bsd.own.mk: rename GCC_NO_* to CC_WNO_*

Rename compiler-warning-disable variables from
GCC_NO_warning
to
CC_WNO_warning
where warning is the full warning name as used by the compiler.

GCC_NO_IMPLICIT_FALLTHRU is CC_WNO_IMPLICIT_FALLTHROUGH

Using the convention CC_compilerflag, where compilerflag
is based on the full compiler flag name.

usr.bin/config: Suppress -Werror=stringop-truncation error.

Add GCC_NO_STRINGOP_TRUNCATION to scan.c to prevent build failure.

Error was reported when build.sh was run with MKLIBCSANITIZER=yes flag.
gcc version 8.3.0

Reviewed by: kamil@

introduce some common variables for use in GCC warning disables:

GCC_NO_FORMAT_TRUNCATION -Wno-format-truncation (GCC 7/8)
GCC_NO_STRINGOP_TRUNCATION -Wno-stringop-truncation (GCC 8)
GCC_NO_STRINGOP_OVERFLOW -Wno-stringop-overflow (GCC 8)
GCC_NO_CAST_FUNCTION_TYPE -Wno-cast-function-type (GCC 8)

use these to turn off warnings for most GCC-8 complaints. many
of these are false positives, most of the real bugs are already
commited, or are yet to come.


we plan to introduce versions of (some?) of these that use the
"-Wno-error=" form, which still displays the warnings but does
not make it an error, and all of the above will be re-considered
as either being "fix me" (warning still displayed) or "warning
is wrong."

branches: 1.10.16;
put back scanner warnings now that flex produces clean code.

- add RCSID's
- fix -Wconversion (except scanner)

branches: 1.8.54;
Using ${NETBSDSRCDIR} requires <bsd.own.mk>

Enable the code that adds global constants to every kernel object file
that uses each configuration parameter.
This will stop kernel objects build with different options (that come from
config files) being linked together or loaded as a LKM.
Currently some options that have an effect on the kernel DDI/KI are passed
directly as parameters to cc.
An aim (for anyone adequately bored) would be to reduce the number of these
constants that appear in each .o file. .o files with the same constants
should be sharable between kernels (even between XEN and i386).

Introduce a new option to config(1): -L. What it does is takes a regular
configuration file (although it is meant to be used mostly with
std.${ARCH}), and prints out a configuration file that includes it and
select every single option and parameter, and define an instance for every
single possible attachment. IOW, selects everything, into a would-be
"LINT" config.

The resulting config really isn't meant to be runnable, but should be
somewhat buildable on most archs. It still needs adjustments because some
options are peculiar (e.g., ACPI_DSDT_FILE wants an existing file as its
value), so it's not yet possible to do "config -L; config; build" in an
automated way.

protect libutil with HOSTPROG

Now that this is part of tools, we don't need the extra build glue.

Install config.5 and config.samples.5.

branches: 1.2.2;
Remove bogus BINDIR assignment.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

Now that this is part of tools, we don't need the extra build glue.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

fix various typos in comments and log messages.

Fix all remaining typos, mainly in comments but also in few definitions and log messages, reported by me in PR kern/54889.
Also fixed some additional typos in comments, found on review of same files or typos.

genric -> generic

branches: 1.31.16; 1.31.18;
A few more.

Libraries should be built in sub-make. libkern is polluting .PATH too much.

About ${CTFCONVERT}.

${MD_OBJS} and friends.

${LD} input files and library (*.a) support.

remove these items:

o Disallow unknown options.
o Kill makeoptions.

because they're overly strict and i use these both regularly
to do stuff that isn't in the base src, and i'm quite sure
i am not the only one.

+ suffix rules

+ copy symtab

One more about linker script, another about genassym.

I don't like -Ttext .

+ kill makeoptions

TODOs about kernel options.

Linker script TODO items.

Insert comment regarding PR/48571 so it doesn't get lost.

config(1) ioconf is insufficient for pseudo-devices

config_{init,fini}_component

+ pseudo(dev) attach

Typo in previous.

More about attribute -> bus, and child configuration.

branches: 1.14.2;
config(1): TODO about namespace (esp. options vs. module)

config(1): TODO item about swapnetbsd.c

Revise a plan about link-set.

(I was ignorant 2 weeks ago.)

config(1): Note a little more about linkage

config(1): TODO about kernel linkage

config(1): More TODO description.

config(1): One more TODO about retiring direct attachment

Spelling fixes.

config(1): More TODO

config(1): More TODO items

Allow easier adding/removing of options without regard to
whether or not they were already defined.

Spelling.

More TODOs.

Put TODO ideas (to keep your fun).

Experimental ``suffix rules + subdirectories'' build support (-S).

Document previously secret -d debug option. Bump date.

Fix two typos.

Document -M.

branches: 1.15.2;
Sort SYNOPSIS. Bump date for previous.

Add two new options, -U and -D, that can be used to define "makeoptions"
on the config command line. While there, rename the undocumented (internal)
parser debug option from -D to -d.
Discussed on tech-toolchain.

branches: 1.13.2; 1.13.8;
Use more markup. Wording.

bump date.

make config -x look at the booted kernel first.

branches: 1.10.6;
New sentence, new line.

.Sx to refer to a section, and add one ref to the lint stuff for clarity.

Use Sx. Remove superfluous .Pp.

.sp -> .Pp

branches: 1.6.14;
Improve the structure of the man page to make it more usual, with the hope
that people might actually read it some day. Yeah, I know, dream on, dream
on.

Thanks schmonz@ for a quick english check.

branches: 1.5.4;
Bump date for previous (year was wrong).

Introduce a new option to config(1): -L. What it does is takes a regular
configuration file (although it is meant to be used mostly with
std.${ARCH}), and prints out a configuration file that includes it and
select every single option and parameter, and define an instance for every
single possible attachment. IOW, selects everything, into a would-be
"LINT" config.

The resulting config really isn't meant to be runnable, but should be
somewhat buildable on most archs. It still needs adjustments because some
options are peculiar (e.g., ACPI_DSDT_FILE wants an existing file as its
value), so it's not yet possible to do "config -L; config; build" in an
automated way.

Reference config(5). Noted by perry@.

XXX options(4), config(9) and probably some others still have to be changed
XXX a bit. A chaque jour suffit sa peine.

branches: 1.2.2;
config(1), not (8)

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

Fix typos in documentation.

config(5): bump date for and fix typo in previous

PR bin/58583

config(5): mention ioconf and pseudo-root

Add barebones descriptions for ioconf and pseudo-root added by pooka@
long time ago for rump and kmods. Definitely could be improved and
expanded, so feel free...

PR bin/58583

branches: 1.48.2;
With config(1) as it exists today, a kernel Makefile is able to implement
logic on kernel options so long as those options are not defflag'd or
defparam'd. This works because such options are automatally added to the
IDENT var in the kernel Makefile as a preprocessor define, and the Makefile
can then do an operation like:

.if !empty(IDENT:M-DSOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Unfortunately, this precludes making it possible to generate a compile-time
dependency on SOMECOOLCPUOPTION, or having SOMECOOLCPUOPTION imply another
kernel config option using the normal config(1) option dependency function.

Fix this by introducing a new option description keyword: mkflagvar. This
keyword marks an already defflag'd option as wanting a kernel Makefile var
defined if that option is selected in the kernel config file. So:

defflag opt_coolcpu.h SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION
mkflagvar SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION

will cause:

KERNEL_OPT_SOMECOOLCPUOPTION="1"
KERNEL_OPT_ANOTHERCOOLCPUOPTION="1"

...to be emitted into the kernel Makefile if those options are, in fact,
selected with "options ..." in the kernel config file, thus allowing for
a compile-time dependency on the option to be generated in addition to
Makefile logic, which now looks like:

.if !empty(KERNEL_OPT_SOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

remove duplicate the article in documentation.

Add missing double p and d for stopped and overriden accordingly.
Fix few more typos along the way, mainly in copy-pasted comments.

s/implicitely/implicitly/

Swtich from -ohang to -tag lists.

-ohang lists are hard to read b/c long config directives are almost
impossible to tell from the following text. Occasional multi-paragraph
descriptions were quite confusing too.

Missed formatting fixes for "no makeoptions".

The tale of two makeoptions.

There are two forms that were both documented in the selections
section, but the form with the condition (and only that form) can be
used only in the descriptions section. Move it to the appropriate .Ss
and add prominent notice to both.

Formatting improvements.

Minor tweak to list formatting.

Where necessary, use .Bl -ohang -compact and explicit .Pp to be able
to group multiple config directives (separate .It) by omitting .Pp
between them.

While here, add a .|" comment before .It to make them more visible
when editing.

Formatting improvements.

Only comb through the first few pages (OBJECTS AND NAMES) for now.

Use .Em for emphasis, not .Ar (both look the same in text output, but
.Ar is very confusing in PostScript). Reduce emphasis - don't
highlight every usage of a term. Where everything is emphasized
nothing is.

Fix PS formatting of locator syntax examples so that square brackets
are literal and don't look meta-syntactic. Explicitly show optional
locator syntax (in literal square brackets).

Remove workaround for ancient HTML generation code.

Don't reference older CFATTACH_DECL

branches: 1.36.2;
New sentence, new line. Bump date for previous.

Extend syntax of config phrase, a quoted string instead of a device name
is passed as root specification string. This can be used to specify a
wedge by name.

Introduce a new syntax, "buildprefix", to specify prefix of files under kernel
build subdirectory. This is not used now that everything is built at the
top of kernel build directory. It will become mandatory for source/object
files put out of kernel source tree to specify corresponding build subdirectory.
Only ``no dots'' relative path is accepted as "buildprefix".

Allow redefinition of options (with only a warning message). Removal
of non-included options was already only a warning.

This also affects redefinition and removal of file-system and makeoption
entries.

Reviewed and OKd by uebayasi@

Need to read closer - there was another occurrence of the same tyepo

Tyepo

Fix a macro usage. Add more markup.

Define `makeoptions' values as parameters with "makeoptions_" prefix too.
C sources are encouraged to include relevant opt_*.h files rather than relying
on -DXXX passed via makefiles.

Revert "Retire optional `rule' argument of `file' command". It is still used
in m68k ports.

Retire optional `rule' argument of `file' command.

This has existed since Rev. 1.1 of Torek's config(8) [1], but probably has
never been used by anyone.

