]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoThis commit was generated by cvs2svn to compensate for changes in r90067,
Maxim Sobolev [Fri, 1 Feb 2002 16:28:01 +0000 (16:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r90067,
which included commits to RCS files with non-trunk default branches.

22 years agoVirgin import (trimmed) of Bzip2 version 1.0.2
Maxim Sobolev [Fri, 1 Feb 2002 16:28:01 +0000 (16:28 +0000)]
Virgin import (trimmed) of Bzip2 version 1.0.2

22 years agoCompile osigreturn() unconditionally since it will always be needed on
Bruce Evans [Fri, 1 Feb 2002 15:44:03 +0000 (15:44 +0000)]
Compile osigreturn() unconditionally since it will always be needed on
some arches and the syscall table is machine-independent.  It was
(bogusly) conditional on COMPAT_43, so this usually makes no difference.

ia64: in addition:
- replace the bogus cloned comment before osigreturn() by a correct one.
  osigreturn() is just a stub fo ia64's.
- fix the formatting of cloned comment before sigreturn().
- fix the return code.  use nosys() instead of returning ENOSYS to get
  the same semantics as if the syscall is not in the syscall table.
  Generating SIGSYS is actually correct here.
- fix style bugs.

powerpc: copy the cleaned up ia64 stub.  This mainly fixes a bogus comment.

sparc64: copy the cleaned up the ia64 stub, since there was no stub before.

22 years agoExpand COMBRD(x) macro.
Yoshihiro Takahashi [Fri, 1 Feb 2002 14:48:44 +0000 (14:48 +0000)]
Expand COMBRD(x) macro.

22 years agoMFen (1.269 --> 1.271).
Jun Kuriyama [Fri, 1 Feb 2002 13:37:35 +0000 (13:37 +0000)]
MFen (1.269 --> 1.271).

22 years agoRefer to the original mode of the file, not the mode of the original
Sheldon Hearn [Fri, 1 Feb 2002 12:37:21 +0000 (12:37 +0000)]
Refer to the original mode of the file, not the mode of the original
file.

PR: docs/34224
Submitted by: "Gary W. Swearingen" <swear@blarg.net>

22 years agoConnect the pam_lastlog(8) and pam_login_access(8) modules to the build.
Dag-Erling Smørgrav [Fri, 1 Feb 2002 08:49:53 +0000 (08:49 +0000)]
Connect the pam_lastlog(8) and pam_login_access(8) modules to the build.

Sponsored by: DARPA, NAI Labs

22 years agoReturn ENONAME if getaddrinfo() is called with AI_NUMERICHOST
Peter Pentchev [Fri, 1 Feb 2002 08:35:16 +0000 (08:35 +0000)]
Return ENONAME if getaddrinfo() is called with AI_NUMERICHOST
and the hostname given is not numeric.

PR: 34390
Submitted by: Serge van den Boom <svdb@stack.nl>
Approved by: silence from -net
MFC after: 1 month

22 years agoFixed slipage in editor.
David E. O'Brien [Fri, 1 Feb 2002 07:26:00 +0000 (07:26 +0000)]
Fixed slipage in editor.

22 years agoCut-n-paste in `list' mode considered harmful.
David E. O'Brien [Fri, 1 Feb 2002 07:25:07 +0000 (07:25 +0000)]
Cut-n-paste in `list' mode considered harmful.

22 years agoAdd -a option (SUSv3) to split(1).
Mike Barcroft [Fri, 1 Feb 2002 06:55:18 +0000 (06:55 +0000)]
Add -a option (SUSv3) to split(1).

Submitted by: Tim J. Robbins <tim@robbins.dropbear.id.au>
MFC after: 1 month

22 years agoSome netgraph parse types (such as for the 'value' field in ng_ksocket's
Archie Cobbs [Fri, 1 Feb 2002 02:21:41 +0000 (02:21 +0000)]
Some netgraph parse types (such as for the 'value' field in ng_ksocket's
'struct ng_ksocket_sockopt') like to peek into the ng_mesg header for
information. Make sure when generating default values that we provide
a valid header to peek into.

