Add virt68k to valid_MACHINE_ARCH.

build.sh: add support for git worktree

Git supports multiple working trees attached to the same repository,
with the git-worktree(1) subcommand.

When building from a Git working tree, the repository metadata in .git
from the top source directory is a file (containing the actual location
of the metadata) instead of the usual directory structure. With this
change, build.sh allows .git to be a directory or a regular file
instead, therefore letting the sources build as expected from a Git
working tree.

To test: (without this patch)

src$ git worktree add ../src-clean-tree
[...]
src$ (cd ../src-clean-tree && ./build.sh tools)
[...]
ERROR: Cannot determine VCS for '.../src-clean-tree/'

XXX: pullup-10 pullup-9

build.sh: use mk.conf (if present) as MAKECONF

Set MAKECONF to mk.conf in the same directory as build.sh
(i.e., the top of the source tree) if mk.conf is present.
This means unprivileged users can benefit from mk.conf(5)
semantics without write privileges to /etc/mk.conf, and
also simplifies per-source-directory configuration.

Fail early if $MAKECONF is empty, since build.sh will fail
anyway after rebuilding nbmake.

Ensure that the computed MAKECONF isn't empty, and then
always set MAKECONF in the makewrapper (nbmake-$MACHINE).

Improve some formatting consistency in BUILDING.mdoc,
(even if BUILDING is generated without markup).

fix build.sh wrong variable use that caused nbmake to be always rebult
(Jan-Benedict Glaw)

build.sh: add show-params operation

Add show-params operation to the makefile target, and reword
params operation to describe what it's done since 2012.

(In 2012-11-04 the Makefile's previous params target was renamed
to show-params and a new params target was added to save a 'params'
file, and build.sh wasn't adapted.)

build.sh: fix MKUNPRIVED comment

I renamed UNPRIVED to MKUNPRIVED 20 years ago, and missed a comment.

build.sh, BUILDING: consistency improvements

Consistently use "show", instead of sometimes using "display" or "print".
Fix mdoc markup (even though BUILDING is raw text).
Document help, -P, -?

build.sh: drop -lrumpdev hack introduced in previous

The existing trick in sys/rump/librump/rumpdev/rump_dev.c was not
updated after the renames in the kernel. kudos to riastradh.

PR kern/56599

build.sh: tweak rumptest and RUMP_LIBSETS

Use --no-whole-archive to keep us as honest as currently possible.
PR kern/56599

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)

handle different flavors of date(1)

for mercurial, use the latest revision instead of limiting the output to 1
(requested by joerg)

Handle git and mercurial for reproducible builds.

ignore cvslatest errors in expert mode

branches: 1.347.2;
Add mipsn64e[bl] targets which use 64 bit binaries instead of n32 ones.

Modify the generated make wrapper to avoid complaints about
non-writable OBJDIR from make (which happen a lot when cross-building
kernels re-using a makewrapper build for some other user).

add "distsets" alias for "distribution sets".

Fix typo in usage

make(1): remove references to the removed lst.lib

New build.sh option: -c <compiler>

Could never remember what the incantation is to do a clang build, so
now it's just `build.sh -c clang'.

Nix trailing whitespace.

nbmake bootstrap: silent configure if MAKEVERBOSE==0

Be consistent with the silencing of configure in tools/
and suppress the output in build.sh configure of nbmake

Add missing MACHINE_ARCH=earmv5hf{,eb} for evbarm.

When MACHINE_ARCH is missing, give direction on how to get a list of valid options

Remove plain "earm" targets, and require an explicit MACHINE_ARCH for
evbarm instead of selecting a default.

Remove evbarm OABI targets

Separate devicetree .dts -> .dtb building from kernel builds. They are now
part of a separate set, "dtb.tgz", and only built when MKDTB=yes. This
defaults to yes for earmv[67]* and aarch64, and no everywhere else.

Allow path components that start with ~

branches: 1.333.2;
Fix long-term broken pattern match when determining if uname -p output is
valid. [^a-z] syntax isn't valid.

on netbsd when defaulting to the native system, also derived MACHINE_ARCH
from the uname -p value already fetched.

this means eg, build.sh on evbarm will build the native platform rather
than "earm" (armv5) -- which was the default even on arm64 hosts.

Tyop in error message...

add aliases for evbarm without saying 'earm':
- copy all the 'evbearm' alias as 'evbarm'.

add build libs (undocumented).

Add some aliases (from phone)

branches: 1.327.2;
Re-add files that were accidentally deleted in my previous commit.

Synch with HEAD

Add initial support for ARMv8 (AARCH64) (by nisimura@ and ryo@)

- sys/arch/evbarm64 is gone and integrated into sys/arch/evbarm. (by skrll@)
- add support fdt. evbarm/conf/GENERIC64 fdt (bcm2837,sunxi,tegra) based generic 64bit kernel config. (by skrll@, jmcneill@)

branches: 1.324.2;
Remove port-acorn26

OK core@

Restore formatting/indentation of the configure call in rebuildmake()
to what it used to be.

Make sure nbmake is always built with the right _PATH_DEFSYSPATH.

Better validation of var name args to -V and -Z, in a way that makes
it trivial to add a list of banned var names for either of those args
should that ever be considered desireable (as the XXX suggests it might.)
I've had this mod locked in my tree for (at least) months - time to set it free.

If we're going xsrc hunting, lets's do it properly. In particular
it is not "good" just because ../xsrc or /usr/xsrc exists, if the user said
-X /no/xsrc/here ... Further, when we find it, remember it.

While here, also look in my favourite place: .../src/xsrc - src/x would be
better (or src/x11) but that's too hard given the cvs module name and src
tarball format.

Simplify.

Accept also ../xsrc in our sanity check, we use this as a default too,
if it exists.

