Hopefully I didn't make too much of a hash of the postinstall support for
this; it currently installs the file if there's none in the destination,
and elsewise notes if the CVS version differs (or is missing) but doesn't
try to fix that.
${rcvar}=yes yet all the other prerequisite tests are still performed.
The existing ``force'' prefix is a sledgehammer that ignores all the
prerequisite checks and always returns a zero exit status; this is a
more gentle approach to the problem of "manipulate this disabled
service without editing rc.conf(5)".
Add "iscsi-target" at 3260/tcp as per iana. Add "iscsi-rfc" at
860/tcp. Modify comment on "iscsi" service to indicate that it
should move to 860/tcp after the iSCSI RFC gets released. iscsi-rfc
should be removed at that point.
The problem with the iscsi service at present is that all through
draft development, everyone used 3260. As part of the RFC publishing,
the port number will be changed to 860. However no one has started
to use the new port number yet, so it's premature to switch yet.
information about the build, and "install-release-info" to install it.
(The latter will be invoked by src/Makefile)
Based on work by Hubert Feyrer <hubertf@> and the "params" target in
src/Makefile.
at boot automatically, so a machine dual booting another OS that uses
the RTC at localtime and NetBSD agree on the current time even if daylight
saving started/ended (without recompiling a kernel twice per year).
Awk code by Matt Thomas.
isn't enabled.
This is how the rc.d system works in conjunction with our current build
and install system; all the rc.d scripts are installed even if the
subsystems they control are not.
* Use "mknod -F netbsd -r" to create nodes, instead of
"rm ; mknod; chmod; chown".
This means permissions & ownership of existing nodes will
not be changed.
This is up to 30% faster when populating an empty /dev,
and nearly 2x faster when re-running on an existing /dev.
* New options:
-f force change of permission & ownership of existing
devices
-m mknod override name/path of mknod program
(which defaults to $TOOL_MKNOD, then "mknod").
-s generate mtree(8) specfile instead of creating devices
* Remove /usr/etc from $PATH; not needed anymore.
* Provide functions to create devices & directories:
mkdev name [b|c] major minor [mode{=600} [gid{=0} [uid{=0}]]]
create device node `name' with the appropriate permissions
lndev src target
create a symlink from src to target
makedir dir mode
create directory with appropriate mode
* UIDs and GIDs are hardcoded in at MAKEDEV generation time.
(Unfortunately there's not a simple way of determining a GID
a la "id -n user" for determining a UID).
This was tested by generating MAKEDEV for each MACHINE,MACHINE_ARCH
combination and comparing the results of "MAKEDEV all" from the
previous version to the new one.
(This testing actually highlighted mistakes in the previous configuration!)
Simplify distrib/common/Makefile.makedev to use "MAKEDEV -s"
instead of -v ETCDIR.
* Parse $NETBSDSRCDIR/etc/{master.passwd,group} to build a list of
user->uid and group->gid mappings, and replace %uid_XXX% and %gid_XXX%
appropriately.
* Whitespace & regex tweaks.
and without Kerberos 4 & 5 (MKKERBEROS=no). Previously checkflist
complained of missing files.
* move kerberos- and kerberos 4-only files into new flists,
distrib/sets/lists/*/krb.*
* make the flist generators grok MKKERBEROS{,4} variables
* fix Makefiles which treat MKKERBEROS=no as MKKERBEROS5=no.
9 out of 10 experts agree that it is ludicrous to build w/
KERBEROS4 and w/o KERBEROS5.
* fix header files, also, which treat MKKERBEROS=no as MKKERBEROS5=no.
* omit some Kerberos-only subdirectories from the build as
MKKERBEROS{,4} indicate
(I acknowledge the sentiment that flists are the wrong way to go,
and that the makefiles should produce the metalog directly. That
sounds to me like the right way to go, but I am not prepared to do
revamp all the makefiles. While my approach is expedient, it fits
painlessly within the current build architecture until we are
delivered from flist purgatory, and it does not postpone our
delivery. Fair enough?)
Uses a hook in spec_strategy() to save data written from a mounted
file system to its block device and a hook in dounmount().
Not enabled by default in any kernel config.
Approved by: Frank van der Linden <fvdl@netbsd.org>
print a summary rather than the full (and not usually desired) output
of netstat -inv. The old behavior can be returned by simply setting
full_netstat to YES in daily.conf.
Original idea by me, cleaner and more correct execution via small awk
script from Greg Woods.
add some example country codes, and use disjoint CNAMES by default, as
discussed in PR misc/23283 (by Adrian 'Dagurashibanipal' von Bidder),
the pool.ntp.org mastermind and maintainer).
- use ttyC[0123] for DCA and APCI with MI com(4)
- use ttyM[0-f] for DCM instead of tty0[0-f] to avoid confusion with MI com(4)
- create the foloowing symlnks to compatibility:
tty0 -> ttyC0 (for DCA)
ttya0 -> ttyC1 (for APCI)
ttya1 -> ttyC2 (for APCI)
ttya2 -> ttyC3 (for APCI)
tty0? -> ttyM? (for DCM)
- note the above symlinks should be removed after the next release
As per discussion on recent port-hp300 (ok'ed by thorpej).