(I can not imagine how a `file' like look like if `rule' is specified.)

[1] http://cvsweb.netbsd.org/bsdweb.cgi/src/usr.sbin/config.new/Attic/files.c?rev=1.1

Paths given to "file" and "object" are relative from "prefix" if defined.

config(1), config(5): Introduce "select"

o Introduce a new selection directive "select" to select an attribute (as a
module) and its dependencies.
o Support "no select" too.
o Stop abusing "options" to select an attribute.
o Bump config(1) version.

branches: 1.24.2;
Bump date for previous.

implement "no ident".

branches: 1.22.2;
Fix some typos found by Rich Neswold and reported in PR 48597.

branches: 1.21.6; 1.21.12;
Remove fs_foo.h support from deffs now that nothing uses it.

Bump date.

Remove mentions that a kernel configuration file must support at least one
file-system because it's no longer true.

branches: 1.18.6;
Disable use of pseudo-devices defined "defpseudo" with interface
attributes, but provide compatibility for older source trees (according
to the "version" in conf/files and other files.*.
reviewed by cube

Bump date for previous.

Seperate pseudo-devs which don't use device_t et al. (the classical
ones, mostly network pseudo interfaces) and those which use autoconf
to attach instances dynamically, or to attach child devices.
Use a new keyword "defpseudodev" for the latter.
Issue a warning if interface attributes are used with "old" pseudodevs,
but let it work as before for now. This will be removed soon.
approved by cube

branches: 1.15.2;
Convert TNF licenses to new 2 clause variant

branches: 1.14.2;
Typo.

Make the "machine" statement define attributes for `machine', `arch' and
all the `subarch' given as parameters. This is useful in combination with
ifdef...

Add support for "ifdef <attribute>" blocks. It goes like this:

ifdef <attr> / ifndef <attr>
elifdef <attr>
elifndef <attr>
else
endif

No indentation is possible for now, it will come later.

Some more licence clean up.

branches: 1.10.4;
Add new syntaxic sugar to provide a value in lint (i.e., config -L)
configurations for defparams. It will be used _only_ by config -L.

This will greatly help automating tests with lint configurations,
hopefully.

Improve formatting, fix typo, bump date for previous.

- Document "source /path" and "build /path".
- Move documentation for "package", because as I learned tonight, it's not
allowed before setmachine. You get to discover scary new stuff about
config(1) each time you look at it.

Mention that newlines mid-statement are allowed, and clarify how
needs-count and needs-flag work, notably for simple attributes.

fix spelling of accommodate; from Zapher.

Fix typo.

Add support for specifying a default value on a 'defparam' line.
If a default is specified then the option is always defined in the
corresponding .h file.
Particularly useful for parameters where the default action isn't the
same a defining the value zero, given that the current use of #if defined(opt)
is open to problems wehere the relevant .h file isn't actually included, so
requiring an option to always have a value makes sense.
Also included (but commented out) is code that adds a global symbol to
the object file to stop objects compiled with different values for the
same option being linked together - I'm not quite happy with it yet!

s/setmachine/machine. Overlook pointed out by Havard Eidnes.

branches: 1.2.2;
Set a more realistic date than the day I started writing it.

Everything you've always wanted to know about config(1) without daring
asking (with reason).

You've entered the config(1) dungeon--More--
An option hits you. You die.

fix various typos, mainly in comments.

Minor formatting tweaks.

`Controller' takes two l's.

Fix some typos found by Rich Neswold and reported in PR 48597.

branches: 1.4.4; 1.4.26;
Convert TNF licenses to new 2 clause variant

branches: 1.3.6;
Some more licence clean up.

branches: 1.2.2; 1.2.12;
Set a more realistic date than the day I started writing it.

Everything you've always wanted to know about config(1) without daring
asking (with reason).

You've entered the config(1) dungeon--More--
An option hits you. You die.

config(1): Make sort order deterministic.

Ensure we break ties in every case. This way, even though we use the
unstable qsort(3) library routine, the output is reproducible, no
matter what algorithm is behind qsort(3).

It would be nice if we could just use a stable sort function here,
but mergesort(3) is nonstandard, so we'd have to add it to
tools/compat, which is a big pain.

Instead, put a tie-breaking rule in every comparison function we use
with qsort, and abort() in the event of ties -- that way, we noisily
refuse to rely on unstable sort order.

While here, dispense with any question of integer overflow, and
sprinkle comments.

PR bin/58115

Bump version.

With config(1) as it exists today, a kernel Makefile is able to implement
logic on kernel options so long as those options are not defflag'd or
defparam'd. This works because such options are automatally added to the
IDENT var in the kernel Makefile as a preprocessor define, and the Makefile
can then do an operation like:

.if !empty(IDENT:M-DSOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Unfortunately, this precludes making it possible to generate a compile-time
dependency on SOMECOOLCPUOPTION, or having SOMECOOLCPUOPTION imply another
kernel config option using the normal config(1) option dependency function.

Fix this by introducing a new option description keyword: mkflagvar. This
keyword marks an already defflag'd option as wanting a kernel Makefile var
defined if that option is selected in the kernel config file. So:

defflag opt_coolcpu.h SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION
mkflagvar SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION

will cause:

KERNEL_OPT_SOMECOOLCPUOPTION="1"
KERNEL_OPT_ANOTHERCOOLCPUOPTION="1"

...to be emitted into the kernel Makefile if those options are, in fact,
selected with "options ..." in the kernel config file, thus allowing for
a compile-time dependency on the option to be generated in addition to
Makefile logic, which now looks like:

.if !empty(KERNEL_OPT_SOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

branches: 1.106.6;
Avoid depending on common symbols.

Keep track where more objects are declared so that we can print where things
have been redefined.

branches: 1.104.2;
Bump config(5) version for OPT.foo.c variables.

branches: 1.103.2;
Keep previous location of device definitions so we can print them in error
messages.

branches: 1.102.2;
use a reference count to avoid deleting psrefs still in use.

Allow multiple attachments methods to the same child+parent combination:

foo* at bar? with baz
foo* at bar? with barf

Do this by scanning the list of iba's and allocating a new cfparent for
each. Keep track of the shared parent+child combinations by using the
same id for them.

When deleting orphans detect parent<->child loops and break them.
"active" is not a boolean, use the right comparison.

Bump for quoting makeoptions with multiple lines.

branches: 1.98.6;
Make attribute deselection work:
- when deselecting attributes, remove files that depend on them
- when deselecting attributes, remove devices that depend on them

Bump version for -no

Accept "-no" as a "no" that does not cause errors if the object deleted
does not exist.

match the updated <sys/types.h> definition

In ioconfname mode, #define IOCONF as the ioconf token.

Avoids having to retype the name to call config_init/fini_component().

Sort files in ${ALLFILES} in the order of parsing of `files.*'.

config(1) reads the first `files.${MACHINE}' when it encounters `machine'.
Then it includes common `files.${MACHINE_SUBARCH}', `files.${MACHINE_ARCH}',
and MI `sys/conf/files' at last. This change makes the first "file" in
`files.${MACHINE}' appear first in ${ALLFILES}.

Make config(1) output only ${ALLFILES}, selected input files, and pass it to
Makefile.kern.inc, where files are filtered by suffix.

*.o files don't need special handling at all now. Makefile.kern.inc knows
*.o input files don't need to get compiled, but need to get linked.

Add generated *.c files (devsw.c and ioconf.c) to ${CFILES} by config(1)
internally. ${MI_CFILES} in Makefile.kern.inc is no longer needed.

After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

Move `all' target definition to Makefile.kern.inc.

Define kernel dependency in Makefile.kern.inc.

Define ${OBJS} in `Makefile.kern.inc'. Relative paths are not truncated
(a/b/c/x.c -> x.c) in `-S' mode. Don't provide some explicit rules for `-S'
mode for safety.

Output relative path for ${CFILES} and ${SFILES} in generated `Makefile'.
`Makefile.kern.inc' knows how to deal with $S prefix, so don't bother it in
config(1).

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

Experimental ``suffix rules + subdirectories'' build support (-S).

Bump version for "buildprefix".

Introduce a new syntax, "buildprefix", to specify prefix of files under kernel
build subdirectory. This is not used now that everything is built at the
top of kernel build directory. It will become mandatory for source/object
files put out of kernel source tree to specify corresponding build subdirectory.
Only ``no dots'' relative path is accepted as "buildprefix".

Keep track of directory of files internally.

Clean up struct files.

Use per-suffix lists where appropriate.

Abstract struct filetype is no longer needed.

Merge struct objects into struct files. Keep per-suffix file lists.

Start merging struct files and struct objects.

Restore the old behavior of "maxusers" to allow the value to be overriden.
Problem reported by John D. Baker.

Revert "Accept only relative paths ...". This will be redone in much better,
stricter way.

Define compile rules in sys/conf/Makefile.kern.inc except those overriden by
"file ... compile-with ...". As rules are still explicit, be careful to not
override by checking .if !target(xxx.o).

Define `makeoptions' values as parameters with "makeoptions_" prefix too.
C sources are encouraged to include relevant opt_*.h files rather than relying
on -DXXX passed via makefiles.

Convert MAXUSERS, take 2.

Define MAXUSERS in opt_param.h and include it from sys/param.h, for more
accurate dependency. Don't define -DMAXUSERS and add it to ${CPPFLAGS}.
config(5)'s "maxusers" syntax doesn't change.

(The version just before revert was OK, but I had something bad in my obj
dirs.)

Revert MAXUSERS changes. Need fresh air.

Define MAXUXERS in opt_param.h. Bump config(1) version.

Revert "Retire optional `rule' argument of `file' command". It is still used
in m68k ports.

Retire optional `rule' argument of `file' command.

This has existed since Rev. 1.1 of Torek's config(8) [1], but probably has
never been used by anyone.

(I can not imagine how a `file' like look like if `rule' is specified.)

[1] http://cvsweb.netbsd.org/bsdweb.cgi/src/usr.sbin/config.new/Attic/files.c?rev=1.1

move driver attach declarations to ioconf.h

Pre-calc file path len/suffix. Misc. clean up.

config(1): Sort objects by weight in modular build

Sort per-module kernel objects (*.ko) by dependency weight. Important modules
are placed in lower addresses. The ``machdep'' module will be always placed
in the lowest.