MFC after: 1 week

22 years ago* Remove __P and convert to ANSI prototypes.
David E. O'Brien [Fri, 1 Feb 2002 01:39:09 +0000 (01:39 +0000)]
* Remove __P and convert to ANSI prototypes.
* Remove 'register'.  (some functions had 7+ register functions...)
* Fix SCM ID's.

22 years ago* Remove __P and convert to ANSI prototypes.
David E. O'Brien [Fri, 1 Feb 2002 01:32:19 +0000 (01:32 +0000)]
* Remove __P and convert to ANSI prototypes.
* Remove 'register'.  (some functions had 7+ register functions...)
* Fix SCM ID's.

22 years agoNote new status of __P. Don't use it.
Warner Losh [Fri, 1 Feb 2002 01:26:24 +0000 (01:26 +0000)]
Note new status of __P.  Don't use it.

Reviewed by: arch@, mckusick (in principle)

22 years agoThere is no need to wrap vendor id bits with '#if 0' if they come pre-wrapped.
David E. O'Brien [Fri, 1 Feb 2002 01:17:22 +0000 (01:17 +0000)]
There is no need to wrap vendor id bits with '#if 0' if they come pre-wrapped.

22 years ago* Remove 'register'. (some functions had 7+ register functions...)
David E. O'Brien [Fri, 1 Feb 2002 01:08:48 +0000 (01:08 +0000)]
* Remove 'register'.  (some functions had 7+ register functions...)
* Fix SCM ID's.

22 years agoFix SCM ID's.
David E. O'Brien [Fri, 1 Feb 2002 00:57:29 +0000 (00:57 +0000)]
Fix SCM ID's.

22 years agoThe SiS 630 chipset is supported by the sis(4) driver.
Bruce A. Mah [Thu, 31 Jan 2002 19:40:51 +0000 (19:40 +0000)]
The SiS 630 chipset is supported by the sis(4) driver.

PR: 34437
Submitted by: Hendrik Scholz <hendrik@scholz.net>

22 years agoGC P_BUFEXHAUST leftovers, we've had a new mechanism to avoid buffer
Matthew Dillon [Thu, 31 Jan 2002 18:39:44 +0000 (18:39 +0000)]
GC P_BUFEXHAUST leftovers, we've had a new mechanism to avoid buffer
cache lockups for over a year now.

MFC after: 0 days

22 years agoNew release notes: an(4) LEAP, Home key, and Linux utility support;
Bruce A. Mah [Thu, 31 Jan 2002 18:38:03 +0000 (18:38 +0000)]
New release notes:  an(4) LEAP, Home key, and Linux utility support;
uudecode(1) -s; top 3.5b12.

Updated release notes:  binutils update to 27 January 2002 snapshot,
OpenSSL 0.9.6c.

22 years agoNew release notes: iir driver (submitted by msmith), dumpdev loader tunable
Bruce A. Mah [Thu, 31 Jan 2002 17:50:46 +0000 (17:50 +0000)]
New release notes:  iir driver (submitted by msmith), dumpdev loader tunable
(submitted by sobomax).

22 years agoFix minor style nit.
David E. O'Brien [Thu, 31 Jan 2002 17:26:37 +0000 (17:26 +0000)]
Fix minor style nit.

22 years agoMFC: gcc.295/config/alpha/freebsd.h revision 1.12:
David E. O'Brien [Thu, 31 Jan 2002 17:20:44 +0000 (17:20 +0000)]
MFC: gcc.295/config/alpha/freebsd.h revision 1.12:
Remove our definition for FUNCTION_PROFILER as it is wrong.

Note that "jsr $28,_mcount" is a macro for
ldq $27, _mcount($29)           !literal!1
jsr $28, ($27), _mcount         !lituse_jsr!1

1. The call to _mcount is added by alpha_expand_prologue after we load the gp.
Our _mcount uses $27 for the incoming address, unlike OSF/1 and Linux,
which use $28.  This probably doesn't matter since we probably don't use $27
within _mcount itself.
2. You can't use this insn with _mcount because it uses the PLT, which clobbers
the return address in $28.  Note that the prologue_mcount pattern carefully
avoids adding the lituse_jsr relocation so that we call through the GOT
directly.

