gnulib, the implementation goes back to the AMD Software Optimizer
guide. A number of platforms will want to replace the C version with
assembler code using native instructions.
The algorithm used is the Jenkins hash. The name (mi_vector_hash)
reflects the nature of the hash function.
Add glue for libc ATF tests and include a test case to make sure that
(mis)alignment and endianess are handled correctly.
Bump libc minor to 169.
Based on PR port-powerpc/40421 from Wojciech Galazka,
with misc tweaks by me.
Note sysinst part is not pulled because there is
no supported storage device yet on this port.
INSTALL_DIR would want to write to the metalog, and it can't do that
if the metalog is inside DESTDIR but DESTDIR doesn't yet exist.
This allows some XXX comments to be removed.
first, mount root and run the various disk providers. Add swap and
check the remaining file systems after that.
This breaks the dependency cycle for lvm, which needs writeable /dev.
Depend on rndctl in cgd.
ddb.onpanic to 1, change it back to 0 in sysctl.conf and make sure
postinstall installs this setting.
This avoids us trying to dump while booting from install CD, but keeps
the default the same once we are far enough through /etc/rc.d. Failing
earlier is unlikely to be recovered by an automatic reboot.
OK: core.
allows you to invoke rndctl(8) during the boot.
* Add rndctl=NO and rndctl_flags="" to /etc/defaults/rc.conf.
* Document rndctl and rndctl_flags variables in rc.conf(5).
output from mount(8) rather than df(1), because mount is more likely
to be available on install media. mount is invoked in a way that
always prints an error message, and the content of the error message
tells us whether fdesc is mounted. This should fix PR 40316 from
David Ross.
* Don't use negative variable names. Instead of nofdesc=false, use
fdesc_mounted=true.
* Add a comment explaining that MAKEDEV may be executed in an environment
that is missing some common commands.
ok'ed by core and releng.
(thanks for agc@, snj@ and i'm sorry for long time patience).
[libc]
- localeio.[ch] and lc*.[ch] in src/lib/libc/locale was replaced by
new locale-db implementation using citrus_db backend,
see src/lib/libc/citrus/citrus_lc_*.[ch].
- add citrus_bcs_strtou?l.c. don't use strtou?l locale implementation
internally, because they're locale-aware function.
- add some stubs for multi-locale issue, see {current,global}_locale.c.
- remove some obsolete file, setrunelocale.c, ___runetype_mb.c.
- remove __savectype() from ctypeio.[ch].
[tools]
- mklocale(1): add new option ``-t'' that generates new style
LC_{MONETARY,NUMERIC,TIME,MESSAGES} locale-db format.
- chrtbl(1): added ctypeio.[ch] for __savectype().
[locale-db]
- added en_US.US-ASCII locale.
- removed some shareable locale definition file:
en_US.US-ASCII -> en_US.ISO8859-1, en_US.UTF-8
zh_CN.eucCN -> zh_CN.GB18030
and more...see src/share/locale/*/Makefile.
- remove obsoleted locale sr_YU, added new locale sr_ME, sr_RS.
- change locale name ja_JP.ISO2022-JP* -> ja_JP.ISO-2022-JP*
for X11's locale.alias file alignments.
- fix regression test, wrong wcs?width(3), NAN/INF usage.
i tested release-build following arch:
i386, amd64, hpc{mips,arm,sh}, sparc64, vax.
citrus_lc_*.[ch] also can read old-plain-text style locale-db.
so that backward compatibility is keeped, but lc*.[ch] can't read
new citrus_db'ed locale-db and localeio.c never check sanity,
so forward compatibility is broken ;-<
old mklocale(1) doesn't know -t option, so you have to rebuild toolchain.
- Make MD poffd(8) retire, and use MI powerd(8) instead of it.
- Make /dev/pow1 retire, because nobody holds /dev/pow0 any longer.
Use /dev/pow0 for pow(4) ioctl.
- POWIOCSSIGNAL ioctl which is for poffd(8) is also obsoleted.
When rc scripts are started from interactive shell, be nice and when user
tries to start an unconfigured service, point him to rc.conf(5) and tell
him about the one* keyword. Also be more user friendly when an unknown
argument is entered.
Example:
# /etc/rc.d/cron start
$cron is not enabled - see rc.conf(5).
Use the following if you wish to perform the operation:
/etc/rc.d/cron onestart
# /etc/rc.d/cron nonsense
/etc/rc.d/cron: unknown directive 'nonsense'.
Usage: /etc/rc.d/cron [fast|force|one](start stop restart rcvar status poll)
Inspired by discussion on tech-userlevel:
http://mail-index.netbsd.org/tech-userlevel/2007/02/13/0010.htmlhttp://mail-index.netbsd.org/tech-userlevel/2007/02/13/0011.htmlhttp://mail-index.netbsd.org/tech-userlevel/2007/02/13/0012.html
Reminded about it and reviewed by <lukem>
- introduce X11FLAVOUR to choose src/x11 vs src/external/mit/xorg
for the X11 to build if MKX11=yes is set. it takes the values
of either Xorg or XFree86.
- default to Xorg on alpha, i386, macppc, shark, sparc and sparc64
- remove MKXORG_WITH_XSRC_XSERVER, unused and never useful
as the console. Now when coming up in single user mode the tset in
/.profile can set the terminal type correctly.
Note: I believe constty should be 'on' and console 'off', but thats a
separate discussion
This may be done either by embedding newlines in the value,
or by using semicolons to represent line breaks (but not both at once).
* Allow shell quoting insode $ifconfig_xxN variables or /etc/ifconfig.xxN
files. This allows something like ifconfig_wi0="ssid 'my network'; dhcp"
as snap_md_post target in etc/etc.${MACHINE}/Makefile.inc rather than
as release target in distrib/${MACHINE}/Makefile so that builds for
release in distrib/${MACHINE}/Makefile don't have to be serialized.
