]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoThis commit was generated by cvs2svn to compensate for changes in r89750,
David Malone [Thu, 24 Jan 2002 17:24:16 +0000 (17:24 +0000)]
This commit was generated by cvs2svn to compensate for changes in r89750,
which included commits to RCS files with non-trunk default branches.

22 years agoimport top_3_5beta12
David Malone [Thu, 24 Jan 2002 17:24:16 +0000 (17:24 +0000)]
import top_3_5beta12

22 years agoAdd entry for EVFILT_NETDEV, which was inadverdently omitted back in Sept.
Jonathan Lemon [Thu, 24 Jan 2002 17:20:55 +0000 (17:20 +0000)]
Add entry for EVFILT_NETDEV, which was inadverdently omitted back in Sept.

22 years agoChange the order in which pam_sm_open_session() updates the logs. This
Dag-Erling Smørgrav [Thu, 24 Jan 2002 17:15:04 +0000 (17:15 +0000)]
Change the order in which pam_sm_open_session() updates the logs.  This
doesn't really make any difference, except it matches wtmp(5) better.

Don't do anything in pam_sm_close_session(); init(8) will take care of
utmp and wtmp when the tty is released.  Clearing them here would make it
possible to create a ghost session by logging in, running 'login -f $USER'
and exiting the subshell.

Sponsored by: DARPA, NAI Labs (but the bugs are all mine)

22 years agoStyle.
Ruslan Ermilov [Thu, 24 Jan 2002 16:53:08 +0000 (16:53 +0000)]
Style.

Approved by: ken

22 years agoDon't set PAM_RHOST, this is a local login.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 16:20:17 +0000 (16:20 +0000)]
Don't set PAM_RHOST, this is a local login.

Sponsored by: DARPA, NAI Labs

22 years agoCorrectly interpret PAM_RHOST being unset as an indicator of a local
Dag-Erling Smørgrav [Thu, 24 Jan 2002 16:18:43 +0000 (16:18 +0000)]
Correctly interpret PAM_RHOST being unset as an indicator of a local
login.

Sponsored by: DARPA, NAI Labs

22 years agoCorrectly interpret PAM_RHOST being unset as an indicator of a local
Dag-Erling Smørgrav [Thu, 24 Jan 2002 16:16:01 +0000 (16:16 +0000)]
Correctly interpret PAM_RHOST being unset as an indicator of a local
login.

22 years agoStyle nits.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 16:14:56 +0000 (16:14 +0000)]
Style nits.

Sponsored by: DARPA, NAI Labs

22 years agoget __time_load_locale() prototype from include file, rather than declare
Alexey Zelkin [Thu, 24 Jan 2002 15:38:59 +0000 (15:38 +0000)]
get __time_load_locale() prototype from include file, rather than declare
own

22 years agoThis commit disables chain caching.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 15:30:03 +0000 (15:30 +0000)]
This commit disables chain caching.

Chain caching is a feature of Linux-PAM, where pam_authenticate() and
pam_open_session() "freeze" the chain so that their companion
primitive (pam_setcred() and pam_close_session() respectively) will
call the exact same modules, skipping those that failed in the
previous call.

There are several reasons not to do this, the most prominent of which
is that it makes it impossible to call pam_setcred() without first
calling pam_authenticate() - which is perfectly valid according to
DCE/RFC 86.0 and XSSO, and is necessary to make 'login -f' work.

Instead of chain caching, implement something similar to the way
Solaris' libpam behaves: pam_setcred treats "sufficient" modules as if
they were "required", i.e. does not break the chain when they succeed.

PAM modules whose pam_sm_setcred() should not be called unless their
pam_sm_authenticate() succeeded can simply set a state variable using
pam_set_data() in pam_sm_authenticate(), and use pam_get_data() to
check it in pam_sm_setcred().

Sponsored by: DARPA, NAI Labs

22 years ago1) Fix a debug statement by filling in its data before the printf in
Guido van Rooij [Thu, 24 Jan 2002 15:10:53 +0000 (15:10 +0000)]
1) Fix a debug statement by filling in its data before the printf in
stead of after
2) Honour NO_TEST_UNIT_READY quirk for atapi devices as well
3) Actually support FujiFilm FinePix 6800 camera's. Will very likely also
work for other FinePix models.

