File:  [Research Unix] / researchv8dc / sys / conf / config_how
Revision 1.1.1.1 (vendor branch): download - view: text, annotated - select for diffs
Tue Apr 24 17:21:59 2018 UTC (8 years, 3 months ago) by root
Branches: belllabs, MAIN
CVS tags: researchv8, HEAD
researchv8 Dan Cross

config makes the following files for each system:
	conf.c
	ioconf.c
	ubglue.s
	makefile
	zillions of header files

to build a system from scratch,
	mkdir /usr/sys/sysname
	cd /usr/sys/sysname
	(create a machine description in `conf')
	/etc/config
	make

to change a configuration, edit `conf,' run /etc/config, and make.

to add a piece to the system:
put the filename in conf/files, with appropriate annotations (device-driver,
pseudo-device, optional, and all that).
if it's a device driver, line discipline, or filesystem handler, add an
entry to conf/devices as well.
if you just want to test things, addenda to `files' and `devices' can be
put in /usr/sys/sysname; they'll be treated as if appended to the ones in
/usr/sys/conf.

the fields in devices are
	[standard] typeofthing
		typeofthing is (character) device, block-device, stream-device,
		file-system, line-discipline.
	major device number (or filesystem number, or line discipline number)
	driver name.  the one you use in the conf file, and in conf/files.
	driver routine prefix.  see below.
	entry points:
		for a block device, some of open close strategy B_TAPE
			(the last puts B_TAPE into d_flags in bdevsw)
		for a character device, open close read write ioctl reset
		for a stream device or a line discipline, info
		for a filesystem, put get free updat read write trinc stat nami mount ioctl
entry points are catenated with the prefix.  example:
device 13		up	up	read write reset
defines device up:
major character device 13
has entry points named upread upwrite upreset

default entry points are generally `nodev'.
exceptions: open close default to `nulldev,' except for stream devices.
reset defaults to nulldev always.
info (obviously) defaults to NULL.

unix.superglobalmegacorp.com

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.