(At this moment, the order is not exact because dependency information is
incomplete in config files (files.*). Thus ``sysmon_power.ko'' comes lower
than ``machdep.ko'' and/or ``uvm.ko''; but still much better than alphabetical
sort.)

config(1): Rename internal flag to enable modular build

config(1): Start calling module as module

config(1), config(5): Introduce "select"

o Introduce a new selection directive "select" to select an attribute (as a
module) and its dependencies.
o Support "no select" too.
o Stop abusing "options" to select an attribute.
o Bump config(1) version.

- add RCSID's
- fix -Wconversion (except scanner)

Keep track of attribute (module) dependency using hash2.

Make hash capable of taking two key strings.

Enable debug for tools.

Bump config(1) version.

config(1) still keeps backward-compatibility; it is more permissive than
before. The tree, however, needs the updated config(1), mainly to strictly
define attribute ("module") ownership and dependencies.

Refactor attr definition and initialization a little.

FIT_FORCESELECT is never set anywhere.

FIT_NOPROLOGUE is never set anywhere.

Steal -M to enable "modular" build.

Define an implicit attribute "netbsd" internally to collect files that don't
belong to any specific attribute.

Eventually, all operations doing "foreach (files)" can be rewritten as "foreach
(attributes) foreach (files)".

If a file is marked by an atom expression, like "file a.c foo", and if the
specified expression is an attribute, mark the file as belonging to the
attribute.

At this moment this information is not used for any purpose, but can be
traced by config -ddd.

Provide a CFGDBG version for the tools build.

Add debug print function.

Bump CONFIG_VERSION.

branches: 1.45.2;
Add two new options, -U and -D, that can be used to define "makeoptions"
on the config command line. While there, rename the undocumented (internal)
parser debug option from -D to -d.
Discussed on tech-toolchain.

branches: 1.44.2; 1.44.8;
If we build several kernel (variants) within the same build directory, bad
things may happen in a parallel build - especially with rules like the
automatic size adjustment for SYMTAB_SPACE, see long standing failure of
evbarm on the build cluster.
Easy fix: .WAIT for each config to complete, before going on with the
next. Low impact, only minor loss of paralellism, and only in cases where
needed.

Remove the NV_OBSOLETE flag, which is no longer needed. While here,
update an outdated comment about condition expressions.

Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Introduce type-safe wrappers around the hash tables. Use them for a
selected set of tables affected by the next nvlist cleanup in the
works.

Move locator lists to their own data structure. This can use more tidying;
it is not clear to me at the moment what the "string" and "num" values
pushed around in locator lists are supposed to actually mean.

Create a struct condexpr type to hold condition expressions, instead
of abusing struct nvlist to make trees.

(These are the a|b and a&b constructs.)

Create a new type struct attrlist to manage lists of attributes,
instead of using struct nvlist.

(struct nvlist holds lists (or trees!) of semi-arbitrary stuff with no
structure and almost no type safety; it should go away.)

Remove cf_swap, which hasn't been used in a long time.

Use __printflike. Build products unchanged.

branches: 1.35.6; 1.35.8;
For the simple cases, augment device-major with information on how
a driver expects /dev/node -> minor mappings to go and include that
information in devsw_conv.
(no, I didn't plow through all the MD majors files)

Some drivers want to include ioconf.h, so generate it if doing
"ioconf" config.

Add a highly experimental pseudo-root feature to be used in conjuction
with the also-experimental ioconf keyword. pseudo-root allows to
specify a root at any point in the device tree without having
attachments from the actual root.

For example, instead of having a config file like this:
mainbus0 at root
bus_a? at mainbus0
bus_b? at bus_a*
device7 at bus_b?

You can have one like this:
pseudo-root bus_b*
device7 at bus_b?

This will produce the relevant ioconf.c glue for device number 7
only instead of the whole 9 yards from root. Perhaps needless to
say, this can be used to generate snippets of config glue for
modules and, let's not deny that my main motivation for doing this,
rump components.

This is part 2/3 of my modifications to config (the last part is
autogenerating source file lists and component Makefiles).

No strong objection from cube (after a little pressuring ;), but
like he said, the implementation will most likely need some more
tweaking and may not work correctly under all pseudo-root uses yet.

Remove fs_foo.h support from deffs now that nothing uses it.

Introduce experimental support for ioconf-only configuration files.
This is done by giving the "ioconf" keyword in the config file.
As a result, config produces only ioconf.c and locators.h. Currently,
only "monolithic" configurations with the device path starting from
root are supported. Eventually, the goal is to support a local
root in the input file from any point along the device tree using
files.* in our kernel tree. This will make autogenerating the
config glue for device modules possible instead of having to write
it by hand like is currently required (yes, it sounds simple to
implement, but ...).

reviewed by cube.
(the next part will demand major discussions with you, so prepare ;)

Now that condmkopttab isn't a hash table anymore, don't initialise it with
ht_new()... So actually rename it to condmkoptions to avoid confusion.

Reported build failure on amd64 (strange that I wouldn't get it on i386) on
current-users by Kurt Schreiner.

Actually allow a full expression for the condition for "makeoptions" in
files.*

Patch from Yorick Hardy.

- Allow substition of options values in the generated Makefile using the
syntax %OPTION%.
- Make it possible to specify a syntax version information in the Makefile.
- Bump version to 20090214.

branches: 1.27.2;
Change major()/minor() to return 32-bit types again, called
devmajor_t/devminor_t, as proposed on tech-kern.
This avoids 64-bit arithmetics and 64-bit printf formats in parts
of the kernel where it is not really useful, and helps clarity.

rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

bump version due to recent rootfs changes

Bump CONFIG_VERSION after the previous fix to main.c.

Do proper config(5) version checking after previous changes [hi drochner!].

The rule is, if you change scan.l or gram.y, you bump the config(5)
version. If you implement the changes under sys/conf/files or affiliate,
you bump the required version in sys/conf/files or in an appropriate place
to minimise annoyance. If the changes makes new config(1) incompatible
with a previous version of config(5), embed it in config(1) using the
CONFIG_MINVERSION definition along with CONFIG_VERSION.

This has been in the tree for what, 3 years now? It's even documented...

branches: 1.22.4; 1.22.6;
Use a common function to generate the "MACHINE GENERATED" comment,
and ensure mkswap.c adds it too.

Add support for "ifdef <attribute>" blocks. It goes like this:

ifdef <attr> / ifndef <attr>
elifdef <attr>
elifndef <attr>
else
endif

No indentation is possible for now, it will come later.

branches: 1.20.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Handle defopt lists properly, as ':=' syntax introduce an additional
element... For that, introduce nvcat(nv1, nv2).

Add new syntaxic sugar to provide a value in lint (i.e., config -L)
configurations for defparams. It will be used _only_ by config -L.

This will greatly help automating tests with lint configurations,
hopefully.

Introduce a new option to config(1): -L. What it does is takes a regular
configuration file (although it is meant to be used mostly with
std.${ARCH}), and prints out a configuration file that includes it and
select every single option and parameter, and define an instance for every
single possible attachment. IOW, selects everything, into a would-be
"LINT" config.

The resulting config really isn't meant to be runnable, but should be
somewhat buildable on most archs. It still needs adjustments because some
options are peculiar (e.g., ACPI_DSDT_FILE wants an existing file as its
value), so it's not yet possible to do "config -L; config; build" in an
automated way.

Coverity CID 175: Avoid memory leak; add new function to destroy a hash table.

Bump version after last feature addition (default value for defparam).

There isn't really any point in the 'union' for nv_str and nv_ptr, all it
does is cause confusion when I try to add default values for defparam.
Also if malloc()/free() aren't efficient enough for us, we ought to fix
them rather than having local free list - so kill the local free lists.

Bump version after deffs change.

Allow deffs to take dependencies like defflag and defopt

Bump version after fix to obsolete syntax.

Introduce two new statements:

obsolete defflag <option> [, <option> [, ...]]
obsolete defparam <option> [, <option> [, ...]]

The two statements actually do the same thing (there could be only one),
but it makes things less cryptic that way. The optional ": deps" part of
a 'defflag' or 'defparam' still has to be dropped when it gets obsoleted.

When the user has 'options OBSOLETE_OPTION' in his configuration file, it
is ignored (that is, opt_*.h files are generated as if it wasn't there),
and the user gets a warning about it.

Bump version to 20060525.

When someone first uses that syntax in the tree, a "version 20060525"
statement should be added before its occurrence, preferably at the top
of sys/conf/files.

Suggested by Matt Thomas a few months ago.

branches: 1.9.2;
Remove special hack for swapnetbsd.c when generating the makefile.
Instead, just add it to the list of files.
Make "prefix foo" lines actually work right when foo is an absolute path,
and make sure the length calculations correspond to the output.
Provide a way to specify a file that will always be included, and a way
to omit the prologue ("$S/") on that file.

Introduce versioning to config(1). This will allow us to provide a way to
error out in a bit more friendly way when the user is trying to use
config(1) on a too old or too recent source tree.

To achieve that, introduce the "version NUMBER" statement which can be use
about anywhere in the config files. Also, use two defines, CONFIG_VERSION
(which is the actual version of binary), and CONFIG_MINVERSION, which is
the minimum version the binary supports.

Allowing a range of versions serves several purposes: first it allows me
to introduce the versioning without requiring it to be used right away in
the kernel tree, which means it will be possible to introduce new features
of config(1) rather progressively in the future. E.g., using 'no pci' in
a config file could only require the new version in that config file, so
that the rest remains compatible.

In the end, an actual bump of the main config system (i.e., in conf/files)
will only be required when e.g., ioconf.c semantics change.

(Mostly-)silently accepted on tech-kern. Error messages turned into
correct and meaningful English thanks to Tracy and Perry.

- Make sure to try aliases of removed device instances when walking
deaddevitab.
- Record the position in the config file of device instances so it is
possible to tell if a device instance was declared before or after its
parent's removal.

E.g.:

child* at parent?
no parent

will have the child instance ignored as an explicit orphan, while

no parent
child* at parent?

will error out because now the child instance is a real orphan.

That let the POSTPONED_ORPHAN regression test pass.

Add a flag to mark broken instances. That might happen when there is a
syntax error in the file, yet we want to somehow gracefully go on in order
to print out all the errors in the file, which means we have to take
special care with those structures.