Based on a debug session about half a year ago with Nik Hibma.
MFC after: 2 weeks

22 years ago* style(9)'fy
Alexey Zelkin [Thu, 24 Jan 2002 15:07:44 +0000 (15:07 +0000)]
* style(9)'fy
* declare prototype for __time_load_locale() in timelocal.h

22 years agoFix problem where with PicoBSD the shell coredumps if it does not find an
David E. O'Brien [Thu, 24 Jan 2002 13:54:19 +0000 (13:54 +0000)]
Fix problem where with PicoBSD the shell coredumps if it does not find an
entry for its terminal type in /etc/termcap.

Submitted by: bde

22 years agoDocument the even_root option.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 13:35:06 +0000 (13:35 +0000)]
Document the even_root option.

Sponsored by: DARPA, NAI Labs

22 years agoDon't let root through unless the "even_root" option was specified.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 12:47:42 +0000 (12:47 +0000)]
Don't let root through unless the "even_root" option was specified.

Sponsored by: DARPA, NAI Labs

22 years agoChange brk's prototype from char *brk(const char *) to int brk(const void *)
David Malone [Thu, 24 Jan 2002 12:11:31 +0000 (12:11 +0000)]
Change brk's prototype from char *brk(const char *) to int brk(const void *)
and sbrk's prototype from char *sbrk(int) to void *sbrk(intptr_t).

This makes us more consistant with NetBSD and standards which include
these functions. Bruce pointed out that ptrdiff_t would probably
have been better than intptr_t, but this doesn't match other
implimentations.

Also remove local declarations of sbrk and unnecessary casting.

PR: 32296
Tested by: Harti Brandt <brandt@fokus.gmd.de>
MFC after: 1 month

22 years agoProviding long filename support on the Macintosh requires over 30
Murray Stokely [Thu, 24 Jan 2002 10:44:16 +0000 (10:44 +0000)]
Providing long filename support on the Macintosh requires over 30
megabytes of additional space on the install CD.  Given our premium
for space, I don't think this is a wise use of 30MB.  HFS support was
not provided on the FreeBSD 4.4 CDs and I received a grand total of 0
complaints.

22 years agoAdd a PAM module that records sessions in utmp/wtmp/lastlog.
Dag-Erling Smørgrav [Thu, 24 Jan 2002 09:45:17 +0000 (09:45 +0000)]
Add a PAM module that records sessions in utmp/wtmp/lastlog.

Sponsored by: DARPA, NAI Labs

22 years agoFix some pastos. Rather shoddy of me...
Dag-Erling Smørgrav [Thu, 24 Jan 2002 09:44:22 +0000 (09:44 +0000)]
Fix some pastos.  Rather shoddy of me...

Sponsored by: DARPA, NAI Labs

22 years agoAdd libfetch.so.2 from a 10-Dec-2001 releng4 build.
David E. O'Brien [Thu, 24 Jan 2002 00:54:44 +0000 (00:54 +0000)]
Add libfetch.so.2 from a 10-Dec-2001 releng4 build.

22 years agoMFen (1.265 --> 1.266).
Jun Kuriyama [Thu, 24 Jan 2002 00:27:53 +0000 (00:27 +0000)]
MFen (1.265 --> 1.266).

22 years agoDetect if a udp socket on STDIN_FILENO is connected by calling
Brian Somers [Wed, 23 Jan 2002 23:36:35 +0000 (23:36 +0000)]
Detect if a udp socket on STDIN_FILENO is connected by calling
getpeername() and don't set PASSIVE mode if it's is.

22 years agoRegenerated file from previous commit to syscalls.master
Andrew Gallatin [Wed, 23 Jan 2002 22:47:47 +0000 (22:47 +0000)]
Regenerated file from previous commit to syscalls.master

22 years agoLinux/alpha uses the same BSDish return mechanism we do for
Andrew Gallatin [Wed, 23 Jan 2002 22:46:14 +0000 (22:46 +0000)]
Linux/alpha uses the same BSDish return mechanism we do for
getpid, getuid, getgid and pipe, since they bootstrapped from
OSF/1 and never cleaned up.  Switch to the native syscalls
on alpha so that the above functions work

MFC after: 7 days

22 years agoCross reference the NETWORK ADDRESS TRANSLATION section when describing
Brian Somers [Wed, 23 Jan 2002 21:11:33 +0000 (21:11 +0000)]
Cross reference the NETWORK ADDRESS TRANSLATION section when describing
the -nat flag.