Add sanity check: passed -x but no xsrc directory

branches: 1.316.4;
- Allow all other targets (like kernel=) to obey mkrepro (-P)
- Print the timestamp in UTC

State the reason script bombed.
Reword recommendation.

branches: 1.314.2;
add a trailing / to force symlink resolution.

Bail out on error computing the timestamp

Add -P for reproducible builds based on the latest cvs timestamp.

With driver modulization of if_loop.c through introduction of
sys/net/if_module.h, build.sh rumptest of -lrumpnet_virtif now also
requires -lrumpdev -lrumpvfs for if_loop.c:config_cfdriver_attach()
and kern_pmf.c:do_sys_sync().

branches: 1.310.2;

When setting TOP the idea is to run the [/usr]/bin/pwd binary, and
exclude any skell builtin. There are two ways to guarantee exec'ing
a binary, using a path, or using exec. To use a path we would need to
find it first (which certainly could be done.) Using exec here is easier.

Provide the correct system Makefile path to the tools make. This is needed
because although the arch-specific make wrapper sets $MAKEFLAGS to include
the proper system Makefile path, configure clears $MAKEFLAGS before it invokes
$MAKE to figure out which dependency style to use. This made the build fail
during the gcc build.

Make evbarm64 (little endian) the default for aarch64.

Fix typo in usage. From Christopher M. Fuhrman in PR 49882.

Fix the build kernels target with -n, disable temporarily runcmd while we
are fetching the allkernels variable.

Add a new build make variable, CONFIGOPTS, which is passed to config(1)
when building kernels. Revert "mkernel=*" and "mkernels" in favor of
this option.

Diff posted on source-changes-d@, approved by (null).

build.sh mkernels: Build all kernels in modular build

Expand TABs in usage.

build.sh mkernel=*: New command to build kernel in modular

Replace '.' and '-' in ${op} with '_', before setting d_${op}=true.

Also remove or simplify older code that did the same thing in a
different way. The old code handled most cases, but did not change
op=kernel.gdb to op=kernel_gdb.

build.sh: New operation "kernels" to build all kernels

iso -> images

Prefer dash to bash when searching for a usable shell.
I have tested a complete build with HOST_SH=dash on NetBSD.

Add riscv32 and riscv64 platforms.

Add or1k MACHINE/MACHINE_ARCH.

Only the -m command line option, not MACHINE from the environment,
is supposed override MACHINE_ARCH from the environment with the
default MACHINE_ARCH for the requested machine.

Add and use an opt_m variable to enforce this.

branches: 1.294.2;
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.

MACHINE_ARCH from the environment may be overridden by getarch (based
on the MACHINE), but MACHINE_ARCH from the -a option should not be
overridden. The previous commit accidentally removed this behaviour.

Add a list-arch operation to build.sh.

This lists all known MACHINE/MACHINE_ARCH values, optionally
filtered by glob patterns given in the "-m" and "-a" options.
For example: build.sh -a '*arm*' list-arch

In the usage message, mention that the -m arg may be an alias.
Also adjust a comment around alias processing.

When searching for a default MACHINE_ARCH for a given MACHINE,
don't break out of the loop too early.

Synchronise several shell_quote implementations, and:
* Elide some unnecessary pairs of quotation marks, to improve readability.
For example, shell_quote "''" is now \'\' instead of ''\'''\'''.
* Don't add quotes around words that contain only safe characters,
to improve readability.
* LC_COLLATE=C to prevent [a-zA-Z] from matching non-ASCII characters.
* Use ${SED} if defined.

hpcarm wants earmv4 because some models only have v4 CPUs.

BUILDINFO part 4: build.sh

Add BUILDINFO to the list of variables saved in the makewrapper script,
and print any non-blank BUILDINFO lines near the beginning of the output
from build.sh.
Index: build.sh

build.sh: add support for git worktree

Git supports multiple working trees attached to the same repository,
with the git-worktree(1) subcommand.

When building from a Git working tree, the repository metadata in .git
from the top source directory is a file (containing the actual location
of the metadata) instead of the usual directory structure. With this
change, build.sh allows .git to be a directory or a regular file
instead, therefore letting the sources build as expected from a Git
working tree.

To test: (without this patch)

src$ git worktree add ../src-clean-tree
[...]
src$ (cd ../src-clean-tree && ./build.sh tools)
[...]
ERROR: Cannot determine VCS for '.../src-clean-tree/'

XXX: pullup-10 pullup-9

build.sh: use mk.conf (if present) as MAKECONF

Set MAKECONF to mk.conf in the same directory as build.sh
(i.e., the top of the source tree) if mk.conf is present.
This means unprivileged users can benefit from mk.conf(5)
semantics without write privileges to /etc/mk.conf, and
also simplifies per-source-directory configuration.

Fail early if $MAKECONF is empty, since build.sh will fail
anyway after rebuilding nbmake.

Ensure that the computed MAKECONF isn't empty, and then
always set MAKECONF in the makewrapper (nbmake-$MACHINE).

Improve some formatting consistency in BUILDING.mdoc,
(even if BUILDING is generated without markup).

fix build.sh wrong variable use that caused nbmake to be always rebult
(Jan-Benedict Glaw)

build.sh: add show-params operation

Add show-params operation to the makefile target, and reword
params operation to describe what it's done since 2012.

(In 2012-11-04 the Makefile's previous params target was renamed
to show-params and a new params target was added to save a 'params'
file, and build.sh wasn't adapted.)

build.sh: fix MKUNPRIVED comment

I renamed UNPRIVED to MKUNPRIVED 20 years ago, and missed a comment.

build.sh, BUILDING: consistency improvements

Consistently use "show", instead of sometimes using "display" or "print".
Fix mdoc markup (even though BUILDING is raw text).
Document help, -P, -?

build.sh: drop -lrumpdev hack introduced in previous

The existing trick in sys/rump/librump/rumpdev/rump_dev.c was not
updated after the renames in the kernel. kudos to riastradh.