When /dev is an fdesc, and /dev/tty is stat()ed without a controlling tty,
a "Device not configured" error is returned.
Filter mtree's stderr to ignore this error.
If fdesc is fixed to not behave in this fashion, this workaround can
be removed; bin/12900 should remain open until that time.
Make ls -A explicit, to help n debugging when not run as root
(-A is implied when ls is run as root)
Ignore dotfiles, as they are not mailboxes (e.g. .jhawk.pop)
* Restore previous functionality which supported running postinstall
with an extracted etc.tgz set as the "source" directory (i.e,
/usr/src isn't available). We do this by generating the MAKEDEV
to ${SCRATCHDIR} if the real source is available, and leverage off
the compare_dir() code that the rest of postinstall uses to support
this functionality.
* Don't special case MACHINE_ARCH==arm; the MAKEDEV.awk should
be doing that since etc/Makefile doesn't have similar special
case code. In any case, AFAICT the only MACHINE_ARCHs are
"arm" and "armeb"; there is no "armel" in NetBSD.
to generate the MAKEDEV to scratch dir and compare that with the currently
installed /dev/MAKEDEV; also add -a MACHINE_ARCH option to explicitly
set machine architecture for non-native builds
problem with postinstall pointed out by Patrick Welche and Hubert Feyrer
in private e-mails
add some more necessary device entries
introduce makedisk_minimal() and use it to create the CD-ROM & md entries;
this procedure only creates 'a' and RAW_PART device entries
remove pf from 'all' target - it's not part of tree, so shouldn't be
created by default
at MAXPARTITIONS/OLDMAXPARTITIONS in kernel sources, so that it wouldn't
need to be specified separately in MAKEDEV.conf
change platform MAKEDEV.conf to contain only MD targets and nothing else;
simplify the parsing in MAKEDEV.awk accordingly
for tty0[0-9]
as pointed out by David Laight
remove spaces between makedisk_p*() functions, so the output script looks
neater once some of them are filtered out
when filtering the template, leave only the wanted makedisk_p*() routine
in the output
as pointed out by David Laight
also make it mandatory for MACHINE and MACHINE_ARCH to be set in
environment; it would be possible to use use uname -p/-m, but better
to catch possible cross-build mistakes sooner than later
this contains information about disk partitions used by platform,
and MD MAKEDEV targets, such as 'init', MD part of 'all',
as well as any other MD-specific targets not covered by MAKEDEV.tmpl
/etc/security should produce no output (and thus suppress the report)
when nothing is wrong.
While we're here, use printf instead of two echos, like the rest of
the script.
any pipes and running other commands (eg dd).
Measurable speeds up creating pseudo ttys.
(there has to be a better way than editing 51 files! - tedious at only
6 key presses per file)
- If DISTRIBUTION_DONE is defined, don't add the 'root' directory to METALOG.
(maybe install -M shouldn't write to the METALOG file if it doesn't install
the file either)
- Don't re-compress the kernels if the compressed file is newer than the
kernel.
Useful if all you want to do is rebuild the ramdisk image.
Replace defined(UNPRIVED) tests with ${MKUNPRIVED} != "no"
Add MKUPDATE; if not no has the same semantics as if UPDATE was defined.
Replace defined(UPDATE) tests with ${MKUPDATE} != "no"
Improve documentation for these and other make flags.
program/tool from "FOO" to "TOOL_FOO". The new variables are:
TOOL_ASN1_COMPILE TOOL_CAP_MKDB TOOL_CAT TOOL_CKSUM TOOL_COMPILE_ET
TOOL_CONFIG TOOL_CRUNCHGEN TOOL_CTAGS TOOL_DB TOOL_EQN TOOL_FGEN
TOOL_GENCAT TOOL_GROFF TOOL_HEXDUMP TOOL_INDXBIB TOOL_INSTALLBOOT
TOOL_INSTALL_INFO TOOL_M4 TOOL_MAKEFS TOOL_MAKEINFO TOOL_MAKEWHATIS
TOOL_MDSETIMAGE TOOL_MENUC TOOL_MKCSMAPPER TOOL_MKESDB
TOOL_MKLOCALE TOOL_MKMAGIC TOOL_MKTEMP TOOL_MSGC TOOL_MTREE
TOOL_PAX TOOL_PIC TOOL_PREPMKBOOTIMAGE TOOL_PWD_MKDB TOOL_REFER
TOOL_ROFF_ASCII TOOL_ROFF_DVI TOOL_ROFF_HTML TOOL_ROFF_PS
TOOL_ROFF_RAW TOOL_RPCGEN TOOL_SOELIM TOOL_SUNLABEL TOOL_TBL
TOOL_UUDECODE TOOL_VGRIND TOOL_ZIC
For each, provide default in <bsd.sys.mk> of the form:
TOOL_FOO?= foo
and for the ${USETOOLS}=="yes" case in <bsd.own.mk>, provide override:
TOOL_FOO= ${TOOLDIR}/bin/${_TOOL_PREFIX}foo
Document all of these in bsd.README.
This cleans up a chunk of potential (and actual) namespace collision
within our build infrastructure, as well as improves consistency in
the share/mk documentation and provision of appropriate defaults for
each of these variables.