XXX: maybe ramdisk kernels should be installed into binary/kernel dir
XXX: as other ports rather than installation dir, and
XXX: kernel config names should be preserved in their filenames.
image is installed properly
- INSTALL_IPAQ should be in BUILD_KERNELS rather than KERNEL_SETS
since it requires ramdisk image by mdsetimage(8) in src/distrib
MD MAKEDEV.conf. Not all ports have vcoda in their majors.foo files
and we shouldn't (re)assign major numbers on all ports before 5.0.
Should close PR port-sgimips/38962.
private non-installed build infrastructure from sys/rump.
breakdown of commit:
* install relevant headers into /usr/include/rump
* build sys/rump/librump/rumpuser and sys/rump/librump/rumpkern
from src/lib and install as librumpuser and librump, respectively
+ this retains the ability to test a librump build with just the
kernel sources at hand
* move sys/rump/fs/lib/libukfs and sys/rump/fs/lib/libp2k to src/lib
for general consumption, they are not kernel-space dwellers anyway
* build and install sys/rump/fs/lib/lib$fs as librumpfs_$fs
* add chapter 3 manual pages for rump, rumpuser, ukfs and p2k
* build and install userspace kernel file system daemons if MKPUFFS=yes
is spexified
* retire fsconsole for now, it will make a comeback with an actually
implemented version shortly
This is consistent with several other rc.d scripts
and fixes problem with /sbin not in PATH.
(Tested several times for over three weeks, but not all
parts of this code were used by me.)
and create sums for it.
This means all the pieces to do network installation are easily accessible
following a release build (and will also come out of the build servers).
netbsd-GENERIC is in binary/kernel/
pxeboot_ia32.bin is in installation/misc/
miniroot.kmod is in installation/miniroot/
- Make MULTIPROCESSOR mandatory on i386.
Installation changes:
- Update installation section of release notes to match reality.
- Rename INSTALL to INSTALL_FLOPPY, retire INSTALL_LARGE.
- Build INSTALL kernel from GENERIC, like on amd64.
- Update boot menu to allow disabling ACPI and/or SMP.
- Remove GENERIC.NOACPI from the installed kernel list.
- TODO: install default boot.cfg in etc.tgz.
- TODO: possibly enable PCI fixup stuff at runtime if ACPI is disabled.
Build changes:
- No longer build ALL, it's for verification, is slow to build and the
build process is already crippled by the number of kernels built.
- No longer build GENERIC.NOACPI.
Only a sensor is only returning unknown data, all the other ones
match what the BIOS returns.
- Improve comments and add a note for contacting me if the values
returned by drivers are not correct.
btuartd(8) should be named btattach(8) for consistency
with other parts of NetBSD
make btattach(8) a single-use tool for less complexity
device specicific initialisation (from btuart(4)) is carried
out prior to activating the line discipline (in btattach(8)),
which simplifies the API somewhat and means that the user
tool and the kernel do not need to be kept in sync.
btuart(4) driver is much reduced; naming is made consistent
and all tsleep() and delay() are removed to userland
like "cmd: not found" are also redirected.
This should fix a problem reported by Chavdar Ivanov in which
the message "mtree: not found" appears during installation.
already-existing device nodes. This is done by invoking pax with its
current working directory set to an empty temporary directory.
* Add "-t mtree" command line option, to force MAKEDEV to use "mtree
-e -U" to create device nodes. Here are some timings from "MAKEDEV
${method} all" on an i386 system; each method was executed three
times, first in an empty directory, then twice in an already-populated
directory:
-m mknod
4.56 real 1.58 user 1.68 sys
2.96 real 1.37 user 1.48 sys
2.95 real 1.43 user 1.44 sys
-p pax
1.80 real 0.18 user 0.62 sys
2.35 real 0.25 user 0.80 sys
2.38 real 0.24 user 0.82 sys
-t mtree
1.62 real 0.22 user 0.44 sys
0.23 real 0.29 user 0.04 sys
0.24 real 0.28 user 0.05 sys
* If no command line options force a choice, then prefer mtree, pax, or
mknod, in that order of preference.
No objections in tech-userlevel to an earlier version of this change.
"retro" computers, but NetBSD also runs a growing number of rare
and retro add-on cards. With this patch, NetBSD supports the IDEC
Supervision/16, a black&white image capture board for the 16-bit
ISA bus. Approximate date of manufacture: 1991. Total instances
known to be in use throughout the world: one.
Coming soon; isvctl(8), the utility program for capturing 8-bit,
512x480 images at speeds of up to 6 frames per second.
This is so that md code that tries to call makedev altq or fd does not fail
when the file exists. This was reported in port-powerpc, when building /dev.
Now "etcupdate -s /path/to/src" should no longer complain about being
unable to make atf-run.hooks. build.sh passes NETBSDSRCDIR in the
environment, so it used to work even without this include.
* Add a nooutput function to be used instead of redirecting to /dev/null,
because /dev/null might not exist early in the boot sequence. This
should fix a bug in which the pax autodetection failed when used from
init(8) with a read-only root file system without a /dev/null.
* Move most of the main program into a new makedev_main function,
to allow reuse by MAKEDEV.local.
In MAKEDEV.local:
* Use the new makedev_main function defined in MAKEDEV. This allows
MAKEDEV.local to support all the options understood by MAKEDEV, instead
of only a subset. This should fix a bug in which the setup() function
could try to use pax mode even though that was previously not supported.
In MAKEDEV.local man page:
* Update list of command line options.
* Don't explain all the options, just refer to MAKEDEV(8) man page.
* Remove BUGS section now that all MAKEDEV options are supported
by MAKEDEV.local.
joerg