Submitted by: Richard Henderson <rth@redhat.com>

22 years agoidentify GLPRISM2 PCI WaveLAN/IEEE 802.11
Alfred Perlstein [Thu, 31 Jan 2002 16:55:13 +0000 (16:55 +0000)]
identify GLPRISM2 PCI WaveLAN/IEEE 802.11

Submitted by: Toni Andjelkovic <toni@soth.at>

22 years agoMerge rev 1.1.1.3 from vendor branch: fix refresh / warning deadlock
Sheldon Hearn [Thu, 31 Jan 2002 14:23:49 +0000 (14:23 +0000)]
Merge rev 1.1.1.3 from vendor branch: fix refresh / warning deadlock

22 years agoFinish revs.1.23 and 1.24 so that MCOUNT_ENTER really actually compiles
Bruce Evans [Thu, 31 Jan 2002 13:49:55 +0000 (13:49 +0000)]
Finish revs.1.23 and 1.24 so that MCOUNT_ENTER really actually compiles
for SMP in the plain profiling case.  It seems to work too.

This error was not detected by LINT because LINT only compiles the
GUPROF profiling case, which is is a superset of the plain profiling
case for !SMP but which is so broken for SMP that the buggy code is
not compiled.

22 years agoMerge rev 1.1.1.2 from the vendor branch: fix leftright scrolling for
Sheldon Hearn [Thu, 31 Jan 2002 13:45:31 +0000 (13:45 +0000)]
Merge rev 1.1.1.2 from the vendor branch: fix leftright scrolling for
long lines

22 years agoAdd $FreeBSD$ CVS Id tag so that I can merge rev 1.1.1.2 from the
Sheldon Hearn [Thu, 31 Jan 2002 13:44:36 +0000 (13:44 +0000)]
Add $FreeBSD$ CVS Id tag so that I can merge rev 1.1.1.2 from the
vendor branch.  This file was taken off the vendor branch in 1998,
before we started enforcing the presence of CVS Id tags. :-(

22 years agoRevert last commit. stdio.h include is for perror(), not strerror()
Chris D. Faulhaber [Thu, 31 Jan 2002 13:32:14 +0000 (13:32 +0000)]
Revert last commit.  stdio.h include is for perror(), not strerror()

22 years agoThis commit was generated by cvs2svn to compensate for changes in r90019,
Sheldon Hearn [Thu, 31 Jan 2002 13:28:37 +0000 (13:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r90019,
which included commits to RCS files with non-trunk default branches.

22 years agoFix core dump in left-right screen scrolling for long lines.
Sheldon Hearn [Thu, 31 Jan 2002 13:28:37 +0000 (13:28 +0000)]
Fix core dump in left-right screen scrolling for long lines.

The nvi maintainer expects this patch to be included in the upcoming
nvi-1.81.6 unless a better fix presents itself.

The MFC reminder below should be taken as a cue to MFC rev 1.1.1.2
of vs_relative.c as well.

PR: bin/26869
Reported by: Alex Semenyaka <alexs@snark.ratmir.ru>
Obtained from: skimo@kotnet.org (via "Alexey V. Neyman" <alex.neyman@auriga.ru>)
MFC after: 1 month

22 years agoModified the rules for vendor ids:
Bruce Evans [Thu, 31 Jan 2002 13:26:29 +0000 (13:26 +0000)]
Modified the rules for vendor ids:
- explictly say not to edit infrastructure for vendor ids (not just the
  ids).
- say to enclose vendor ids and their infrastucture in ``#if 0'', and
  partly explain why.
- don't set a bad example by mangling the Berkeley id infrastructure from
  ``static char sccsid[] ...'' to ``__RCSID(...)''.
- show a blank line between the vendor id cruft and the FreeeBSD if cruft
  in the example.
- relaxed the rule about adding "From: " to say that "From: " is actually
  useful if the file has been renamed.
- minor English improvements.

Discussed with: obrien

22 years agostrerror(3) is in string.h, not stdio.h
Chris D. Faulhaber [Thu, 31 Jan 2002 13:20:48 +0000 (13:20 +0000)]
strerror(3) is in string.h, not stdio.h

MFC after: 1 week

22 years agoFixed world breakage due to missing include of <sys/cdefs.h> in previous
Bruce Evans [Thu, 31 Jan 2002 12:58:36 +0000 (12:58 +0000)]
Fixed world breakage due to missing include of <sys/cdefs.h> in previous
commit.