Requested by: eivind

22 years agoNew release note: smmsp/mailnull users.
Bruce A. Mah [Wed, 23 Jan 2002 18:49:43 +0000 (18:49 +0000)]
New release note:  smmsp/mailnull users.

Reviewed by: gshapiro

22 years agoAdd a PAM module that provides an account management component for checking
Dag-Erling Smørgrav [Wed, 23 Jan 2002 17:42:16 +0000 (17:42 +0000)]
Add a PAM module that provides an account management component for checking
either PAM_RHOST or PAM_TTY against /etc/login.access.o

This uncovers a problem with PAM_RHOST, in that if we always set it, there
is no way to distinguish between a user logging in locally and a user
logging in using 'ssh localhost'.  This will be fixed by first making sure
that all PAM modules can handle PAM_RHOST being unset (which is currently
not the case), and then modifying su(1) and login(1) to not set it for
local logins.

Sponsored by: DARPA, NAI Labs

22 years agoAdd an AUTHORS section crediting ThinkSec, DARPA and NAI Labs.
Dag-Erling Smørgrav [Wed, 23 Jan 2002 17:16:00 +0000 (17:16 +0000)]
Add an AUTHORS section crediting ThinkSec, DARPA and NAI Labs.

Sponsored by: DARPA, NAI Labs

22 years agoAdd pam_ssh support to the static PAM library, libpam.a:
Ruslan Ermilov [Wed, 23 Jan 2002 15:54:17 +0000 (15:54 +0000)]
Add pam_ssh support to the static PAM library, libpam.a:

- Spam /usr/lib some more by making libssh a standard library.
- Tweak ${LIBPAM} and ${MINUSLPAM}.
- Garbage collect unused libssh_pic.a.
- Add fake -lz dependency to secure/ makefiles needed for
  dynamic linkage with -lssh.

Reviewed by: des, markm
Approved by: markm

22 years agoBase the comparison on UIDs, not on user names.
Dag-Erling Smørgrav [Wed, 23 Jan 2002 15:16:01 +0000 (15:16 +0000)]
Base the comparison on UIDs, not on user names.

Sponsored by: DARPA, NAI Labs

22 years agoMake libssh.so useable (undefined reference to IPv4or6).
Ruslan Ermilov [Wed, 23 Jan 2002 15:06:47 +0000 (15:06 +0000)]
Make libssh.so useable (undefined reference to IPv4or6).

Reviewed by: des, markm
Approved by: markm

22 years agodhclient-script.conf lives in /etc and not /sbin.
Giorgos Keramidas [Wed, 23 Jan 2002 13:56:41 +0000 (13:56 +0000)]
dhclient-script.conf lives in /etc and not /sbin.

PR: docs/32008
Submitted by: John E Hein <jhein@timing.com>
Approved by: obrien

22 years agoAdd local/share/java/classes, local/share/sgml, local/share/xml
Dag-Erling Smørgrav [Wed, 23 Jan 2002 13:02:16 +0000 (13:02 +0000)]
Add local/share/java/classes, local/share/sgml, local/share/xml

Approved by: ru, silence on -ports
MFC after: 1 week

22 years agoThe sixth argument to the NET_RT_IFLIST sysctl is actually 0 for
Ruslan Ermilov [Wed, 23 Jan 2002 12:48:08 +0000 (12:48 +0000)]
The sixth argument to the NET_RT_IFLIST sysctl is actually 0 for
all interfaces, and ifnet.if_index value for a single interface.

22 years agoCG hard sentence breaks.
Maxim Sobolev [Wed, 23 Jan 2002 09:35:55 +0000 (09:35 +0000)]
CG hard sentence breaks.

Submitted by: ru

22 years agoin fget() return EINVAL when the descriptor requested is negative.
Alfred Perlstein [Wed, 23 Jan 2002 08:40:35 +0000 (08:40 +0000)]
in fget() return EINVAL when the descriptor requested is negative.

22 years agoRemove holdfp() prototype now that the function no longer exists
Alfred Perlstein [Wed, 23 Jan 2002 08:28:55 +0000 (08:28 +0000)]
Remove holdfp() prototype now that the function no longer exists

22 years agoforced commit, Previous revision also removed the holdfp() function
Alfred Perlstein [Wed, 23 Jan 2002 08:28:15 +0000 (08:28 +0000)]
forced commit, Previous revision also removed the holdfp() function
from the kernel.