PR kern/56599

build.sh: tweak rumptest and RUMP_LIBSETS

Use --no-whole-archive to keep us as honest as currently possible.
PR kern/56599

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)

handle different flavors of date(1)

for mercurial, use the latest revision instead of limiting the output to 1
(requested by joerg)

Handle git and mercurial for reproducible builds.

ignore cvslatest errors in expert mode

branches: 1.347.2;
Add mipsn64e[bl] targets which use 64 bit binaries instead of n32 ones.

Modify the generated make wrapper to avoid complaints about
non-writable OBJDIR from make (which happen a lot when cross-building
kernels re-using a makewrapper build for some other user).

add "distsets" alias for "distribution sets".

Fix typo in usage

make(1): remove references to the removed lst.lib

New build.sh option: -c <compiler>

Could never remember what the incantation is to do a clang build, so
now it's just `build.sh -c clang'.

Nix trailing whitespace.

nbmake bootstrap: silent configure if MAKEVERBOSE==0

Be consistent with the silencing of configure in tools/
and suppress the output in build.sh configure of nbmake

Add missing MACHINE_ARCH=earmv5hf{,eb} for evbarm.

When MACHINE_ARCH is missing, give direction on how to get a list of valid options

Remove plain "earm" targets, and require an explicit MACHINE_ARCH for
evbarm instead of selecting a default.

Remove evbarm OABI targets

Separate devicetree .dts -> .dtb building from kernel builds. They are now
part of a separate set, "dtb.tgz", and only built when MKDTB=yes. This
defaults to yes for earmv[67]* and aarch64, and no everywhere else.

Allow path components that start with ~

branches: 1.333.2;
Fix long-term broken pattern match when determining if uname -p output is
valid. [^a-z] syntax isn't valid.

on netbsd when defaulting to the native system, also derived MACHINE_ARCH
from the uname -p value already fetched.

this means eg, build.sh on evbarm will build the native platform rather
than "earm" (armv5) -- which was the default even on arm64 hosts.

Tyop in error message...

add aliases for evbarm without saying 'earm':
- copy all the 'evbearm' alias as 'evbarm'.

add build libs (undocumented).

Add some aliases (from phone)

branches: 1.327.2;
Re-add files that were accidentally deleted in my previous commit.

Synch with HEAD

Add initial support for ARMv8 (AARCH64) (by nisimura@ and ryo@)

- sys/arch/evbarm64 is gone and integrated into sys/arch/evbarm. (by skrll@)
- add support fdt. evbarm/conf/GENERIC64 fdt (bcm2837,sunxi,tegra) based generic 64bit kernel config. (by skrll@, jmcneill@)

branches: 1.324.2;
Remove port-acorn26

OK core@

Restore formatting/indentation of the configure call in rebuildmake()
to what it used to be.

Make sure nbmake is always built with the right _PATH_DEFSYSPATH.

Better validation of var name args to -V and -Z, in a way that makes
it trivial to add a list of banned var names for either of those args
should that ever be considered desireable (as the XXX suggests it might.)
I've had this mod locked in my tree for (at least) months - time to set it free.

If we're going xsrc hunting, lets's do it properly. In particular
it is not "good" just because ../xsrc or /usr/xsrc exists, if the user said
-X /no/xsrc/here ... Further, when we find it, remember it.

While here, also look in my favourite place: .../src/xsrc - src/x would be
better (or src/x11) but that's too hard given the cvs module name and src
tarball format.

Simplify.

Accept also ../xsrc in our sanity check, we use this as a default too,
if it exists.

Add sanity check: passed -x but no xsrc directory

branches: 1.316.4;
- Allow all other targets (like kernel=) to obey mkrepro (-P)
- Print the timestamp in UTC

State the reason script bombed.
Reword recommendation.

branches: 1.314.2;
add a trailing / to force symlink resolution.

Bail out on error computing the timestamp

Add -P for reproducible builds based on the latest cvs timestamp.

With driver modulization of if_loop.c through introduction of
sys/net/if_module.h, build.sh rumptest of -lrumpnet_virtif now also
requires -lrumpdev -lrumpvfs for if_loop.c:config_cfdriver_attach()
and kern_pmf.c:do_sys_sync().

branches: 1.310.2;

When setting TOP the idea is to run the [/usr]/bin/pwd binary, and
exclude any skell builtin. There are two ways to guarantee exec'ing
a binary, using a path, or using exec. To use a path we would need to
find it first (which certainly could be done.) Using exec here is easier.

Provide the correct system Makefile path to the tools make. This is needed
because although the arch-specific make wrapper sets $MAKEFLAGS to include
the proper system Makefile path, configure clears $MAKEFLAGS before it invokes
$MAKE to figure out which dependency style to use. This made the build fail
during the gcc build.

Make evbarm64 (little endian) the default for aarch64.

Fix typo in usage. From Christopher M. Fuhrman in PR 49882.

Fix the build kernels target with -n, disable temporarily runcmd while we
are fetching the allkernels variable.

Add a new build make variable, CONFIGOPTS, which is passed to config(1)
when building kernels. Revert "mkernel=*" and "mkernels" in favor of
this option.

Diff posted on source-changes-d@, approved by (null).

build.sh mkernels: Build all kernels in modular build

Expand TABs in usage.

build.sh mkernel=*: New command to build kernel in modular

Replace '.' and '-' in ${op} with '_', before setting d_${op}=true.

Also remove or simplify older code that did the same thing in a
different way. The old code handled most cases, but did not change
op=kernel.gdb to op=kernel_gdb.

build.sh: New operation "kernels" to build all kernels

iso -> images

Prefer dash to bash when searching for a usable shell.
I have tested a complete build with HOST_SH=dash on NetBSD.

