Use Warner Losh's "hint" driver to decode ascii strings to fill the
resource table at boot time.
config(8) no longer generates an ioconf.c table - ie: the configuration
no longer has to be compiled into the kernel. You can reconfigure your
isa devices with the likes of this at loader(8) time:
set hint.ed.0.port=0x320
userconfig will be rewritten to use this style interface one day and will
move to /boot/userconfig.4th or something like that.
It is still possible to statically compile in a set of hints into a kernel
if you do not wish to use loader(8). See the "hints" directive in GENERIC
as an example.
All device wiring has been moved out of config(8). There is a set of
helper scripts (see i386/conf/gethints.pl, and the same for alpha and pc98)
that extract the 'at isa? port foo irq bar' from the old files and produces
a hints file. If you install this file as /boot/device.hints (and update
/boot/defaults/loader.conf - You can do a build/install in sys/boot) then
loader will load it automatically for you. You can also compile in the
hints directly with: hints "device.hints" as well.
There are a few things that I'm not too happy with yet. Under this scheme,
things like LINT would no longer be useful as "documentation" of settings.
I have renamed this file to 'NOTES' and stored the example hints strings
in it. However... this is not something that config(8) understands, so
there is a script that extracts the build-specific data from the
documentation file (NOTES) to produce a LINT that can be config'ed and
built. A stack of man4 pages will need updating. :-/
Also, since there is no longer a difference between 'device' and
'pseudo-device' I collapsed the two together, and the resulting 'device'
takes a 'number of units' for devices that still have it statically
allocated. eg: 'device fe 4' will compile the fe driver with NFE set
to 4. You can then set hints for 4 units (0 - 3). Also note that
'device fe0' will be interpreted as "zero units of 'fe'" which would be
bad, so there is a config warning for this. This is only needed for
old drivers that still have static limits on numbers of units.
All the statically limited drivers that I could find were marked.
Please exercise EXTREME CAUTION when transitioning!
Moral support by: phk, msmith, dfr, asmodai, imp, and others
the need to specify the unit number of unwired devices. ie: instead
of saying "device fxp0" we can say "device fxp" which is much closer
to what it actually means. The former (fxp0) implied something about
reserving the 0th unit, but it does not and never did - it was a
figment of config(8)'s imagination that we had to work around..
"device fxp0" simply means "compile in the fxp device driver", so we
may as well just write it as "device fxp" which is closer to what it
really means.
Doing this also saves us from filling up the ioconf.c tables with
meaningless entries.
- redo the "at" configuration system so that it just syntax checks
to make sure the device you're configuring something "at" appears to
exist. Nuke a bunch of complexity that was responsible for creating
"clones" of wildcard devices and some wierd stuff in a few places
including the scbus config tables etc.
- merge "controller" and "device" - there is no difference as far as
the kernel is concernend, it's just something there to make life
difficult for config file writers. "controller" is now an alias for
"device".
- emit full scsi config into the resource tables. We could trivially
change cam to use that rather than it's own "special" table for wiring
and static configuration. ATA could use this too for static wiring.
- try and emulate some of the quirks of the old system where it made
sense. Some were too strange though and I'd be very suprised if they
were features and not outright bugs. nexus handling is still strange.
One thing in particular is that some of the wierd entries in the
newbus devtables is now gone as it was a quirk side effect of the
wildcard/question-mark cloning above.
GENERIC and LINT still build etc.
directory than the default one. If the option is not given, then the
output of config is exactly as before. Only when an alternate output
directory has been specified will config modify its behavior.
Additional changed:
o Remove the now conflicting and unused NODEV define. It
conflicts with NODEV in sys/param.h.
o Rename the now conflicting MACHINE token to ARCH. It
conflicts with MACHINE in sys/param.h.
o Fix some easy style bugs.
o Fix some easy grammar bugs in the manpage.
Approved by: peter, archie
'makeoptions KERNEL=kernelname'. Warn about any trailing stuff as it's
not handled here. This is a simple bandaid, hopefully to head off some
complaints from certain people.
- make this work: options FOO123=456 *without quotes*
- grumble (but accept) vector xxxintr, and tty/net/bio/cam flags.
- complain if a device is specified twice (eg: 2 x psm0)
- don't require quotes around: port IO_COM2
- recognize negative numbers. (ie: options CAM_DEBUG_UNIT=-1)
- GC some more unused stuff (we don't have composite disks from config(8)).
