Commit Graph

5177 Commits

Author SHA1 Message Date
agc 0a602aa3de Handle multiple ranges on the command line and in the defaults file,
based on some code from Todd C.  Miller, which in turn was based on a
patch from Brian Poole <raj@cerias.purdue.edu>.

Look first in any uid ranges specified on the command line, in the order
they were given on the command line, and then in any ranges specified in
the defaults file.

With thanks to Brian for nudging me a number of times to fix this.
2002-05-03 10:31:14 +00:00
agc 2de2bc02d6 Fix for skeleton directory changes to defaults file from Brian Poole
<raj@cerias.purdue.edu>

Previous behaviour as follows:

# useradd -D -k /foo/bar
group           users
base_dir        /home
skel_dir        /foo/bar
shell           /bin/csh
class
inactive        0
expire          Null (unset)
range           10200..10300
range           10600..10700
# useradd -D
group           users
base_dir        /home
skel_dir        /etc/skel
shell           /bin/csh
class
inactive        0
expire          Null (unset)
range           10200..10300
range           10600..10700
2002-05-03 10:11:16 +00:00
agc 3db3295859 Fix from Brian Poole <raj@cerias.purdue.edu> for improper handling of
UID ranges. Previous behaviour is demonstrated below...

# useradd -D
group           users
base_dir        /home
skel_dir        /etc/skel
shell           /bin/csh
class
inactive        0
expire          Null (unset)
range           10200..10300
# useradd -D -r 10200..10300 -r 10400..10500
# useradd -D
group           users
base_dir        /home
skel_dir        /etc/skel
shell           /bin/csh
class
inactive        0
expire          Null (unset)
range           10400..10500