Add riscv32 and riscv64 platforms.

Add or1k MACHINE/MACHINE_ARCH.

Only the -m command line option, not MACHINE from the environment,
is supposed override MACHINE_ARCH from the environment with the
default MACHINE_ARCH for the requested machine.

Add and use an opt_m variable to enforce this.

branches: 1.294.2;
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.

MACHINE_ARCH from the environment may be overridden by getarch (based
on the MACHINE), but MACHINE_ARCH from the -a option should not be
overridden. The previous commit accidentally removed this behaviour.

Add a list-arch operation to build.sh.

This lists all known MACHINE/MACHINE_ARCH values, optionally
filtered by glob patterns given in the "-m" and "-a" options.
For example: build.sh -a '*arm*' list-arch

In the usage message, mention that the -m arg may be an alias.
Also adjust a comment around alias processing.

When searching for a default MACHINE_ARCH for a given MACHINE,
don't break out of the loop too early.

Synchronise several shell_quote implementations, and:
* Elide some unnecessary pairs of quotation marks, to improve readability.
For example, shell_quote "''" is now \'\' instead of ''\'''\'''.
* Don't add quotes around words that contain only safe characters,
to improve readability.
* LC_COLLATE=C to prevent [a-zA-Z] from matching non-ASCII characters.
* Use ${SED} if defined.

hpcarm wants earmv4 because some models only have v4 CPUs.

BUILDINFO part 4: build.sh

Add BUILDINFO to the list of variables saved in the makewrapper script,
and print any non-blank BUILDINFO lines near the beginning of the output
from build.sh.
Index: build.sh

build.sh: use mk.conf (if present) as MAKECONF

Set MAKECONF to mk.conf in the same directory as build.sh
(i.e., the top of the source tree) if mk.conf is present.
This means unprivileged users can benefit from mk.conf(5)
semantics without write privileges to /etc/mk.conf, and
also simplifies per-source-directory configuration.

Fail early if $MAKECONF is empty, since build.sh will fail
anyway after rebuilding nbmake.

Ensure that the computed MAKECONF isn't empty, and then
always set MAKECONF in the makewrapper (nbmake-$MACHINE).

Improve some formatting consistency in BUILDING.mdoc,
(even if BUILDING is generated without markup).

fix build.sh wrong variable use that caused nbmake to be always rebult
(Jan-Benedict Glaw)

build.sh: add show-params operation

Add show-params operation to the makefile target, and reword
params operation to describe what it's done since 2012.

(In 2012-11-04 the Makefile's previous params target was renamed
to show-params and a new params target was added to save a 'params'
file, and build.sh wasn't adapted.)

build.sh: fix MKUNPRIVED comment

I renamed UNPRIVED to MKUNPRIVED 20 years ago, and missed a comment.

build.sh, BUILDING: consistency improvements

Consistently use "show", instead of sometimes using "display" or "print".
Fix mdoc markup (even though BUILDING is raw text).
Document help, -P, -?

build.sh: drop -lrumpdev hack introduced in previous

The existing trick in sys/rump/librump/rumpdev/rump_dev.c was not
updated after the renames in the kernel. kudos to riastradh.

PR kern/56599

build.sh: tweak rumptest and RUMP_LIBSETS

Use --no-whole-archive to keep us as honest as currently possible.
PR kern/56599

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)

handle different flavors of date(1)

for mercurial, use the latest revision instead of limiting the output to 1
(requested by joerg)

Handle git and mercurial for reproducible builds.

ignore cvslatest errors in expert mode

branches: 1.347.2;
Add mipsn64e[bl] targets which use 64 bit binaries instead of n32 ones.

Modify the generated make wrapper to avoid complaints about
non-writable OBJDIR from make (which happen a lot when cross-building
kernels re-using a makewrapper build for some other user).

add "distsets" alias for "distribution sets".

Fix typo in usage

make(1): remove references to the removed lst.lib

New build.sh option: -c <compiler>

Could never remember what the incantation is to do a clang build, so
now it's just `build.sh -c clang'.

Nix trailing whitespace.

nbmake bootstrap: silent configure if MAKEVERBOSE==0

Be consistent with the silencing of configure in tools/
and suppress the output in build.sh configure of nbmake

Add missing MACHINE_ARCH=earmv5hf{,eb} for evbarm.

When MACHINE_ARCH is missing, give direction on how to get a list of valid options

Remove plain "earm" targets, and require an explicit MACHINE_ARCH for
evbarm instead of selecting a default.

Remove evbarm OABI targets

Separate devicetree .dts -> .dtb building from kernel builds. They are now
part of a separate set, "dtb.tgz", and only built when MKDTB=yes. This
defaults to yes for earmv[67]* and aarch64, and no everywhere else.

Allow path components that start with ~

branches: 1.333.2;
Fix long-term broken pattern match when determining if uname -p output is
valid. [^a-z] syntax isn't valid.

on netbsd when defaulting to the native system, also derived MACHINE_ARCH
from the uname -p value already fetched.

this means eg, build.sh on evbarm will build the native platform rather
than "earm" (armv5) -- which was the default even on arm64 hosts.

Tyop in error message...

add aliases for evbarm without saying 'earm':
- copy all the 'evbearm' alias as 'evbarm'.

add build libs (undocumented).

Add some aliases (from phone)

branches: 1.327.2;
Re-add files that were accidentally deleted in my previous commit.

Synch with HEAD

Add initial support for ARMv8 (AARCH64) (by nisimura@ and ryo@)

- sys/arch/evbarm64 is gone and integrated into sys/arch/evbarm. (by skrll@)
- add support fdt. evbarm/conf/GENERIC64 fdt (bcm2837,sunxi,tegra) based generic 64bit kernel config. (by skrll@, jmcneill@)

branches: 1.324.2;
Remove port-acorn26

OK core@

Restore formatting/indentation of the configure call in rebuildmake()
to what it used to be.