- various other nits (snprintf paranoia etc)
I zapped the MACHINE_MIPS stuff, it isn't likely to be useful apart from
recognition of the machine name. It would be reasonable to expect new
ports would look something like the alpha/i386 from a config perspective.
only worked for configurations with "swap on generic".
usr.sbin/config/config.y:
- ignore all "swap [on] device ...' specifications except for
warning about them. They haven't done anything related to swap
for almost 4 years, and were previously silently ignored,
except for "swap on generic" which stopped swap${KERNEL}.c
from being generated. Code to support swapping is now deader
than before.
usr.sbin/config/mkswapconf.c:
- don't generate a dummy setconf() function in swap${KERNEL}.c.
sys/i386/conf/files.i386:
- swapgeneric.c is now standard. It should be merged into autoconf.c
so that it doesn't conflict with swap${KERNEL}.c for kernels named
"generic".
sys/i386/i386/autoconf.c:
- don't call setroot() for mfs roots. Since setroot() doesn't do anything
harmful, this was just a waste of time, except possibly for booting with
-a it may have helped prevent an undesireable call to setconf() by
finding a bogus rootdev.
- honor -a for ffs roots. -a now overrides all other ways of specifying
the root device. Previously, -r had precedence over -a, and the -a
handling was usually a no-op.
- don't honor -a for non-ffs roots, since it would currently just get in
the way of a clean panic.
sys/i386/i386/swapgeneric.c:
- don't declare things that are now always declared in swap${KERNEL}.c.
Don't decide things that are now decided in autoconf.c. Code to
support the "generic" case is now dead instead of useless.
was used as if it is 1-based. This happened to give the correct result
for options without values because of a compensating error in newline
lexing. Didn't fix the latter, so line numbers in yyerror() may still
be 1 too high in some cases.
non-standard and not used. "port auto" is equal to "port?" or missing "port"
keyword now. "port none" is really probe routine task (return -1 for
no ports).
version of strdup() by a macro, killed many calls to strdup(), thus
potentially wasting less malloc'ed space (their args were never be
free()ed desptie despite of being malloc'ed). Probably still a huge
memory leak at all... Also killed two totally useless variables.
I've tested it as i could, but wouldn't be surprised if unexpected
problems showed up. So watch out this space!
conservative part of the tidyup, like fixing potential buffer overflow
conditions. It is believed to be safe to go into 2.2.
Pointed out by: lozenko@cc.acnit.ac.ru (Evgeny A. Lozenko)
Note that this code is dormant unless the options files exist.
Also, parsing of quoted options in the config files is improved.
What this allows, is all the options in LINT to be specified to be
configured as #defines in a file rather than on the CC command line at
kernel build time. This means that 'make depend' will catch dependencies
on actual *options*, meaning that you can run 'config' and 'make depend'
in complete safety WITHOUT removing the compile directory each time.
Unfortunately, this requires a pass over the source to get the individual
files to #include the new .h files that would be generated by config.
This has a small compile time penalty (appears up to about 2% slower)
from a "fresh" build. Of course, you should not be needing to do complete
rebuilds very often once this was completed, so it would be an overall
win for most people.
Since this code is dormant and we've got a lot of other things happening
on the kernel tree at the moment (prototypes, devfs, static declarations
etc) I am not planning on doing any changes to activate this feature just
yet.
This is performed by using a line similar to:
controller scbus0 at ahc0 bus 1
to wire scbus0 to the second bus on an adaptec 2742T controller.
Reviewed by: Peter Dufault(dufault@hda.com), Rod Grimes(rgrimes@FreeBSD.org)
others. The flag can be put in descriptive locations, e.g.:
device sb0 at isa? port 0x220 irq 7 conflicts drq 1 vector sbintr
or
device psm0 at isa? port "IO_KBD" conflicts tty irq 12 vector psmintr
But is nonetheless boolean only. You can't turn conflict checking off for
only a given type of conflict. I didn't deem it worth the trouble at this
stage, and it's far better than the ALLOW_CONFLICT_* that preceeded it.
Support slice numbers in device names. The syntax is `<driver name>
[<unit number>] ['s' <slice number>] [<partition letter>]'. Only
`['s' <slice number>]' is new here. The slice number defaults to 0
so that there is no change in the output from config if this new
feature is not used.
Replace some magic disk numbers by `dk' slice and label macros.
mkswapconf.c:
Improve the output formatting:
Generate <> style includes.
Print minor numbers in hex so that slice numbers are easy to see and edit.
Print the rootdev and dumpdev names in comments like the swapdev names.