Reported by Simon Burge in private mail.

- Keep track of the file in which we find the device instance
- Introduce xwarn() for delayed warnings
- Use xerror() and xwarn() in fixdevis() to notify about orphans

That way the correct file is printed when listing orphaned devices.

Reported by Juergen Hannken-Illjes in private mail.

o Keep track of negated devices in deaddevitab
o Rework do_kill_orphans() to use that information and mark explicitely
orphaned devices (i.e., the ones whose missing ancestor has been
negated)
o Make a distinction between erroneous orphans and explicit orphans.
Error out on the former, ignore the later (but print a warning when -v
is used)

Yes, now config(1) will actually stop if you comment out a parent. That
should help people still hoping adjustkernel is relevant these days :)

Rework the way orphan device instances are handled. To achieve that, keep
track of instances attaching at root, and walk down the tree of active
device instances. Then, all instances that are not marked active are
found as orphans.

Doing it that way allows us to simply ignore orphan devices, instead of
warning about them and still keep them in the configuration. Now, orphaned
instances are considered as never having existed.

In the end, this allows 'no <device> at <attachment>' to be much more
efficient, as the user doesn't have to negate all descendents of the
instance s/he actually wants to negate. Warnings are still emitted,
though.

While there, make official a side-effect of the previous lack of action
against orphaned instances: config(1) used to warn about instances that
attach at a numbered device when no instance of that device with that
number existed, even though there was a starred instance of the device.

E.g. (provided by Alan Barrett):

pciide* at pci? dev ? function ? flags 0x0000
wdc0 at isa? port 0x1f0 irq 14 flags 0x00
wdc1 at isa? port 0x170 irq 15 flags 0x00
atabus* at ata?
wd0 at atabus0 drive 0

With this commit, config(1) will no longer warn about 'wd0 at atabus0'.

Move package/cinclude/include handling from the parser to the scanner,
where it belongs. This has the side effect of fixing PR toolchain/30903.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

config(1): Make sort order deterministic.

Ensure we break ties in every case. This way, even though we use the
unstable qsort(3) library routine, the output is reproducible, no
matter what algorithm is behind qsort(3).

It would be nice if we could just use a stable sort function here,
but mergesort(3) is nonstandard, so we'd have to add it to
tools/compat, which is a big pain.

Instead, put a tie-breaking rule in every comparison function we use
with qsort, and abort() in the event of ties -- that way, we noisily
refuse to rely on unstable sort order.

While here, dispense with any question of integer overflow, and
sprinkle comments.

PR bin/58115

branches: 1.37.6;
Keep track where more objects are declared so that we can print where things
have been redefined.

branches: 1.36.14; 1.36.16;
Make attribute deselection work:
- when deselecting attributes, remove files that depend on them
- when deselecting attributes, remove devices that depend on them

Fix previous.

Fix segmentation fault.

Sort files in ${ALLFILES} in the order of parsing of `files.*'.

config(1) reads the first `files.${MACHINE}' when it encounters `machine'.
Then it includes common `files.${MACHINE_SUBARCH}', `files.${MACHINE_ARCH}',
and MI `sys/conf/files' at last. This change makes the first "file" in
`files.${MACHINE}' appear first in ${ALLFILES}.

*.o files don't need special handling at all now. Makefile.kern.inc knows
*.o input files don't need to get compiled, but need to get linked.

Now Makefile.kern.inc handles swap<kernel>.c compiling/linking nicely,
config(1) can become less hackish.

Share more code by making *.o handling less special.

Add generated *.c files (devsw.c and ioconf.c) to ${CFILES} by config(1)
internally. ${MI_CFILES} in Makefile.kern.inc is no longer needed.

After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

Experimental ``suffix rules + subdirectories'' build support (-S).

Introduce a new syntax, "buildprefix", to specify prefix of files under kernel
build subdirectory. This is not used now that everything is built at the
top of kernel build directory. It will become mandatory for source/object
files put out of kernel source tree to specify corresponding build subdirectory.
Only ``no dots'' relative path is accepted as "buildprefix".

Keep track of directory of files internally.

Use per-suffix lists where appropriate.

Merge struct objects into struct files. Keep per-suffix file lists.

Revert "Retire optional `rule' argument of `file' command". It is still used
in m68k ports.

Retire optional `rule' argument of `file' command.

This has existed since Rev. 1.1 of Torek's config(8) [1], but probably has
never been used by anyone.

(I can not imagine how a `file' like look like if `rule' is specified.)

[1] http://cvsweb.netbsd.org/bsdweb.cgi/src/usr.sbin/config.new/Attic/files.c?rev=1.1

Error out if a given path is absolute.

Pre-calc file path len/suffix. Misc. clean up.

- add RCSID's
- fix -Wconversion (except scanner)

FIT_FORCESELECT is never set anywhere.

Define an implicit attribute "netbsd" internally to collect files that don't
belong to any specific attribute.

Eventually, all operations doing "foreach (files)" can be rewritten as "foreach
(attributes) foreach (files)".

If a file is marked by an atom expression, like "file a.c foo", and if the
specified expression is an attribute, mark the file as belonging to the
attribute.

At this moment this information is not used for any purpose, but can be
traced by config -ddd.

Debug print where things are selected (by user).

branches: 1.12.2;
avoid error path SIGSEGV

branches: 1.11.2; 1.11.8;
Create a struct condexpr type to hold condition expressions, instead
of abusing struct nvlist to make trees.

(These are the a|b and a&b constructs.)

branches: 1.10.6;
Actually allow a full expression for the condition for "makeoptions" in
files.*

Patch from Yorick Hardy.

branches: 1.9.2;
Change major()/minor() to return 32-bit types again, called
devmajor_t/devminor_t, as proposed on tech-kern.
This avoids 64-bit arithmetics and 64-bit printf formats in parts
of the kernel where it is not really useful, and helps clarity.

rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Be more informative about duplicated device names.

branches: 1.6.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

CID-4275: don't leak `fixdevmtab'

Coverity CID 175: Avoid memory leak; add new function to destroy a hash table.

Now that this is part of tools, we don't need the extra build glue.

Remove special hack for swapnetbsd.c when generating the makefile.
Instead, just add it to the list of files.
Make "prefix foo" lines actually work right when foo is an absolute path,
and make sure the length calculations correspond to the output.
Provide a way to specify a file that will always be included, and a way
to omit the prologue ("$S/") on that file.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

With config(1) as it exists today, a kernel Makefile is able to implement
logic on kernel options so long as those options are not defflag'd or
defparam'd. This works because such options are automatally added to the
IDENT var in the kernel Makefile as a preprocessor define, and the Makefile
can then do an operation like:

.if !empty(IDENT:M-DSOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Unfortunately, this precludes making it possible to generate a compile-time
dependency on SOMECOOLCPUOPTION, or having SOMECOOLCPUOPTION imply another
kernel config option using the normal config(1) option dependency function.

Fix this by introducing a new option description keyword: mkflagvar. This
keyword marks an already defflag'd option as wanting a kernel Makefile var
defined if that option is selected in the kernel config file. So:

defflag opt_coolcpu.h SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION
mkflagvar SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION

will cause:

KERNEL_OPT_SOMECOOLCPUOPTION="1"
KERNEL_OPT_ANOTHERCOOLCPUOPTION="1"

...to be emitted into the kernel Makefile if those options are, in fact,
selected with "options ..." in the kernel config file, thus allowing for
a compile-time dependency on the option to be generated in addition to
Makefile logic, which now looks like:

.if !empty(KERNEL_OPT_SOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Add debug output for ENDDEFS.

This makes it more clear in the debug output where config switched
from definitions to selections.

Keep track where more objects are declared so that we can print where things
have been redefined.

branches: 1.54.14; 1.54.16;
Accept "-no" as a "no" that does not cause errors if the object deleted
does not exist.

Extend syntax of config phrase, a quoted string instead of a device name
is passed as root specification string. This can be used to specify a
wedge by name.

Introduce a new syntax, "buildprefix", to specify prefix of files under kernel
build subdirectory. This is not used now that everything is built at the
top of kernel build directory. It will become mandatory for source/object
files put out of kernel source tree to specify corresponding build subdirectory.
Only ``no dots'' relative path is accepted as "buildprefix".

Merge struct objects into struct files. Keep per-suffix file lists.

Restore the old behavior of "maxusers" to allow the value to be overriden.
Problem reported by John D. Baker.

Convert MAXUSERS, take 2.

Define MAXUSERS in opt_param.h and include it from sys/param.h, for more
accurate dependency. Don't define -DMAXUSERS and add it to ${CPPFLAGS}.
config(5)'s "maxusers" syntax doesn't change.

(The version just before revert was OK, but I had something bad in my obj
dirs.)

Revert "Retire optional `rule' argument of `file' command". It is still used
in m68k ports.

Retire optional `rule' argument of `file' command.

This has existed since Rev. 1.1 of Torek's config(8) [1], but probably has
never been used by anyone.

(I can not imagine how a `file' like look like if `rule' is specified.)

[1] http://cvsweb.netbsd.org/bsdweb.cgi/src/usr.sbin/config.new/Attic/files.c?rev=1.1

makedev returns ~uint64_t, but val is int64_t. Cast for -Wconversion.

config(1), config(5): Introduce "select"

o Introduce a new selection directive "select" to select an attribute (as a
module) and its dependencies.
o Support "no select" too.
o Stop abusing "options" to select an attribute.
o Bump config(1) version.

- add RCSID's
- fix -Wconversion (except scanner)

Allow dependency on undefined attributes, so that attribute definitions can
be written out-of-order, like:

# b is not defined yet
define a: b
file a.c a

# b is defined here
define b
fine b.c

Split plain/interface/devclass attribute initializers.

Sort definition/selection rules to clarify which is which.

The word "configuration" is too ambiguous; use "selection" meaning that
user selects some of provided "definitions".

branches: 1.39.2;
implement "no ident".

branches: 1.38.2;
ioconf/pseudo-root have been in here unchanged for 3+ years and their use
has spread from rump kernels into dozens of kernel modules. Let's pretend
we don't want to see config yell about them being experimental every time.

branches: 1.37.2;
Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Move locator lists to their own data structure. This can use more tidying;
it is not clear to me at the moment what the "string" and "num" values
pushed around in locator lists are supposed to actually mean.

More naming improvements.