Fixed related style bugs:
basename.c: misplaced '#if 0'
dirname.c: misplaced '#if 0'
getgrent.c: missing '#if 0', and tab lossage in vendor id (the previous
            commit fixed the complete corruption of the vendor id but
    lost a tab)
getpwent.c: missing '#if 0'

22 years agoAdd support of PhotoClip USB Camera (http://www.myphotoclip.com):
Maxim Sobolev [Thu, 31 Jan 2002 11:39:17 +0000 (11:39 +0000)]
Add support of PhotoClip USB Camera (http://www.myphotoclip.com):
- Vendor&Device IDs for USB product,
- quirk for SCSI CAM.

PR: 34481
Submitted by: Olexander Kunytsa <kunia@x-telecom.net>
MFC in: 3 days

22 years agoFix irq/drq handling. IRQ and DRQ resource information can be get
Takanori Watanabe [Thu, 31 Jan 2002 09:18:27 +0000 (09:18 +0000)]
Fix irq/drq handling. IRQ and DRQ resource information can be get
in one object for one resource. Array of values in a object means
possible values for the object.

22 years agoMFi386: revision 1.489.
Yoshihiro Takahashi [Thu, 31 Jan 2002 08:35:38 +0000 (08:35 +0000)]
MFi386: revision 1.489.

22 years agoChanged iskanji[12] to inline functions.
Yoshihiro Takahashi [Thu, 31 Jan 2002 08:29:46 +0000 (08:29 +0000)]
Changed iskanji[12] to inline functions.

22 years agoMerged from sys/dev/sio/{sio.c,sio_isa.c} revisions 1.363 and 1.5,
Yoshihiro Takahashi [Thu, 31 Jan 2002 08:26:45 +0000 (08:26 +0000)]
Merged from sys/dev/sio/{sio.c,sio_isa.c} revisions 1.363 and 1.5,
respectively.

22 years agoAdd pc98 support.
Yoshihiro Takahashi [Thu, 31 Jan 2002 08:24:36 +0000 (08:24 +0000)]
Add pc98 support.

22 years agoRemove obsolete COMBRD* definitions.
Yoshihiro Takahashi [Thu, 31 Jan 2002 08:23:30 +0000 (08:23 +0000)]
Remove obsolete COMBRD* definitions.

22 years agoMFi386: revision 1.180
Yoshihiro Takahashi [Thu, 31 Jan 2002 04:28:52 +0000 (04:28 +0000)]
MFi386: revision 1.180

22 years agoMFi386: revision 1.52
Yoshihiro Takahashi [Thu, 31 Jan 2002 04:22:20 +0000 (04:22 +0000)]
MFi386: revision 1.52

22 years agoFix error handling.
Takanori Watanabe [Thu, 31 Jan 2002 01:23:22 +0000 (01:23 +0000)]
Fix error handling.

PR:30665
Submitted by:TOMITA Yoshinori <yoshint@flab.fujitsu.co.jp>

22 years agoIn revision 1.351, All files/directories in ${CHROOTDIR}/usr/ports/distfiles
Makoto Matsushita [Thu, 31 Jan 2002 00:38:09 +0000 (00:38 +0000)]
In revision 1.351, All files/directories in ${CHROOTDIR}/usr/ports/distfiles
are removed, because ports.tgz includes distfiles by accident.

However, removing files/directories is *not* mandatory.  The only
requirement is "ports.tgz doesn't include distfiles/* but distfiles
directory itself".  If this is correct, the right way is to set an
option to tar(1), not to include these files.

This commit is based on PR: 34194, but a little bit of modification
is made by me (original patch doesn't include 'distfiles' directory).

PR: 34194
Reviewed by: Adrian Steinmann <ast@marabu.ch> (PR submitter)
MFC after: 9 days

22 years agoAllow this to compile again
Julian Elischer [Wed, 30 Jan 2002 23:38:31 +0000 (23:38 +0000)]
Allow this to compile again
not tested under LINT by: msmith

22 years agoinclude sys/lock.h and sys/mutex.h to make compile.
Alfred Perlstein [Wed, 30 Jan 2002 23:28:25 +0000 (23:28 +0000)]
include sys/lock.h and sys/mutex.h to make compile.