22 years agomake pread use fget_read instead of holdfp.
Alfred Perlstein [Wed, 23 Jan 2002 08:22:59 +0000 (08:22 +0000)]
make pread use fget_read instead of holdfp.

22 years agoRoll to latest production level firmware.
Matt Jacob [Wed, 23 Jan 2002 06:42:54 +0000 (06:42 +0000)]
Roll to latest production level firmware.

MFC after: 1 day

22 years agoimprove sndstat output of feederchains so it can be understood without
Cameron Grant [Wed, 23 Jan 2002 06:02:15 +0000 (06:02 +0000)]
improve sndstat output of feederchains so it can be understood without
reading the feeder sourcecode

22 years agoprint warnings if a pcm*.buffersize hint is out of range or a non-power-of-2
Cameron Grant [Wed, 23 Jan 2002 05:49:41 +0000 (05:49 +0000)]
print warnings if a pcm*.buffersize hint is out of range or a non-power-of-2

22 years agofix some comments accidentally hit by search/replace several revisions ago
Cameron Grant [Wed, 23 Jan 2002 05:41:35 +0000 (05:41 +0000)]
fix some comments accidentally hit by search/replace several revisions ago

22 years agoset the speeds the right way round for recording using the rate feeder
Cameron Grant [Wed, 23 Jan 2002 05:35:12 +0000 (05:35 +0000)]
set the speeds the right way round for recording using the rate feeder

22 years agoadd more error checking in chn_init() and chn_reset()
Cameron Grant [Wed, 23 Jan 2002 05:21:27 +0000 (05:21 +0000)]
add more error checking in chn_init() and chn_reset()

22 years agomake the feederchain builder work for recording. this has not been tested
Cameron Grant [Wed, 23 Jan 2002 05:10:56 +0000 (05:10 +0000)]
make the feederchain builder work for recording.  this has not been tested
extensively as none of my testboxes have speakers or an audio source at
present, but the chains built look correct and reading /dev/audio (ulaw,
translated from signed 16 bit little-endian) gives values within the
expected range for silence.

22 years agodon't bother checking if an unsigned parameter is less than 0 in a KASSERT
Cameron Grant [Wed, 23 Jan 2002 04:50:51 +0000 (04:50 +0000)]
don't bother checking if an unsigned parameter is less than 0 in a KASSERT

22 years agoadd more error checking to open of /dev/dsp* and /dev/audio* - if the
Cameron Grant [Wed, 23 Jan 2002 04:44:03 +0000 (04:44 +0000)]
add more error checking to open of /dev/dsp* and /dev/audio* - if the
default format for the chosen subdevice cannot be obtained return an error
instead of returning success with an indeteterminate format selected.

note that this should never happen once the feederchain builder works for
recording.

22 years agodon't allow mmap beyond the end of the buffer
Cameron Grant [Wed, 23 Jan 2002 04:32:18 +0000 (04:32 +0000)]
don't allow mmap beyond the end of the buffer

Submitted by: Philippe Anel <philippe.anel@noos.fr> (partially)

22 years agoremove modules no longer built
Cameron Grant [Wed, 23 Jan 2002 03:43:16 +0000 (03:43 +0000)]
remove modules no longer built

22 years agoconsolidate bits that don't need to be seperate modules into the modules
Cameron Grant [Wed, 23 Jan 2002 03:32:36 +0000 (03:32 +0000)]
consolidate bits that don't need to be seperate modules into the modules
that want them

22 years agoAdd a stub for softdep_request_cleanup() so that compilation without
Kirk McKusick [Wed, 23 Jan 2002 02:18:56 +0000 (02:18 +0000)]
Add a stub for softdep_request_cleanup() so that compilation without
SOFTUPDATES option works properly.

Submitted by: Benno Rice <benno@jeamland.net>

22 years agoAdd missing destroy_dev().
Wes Peters [Wed, 23 Jan 2002 01:11:52 +0000 (01:11 +0000)]
Add missing destroy_dev().