Create a struct condexpr type to hold condition expressions, instead
of abusing struct nvlist to make trees.

(These are the a|b and a&b constructs.)

Rename a few more things; try to use the same terminology as config(5).

Create a new type struct attrlist to manage lists of attributes,
instead of using struct nvlist.

(struct nvlist holds lists (or trees!) of semi-arbitrary stuff with no
structure and almost no type safety; it should go away.)

Strengthen the logic that wraps allocations going on the parser stack.
(This allows cleaning up when a syntax error occurs, because yacc just
dumps the parser stack on the floor.)

The new logic can handle arbitrary data types instead of being limited
to struct nvlist.

Write out the expression grammar for dependency constraints instead of
using precedence rules. Precedence rules (besides being a hack) often
lead to unexpected/unwanted behavior in corner cases.

Sort productions in grammar. No functional change.

Simplify some things. Rename some productions for clarity.

A bit more of previous.

Reformat according to (my) basic standards for yacc grammars.
No functional change.

Comments.

branches: 1.24.6;
For the simple cases, augment device-major with information on how
a driver expects /dev/node -> minor mappings to go and include that
information in devsw_conv.
(no, I didn't plow through all the MD majors files)

print ioconf WARNING in caps too

Add a highly experimental pseudo-root feature to be used in conjuction
with the also-experimental ioconf keyword. pseudo-root allows to
specify a root at any point in the device tree without having
attachments from the actual root.

For example, instead of having a config file like this:
mainbus0 at root
bus_a? at mainbus0
bus_b? at bus_a*
device7 at bus_b?

You can have one like this:
pseudo-root bus_b*
device7 at bus_b?

This will produce the relevant ioconf.c glue for device number 7
only instead of the whole 9 yards from root. Perhaps needless to
say, this can be used to generate snippets of config glue for
modules and, let's not deny that my main motivation for doing this,
rump components.

This is part 2/3 of my modifications to config (the last part is
autogenerating source file lists and component Makefiles).

No strong objection from cube (after a little pressuring ;), but
like he said, the implementation will most likely need some more
tweaking and may not work correctly under all pseudo-root uses yet.

Remove fs_foo.h support from deffs now that nothing uses it.

Introduce experimental support for ioconf-only configuration files.
This is done by giving the "ioconf" keyword in the config file.
As a result, config produces only ioconf.c and locators.h. Currently,
only "monolithic" configurations with the device path starting from
root are supported. Eventually, the goal is to support a local
root in the input file from any point along the device tree using
files.* in our kernel tree. This will make autogenerating the
config glue for device modules possible instead of having to write
it by hand like is currently required (yes, it sounds simple to
implement, but ...).

reviewed by cube.
(the next part will demand major discussions with you, so prepare ;)

Actually allow a full expression for the condition for "makeoptions" in
files.*

Patch from Yorick Hardy.

branches: 1.18.2;
rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Disable use of pseudo-devices defined "defpseudo" with interface
attributes, but provide compatibility for older source trees (according
to the "version" in conf/files and other files.*.
reviewed by cube

Seperate pseudo-devs which don't use device_t et al. (the classical
ones, mostly network pseudo interfaces) and those which use autoconf
to attach instances dynamically, or to attach child devices.
Use a new keyword "defpseudodev" for the latter.
Issue a warning if interface attributes are used with "old" pseudodevs,
but let it work as before for now. This will be removed soon.
approved by cube

branches: 1.15.6; 1.15.8;
Don't defattr machinearch if it's NULL. Pointed out by
Juergen Hannken-Illjes.

Make the "machine" statement define attributes for `machine', `arch' and
all the `subarch' given as parameters. This is useful in combination with
ifdef...

branches: 1.13.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Handle defopt lists properly, as ':=' syntax introduce an additional
element... For that, introduce nvcat(nv1, nv2).

Add new syntaxic sugar to provide a value in lint (i.e., config -L)
configurations for defparams. It will be used _only_ by config -L.

This will greatly help automating tests with lint configurations,
hopefully.

Add support for specifying a default value on a 'defparam' line.
If a default is specified then the option is always defined in the
corresponding .h file.
Particularly useful for parameters where the default action isn't the
same a defining the value zero, given that the current use of #if defined(opt)
is open to problems wehere the relevant .h file isn't actually included, so
requiring an option to always have a value makes sense.
Also included (but commented out) is code that adds a global symbol to
the object file to stop objects compiled with different values for the
same option being linked together - I'm not quite happy with it yet!

Allow deffs to take dependencies like defflag and defopt

Now that this is part of tools, we don't need the extra build glue.

D'uh. Forgot the file name argument in the "obsolete" statements.

Introduce two new statements:

obsolete defflag <option> [, <option> [, ...]]
obsolete defparam <option> [, <option> [, ...]]

The two statements actually do the same thing (there could be only one),
but it makes things less cryptic that way. The optional ": deps" part of
a 'defflag' or 'defparam' still has to be dropped when it gets obsoleted.

When the user has 'options OBSOLETE_OPTION' in his configuration file, it
is ignored (that is, opt_*.h files are generated as if it wasn't there),
and the user gets a warning about it.

Bump version to 20060525.

When someone first uses that syntax in the tree, a "version 20060525"
statement should be added before its occurrence, preferably at the top
of sys/conf/files.

Suggested by Matt Thomas a few months ago.

branches: 1.5.2;
Add 'no config <name>'.

Introduce versioning to config(1). This will allow us to provide a way to
error out in a bit more friendly way when the user is trying to use
config(1) on a too old or too recent source tree.

To achieve that, introduce the "version NUMBER" statement which can be use
about anywhere in the config files. Also, use two defines, CONFIG_VERSION
(which is the actual version of binary), and CONFIG_MINVERSION, which is
the minimum version the binary supports.

Allowing a range of versions serves several purposes: first it allows me
to introduce the versioning without requiring it to be used right away in
the kernel tree, which means it will be possible to introduce new features
of config(1) rather progressively in the future. E.g., using 'no pci' in
a config file could only require the new version in that config file, so
that the rest remains compatible.

In the end, an actual bump of the main config system (i.e., in conf/files)
will only be required when e.g., ioconf.c semantics change.

(Mostly-)silently accepted on tech-kern. Error messages turned into
correct and meaningful English thanks to Tracy and Perry.

Introduce two new statements:

no device at <attachment>

<attachment> can take two forms: either numbered/wildcarded, in which
case only exactly matching instances will be removed, or plain (with
no number or wildcard), in which case all matching instances will be
removed.

When <attachment> is a plain interface attribute, all instances using
that attribute (either directly or through an explicit device) will be
removed.

E.g.:
auich* at pci? dev ? function ?
audio0 at audiobus?
audio1 at auich?
audio* at auich0

no device at auich0 -> removes audio*
no device at auich? -> removes audio1
no device at auich -> removes audio1 _and_ audio*
no device at audiobus? -> removes audio0
no device at audiobus -> removes audio0, audio1 and audio*

no <device>

As in the previous case, <device> can either be numbered/starred, in
which case all exactly matching instances are removed, or plain, in
which case all instances of the device are removed.

E.g.: (continuing previous example)

no audio* -> removes 'audio* at auich0'
no audio -> removes all audio instances

Move package/cinclude/include handling from the parser to the scanner,
where it belongs. This has the side effect of fixing PR toolchain/30903.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

clear: trailing whitespaces

branches: 1.11.28;
- add RCSID's
- fix -Wconversion (except scanner)

Fix typos in previous.

Make hash capable of taking two key strings.

branches: 1.8.10;
Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Introduce type-safe wrappers around the hash tables. Use them for a
selected set of tables affected by the next nvlist cleanup in the
works.

branches: 1.6.6;
Fix WARNS=4 issues (-Wcast-qual -Wsign-compare -Wshadow)

branches: 1.5.20;
Fix a typo in ht_free() which caused the last list of the hashtable to be ignored.
It caused config(1) to crash on the following assert() if the list was used.

This bug shows up after the recent change in files.c which fixed a memory leak.

pointed out by dogcow@
reviewed by cube@

Coverity CID 175: Avoid memory leak; add new function to destroy a hash table.

There isn't really any point in the 'union' for nv_str and nv_ptr, all it
does is cause confusion when I try to add default values for defparam.
Also if malloc()/free() aren't efficient enough for us, we ought to fix
them rather than having local free list - so kill the local free lists.

Now that this is part of tools, we don't need the extra build glue.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

clear: trailing whitespaces

branches: 1.16.4;
PR/56453: Wataru Ashihara: config(1): Assertion fails in config -L
Only compare pointers when one is found.

- add RCSID's
- fix -Wconversion (except scanner)

branches: 1.14.10;
Use C89 function definitions

Remove the NV_OBSOLETE flag, which is no longer needed. While here,
update an outdated comment about condition expressions.

Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Introduce type-safe wrappers around the hash tables. Use them for a
selected set of tables affected by the next nvlist cleanup in the
works.

Move locator lists to their own data structure. This can use more tidying;
it is not clear to me at the moment what the "string" and "num" values
pushed around in locator lists are supposed to actually mean.

Create a new type struct attrlist to manage lists of attributes,
instead of using struct nvlist.

(struct nvlist holds lists (or trees!) of semi-arbitrary stuff with no
structure and almost no type safety; it should go away.)

branches: 1.8.6;
For {pseudo-,}devices, check if an instance of the pseudo or the considered
attachment of the device has already been done by the template config file.
If so, don't emit a line. Reported by John Nemeth.

rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Convert to new 2 clause license

branches: 1.5.6;
Some more licence clean up.

branches: 1.4.4;
Change the name of the resulting kernel binary, because "lint" confuses
our Makefiles. While there, add a couple comments to explain how all
attachments are found.

Add new syntaxic sugar to provide a value in lint (i.e., config -L)
configurations for defparams. It will be used _only_ by config -L.

This will greatly help automating tests with lint configurations,
hopefully.

Print pseudo-devices in their own section and print all of them (d'oh).

Introduce a new option to config(1): -L. What it does is takes a regular
configuration file (although it is meant to be used mostly with
std.${ARCH}), and prints out a configuration file that includes it and
select every single option and parameter, and define an instance for every
single possible attachment. IOW, selects everything, into a would-be
"LINT" config.

