXXXX" instead of "netbsd.XXXXX".

Use _KERNEL.${_K}.${_F} instead of _KERNEL.${_K}, as this allows
MDSETTARGETS to contain the same kernel (_K) more than once, as long
as the filename (_F) is different.

support MDSET_NOSTRIP.${FILENAME}

- whitespace consistency cleanup
- build (mdset/nm/strip) into ${filename} as one rule, and move ${filename}.gz
into separate rule
- rename MDSET.${filename}.nosymbols -> MDSET_NOSYMBOLS.${filename}
- rename MDSET.${filename}.post -> MDSET_POST.${filename}, and don't do
suffix handling here
- rework suffix handling:
- rename MDSET.${filename}.suffixes -> MDSET_SUFFIXES.${filename},
and change from list of suffixes to list of "suffix cmdtobuildsuffix",
where cmdtobuildsuffix is the name of the variable containing the
command to build ${filename}.${suffix} (available as "${.TARGET}").
- for each ${filename}.${suffix}, depend upon ${filename} and use
${cmdtobuildsuffix} to create the former.
- add separate ${filename}.${suffix}.gz rule
- mipsco & pmax: rename "install.gz" to "netbsd-INSTALL.gz", for
consistency and so the target to build the mdset kernel doesn't get
confused with general "install" target

Use release:: instead of release: when multiple release targets may exist

add support for MDSET.${FILENAME}.nosymbols

rather than blocking parallel mdset builds with .WAIT, replace usage of
"netbsd.tmp" with "${.TARGET:R}" (et al)

yet another distrib overhaul:
- move guts of distrib/Makefile.inc to distrib/common/Makefile.distrib
(fixes problem caused by implicit include of ../Makefile.inc in certain
submake conditions triggered by makefiles not yet in tree)
- removed mkdir of ${RELEASEDIR}/*; rely upon "snap_pre" target of
etc/Makefile to create all the release directories
- renamed RELINSTALL to RELEASE_INSTALL
- renamed FLOPPYINSTDIR to FLOPPY_RELEASEDIR
- renamed MDSETDIR to MDSET_RELEASEDIR
- removed ITARGET
- move release target from top level to appropriate subdirectory
- ensure release target has correct depends
- replace miniroot's IMAGE_MD_POST with common/Makefile.image IMAGEPOSTBUILD
- Makefile.image: add realall: ${IMAGE}

replace _SRC_TOP_ with NETBSDSRCDIR, since the latter is the "public" version

Fix kernel suffix processing:
* Don't put .WAIT in ${KERNELS}.
* Make sure to add suffixed kernel names to ${KERNELDEPS}.

Makefile.mdset: add ${KERNELDEPS} variable which is ${KERNELS} with
.WAIT between the items, and remove .WAIT from ${KERNELS}.
Use ${KERNELDEPS} in make dependencies.

improve comments and status messages

- Rename Makefile.mdset variables:
MDEXTRA_SUFFIXES.${KERNEL} -> MDSET.${FILENAME}.suffixes
${KERNEL.${_K}}.post -> MDSET.${FILENAME}.post
This allows "default" rules for for MDSETTARGETS with FILENAMEs of "-"
by defined MDSET.-.post & MDSET.-.suffixes
- Document the above
- Use the above appropriately in various Makefiles
- Be a be quieter when building these kernels
- Add .WAIT syncs so that multiple kernels aren't built in parallel, because
the temporary filename is the same, and it also avoid hairy magic in
syncing the targets generated from MDSET.${FILENAME}.suffixes with
the master target for that ${FILENAME}

Add support for optional extra install kernel suffixes (e.g. .srec
or .bin), provided by the MDEXTRA_SUFFIXES.${kernel} variable.

Makefile.mdset:
replace
TARGETS = KERNEL IMAGE [...]
with
MDTARGETS = KERNEL IMAGE FILENAME [...]
If FILENAME is "-", use "netbsd.${KERNEL}" as the target filename

The rest:
Update for new MDTARGETS syntax. In some cases, the kernel
name from the "old" method will be used to override the default.

document which variables get modified by this fragment

- don't pull in <bsd.prog.mk> in Makefile.mdset; do that in the caller instead
- renamed optional post-strip action variable from ${KERNEL}.post to
netbsd.${KERNEL}.post

Makefile snippet to do the ${MDSETIMAGE} / ${NM} / ${STRIP} / gzip
game on various kernels and fs images.