Submitted by: Maxime Henrion <mux@sneakerz.org>
Reviewed by: msmith@
MFC after: 3 weeks

22 years agoMFen (1.264 --> 1.265).
Jun Kuriyama [Wed, 23 Jan 2002 00:46:49 +0000 (00:46 +0000)]
MFen (1.264 --> 1.265).
Refine translation (by y-koga@jp).

22 years agoUndo the work-around for the sendfile bug where nbytes needed the hdr/trl
David Greenman [Tue, 22 Jan 2002 23:35:09 +0000 (23:35 +0000)]
Undo the work-around for the sendfile bug where nbytes needed the hdr/trl
size added to it in order for it to work properly when nbytes != 0.

Reviewed by: alfred
MFC after: 3 days

22 years agoThe mode of files created by ctm_rmail was always 0600, even if the
Ian Dowse [Tue, 22 Jan 2002 22:54:52 +0000 (22:54 +0000)]
The mode of files created by ctm_rmail was always 0600, even if the
umask was less restrictive. This was caused by the use of mkstemp()
which internally passes a mode of 0600 to open(). Fix this by
explicitly chmod'ing the files to (0666 & ~umask).

PR: bin/16119
Submitted by: Sascha Blank <blank@uni-trier.de>

22 years agoAdd ngctl "write" command.
Archie Cobbs [Tue, 22 Jan 2002 22:01:25 +0000 (22:01 +0000)]
Add ngctl "write" command.

MFC after: 1 week

22 years agoRemove bogus _POSIX_ACL_PATH_MAX definition.
Brian Feldman [Tue, 22 Jan 2002 21:33:12 +0000 (21:33 +0000)]
Remove bogus _POSIX_ACL_PATH_MAX definition.

22 years agoFix a typo and fix indenting for struct sockaddr to match other structs.
Jens Schweikhardt [Tue, 22 Jan 2002 21:26:15 +0000 (21:26 +0000)]
Fix a typo and fix indenting for struct sockaddr to match other structs.
No content changes.
MFC after: 3 weeks

22 years agoDon't use `you'.
Maxim Sobolev [Tue, 22 Jan 2002 21:11:18 +0000 (21:11 +0000)]
Don't use `you'.

Submitted by: ru

22 years agoRestore C99 standard conformance information, isblank() _is_ in final
Andrey A. Chernov [Tue, 22 Jan 2002 20:14:35 +0000 (20:14 +0000)]
Restore C99 standard conformance information, isblank() _is_ in final
standard document

Pointed by: "Jacques A. Vidrine" <n@nectar.cc>

22 years agoShow arguments of command line options
Philippe Charnier [Tue, 22 Jan 2002 19:49:48 +0000 (19:49 +0000)]
Show arguments of command line options
Reviewed by: ru

22 years agoThe ENDPTS_EQ macro was comparing the one of the fports to itself. Fix.
Jonathan Lemon [Tue, 22 Jan 2002 17:54:28 +0000 (17:54 +0000)]
The ENDPTS_EQ macro was comparing the one of the fports to itself.  Fix.

Submitted by: emy@boostworks.com

22 years agoInitialize the sysctl_ctx list early, which avoids a panic in case other
Jonathan Lemon [Tue, 22 Jan 2002 17:51:44 +0000 (17:51 +0000)]
Initialize the sysctl_ctx list early, which avoids a panic in case other
allocatiosn fail and fxp_release() is called.

22 years agoNull commit - the previous log message should have read:
David Greenman [Tue, 22 Jan 2002 17:38:58 +0000 (17:38 +0000)]
Null commit -  the previous log message should have read:

  Fixed bug in calculation of amount of file sent when nbytes !=0 and
  headers or trailers are supplied. Reported by Vladislav Shabanov
  <vs@rambler-co.ru>.

22 years agoUse the proper type (gid_t) for (group)->gr_gid to be orthogonal
Mark Murray [Tue, 22 Jan 2002 17:32:53 +0000 (17:32 +0000)]
Use the proper type (gid_t) for (group)->gr_gid to be orthogonal
with uid_t usage and (user)->pw_uid.

PR: 3242

22 years agoFixed bug in calculation of amount of file to send when nbytes !=0 and
David Greenman [Tue, 22 Jan 2002 17:32:10 +0000 (17:32 +0000)]
Fixed bug in calculation of amount of file to send when nbytes !=0 and
headers or trailers are supplied. Reported by Vladislav Shabanov
<vs@rambler-co.ru>.