The resulting config really isn't meant to be runnable, but should be
somewhat buildable on most archs. It still needs adjustments because some
options are peculiar (e.g., ACPI_DSDT_FILE wants an existing file as its
value), so it's not yet possible to do "config -L; config; build" in an
automated way.

clear: trailing whitespaces

branches: 1.101.2;
With config(1) as it exists today, a kernel Makefile is able to implement
logic on kernel options so long as those options are not defflag'd or
defparam'd. This works because such options are automatally added to the
IDENT var in the kernel Makefile as a preprocessor define, and the Makefile
can then do an operation like:

.if !empty(IDENT:M-DSOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Unfortunately, this precludes making it possible to generate a compile-time
dependency on SOMECOOLCPUOPTION, or having SOMECOOLCPUOPTION imply another
kernel config option using the normal config(1) option dependency function.

Fix this by introducing a new option description keyword: mkflagvar. This
keyword marks an already defflag'd option as wanting a kernel Makefile var
defined if that option is selected in the kernel config file. So:

defflag opt_coolcpu.h SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION
mkflagvar SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION

will cause:

KERNEL_OPT_SOMECOOLCPUOPTION="1"
KERNEL_OPT_ANOTHERCOOLCPUOPTION="1"

...to be emitted into the kernel Makefile if those options are, in fact,
selected with "options ..." in the kernel config file, thus allowing for
a compile-time dependency on the option to be generated in addition to
Makefile logic, which now looks like:

.if !empty(KERNEL_OPT_SOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Avoid depending on common symbols.

Keep track where more objects are declared so that we can print where things
have been redefined.

branches: 1.98.2;
yydebug is now defined only if YYDEBUG is

branches: 1.97.2; 1.97.4;
- make the level computation consistent
- keep going only if things changed.

use a reference count to avoid deleting psrefs still in use.

- Instead of checking the recursion level before we recurse, check in on
function entry.
- Always decrement the level and reset levelparent on exit.

add more debugging, no functional change.

avoid creating infinite loops.

When deleting orphans detect parent<->child loops and break them.
"active" is not a boolean, use the right comparison.

branches: 1.91.6;
Drop main() prototype.

Accept "-no" as a "no" that does not cause errors if the object deleted
does not exist.

*.o files don't need special handling at all now. Makefile.kern.inc knows
*.o input files don't need to get compiled, but need to get linked.

After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

Fix build.

In `-S' mode, create symlinks from outside of $S to build directory, for
files that are placed outside of $S, so that `Makefile.kern.inc' can see all
files under either $S or build directory.

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

Experimental ``suffix rules + subdirectories'' build support (-S).

Prepare to build subdirectories. Not enabled yet.

Allow redefinition of options (with only a warning message). Removal
of non-included options was already only a warning.

This also affects redefinition and removal of file-system and makeoption
entries.

Reviewed and OKd by uebayasi@

Restore the old behavior of "maxusers" to allow the value to be overriden.
Problem reported by John D. Baker.

Define `makeoptions' values as parameters with "makeoptions_" prefix too.
C sources are encouraged to include relevant opt_*.h files rather than relying
on -DXXX passed via makefiles.

Revert MAXUSERS. CVS is hard.

Fix previous.

Global paths looked up at output stage (mk*()) are relative to `srcdir'.
Ignore "prefix" by not calling sourcepath().

Add missed munmap(2) in extract_config

remove stray newlines from errors/warnings

appease gcc-4.5.4

config(1): Rename internal flag to enable modular build

To make -Wconversion happy, cast return value of tolower/toupper to
char.

config(1), config(5): Introduce "select"

o Introduce a new selection directive "select" to select an attribute (as a
module) and its dependencies.
o Support "no select" too.
o Stop abusing "options" to select an attribute.
o Bump config(1) version.

- add RCSID's
- fix -Wconversion (except scanner)

Keep track of attribute (module) dependency using hash2.

Revert implicit attribute (module) declaration for options (flag/param).

Flag and param are to change contents of attributes (modules). These don't
have code (*.[coS]).

Filesystem definition still defines an attribute (module) implicitly.

More debug messages.

After yyparse() is done, print not file but stage for debug.

Enable debug for tools.

Remove an unsed var.

Revert a mistakenly added, superflous attribute match test that caused some
valid options to get lost.

Select attributes for filesystem options too.

If an option is selected, select the matching (lowercased) attribute. Handle
dependencies too.

Implicitly define attribute for options.

Implicitly define attribute for filesystem.

Steal -M to enable "modular" build.

Always select the fallback allattr.

Debug print where things are selected (by user).

Add debug print function.

branches: 1.54.2;
Treat an undefined option as a warning instead of an error when undoing an
option.

Allows one to win with declarations like:
no options MSGBUFSIZE
options MSGBUFSIZE="128*1024"

Briefly discussed with cube@, lukem@ and martin@.

Sort options in usage.

Add two new options, -U and -D, that can be used to define "makeoptions"
on the config command line. While there, rename the undocumented (internal)
parser debug option from -D to -d.
Discussed on tech-toolchain.

branches: 1.51.2;
go back to using file:line: from uwe.

make config errors look more like other programs: <file>,<line>:
instead of <file>:<line>:

branches: 1.49.2;
Don't support sysctl for the tools version of config. It is too complicated
to do so, and config -x is not used during the build.

make config -x look at the booted kernel first.

Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Introduce type-safe wrappers around the hash tables. Use them for a
selected set of tables affected by the next nvlist cleanup in the
works.

Create a struct condexpr type to hold condition expressions, instead
of abusing struct nvlist to make trees.

(These are the a|b and a&b constructs.)

Create a new type struct attrlist to manage lists of attributes,
instead of using struct nvlist.

(struct nvlist holds lists (or trees!) of semi-arbitrary stuff with no
structure and almost no type safety; it should go away.)

branches: 1.43.2;
When making the symlink for machine -> arch/foo/include
also make foo -> arch/foo/include. This allows one to include <foo/bar.h>
include <machine/bar.h> which can be handy.

Some drivers want to include ioconf.h, so generate it if doing
"ioconf" config.

Add a highly experimental pseudo-root feature to be used in conjuction
with the also-experimental ioconf keyword. pseudo-root allows to
specify a root at any point in the device tree without having
attachments from the actual root.

For example, instead of having a config file like this:
mainbus0 at root
bus_a? at mainbus0
bus_b? at bus_a*
device7 at bus_b?

You can have one like this:
pseudo-root bus_b*
device7 at bus_b?

This will produce the relevant ioconf.c glue for device number 7
only instead of the whole 9 yards from root. Perhaps needless to
say, this can be used to generate snippets of config glue for
modules and, let's not deny that my main motivation for doing this,
rump components.

This is part 2/3 of my modifications to config (the last part is
autogenerating source file lists and component Makefiles).

No strong objection from cube (after a little pressuring ;), but
like he said, the implementation will most likely need some more
tweaking and may not work correctly under all pseudo-root uses yet.

reflect previous change in comment

Remove fs_foo.h support from deffs now that nothing uses it.

Fix off-by-one (config -x works again) and improve editing artifacts
(actually the whole "config -x must be used alone" change was made
unnecessary along the way. cest la vie).

Introduce experimental support for ioconf-only configuration files.
This is done by giving the "ioconf" keyword in the config file.
As a result, config produces only ioconf.c and locators.h. Currently,
only "monolithic" configurations with the device path starting from
root are supported. Eventually, the goal is to support a local
root in the input file from any point along the device tree using
files.* in our kernel tree. This will make autogenerating the
config glue for device modules possible instead of having to write
it by hand like is currently required (yes, it sounds simple to
implement, but ...).

reviewed by cube.
(the next part will demand major discussions with you, so prepare ;)

Now that condmkopttab isn't a hash table anymore, don't initialise it with
ht_new()... So actually rename it to condmkoptions to avoid confusion.

Reported build failure on amd64 (strange that I wouldn't get it on i386) on
current-users by Kurt Schreiner.

Actually allow a full expression for the condition for "makeoptions" in
files.*

Patch from Yorick Hardy.

Error out if one f the mkxxx() prints an error.

branches: 1.33.2;
rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Don't complain about a missing 'file-system' directive. File system code
can be provided by modules.

Adjust previous; apb@ points out that _PATH_TMP is provided by compat_defs.h;
thus, this shouldn't need a paranoid fallback definition.
And, use _PATH_TMP also in the second place that was using a hardwired "/tmp".

Fix ridiculously small filename buffer for temporary file and resulting
truncation if $TMPDIR is set; also, fix an error path to not leave the
temporary file behind. From Gao Ya'nan in PR 39746.
While here, use _PATH_TMP instead of a hardwired "/tmp".

In find_declared_option, use the correct hash for file-systems. That way,
information about dependencies are actually processed.

Remove the \n and tabs from the __COPYRIGHT() strings.
Tweak to use a consistent format.

Disable use of pseudo-devices defined "defpseudo" with interface
attributes, but provide compatibility for older source trees (according
to the "version" in conf/files and other files.*.
reviewed by cube

Seperate pseudo-devs which don't use device_t et al. (the classical
ones, mostly network pseudo interfaces) and those which use autoconf
to attach instances dynamically, or to attach child devices.
Use a new keyword "defpseudodev" for the latter.
Issue a warning if interface attributes are used with "old" pseudodevs,
but let it work as before for now. This will be removed soon.
approved by cube

branches: 1.25.4; 1.25.6;
convert __attribute__s to applicable cdefs.h macros

Don't pollute fsopttab with duplicate options for the sake of a test that
has probably been gone for a very long time. Produces identical output.

branches: 1.23.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Add new syntaxic sugar to provide a value in lint (i.e., config -L)
configurations for defparams. It will be used _only_ by config -L.

This will greatly help automating tests with lint configurations,
hopefully.

One %s is enough to print one string.

Introduce a new option to config(1): -L. What it does is takes a regular
configuration file (although it is meant to be used mostly with
std.${ARCH}), and prints out a configuration file that includes it and
select every single option and parameter, and define an instance for every
single possible attachment. IOW, selects everything, into a would-be
"LINT" config.

The resulting config really isn't meant to be runnable, but should be
somewhat buildable on most archs. It still needs adjustments because some
options are peculiar (e.g., ACPI_DSDT_FILE wants an existing file as its
value), so it's not yet possible to do "config -L; config; build" in an
automated way.

Make deffilesystem check all defined options, not just defopttab (which
ought to be empty nowadays).