Make sure nbmake is always built with the right _PATH_DEFSYSPATH.

Better validation of var name args to -V and -Z, in a way that makes
it trivial to add a list of banned var names for either of those args
should that ever be considered desireable (as the XXX suggests it might.)
I've had this mod locked in my tree for (at least) months - time to set it free.

If we're going xsrc hunting, lets's do it properly. In particular
it is not "good" just because ../xsrc or /usr/xsrc exists, if the user said
-X /no/xsrc/here ... Further, when we find it, remember it.

While here, also look in my favourite place: .../src/xsrc - src/x would be
better (or src/x11) but that's too hard given the cvs module name and src
tarball format.

Simplify.

Accept also ../xsrc in our sanity check, we use this as a default too,
if it exists.

Add sanity check: passed -x but no xsrc directory

branches: 1.316.4;
- Allow all other targets (like kernel=) to obey mkrepro (-P)
- Print the timestamp in UTC

State the reason script bombed.
Reword recommendation.

branches: 1.314.2;
add a trailing / to force symlink resolution.

Bail out on error computing the timestamp

Add -P for reproducible builds based on the latest cvs timestamp.

With driver modulization of if_loop.c through introduction of
sys/net/if_module.h, build.sh rumptest of -lrumpnet_virtif now also
requires -lrumpdev -lrumpvfs for if_loop.c:config_cfdriver_attach()
and kern_pmf.c:do_sys_sync().

branches: 1.310.2;

When setting TOP the idea is to run the [/usr]/bin/pwd binary, and
exclude any skell builtin. There are two ways to guarantee exec'ing
a binary, using a path, or using exec. To use a path we would need to
find it first (which certainly could be done.) Using exec here is easier.

Provide the correct system Makefile path to the tools make. This is needed
because although the arch-specific make wrapper sets $MAKEFLAGS to include
the proper system Makefile path, configure clears $MAKEFLAGS before it invokes
$MAKE to figure out which dependency style to use. This made the build fail
during the gcc build.

Make evbarm64 (little endian) the default for aarch64.

Fix typo in usage. From Christopher M. Fuhrman in PR 49882.

Fix the build kernels target with -n, disable temporarily runcmd while we
are fetching the allkernels variable.

Add a new build make variable, CONFIGOPTS, which is passed to config(1)
when building kernels. Revert "mkernel=*" and "mkernels" in favor of
this option.

Diff posted on source-changes-d@, approved by (null).

build.sh mkernels: Build all kernels in modular build

Expand TABs in usage.

build.sh mkernel=*: New command to build kernel in modular

Replace '.' and '-' in ${op} with '_', before setting d_${op}=true.

Also remove or simplify older code that did the same thing in a
different way. The old code handled most cases, but did not change
op=kernel.gdb to op=kernel_gdb.

build.sh: New operation "kernels" to build all kernels

iso -> images

Prefer dash to bash when searching for a usable shell.
I have tested a complete build with HOST_SH=dash on NetBSD.

Add riscv32 and riscv64 platforms.

Add or1k MACHINE/MACHINE_ARCH.

Only the -m command line option, not MACHINE from the environment,
is supposed override MACHINE_ARCH from the environment with the
default MACHINE_ARCH for the requested machine.

Add and use an opt_m variable to enforce this.

branches: 1.294.2;
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.

MACHINE_ARCH from the environment may be overridden by getarch (based
on the MACHINE), but MACHINE_ARCH from the -a option should not be
overridden. The previous commit accidentally removed this behaviour.

Add a list-arch operation to build.sh.

This lists all known MACHINE/MACHINE_ARCH values, optionally
filtered by glob patterns given in the "-m" and "-a" options.
For example: build.sh -a '*arm*' list-arch

In the usage message, mention that the -m arg may be an alias.
Also adjust a comment around alias processing.

When searching for a default MACHINE_ARCH for a given MACHINE,
don't break out of the loop too early.

Synchronise several shell_quote implementations, and:
* Elide some unnecessary pairs of quotation marks, to improve readability.
For example, shell_quote "''" is now \'\' instead of ''\'''\'''.
* Don't add quotes around words that contain only safe characters,
to improve readability.
* LC_COLLATE=C to prevent [a-zA-Z] from matching non-ASCII characters.
* Use ${SED} if defined.

hpcarm wants earmv4 because some models only have v4 CPUs.

BUILDINFO part 4: build.sh

Add BUILDINFO to the list of variables saved in the makewrapper script,
and print any non-blank BUILDINFO lines near the beginning of the output
from build.sh.
Index: build.sh

fix build.sh wrong variable use that caused nbmake to be always rebult
(Jan-Benedict Glaw)

build.sh: add show-params operation

Add show-params operation to the makefile target, and reword
params operation to describe what it's done since 2012.

(In 2012-11-04 the Makefile's previous params target was renamed
to show-params and a new params target was added to save a 'params'
file, and build.sh wasn't adapted.)

build.sh: fix MKUNPRIVED comment

I renamed UNPRIVED to MKUNPRIVED 20 years ago, and missed a comment.

build.sh, BUILDING: consistency improvements

Consistently use "show", instead of sometimes using "display" or "print".
Fix mdoc markup (even though BUILDING is raw text).
Document help, -P, -?

build.sh: drop -lrumpdev hack introduced in previous

The existing trick in sys/rump/librump/rumpdev/rump_dev.c was not
updated after the renames in the kernel. kudos to riastradh.