Noticed by: Vincent Poy <vince@oahu.WURLDLINK.NET>

22 years agoo Reserve two interface ioctl values, 38, and 38, for SIOCGIFMAC and
Robert Watson [Wed, 30 Jan 2002 21:37:32 +0000 (21:37 +0000)]
o Reserve two interface ioctl values, 38, and 38, for SIOCGIFMAC and
  SIOCSIFMAC, which are for the Mandatory Access Control
  implementation.  This will prevent collisions on the p4 development
  branch.

Obtained from: TrustedBSD Project

22 years agoFix FreeBSD IDs.
David E. O'Brien [Wed, 30 Jan 2002 21:36:57 +0000 (21:36 +0000)]
Fix FreeBSD IDs.

22 years agofget_locked fixes.
Alfred Perlstein [Wed, 30 Jan 2002 19:53:57 +0000 (19:53 +0000)]
fget_locked fixes.
  sort declaration.
  give paramters names.
  remove bogus check for negative fd (u_int) cast does this for us.

Pointed out by: bde

22 years agoRemove unused variables in select(2) from previous delta.
Alfred Perlstein [Wed, 30 Jan 2002 19:48:25 +0000 (19:48 +0000)]
Remove unused variables in select(2) from previous delta.

Pointed out by: bde

22 years agoAdd pam_self(8) so users can login(1) as themselves without authentication,
Dag-Erling Smørgrav [Wed, 30 Jan 2002 19:13:23 +0000 (19:13 +0000)]
Add pam_self(8) so users can login(1) as themselves without authentication,
pam_login_access(8) and pam_securetty(8) to enforce various checks
previously done by login(1) but now handled by PAM, and pam_lastlog(8) to
record login sessions in utmp / wtmp / lastlog.

Sponsored by: DARPA, NAI Labs

22 years agoStill with asbestos longjohns on, completely PAMify login(1) and remove
Dag-Erling Smørgrav [Wed, 30 Jan 2002 19:10:21 +0000 (19:10 +0000)]
Still with asbestos longjohns on, completely PAMify login(1) and remove
code made redundant by various PAM modules (primarily pam_unix(8)).

Sponsored by: DARPA, NAI Labs

22 years agoWith asbestos longjohns on, integrate most of the checks normally done by
Dag-Erling Smørgrav [Wed, 30 Jan 2002 19:09:11 +0000 (19:09 +0000)]
With asbestos longjohns on, integrate most of the checks normally done by
login(1) (password & account expiry, hosts.access etc.) into pam_unix(8).

Sponsored by: DARPA, NAI Labs

22 years agoUse pam_self(8) to allow users to su(1) to themselves without authentication.
Dag-Erling Smørgrav [Wed, 30 Jan 2002 19:04:39 +0000 (19:04 +0000)]
Use pam_self(8) to allow users to su(1) to themselves without authentication.

Sponsored by: DARPA, NAI Labs

22 years agoMove the code from pam_sm_authenticate() to pam_sm_acct_mgmt(). Simplify
Dag-Erling Smørgrav [Wed, 30 Jan 2002 19:03:16 +0000 (19:03 +0000)]
Move the code from pam_sm_authenticate() to pam_sm_acct_mgmt().  Simplify
it a little and try to make it more resilient to various possible failure
conditions.  Change the man page accordingly, and take advantage of this
opportunity to simplify its language.

Sponsored by: DARPA, NAI Labs

22 years agoBacked out the main part of revs.1.14-16. Don't disable interrupts in
Bruce Evans [Wed, 30 Jan 2002 18:51:24 +0000 (18:51 +0000)]
Backed out the main part of revs.1.14-16.  Don't disable interrupts in
the packet transfer routines, since rev.1.468 of machdep.c does this
better.  I'm surprised that disabling interrupts helped much.  Disabling
them in the packet receive routine is too late.

Fixed some minor style bugs in rev.1.14.