Include <sys/cdefs.h> and use __used on config[] instead of the ifdef dance.
Prevents gcc4 from g/c'ing embedded config from VAX kernels.

Instead of trying to check the return value of every fprintf() and fputs()
call - which makes the code completely impossibly to follow, call fflush()
and ferror() just prior to calling fclose().
This has the advantage of actually detecting the any write errors, since
the output is block bufferred and will typically not happen during fprintf()
but only during the fclose() - where it is difficult to report.
I also singlilarly refuse to add (void) casts to every printf call in the
system - since it almost never makes any sense to look at the return value
(unless you want to know how many bytes were actually writtem).

Remove unused local variable.

Allow deffs to take dependencies like defflag and defopt

Now that this is part of tools, we don't need the extra build glue.

Introduce two new statements:

obsolete defflag <option> [, <option> [, ...]]
obsolete defparam <option> [, <option> [, ...]]

The two statements actually do the same thing (there could be only one),
but it makes things less cryptic that way. The optional ": deps" part of
a 'defflag' or 'defparam' still has to be dropped when it gets obsoleted.

When the user has 'options OBSOLETE_OPTION' in his configuration file, it
is ignored (that is, opt_*.h files are generated as if it wasn't there),
and the user gets a warning about it.

Bump version to 20060525.

When someone first uses that syntax in the tree, a "version 20060525"
statement should be added before its occurrence, preferably at the top
of sys/conf/files.

Suggested by Matt Thomas a few months ago.

s/vax/__vax__/

deal with the vax gcc that does not understand __attribute__ used.

branches: 1.10.2;
gcc-4.0 needs attribute used. Variables with attribute unused get removed.

Close file on error. CID 176.

The loop detection code could not detect the negation of an instance that
made a device its own parent. Add a test that checks that and stop looping
in that special case (after all, everything is already being handled by the
parent instance).

Reported by Jukka Salmi on current-user.

Invert the meaning of a comment to match reality. Pointed out by uwe@.

Stop walking the tree when two conditions are met: we've seen the device
already (one of its instances has been changed), and we have made no change
on any of the instances.

Previously, it stopped as soon as it detected the device had been seen.
While all the instances of the device at stake were eventually seen, the
same wasn't true for its children...

Fixes hpcmips's GENERIC.

- Make sure to try aliases of removed device instances when walking
deaddevitab.
- Record the position in the config file of device instances so it is
possible to tell if a device instance was declared before or after its
parent's removal.

E.g.:

child* at parent?
no parent

will have the child instance ignored as an explicit orphan, while

no parent
child* at parent?

will error out because now the child instance is a real orphan.

That let the POSTPONED_ORPHAN regression test pass.

Don't return immediately when detecting that a devbase has already been
seen: aliases might have to be activated, too. Fixes GENERIC of alpha
and amiga.

o Keep track of negated devices in deaddevitab
o Rework do_kill_orphans() to use that information and mark explicitely
orphaned devices (i.e., the ones whose missing ancestor has been
negated)
o Make a distinction between erroneous orphans and explicit orphans.
Error out on the former, ignore the later (but print a warning when -v
is used)

Yes, now config(1) will actually stop if you comment out a parent. That
should help people still hoping adjustkernel is relevant these days :)

Rework the way orphan device instances are handled. To achieve that, keep
track of instances attaching at root, and walk down the tree of active
device instances. Then, all instances that are not marked active are
found as orphans.

Doing it that way allows us to simply ignore orphan devices, instead of
warning about them and still keep them in the configuration. Now, orphaned
instances are considered as never having existed.

In the end, this allows 'no <device> at <attachment>' to be much more
efficient, as the user doesn't have to negate all descendents of the
instance s/he actually wants to negate. Warnings are still emitted,
though.

While there, make official a side-effect of the previous lack of action
against orphaned instances: config(1) used to warn about instances that
attach at a numbered device when no instance of that device with that
number existed, even though there was a starred instance of the device.

E.g. (provided by Alan Barrett):

pciide* at pci? dev ? function ? flags 0x0000
wdc0 at isa? port 0x1f0 irq 14 flags 0x00
wdc1 at isa? port 0x170 irq 15 flags 0x00
atabus* at ata?
wd0 at atabus0 drive 0

With this commit, config(1) will no longer warn about 'wd0 at atabus0'.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

clear: trailing whitespaces

branches: 1.15.26;
Only emit bdevsw externs for entries in the bdev table, rather than for
entries in the cdev table.

branches: 1.14.2;
After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

merge a bit of duplicated code.

- add RCSID's
- fix -Wconversion (except scanner)

branches: 1.10.24;
Emit compatible struct devsw_conv when version is older than 20100430.

For the simple cases, augment device-major with information on how
a driver expects /dev/node -> minor mappings to go and include that
information in devsw_conv.
(no, I didn't plow through all the MD majors files)

output __arraycount instead of homegrown macro

Change major()/minor() to return 32-bit types again, called
devmajor_t/devminor_t, as proposed on tech-kern.
This avoids 64-bit arithmetics and 64-bit printf formats in parts
of the kernel where it is not really useful, and helps clarity.

Remove clause 3 and 4 from TNF licenses

branches: 1.5.4;
Use a common function to generate the "MACHINE GENERATED" comment,
and ensure mkswap.c adds it too.

branches: 1.4.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Instead of trying to check the return value of every fprintf() and fputs()
call - which makes the code completely impossibly to follow, call fflush()
and ferror() just prior to calling fclose().
This has the advantage of actually detecting the any write errors, since
the output is block bufferred and will typically not happen during fprintf()
but only during the fclose() - where it is difficult to report.
I also singlilarly refuse to add (void) casts to every printf call in the
system - since it almost never makes any sense to look at the return value
(unless you want to know how many bytes were actually writtem).

allow configuration without swap.

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

clear: trailing whitespaces

branches: 1.29.28;
After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

move driver attach declarations to ioconf.h

appease gcc-4.5.4

To make -Wconversion happy, cast return value of tolower/toupper to
char.

remove overflow test that triggered for negative values; we only care
about the hash here.

print more info about the overflow

- add RCSID's
- fix -Wconversion (except scanner)

branches: 1.21.10;
Give option definitions their own data structure instead of using nvlists.
(and using messy hacks to make up for nvlists not holding quite the right
things)

Introduce type-safe wrappers around the hash tables. Use them for a
selected set of tables affected by the next nvlist cleanup in the
works.

Move locator lists to their own data structure. This can use more tidying;
it is not clear to me at the moment what the "string" and "num" values
pushed around in locator lists are supposed to actually mean.

branches: 1.18.6;
Some drivers want to include ioconf.h, so generate it if doing
"ioconf" config.

Introduce experimental support for ioconf-only configuration files.
This is done by giving the "ioconf" keyword in the config file.
As a result, config produces only ioconf.c and locators.h. Currently,
only "monolithic" configurations with the device path starting from
root are supported. Eventually, the goal is to support a local
root in the input file from any point along the device tree using
files.* in our kernel tree. This will make autogenerating the
config glue for device modules possible instead of having to write
it by hand like is currently required (yes, it sounds simple to
implement, but ...).

reviewed by cube.
(the next part will demand major discussions with you, so prepare ;)

Dependency on config_time.src was removed today from Makefile.kern.inc, so
starting from that date it's not necessary to create it. (Of course that
won't happen until config version is bumped for some other reason.)

Fix WARNS=4 issues (-Wcast-qual -Wsign-compare -Wshadow)

branches: 1.14.2;
rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Don't pollute fsopttab with duplicate options for the sake of a test that
has probably been gone for a very long time. Produces identical output.

branches: 1.12.4;
Enable the code that adds global constants to every kernel object file
that uses each configuration parameter.
This will stop kernel objects build with different options (that come from
config files) being linked together or loaded as a LKM.
Currently some options that have an effect on the kernel DDI/KI are passed
directly as parameters to cc.
An aim (for anyone adequately bored) would be to reduce the number of these
constants that appear in each .o file. .o files with the same constants
should be sharable between kernels (even between XEN and i386).

Use '.equ' instead of '.set' when defining global constants.
On alpha '.set' has a non-standard purpose.

- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Instead of trying to check the return value of every fprintf() and fputs()
call - which makes the code completely impossibly to follow, call fflush()
and ferror() just prior to calling fclose().
This has the advantage of actually detecting the any write errors, since
the output is block bufferred and will typically not happen during fprintf()
but only during the fclose() - where it is difficult to report.
I also singlilarly refuse to add (void) casts to every printf call in the
system - since it almost never makes any sense to look at the return value
(unless you want to know how many bytes were actually writtem).

make this compile again.

Add support for specifying a default value on a 'defparam' line.
If a default is specified then the option is always defined in the
corresponding .h file.
Particularly useful for parameters where the default action isn't the
same a defining the value zero, given that the current use of #if defined(opt)
is open to problems wehere the relevant .h file isn't actually included, so
requiring an option to always have a value makes sense.
Also included (but commented out) is code that adds a global symbol to
the object file to stop objects compiled with different values for the
same option being linked together - I'm not quite happy with it yet!

Now that this is part of tools, we don't need the extra build glue.

Introduce two new statements:

obsolete defflag <option> [, <option> [, ...]]
obsolete defparam <option> [, <option> [, ...]]

The two statements actually do the same thing (there could be only one),
but it makes things less cryptic that way. The optional ": deps" part of
a 'defflag' or 'defparam' still has to be dropped when it gets obsoleted.

When the user has 'options OBSOLETE_OPTION' in his configuration file, it
is ignored (that is, opt_*.h files are generated as if it wasn't there),
and the user gets a warning about it.

Bump version to 20060525.

When someone first uses that syntax in the tree, a "version 20060525"
statement should be added before its occurrence, preferably at the top
of sys/conf/files.

Suggested by Matt Thomas a few months ago.

branches: 1.4.2;
Fix various resource leaks, all happenning on error situations (IOW,
shortly before exit(1)). CIDs 181, 182, 183, 184 and 185.

in locators.h, also emit definitions about the number of locators per
interface attribute

don't put the xxxcf_locnames[] into the global namespace

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

config(1): Make sort order deterministic.

Ensure we break ties in every case. This way, even though we use the
unstable qsort(3) library routine, the output is reproducible, no
matter what algorithm is behind qsort(3).