PR kern/56599

build.sh: tweak rumptest and RUMP_LIBSETS

Use --no-whole-archive to keep us as honest as currently possible.
PR kern/56599

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)

handle different flavors of date(1)

for mercurial, use the latest revision instead of limiting the output to 1
(requested by joerg)

Handle git and mercurial for reproducible builds.

ignore cvslatest errors in expert mode

branches: 1.347.2;
Add mipsn64e[bl] targets which use 64 bit binaries instead of n32 ones.

Modify the generated make wrapper to avoid complaints about
non-writable OBJDIR from make (which happen a lot when cross-building
kernels re-using a makewrapper build for some other user).

add "distsets" alias for "distribution sets".

Fix typo in usage

make(1): remove references to the removed lst.lib

New build.sh option: -c <compiler>

Could never remember what the incantation is to do a clang build, so
now it's just `build.sh -c clang'.

Nix trailing whitespace.

nbmake bootstrap: silent configure if MAKEVERBOSE==0

Be consistent with the silencing of configure in tools/
and suppress the output in build.sh configure of nbmake

Add missing MACHINE_ARCH=earmv5hf{,eb} for evbarm.

When MACHINE_ARCH is missing, give direction on how to get a list of valid options

Remove plain "earm" targets, and require an explicit MACHINE_ARCH for
evbarm instead of selecting a default.

Remove evbarm OABI targets

Separate devicetree .dts -> .dtb building from kernel builds. They are now
part of a separate set, "dtb.tgz", and only built when MKDTB=yes. This
defaults to yes for earmv[67]* and aarch64, and no everywhere else.

Allow path components that start with ~

branches: 1.333.2;
Fix long-term broken pattern match when determining if uname -p output is
valid. [^a-z] syntax isn't valid.

on netbsd when defaulting to the native system, also derived MACHINE_ARCH
from the uname -p value already fetched.

this means eg, build.sh on evbarm will build the native platform rather
than "earm" (armv5) -- which was the default even on arm64 hosts.

Tyop in error message...

add aliases for evbarm without saying 'earm':
- copy all the 'evbearm' alias as 'evbarm'.

add build libs (undocumented).

Add some aliases (from phone)

branches: 1.327.2;
Re-add files that were accidentally deleted in my previous commit.

Synch with HEAD

Add initial support for ARMv8 (AARCH64) (by nisimura@ and ryo@)

- sys/arch/evbarm64 is gone and integrated into sys/arch/evbarm. (by skrll@)
- add support fdt. evbarm/conf/GENERIC64 fdt (bcm2837,sunxi,tegra) based generic 64bit kernel config. (by skrll@, jmcneill@)

branches: 1.324.2;
Remove port-acorn26

OK core@

Restore formatting/indentation of the configure call in rebuildmake()
to what it used to be.

Make sure nbmake is always built with the right _PATH_DEFSYSPATH.

Better validation of var name args to -V and -Z, in a way that makes
it trivial to add a list of banned var names for either of those args
should that ever be considered desireable (as the XXX suggests it might.)
I've had this mod locked in my tree for (at least) months - time to set it free.

If we're going xsrc hunting, lets's do it properly. In particular
it is not "good" just because ../xsrc or /usr/xsrc exists, if the user said
-X /no/xsrc/here ... Further, when we find it, remember it.

While here, also look in my favourite place: .../src/xsrc - src/x would be
better (or src/x11) but that's too hard given the cvs module name and src
tarball format.

Simplify.

Accept also ../xsrc in our sanity check, we use this as a default too,
if it exists.

Add sanity check: passed -x but no xsrc directory

branches: 1.316.4;
- Allow all other targets (like kernel=) to obey mkrepro (-P)
- Print the timestamp in UTC

State the reason script bombed.
Reword recommendation.

branches: 1.314.2;
add a trailing / to force symlink resolution.

Bail out on error computing the timestamp

Add -P for reproducible builds based on the latest cvs timestamp.

With driver modulization of if_loop.c through introduction of
sys/net/if_module.h, build.sh rumptest of -lrumpnet_virtif now also
requires -lrumpdev -lrumpvfs for if_loop.c:config_cfdriver_attach()
and kern_pmf.c:do_sys_sync().

branches: 1.310.2;

When setting TOP the idea is to run the [/usr]/bin/pwd binary, and
exclude any skell builtin. There are two ways to guarantee exec'ing
a binary, using a path, or using exec. To use a path we would need to
find it first (which certainly could be done.) Using exec here is easier.

Provide the correct system Makefile path to the tools make. This is needed
because although the arch-specific make wrapper sets $MAKEFLAGS to include
the proper system Makefile path, configure clears $MAKEFLAGS before it invokes
$MAKE to figure out which dependency style to use. This made the build fail
during the gcc build.

Make evbarm64 (little endian) the default for aarch64.

Fix typo in usage. From Christopher M. Fuhrman in PR 49882.

Fix the build kernels target with -n, disable temporarily runcmd while we
are fetching the allkernels variable.

Add a new build make variable, CONFIGOPTS, which is passed to config(1)
when building kernels. Revert "mkernel=*" and "mkernels" in favor of
this option.

Diff posted on source-changes-d@, approved by (null).

build.sh mkernels: Build all kernels in modular build

Expand TABs in usage.

build.sh mkernel=*: New command to build kernel in modular

Replace '.' and '-' in ${op} with '_', before setting d_${op}=true.

Also remove or simplify older code that did the same thing in a
different way. The old code handled most cases, but did not change
op=kernel.gdb to op=kernel_gdb.

build.sh: New operation "kernels" to build all kernels

iso -> images

Prefer dash to bash when searching for a usable shell.
I have tested a complete build with HOST_SH=dash on NetBSD.

Add riscv32 and riscv64 platforms.

Add or1k MACHINE/MACHINE_ARCH.

Only the -m command line option, not MACHINE from the environment,
is supposed override MACHINE_ARCH from the environment with the
default MACHINE_ARCH for the requested machine.