The second command should have put 2 ranges back into the config file,
but it really only put one because the first range on the command line
was marked as a duplicate (but when it got to writing the config file,
it only wrote command line ranges, so the 10200 got skipped).  Fix
this by initializing defrc to 0 and then only looking after defrc for
duplicate ranges.
2002-05-03 10:05:28 +00:00
agc 00ee66cd54 Modifications from Brian Poole <raj@cerias.purdue.edu>, via OpenBSD:
+ if the cd built-in fails, don't try to copy the directory hierarchy
  with pax - s/;/&&/ in a shell command
+ clean up after ourselves if a rename fails - remove the newly-created
  group file.
2002-05-03 08:07:02 +00:00
lukem 6206a1d7eb remove commented out debug printf() accidentally committed in the last rev.
noted by YAMAMOTO Takashi<yamt@mwd.biglobe.ne.jp> in private email.
2002-05-02 13:53:10 +00:00
lukem 033fe17be5 tweak asserts 2002-05-02 13:03:32 +00:00
lukem 80ae5fb69d implement "raw" file system type. this can be used when the primary
bootstrap is being installed at a particular block offset and the underlying
filesystem doesn't matter (e.g, alpha ustarboot in ustar floppies)
2002-04-30 14:45:12 +00:00
lukem d9e55fca47 display sector start & count in decimal rather than hex 2002-04-30 14:24:33 +00:00
lukem 1a478e4092 add some more assert()s 2002-04-30 14:21:17 +00:00
lukem fa2a19276b consistently use sizeof(bb) instead of SUN68K_BOOT_BLOCK_MAX_SIZE 2002-04-29 13:07:41 +00:00
lukem 8feb052770 s/_SRC_TOP_/NETBSDSRCDIR/ 2002-04-26 16:23:14 +00:00
itojun cd1e16de59 upgrade to KAME racoon as of 2002/4/26.
file descriptor leak fix.
null encryption algorithm key length fix (should use 0).
couple of null-pointer reference fixes.
set port # to 500 in ID payload (possible interop issue - spec is unclear).
correctly match address pair on informational exchange
2002-04-26 02:25:13 +00:00
tv 4887cf3412 HAVE_CONFIG_Hize. Also get *_boot.h from "../../sys/" when HAVE_CONFIG_H,
rather than creating more stubs.
2002-04-25 18:11:53 +00:00
atatat 29bf0c2821 Implement CRON_TZ and CRON_WITHIN to provide finer control over job
execution.  CRON_TZ sets the time zone within which a job is
considered for execution (but not in which it runs), and CRON_WITHIN
allows jobs to be skipped whose execution is delayed for any reason
(eg, the system was sleeping or the load average is too high for
timely execution).
2002-04-25 14:45:05 +00:00
itojun 3f18342704 handle ECONNABORTED at accept(2). correct error handling for connect(2)
sync w/ kame
2002-04-24 12:14:42 +00:00
lukem 2c1cfc8e8c - in <bsd.files.mk>, don't clear FILES after using it, as that prevents
make -V FILES
  from being useful (and given that every other variable can be
  extracted using make -V, the behaviour was unusually inconsistent
  given that the original reason for clearing it doesn't seem to be
  relevant anymore)
- use <bsd.prog.mk> instead of directly including <bsd.files.mk>
  (and possibly <bsd.man.mk> or <bsd.own.mk>)
- remove obsolete NOPROG
2002-04-24 08:18:45 +00:00
lukem 2fddf94b34 convert from "sun68k_bbinfo.h" to <dev/sun/sun_boot.h> 2002-04-24 01:35:25 +00:00
hubertf 19f6d0bbcb Add more error checking, and tell the user what's up when he runs
into descriptor-shortage. Can be caused when installing deeply-nested
binary pkgs like kde.
2002-04-23 10:14:59 +00:00
lukem 1608f561f2 after consideration, don't bother with compat symlinks, as it might
cause end-user confusion if the syntax changed between
/usr/mdec/installboot and /usr/sbin/installboot
2002-04-23 02:39:33 +00:00
bjh21 1723b1333c When printf'ing a size_t, cast it to unsigned long and use a %lu format.
This gets rid of a warning when compiling for ARM.
2002-04-22 23:24:10 +00:00
bjh21 a3e433c6fa include <bsd.own.mk> for _SRC_TOP_ 2002-04-22 23:23:05 +00:00
fredette 7493ba97a6 Install the compatability symlink for sun2.
In the sun68k handler, add some sync()s to make sure we
get a coherent picture of the filesystem.  Maybe should
be in the MI filesystem code?
2002-04-22 21:11:45 +00:00
lukem 878d42b158 Add sun2 backend from Matthew Fredette.
(Implemented with "sun68k.c" as it should eventually support "sun3" as well.)
2002-04-22 17:17:35 +00:00
lukem ed5fff8877 formatting consistency frobs 2002-04-22 16:03:07 +00:00
wiz 554a5b8a3a Don't default the portname to devices that don't exist on NetBSD. 2002-04-22 13:08:34 +00:00
lukem 8eb8919e12 - Add code from Matt Fredette <fredette@theory.lcs.mit.edu> to implement
fstype checking, and back-end "ffs" support.
- Consistently use at least uint32_t for blocknumbers (vs. int / long / ...)
- Consistently use uintXX_t instead of u_intXX_t.
- Move various duplicated prototypes into "installboot.h"
2002-04-19 07:08:51 +00:00
drochner a6d922db91 pass correct usedevicename to telephony record/answer scripts 2002-04-17 15:27:15 +00:00
drochner 1b7bad835c fix the check for invalid usrdevicename 2002-04-17 15:26:13 +00:00
fredette 4897886acb Small fix to a warning message that was displaying the wrong filename. 2002-04-17 04:00:44 +00:00
msaitoh 959914993f fix a check of the cpu types from __sh3__ to __sh__ 2002-04-16 18:36:08 +00:00
drochner e4a8393a6b initialize a global pointer before re-reading the config file,
otherwise we'd take free()d memory as configuration data
2002-04-16 17:07:58 +00:00
wiz d62f6fdd66 this: singular, these: plural. 2002-04-16 07:28:09 +00:00
martin d200477ffe Yet another nit pick. 2002-04-16 06:11:39 +00:00
tron 0f55b8bb08 Use vfork(2) and exec(2) instead of system for performance and security
reasons.
2002-04-15 21:08:41 +00:00
tron 7b03e17cb9 Don't invoke the up- and down-script for link local IPv6 addresses. 2002-04-15 20:42:37 +00:00
martin cd924e0a08 nit pick. 2002-04-15 18:08:22 +00:00
tron c28d45defe Invoke "ip-up" and "ip-down" script with parameters exactly matching
those used by pppd(8).
2002-04-15 17:32:18 +00:00
martin 573dfcd430 Fix copyright notice, sprinkle some $NetBSD$. 2002-04-14 11:41:42 +00:00
lukem d280442683 - support optional secondary bootstrap arg for install (unused by
existing backends, but future ones need it)
- use "bootstrap" instead of "boot block" in various messages where appropriate
- rename some members of ib_params
2002-04-12 06:50:41 +00:00
lukem 050f6eec3e run fsync on the fsfd before closing 2002-04-12 03:15:20 +00:00
wiz b9337deef4 ispell, and grammar fixes. 2002-04-11 12:34:31 +00:00
lukem aa5396b0ec check size of reading bootblock against the correct size... 2002-04-11 08:02:27 +00:00
lukem 38cd7942e5 - add no_parseopt(), no_setboot() and no_clearboot() methods for back-ends
to use when they don't support a particular method
- don't automatically generate machines.c anymore
- add entries for i386 (temporarily unsupported) and shark
  (permanently unsupported)
2002-04-11 07:56:13 +00:00
martin 81bc5473a1 Fix a stupid bug when setting authentication for multiple PPP interfaces.
Add some magic to UP/DOWN interface using ISDN on startup/exit and a
configuration entry (autoupdown=yes/no) to prevent this.
2002-04-10 23:35:07 +00:00
enami c7fb252eba - Don't assume that /usr is available. Fixes (one of?) ``fgrep: not found''.
- Don't append ``dot'' when restoring resovl.conf.
2002-04-10 10:11:41 +00:00
bjh21 dd9c5ed338 Don't use __RCSID unless it's defined. 2002-04-09 19:17:11 +00:00
jdolecek 7ddad9fe53 ignore potential error return from rm -rf results
this fixes part of misc/16227 by Stoned Elipot
2002-04-09 17:17:24 +00:00
thorpej a819d2296f Use __RCSID() and __COPYRIGHT(). 2002-04-09 02:56:17 +00:00
thorpej 2597d53f5a Use __RCSID(). 2002-04-09 02:35:31 +00:00
thorpej 2e5904a52c Need <string.h> for memcmp(). 2002-04-09 02:06:29 +00:00