It would be nice if we could just use a stable sort function here,
but mergesort(3) is nonstandard, so we'd have to add it to
tools/compat, which is a big pain.

Instead, put a tie-breaking rule in every comparison function we use
with qsort, and abort() in the event of ties -- that way, we noisily
refuse to rely on unstable sort order.

While here, dispense with any question of integer overflow, and
sprinkle comments.

PR bin/58115

branches: 1.35.6; 1.35.14;
Don't print instances we've already printed.

Allow multiple attachments methods to the same child+parent combination:

foo* at bar? with baz
foo* at bar? with barf

Do this by scanning the list of iba's and allocating a new cfparent for
each. Keep track of the shared parent+child combinations by using the
same id for them.

In ioconfname mode, #define IOCONF as the ioconf token.

Avoids having to retype the name to call config_init/fini_component().

After thought, revert "Generate *.c files under conf/". Generate *.c files
under top build directory. *.c files are never placed just under $S/. Keep
this exclusiveness.

In `-S' mode, generate *.c files under conf/ subdirectory. Register generated
.c files to the `files' list internally.

Global paths looked up at output stage (mk*()) are relative to `srcdir'.
Ignore "prefix" by not calling sourcepath().

move driver attach declarations to ioconf.h

config(1): Don't emit unused empty loc[]

config(1): Emit pdevinit[] into .rodata

config(1): Set NULL to unused cfdata::cf_loc

config(1): Better indent of cfdata in ioconf.c

config(1): KNF generated ioconf.c

config(1): Emit only referenced interface attributes in ioconf.c

- add RCSID's
- fix -Wconversion (except scanner)

branches: 1.21.10;
Move locator lists to their own data structure. This can use more tidying;
it is not clear to me at the moment what the "string" and "num" values
pushed around in locator lists are supposed to actually mean.

Create a new type struct attrlist to manage lists of attributes,
instead of using struct nvlist.

(struct nvlist holds lists (or trees!) of semi-arbitrary stuff with no
structure and almost no type safety; it should go away.)

branches: 1.19.4;
Fix build on Solaris 10.

branches: 1.18.2;
Put cfdata* in a similar namespace with cfdriver** cfattachinit*.
This was nagging me already yesterday, but I decided to leave it
alone for better compat with old ioconf.c builds. But as it turns
out, new code depends on newly built ioconf.c with a new config(1)
anyway, so renaming is not an issue (at least not at this stage).

While renaming, namespace cfdriver and cfattach in "ioconf" rather
than "comp", since the former reflects the config(5) keyword.

Generate cfdriver and cfattach lists for ioconf configs.

... aaand make the emit-order slightly more sensible for ioconf.c to
actually work again.

Introduce experimental support for ioconf-only configuration files.
This is done by giving the "ioconf" keyword in the config file.
As a result, config produces only ioconf.c and locators.h. Currently,
only "monolithic" configurations with the device path starting from
root are supported. Eventually, the goal is to support a local
root in the input file from any point along the device tree using
files.* in our kernel tree. This will make autogenerating the
config glue for device modules possible instead of having to write
it by hand like is currently required (yes, it sounds simple to
implement, but ...).

reviewed by cube.
(the next part will demand major discussions with you, so prepare ;)

Fix WARNS=4 issues (-Wcast-qual -Wsign-compare -Wshadow)

branches: 1.13.2;
Change major()/minor() to return 32-bit types again, called
devmajor_t/devminor_t, as proposed on tech-kern.
This avoids 64-bit arithmetics and 64-bit printf formats in parts
of the kernel where it is not really useful, and helps clarity.

Better whitespace in ioconf.c output.

rename nv_int -> nv_num and make it long long, so that dev_t will fit when
we change it.

Use a common function to generate the "MACHINE GENERATED" comment,
and ensure mkswap.c adds it too.

branches: 1.9.4;
- rename {x,}error -> cfg{x,}error and {x,}warn -> cfg{x,}warn so that we
can use err(3) and warn(3) to display errors instead of using inconsistent
fprintf's.
- remove emalloc/estrdup etc. We have them in libutil.

Instead of trying to check the return value of every fprintf() and fputs()
call - which makes the code completely impossibly to follow, call fflush()
and ferror() just prior to calling fclose().
This has the advantage of actually detecting the any write errors, since
the output is block bufferred and will typically not happen during fprintf()
but only during the fclose() - where it is difficult to report.
I also singlilarly refuse to add (void) casts to every printf call in the
system - since it almost never makes any sense to look at the return value
(unless you want to know how many bytes were actually writtem).

build complete initializations.

Make sure that only specisications for _interface_ attributes are
emitted, the rest is useless at runtime.
saves some bloat in ioconf.c

o Keep track of negated devices in deaddevitab
o Rework do_kill_orphans() to use that information and mark explicitely
orphaned devices (i.e., the ones whose missing ancestor has been
negated)
o Make a distinction between erroneous orphans and explicit orphans.
Error out on the former, ignore the later (but print a warning when -v
is used)

Yes, now config(1) will actually stop if you comment out a parent. That
should help people still hoping adjustkernel is relevant these days :)

Rework the way orphan device instances are handled. To achieve that, keep
track of instances attaching at root, and walk down the tree of active
device instances. Then, all instances that are not marked active are
found as orphans.

Doing it that way allows us to simply ignore orphan devices, instead of
warning about them and still keep them in the configuration. Now, orphaned
instances are considered as never having existed.

In the end, this allows 'no <device> at <attachment>' to be much more
efficient, as the user doesn't have to negate all descendents of the
instance s/he actually wants to negate. Warnings are still emitted,
though.

While there, make official a side-effect of the previous lack of action
against orphaned instances: config(1) used to warn about instances that
attach at a numbered device when no instance of that device with that
number existed, even though there was a starred instance of the device.

E.g. (provided by Alan Barrett):

pciide* at pci? dev ? function ? flags 0x0000
wdc0 at isa? port 0x1f0 irq 14 flags 0x00
wdc1 at isa? port 0x170 irq 15 flags 0x00
atabus* at ata?
wd0 at atabus0 drive 0

With this commit, config(1) will no longer warn about 'wd0 at atabus0'.

Replace the "locnames", attached to cfdata, which was solely good for
userconf, by more complete information (including default values) about
interface attributes, attached to the drivers which provide them.

don't put the xxxcf_locnames[] into the global namespace

config and genassym are not sysadmin tools, they are development tools.
As such, they don't belong in /usr/sbin, but rather /usr/bin. Move them
there.

config: fix lint warning about "effectively discards 'const'"

No binary change.

branches: 1.73.2;
config(1): Make sort order deterministic.

Ensure we break ties in every case. This way, even though we use the
unstable qsort(3) library routine, the output is reproducible, no
matter what algorithm is behind qsort(3).

It would be nice if we could just use a stable sort function here,
but mergesort(3) is nonstandard, so we'd have to add it to
tools/compat, which is a big pain.

Instead, put a tie-breaking rule in every comparison function we use
with qsort, and abort() in the event of ties -- that way, we noisily
refuse to rely on unstable sort order.

While here, dispense with any question of integer overflow, and
sprinkle comments.

PR bin/58115

With config(1) as it exists today, a kernel Makefile is able to implement
logic on kernel options so long as those options are not defflag'd or
defparam'd. This works because such options are automatally added to the
IDENT var in the kernel Makefile as a preprocessor define, and the Makefile
can then do an operation like:

.if !empty(IDENT:M-DSOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

Unfortunately, this precludes making it possible to generate a compile-time
dependency on SOMECOOLCPUOPTION, or having SOMECOOLCPUOPTION imply another
kernel config option using the normal config(1) option dependency function.

Fix this by introducing a new option description keyword: mkflagvar. This
keyword marks an already defflag'd option as wanting a kernel Makefile var
defined if that option is selected in the kernel config file. So:

defflag opt_coolcpu.h SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION
mkflagvar SOMECOOLCPUOPTION ANOTHERCOOLCPUOPTION

will cause:

KERNEL_OPT_SOMECOOLCPUOPTION="1"
KERNEL_OPT_ANOTHERCOOLCPUOPTION="1"

...to be emitted into the kernel Makefile if those options are, in fact,
selected with "options ..." in the kernel config file, thus allowing for
a compile-time dependency on the option to be generated in addition to
Makefile logic, which now looks like:

.if !empty(KERNEL_OPT_SOMECOOLCPUOPTION)
CFLAGS+= -mcpu=somecoolcpu
.endif

branches: 1.71.2; 1.71.10;
Tag each .c file with the options that might have brought it in.

branches: 1.70.4; 1.70.6;
un-c99

Allow multiline makeoptions to work by quoting the newline..

branches: 1.68.8;
Sort files in ${ALLFILES} in the order of parsing of `files.*'.

config(1) reads the first `files.${MACHINE}' when it encounters `machine'.
Then it includes common `files.${MACHINE_SUBARCH}', `files.${MACHINE_ARCH}',
and MI `sys/conf/files' at last. This change makes the first "file" in
`files.${MACHINE}' appear first in ${ALLFILES}.

Make config(1) output only ${ALLFILES}, selected input files, and pass it to
Makefile.kern.inc, where files are filtered by suffix.

Output ${ALLFILES}, the list of selected *.[csSo] files, sorted in the exact
order config(1) parses "file ..." commands in config(5) files. Define
${CFILES}, ${SFILES}, and ${OFILES} as subset of ${ALLFILES} selected by
suffix.

Now Makefile.kern.inc handles swap<kernel>.c compiling/linking nicely,
config(1) can become less hackish.

Share more code by making *.o handling less special.

Pretty-print `Makefile' output.

Move `all' target definition to Makefile.kern.inc.

Define kernel dependency in Makefile.kern.inc.

Fix build.

Define ${OBJS} in `Makefile.kern.inc'. Relative paths are not truncated
(a/b/c/x.c -> x.c) in `-S' mode. Don't provide some explicit rules for `-S'
mode for safety.

Output relative path for ${CFILES} and ${SFILES} in generated `Makefile'.
`Makefile.kern.inc' knows how to deal with $S prefix, so don't bother it in
config(1).

Simplify generated `Makefile' by calculating ${OBJS} from merging ${CFILES},
${SFILES}, and ${OFILES} except swap<kernel>.o excluded.

Fix build.

Experimental ``suffix rules + subdirectories'' build support (-S).

Use per-suffix lists where appropriate.