Add and use an opt_m variable to enforce this.

branches: 1.294.2;
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.

MACHINE_ARCH from the environment may be overridden by getarch (based
on the MACHINE), but MACHINE_ARCH from the -a option should not be
overridden. The previous commit accidentally removed this behaviour.

Add a list-arch operation to build.sh.

This lists all known MACHINE/MACHINE_ARCH values, optionally
filtered by glob patterns given in the "-m" and "-a" options.
For example: build.sh -a '*arm*' list-arch

In the usage message, mention that the -m arg may be an alias.
Also adjust a comment around alias processing.

When searching for a default MACHINE_ARCH for a given MACHINE,
don't break out of the loop too early.

Synchronise several shell_quote implementations, and:
* Elide some unnecessary pairs of quotation marks, to improve readability.
For example, shell_quote "''" is now \'\' instead of ''\'''\'''.
* Don't add quotes around words that contain only safe characters,
to improve readability.
* LC_COLLATE=C to prevent [a-zA-Z] from matching non-ASCII characters.
* Use ${SED} if defined.

hpcarm wants earmv4 because some models only have v4 CPUs.

BUILDINFO part 4: build.sh

Add BUILDINFO to the list of variables saved in the makewrapper script,
and print any non-blank BUILDINFO lines near the beginning of the output
from build.sh.
Index: build.sh

build.sh: drop -lrumpdev hack introduced in previous

The existing trick in sys/rump/librump/rumpdev/rump_dev.c was not
updated after the renames in the kernel. kudos to riastradh.

PR kern/56599

build.sh: tweak rumptest and RUMP_LIBSETS

Use --no-whole-archive to keep us as honest as currently possible.
PR kern/56599

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)

handle different flavors of date(1)

for mercurial, use the latest revision instead of limiting the output to 1
(requested by joerg)

Handle git and mercurial for reproducible builds.

ignore cvslatest errors in expert mode

branches: 1.347.2;
Add mipsn64e[bl] targets which use 64 bit binaries instead of n32 ones.

Modify the generated make wrapper to avoid complaints about
non-writable OBJDIR from make (which happen a lot when cross-building
kernels re-using a makewrapper build for some other user).

add "distsets" alias for "distribution sets".

Fix typo in usage

make(1): remove references to the removed lst.lib

New build.sh option: -c <compiler>

Could never remember what the incantation is to do a clang build, so
now it's just `build.sh -c clang'.

Nix trailing whitespace.

nbmake bootstrap: silent configure if MAKEVERBOSE==0

Be consistent with the silencing of configure in tools/
and suppress the output in build.sh configure of nbmake

Add missing MACHINE_ARCH=earmv5hf{,eb} for evbarm.

When MACHINE_ARCH is missing, give direction on how to get a list of valid options

Remove plain "earm" targets, and require an explicit MACHINE_ARCH for
evbarm instead of selecting a default.

Remove evbarm OABI targets

Separate devicetree .dts -> .dtb building from kernel builds. They are now
part of a separate set, "dtb.tgz", and only built when MKDTB=yes. This
defaults to yes for earmv[67]* and aarch64, and no everywhere else.

Allow path components that start with ~

branches: 1.333.2;
Fix long-term broken pattern match when determining if uname -p output is
valid. [^a-z] syntax isn't valid.

on netbsd when defaulting to the native system, also derived MACHINE_ARCH
from the uname -p value already fetched.

this means eg, build.sh on evbarm will build the native platform rather
than "earm" (armv5) -- which was the default even on arm64 hosts.

Tyop in error message...

add aliases for evbarm without saying 'earm':
- copy all the 'evbearm' alias as 'evbarm'.

add build libs (undocumented).

Add some aliases (from phone)

branches: 1.327.2;
Re-add files that were accidentally deleted in my previous commit.

Synch with HEAD

Add initial support for ARMv8 (AARCH64) (by nisimura@ and ryo@)

- sys/arch/evbarm64 is gone and integrated into sys/arch/evbarm. (by skrll@)
- add support fdt. evbarm/conf/GENERIC64 fdt (bcm2837,sunxi,tegra) based generic 64bit kernel config. (by skrll@, jmcneill@)

branches: 1.324.2;
Remove port-acorn26

OK core@

Restore formatting/indentation of the configure call in rebuildmake()
to what it used to be.

Make sure nbmake is always built with the right _PATH_DEFSYSPATH.

Better validation of var name args to -V and -Z, in a way that makes
it trivial to add a list of banned var names for either of those args
should that ever be considered desireable (as the XXX suggests it might.)
I've had this mod locked in my tree for (at least) months - time to set it free.

If we're going xsrc hunting, lets's do it properly. In particular
it is not "good" just because ../xsrc or /usr/xsrc exists, if the user said
-X /no/xsrc/here ... Further, when we find it, remember it.

While here, also look in my favourite place: .../src/xsrc - src/x would be
better (or src/x11) but that's too hard given the cvs module name and src
tarball format.

Simplify.

Accept also ../xsrc in our sanity check, we use this as a default too,
if it exists.

Add sanity check: passed -x but no xsrc directory

branches: 1.316.4;
- Allow all other targets (like kernel=) to obey mkrepro (-P)
- Print the timestamp in UTC

State the reason script bombed.
Reword recommendation.

branches: 1.314.2;
add a trailing / to force symlink resolution.

Bail out on error computing the timestamp

Add -P for reproducible builds based on the latest cvs timestamp.

With driver modulization of if_loop.c through introduction of
sys/net/if_module.h, build.sh rumptest of -lrumpnet_virtif now also
requires -lrumpdev -lrumpvfs for if_loop.c:config_cfdriver_attach()
and kern_pmf.c:do_sys_sync().

branches: 1.310.2;

When setting TOP the idea is to run the [/usr]/bin/pwd binary, and
exclude any skell builtin. There are two ways to guarantee exec'ing
a binary, using a path, or using exec. To use a path we would need to
find it first (which certainly could be done.) Using exec here is easier.