PR: 33771
Submitted by: Maxim Konovalov <maxim@macomnet.ru>
MFC after: 3 days

22 years agoDo not taint ::/124 for localhost reverse table.
Hajimu UMEMOTO [Tue, 22 Jan 2002 17:22:41 +0000 (17:22 +0000)]
Do not taint ::/124 for localhost reverse table.

22 years agoReincarnate SETUID code in man(1), not compiled in by default.
Ruslan Ermilov [Tue, 22 Jan 2002 15:15:38 +0000 (15:15 +0000)]
Reincarnate SETUID code in man(1), not compiled in by default.

The code will be fixed for all known security vulnerabilities,
and a make.conf(5) knob (ENABLE_SUID_MAN) will be provided for
those who still want it installed setuid for whatever reasons.

22 years agoRevert revision 1.57 -- ache@ axed /var/spool/uucp in BSD.var.dist,v 1.55.
Ruslan Ermilov [Tue, 22 Jan 2002 14:55:25 +0000 (14:55 +0000)]
Revert revision 1.57 -- ache@ axed /var/spool/uucp in BSD.var.dist,v 1.55.

22 years agoFix the description of the O_NONBLOCK flag to match reality.
Ruslan Ermilov [Tue, 22 Jan 2002 14:18:55 +0000 (14:18 +0000)]
Fix the description of the O_NONBLOCK flag to match reality.

Prodded by: Maxim Konovalov <maxim@macomnet.ru>
Obtained from: BSD/OS

22 years agoFix a typo I made in revision 1.5.
Ruslan Ermilov [Tue, 22 Jan 2002 12:38:43 +0000 (12:38 +0000)]
Fix a typo I made in revision 1.5.

Submitted by: trevor

22 years agoIn certain cases sbuf_printf() and sbuf_vprintf() could mistakely
Poul-Henning Kamp [Tue, 22 Jan 2002 11:22:55 +0000 (11:22 +0000)]
In certain cases sbuf_printf() and sbuf_vprintf() could mistakely
make extendable sbufs as overflowed.

Approved by: des

22 years agoMention that a minimal version of camcontrol(8) is now available in the
Joerg Wunsch [Tue, 22 Jan 2002 11:13:43 +0000 (11:13 +0000)]
Mention that a minimal version of camcontrol(8) is now available in the
installation environment (currently only for i386 releases).

22 years agoTiny style change. Whitespace only change.
Giorgos Keramidas [Tue, 22 Jan 2002 10:44:40 +0000 (10:44 +0000)]
Tiny style change.  Whitespace only change.
Use tab after #define ENOTSUP.

Reviewed by: mike

22 years agoFinish cleanup in chroot.c CSRG revision 5.6 by Keith Bostic
Ruslan Ermilov [Tue, 22 Jan 2002 10:44:02 +0000 (10:44 +0000)]
Finish cleanup in chroot.c CSRG revision 5.6 by Keith Bostic
(never installed setuid, so don't need to reset the uid).

PR: bin/34159

22 years agoFinish cleanup in kvm.c revisions 1.10 and 1.11 -- mark sf (swapfile)
Ruslan Ermilov [Tue, 22 Jan 2002 10:07:03 +0000 (10:07 +0000)]
Finish cleanup in kvm.c revisions 1.10 and 1.11 -- mark sf (swapfile)
argument to kvm_open() and kvm_openfiles() as unused.

BSD didn't read swap since kvm.c CSRG revision 5.21 (u-area is pageable
under new VM.  no need to read from swap.)

The old !NEWVM code was removed in CSRG revision 5.23 (~ten years ago).

22 years agoThis patch fixes a long standing complaint with soft updates in
Kirk McKusick [Tue, 22 Jan 2002 06:17:22 +0000 (06:17 +0000)]
This patch fixes a long standing complaint with soft updates in
which small and/or nearly full filesystems would fail with `file
system full' messages when trying to replace a number of existing
files (for example during a system installation). When the allocation
routines are about to fail with a file system full condition, they
make a call to softdep_request_cleanup() which attempts to accelerate
the flushing of pending deletion requests in an effort to free up
space. In the face of filesystem I/O requests that exceed the
available disk transfer capacity, the cleanup request could take
an unbounded amount of time. Thus, the softdep_request_cleanup()
routine will only try for tickdelay seconds (default 2 seconds)
before giving up and returning a filesystem full error. Under typical
conditions, the softdep_request_cleanup() routine is able to free
up space in under fifty milliseconds.