Provide the correct system Makefile path to the tools make. This is needed
because although the arch-specific make wrapper sets $MAKEFLAGS to include
the proper system Makefile path, configure clears $MAKEFLAGS before it invokes
$MAKE to figure out which dependency style to use. This made the build fail
during the gcc build.

Make evbarm64 (little endian) the default for aarch64.

Fix typo in usage. From Christopher M. Fuhrman in PR 49882.

Fix the build kernels target with -n, disable temporarily runcmd while we
are fetching the allkernels variable.

Add a new build make variable, CONFIGOPTS, which is passed to config(1)
when building kernels. Revert "mkernel=*" and "mkernels" in favor of
this option.

Diff posted on source-changes-d@, approved by (null).

build.sh mkernels: Build all kernels in modular build

Expand TABs in usage.

build.sh mkernel=*: New command to build kernel in modular

Replace '.' and '-' in ${op} with '_', before setting d_${op}=true.

Also remove or simplify older code that did the same thing in a
different way. The old code handled most cases, but did not change
op=kernel.gdb to op=kernel_gdb.

build.sh: New operation "kernels" to build all kernels

iso -> images

Prefer dash to bash when searching for a usable shell.
I have tested a complete build with HOST_SH=dash on NetBSD.

Add riscv32 and riscv64 platforms.

Add or1k MACHINE/MACHINE_ARCH.

Only the -m command line option, not MACHINE from the environment,
is supposed override MACHINE_ARCH from the environment with the
default MACHINE_ARCH for the requested machine.

Add and use an opt_m variable to enforce this.

branches: 1.294.2;
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.

MACHINE_ARCH from the environment may be overridden by getarch (based
on the MACHINE), but MACHINE_ARCH from the -a option should not be
overridden. The previous commit accidentally removed this behaviour.

Add a list-arch operation to build.sh.

This lists all known MACHINE/MACHINE_ARCH values, optionally
filtered by glob patterns given in the "-m" and "-a" options.
For example: build.sh -a '*arm*' list-arch

In the usage message, mention that the -m arg may be an alias.
Also adjust a comment around alias processing.

When searching for a default MACHINE_ARCH for a given MACHINE,
don't break out of the loop too early.

Synchronise several shell_quote implementations, and:
* Elide some unnecessary pairs of quotation marks, to improve readability.
For example, shell_quote "''" is now \'\' instead of ''\'''\'''.
* Don't add quotes around words that contain only safe characters,
to improve readability.
* LC_COLLATE=C to prevent [a-zA-Z] from matching non-ASCII characters.
* Use ${SED} if defined.

hpcarm wants earmv4 because some models only have v4 CPUs.

BUILDINFO part 4: build.sh

Add BUILDINFO to the list of variables saved in the makewrapper script,
and print any non-blank BUILDINFO lines near the beginning of the output
from build.sh.
Index: build.sh

Avoid the dependency on a populated tooldir (or building the tools)
when simply doing mkrepro-timestamp and the current repository setups
does not actually require it.

branches: 1.365.2;
build.sh: usage improvements

Use UPPERCASE as the convention to describe an option argument,
as it's more noticable than lowercase in the help description.

Expand tabs in the help.

Editorial improvements.

Deprecate extsrc/, EXTSRCSRCDIR, MKEXTSRC

Remove support for building extsrc/:
1. Makefile: remove do-extsrc target.
2. build.sh: remove options -y and -Y extsrcdir.
3. distrib/sets: remove support for extsrc in various tools
including the options -L ext and -y, and the extsrc sets.
4. doc/BUILDING.mdoc: remove docs for extsrc/, EXTSRCSRCDIR, MKEXTSRC (etc)
5. bsd.own.mk and various Makefiles: remove support for extsrc/,
EXTSRCSRCDIR, MKEXTSRC.

As proposed on tech-kern and tech-userlevel on 2022-01-07
and followed up on 2022-08-21.

build.sh: improve help and usage

Split help into synopsis and help.
Only display the synopsis for usage errors (not the full help).
Print usage errors to stderr not stdout.
Don't print a usage error for -? before the help message.
Improve error messages for unknown options and missing arguments.
Remove trailing "." present in some error messages.
Alway use 'quote' instead of `quote' in some error messages.

build.sh: improve help versus usage

Rework help / usage semantics:
- send help to stdout, exit 0.
- send usage to stderr, exit 1.

Don't try to install netbsd-CONFIG.debug files build with kernel=

Enable split-debug files for kernels. Enabled by default with MKDEBUG=yes.

remove duplicate the article in comments.

Patch the build script and toolchain to allow passing through
HOST_CPPFLAGS, HOST_CXXFLAGS, but not LIBRARY_PATH because it breaks
the build. These variables are necessary to support building NetBSD
from a GNU Guix or NixOS host, where /usr/include, /lib, and all
but /bin/sh do not exist. In many cases, support for HOST_CPPFLAGS
was incomplete. From Ryan Sundberg

Fix timestamp extraction logic for Mercurial repos to use UTC.
Extend logic to also cover "hg archive".

Add a new operation "mkrepro-timestamp" to extract the timestamp a
build with -P would use. Example usage:

./build.sh -T /usr/tools -P mkrepro-timestamp

This allows us to extract this information once, make other use of it,
and replicate it on other machines with -V MKREPRO=yes -V MKREPRO_TIMESTAMP=..

Fix rumptest target && enable separate rump target. From Damien Zammit

fix various typos in comments.

Good bye oabi! Tested building earmv7hf.

kre@ does not like the subshell :-)