22 years agoMFen (1.263 --> 1.264).
Jun Kuriyama [Tue, 22 Jan 2002 06:04:46 +0000 (06:04 +0000)]
MFen (1.263 --> 1.264).
Fix typo.

22 years agohw.physmem is more properly a loader tunable, not an environment
Bruce A. Mah [Tue, 22 Jan 2002 05:21:21 +0000 (05:21 +0000)]
hw.physmem is more properly a loader tunable, not an environment
variable (which is ambiguous).

Submitted by: dwhite

22 years agoo Remove the -V [version number] option, since our version of at(1) no
Mike Barcroft [Tue, 22 Jan 2002 03:04:15 +0000 (03:04 +0000)]
o Remove the -V [version number] option, since our version of at(1) no
  longer resembles the original.
o Remove references to `you' in the manual.

Submitted by: Joe Halpin <joe.halpin@attbi.com>

22 years agoDon't read the sigcontext pointer directly from userspace. If the process
Andrew Gallatin [Tue, 22 Jan 2002 02:40:39 +0000 (02:40 +0000)]
Don't read the sigcontext pointer directly from userspace.  If the process
is swapped out, this can cause a system panic.  Copy it in, instead

tested by: Bernd Walter <ticso@cicely8.cicely.de>

22 years agoList bit 18 (reserved, apparently present on thunderbird cpus)
Peter Wemm [Tue, 22 Jan 2002 01:28:32 +0000 (01:28 +0000)]
List bit 18 (reserved, apparently present on thunderbird cpus)
and bit 19 (athlon XP/MP rev 0x662 and later) for amd_features.

Submitted by:  dwcjr

22 years agoThe idea of mapping non-existen characters to space (0x20) was nice alone,
Andrey A. Chernov [Mon, 21 Jan 2002 21:52:02 +0000 (21:52 +0000)]
The idea of mapping non-existen characters to space (0x20) was nice alone,
but those maps also used as backward maps for Paste, so space becomes mapped
to last non-existen character on Paste as result.

Fix it by mapping non-existen characters to another non-existen one, i.e. to
0x00, so unused 0x00 can be backward-mapped to some junk without real harm.

Pointed by: Alexander Kabaev <ak03@gte.com>

22 years agoGah! last commit botched indentation, fix indentation and some other
Poul-Henning Kamp [Mon, 21 Jan 2002 20:57:03 +0000 (20:57 +0000)]
Gah! last commit botched indentation, fix indentation and some other
white-space nits while at it.

22 years agoRestructure slightly, eliminating some repetitive source lines and
Poul-Henning Kamp [Mon, 21 Jan 2002 20:50:06 +0000 (20:50 +0000)]
Restructure slightly, eliminating some repetitive source lines and
making GEOM patches simpler and more readable at the same time.

22 years agoLink pam_opieaccess, pam_self and pam_ssh into the static library.
Dag-Erling Smørgrav [Mon, 21 Jan 2002 20:43:01 +0000 (20:43 +0000)]
Link pam_opieaccess, pam_self and pam_ssh into the static library.

Sponsored by: DARPA, NAI Labs

22 years agoAdd the necessary dependencies and linker flags for linking with a
Dag-Erling Smørgrav [Mon, 21 Jan 2002 20:42:25 +0000 (20:42 +0000)]
Add the necessary dependencies and linker flags for linking with a
static PAM library that includes pam_ssh.

Sponsored by: DARPA, NAI Labs

22 years agoCorrect typos and macro usage.
Ollivier Robert [Mon, 21 Jan 2002 20:12:02 +0000 (20:12 +0000)]
Correct typos and macro usage.

PR: 33179, 33181, 33295
Submitted by: Norihiro Kumagai <kumagai@attbi.com>
Reviewed by: sheldonh (MAINTAINER)
MFC after: 2 days

22 years ago- Check the address family of the destination cached in a PCB.
Hajimu UMEMOTO [Mon, 21 Jan 2002 20:04:22 +0000 (20:04 +0000)]
- Check the address family of the destination cached in a PCB.
- Clear the cached destination before getting another cached route.
  Otherwise, garbage in the padding space (which might be filled in if it was
  used for IPv4) could annoy rtalloc.

Obtained from: KAME

22 years ago- Check the address family of a cached destination, in case of
Hajimu UMEMOTO [Mon, 21 Jan 2002 20:02:36 +0000 (20:02 +0000)]
- Check the address family of a cached destination, in case of
  sharing the cache with IPv4.
- Check if the cached route is up in in6_selectsrc().

Obtained from: KAME

22 years agoRemove my workaround fallback since PAM now do it properly.
Andrey A. Chernov [Mon, 21 Jan 2002 19:07:15 +0000 (19:07 +0000)]
Remove my workaround fallback since PAM now do it properly.

22 years agoOn second thought, getpwnam() failure should be treated just as if the user
Dag-Erling Smørgrav [Mon, 21 Jan 2002 19:05:45 +0000 (19:05 +0000)]
On second thought, getpwnam() failure should be treated just as if the user
existed, but had no OPIE key, i.e. PAM_IGNORE.

Pointed out by: ache
Sponsored by: DARPA, NAI Labs

22 years agoReturn PAM_SERVICE_ERR rather than PAM_USER_UNKNOWN if getpwnam() fails, as
Dag-Erling Smørgrav [Mon, 21 Jan 2002 18:53:03 +0000 (18:53 +0000)]
Return PAM_SERVICE_ERR rather than PAM_USER_UNKNOWN if getpwnam() fails, as
PAM_USER_UNKNOWN will break the chain, revealing to an attacker that the
user does not exist.

Sponsored by: DARPA, NAI Labs

22 years agoEnable OPIE by default, using the no_fake_prompts option to hide it from
Dag-Erling Smørgrav [Mon, 21 Jan 2002 18:51:24 +0000 (18:51 +0000)]
Enable OPIE by default, using the no_fake_prompts option to hide it from
users who don't wish to use it.  If the admin is worried about leaking
information about which users exist and which have OPIE enabled, the
no_fake_prompts option can simply be removed.

Also insert the appropriate pam_opieaccess lines after pam_opie to break
the chain in case the user is logging in from an untrusted host, or has a
.opiealways file.  The entire opieaccess / opiealways concept is slightly
unpammish, but admins familiar with OPIE will expect it to work.

Reviewed by: ache, markm
Sponsored by: DARPA, NAI Labs

22 years agoFurther changes to allow enabling pam_opie(8) by default:
Dag-Erling Smørgrav [Mon, 21 Jan 2002 18:46:25 +0000 (18:46 +0000)]
Further changes to allow enabling pam_opie(8) by default:

 - Ignore the {try,use}_first_pass options by clearing PAM_AUTHTOK before
   challenging the user.  These options are meaningless for pam_opie(8)
   since the user can't possibly know the right response before she sees
   the challenge.

 - Introduce the no_fake_prompts option.  If this option is set, pam_opie(8)
   will fail - rather than present a bogus challenge - if the target user
   does not have an OPIE key.  With this option, users who haven't set up
   OPIE won't have to wonder what that "weird otp-md5 s**t" means :)

Reviewed by: ache, markm
Sponsored by: DARPA, NAI Labs

22 years agoWhen running on a local terminal, set PAM_RHOST to the local hostname.
Dag-Erling Smørgrav [Mon, 21 Jan 2002 16:19:38 +0000 (16:19 +0000)]
When running on a local terminal, set PAM_RHOST to the local hostname.

Sponsored by: DARPA, NAI Labs

22 years agoRFC1122 requires that addresses of the form { 127, <any> } MUST NOT
Ruslan Ermilov [Mon, 21 Jan 2002 13:59:42 +0000 (13:59 +0000)]
RFC1122 requires that addresses of the form { 127, <any> } MUST NOT
appear outside a host.

PR: 30792, 33996
Obtained from: ip_input.c
MFC after: 1 week

22 years agoAdd a new module, pam_opieaccess(8), which is responsible for checking
Dag-Erling Smørgrav [Mon, 21 Jan 2002 13:43:53 +0000 (13:43 +0000)]
Add a new module, pam_opieaccess(8), which is responsible for checking
/etc/opieaccess and ~/.opiealways so we can decide what to do after
pam_opie(8) fails.

Sponsored by: DARPA, NAI Labs
Reviewed by: ache, markm