22 years agoBacked out the last vestiges of rev.1.51. Don't enter a critical
Bruce Evans [Wed, 30 Jan 2002 18:23:31 +0000 (18:23 +0000)]
Backed out the last vestiges of rev.1.51.  Don't enter a critical
region in Debugger(), since rev.1.468 of machdep.c does this better.
Other cosmetic backouts.

22 years agoCleaned up the 0ldSiG magic check before removing it. Just use fuword()
Bruce Evans [Wed, 30 Jan 2002 17:47:12 +0000 (17:47 +0000)]
Cleaned up the 0ldSiG magic check before removing it.  Just use fuword()
to fetch the magic word instead of useracc() plus a direct access.
This is more efficient as well as simpler and less incorrect:
- it was inefficent because useracc() takes much longer than just
  accessing the data using a correct access method, at least on i386's.
- it was incorrect because direct access is incorrect unless the address
  has been mapped.  This and nearby direct accesses are mostly handled
  better for other arches because they have to be (direct accesses don't
  work).
- using magic in sigreturn is still fundamentally broken because false
  matches are possible.  On i386's, a false match occurs when %eip in a
  new signal context happens to equal the magic value.  This is not
  handled better for other arches.

22 years agoAdd support for different serial clock frequencies and not just the
John Hay [Wed, 30 Jan 2002 17:22:04 +0000 (17:22 +0000)]
Add support for different serial clock frequencies and not just the
standard one of 1.8432MHz. This will be used by the puc (PCI
"universal" communication card) device driver.

Reviewed by: bde

22 years agoRegenerate just this file after unbreaking makesyscalls.sh. The previous
Bruce Evans [Wed, 30 Jan 2002 15:14:24 +0000 (15:14 +0000)]
Regenerate just this file after unbreaking makesyscalls.sh.  The previous
commit broke the world in libc.

22 years agoOops, fix previous commit to not generate a C comment in syscall.mk.
Bruce Evans [Wed, 30 Jan 2002 15:12:12 +0000 (15:12 +0000)]
Oops, fix previous commit to not generate a C comment in syscall.mk.

22 years agoRemoved the one use of the mandatory keyword (for npx).
Bruce Evans [Wed, 30 Jan 2002 14:35:58 +0000 (14:35 +0000)]
Removed the one use of the mandatory keyword (for npx).

npx is no more mandatory than sc.  Its mandatoryness went away in
rev.1.226 of i386/machdep.c 9 months before it was made mandatory in
rev.1.24 of config/mkmakefile.c.

This change is mainly to test building of minimal kernel configurations.
npx should really be even more standard than clk.  It was optional mainly
so that the usual device driver configuration info could be specified in
the usual way in config files, but this hasn't been necessary for a few
years.

22 years agoUpdate some release dates.
Josef Karthauser [Wed, 30 Jan 2002 13:23:16 +0000 (13:23 +0000)]
Update some release dates.

PR: misc/34432

22 years agoDon't include <isa/isavar.h> or compile code depending on it when isa
Bruce Evans [Wed, 30 Jan 2002 12:41:12 +0000 (12:41 +0000)]
Don't include <isa/isavar.h> or compile code depending on it when isa
is not configured.  Including <isa/isavar.h> when it is not used is
harmful as well as bogus, since it includes "isa_if.h" which is not
generated when isa is not configured.

This was fixed in 1999 but was broken by unconditionalizing PNPBIOS.

22 years agoRemoved unused includes. In particular, don't include <isa/isavar.h> since
Bruce Evans [Wed, 30 Jan 2002 12:23:49 +0000 (12:23 +0000)]
Removed unused includes.  In particular, don't include <isa/isavar.h> since
its only effect is to break the optionality of the isa option.

Sorted includes.

22 years agoRegenerate _after_ the commit to syscalls.master.
Bruce Evans [Wed, 30 Jan 2002 10:29:12 +0000 (10:29 +0000)]
Regenerate _after_ the commit to syscalls.master.

22 years agoEscape $FreeBSD$ in a different way to avoid using the bogus escapes \$
Bruce Evans [Wed, 30 Jan 2002 10:22:05 +0000 (10:22 +0000)]
Escape $FreeBSD$ in a different way to avoid using the bogus escapes \$
and \F.  Awk just started warning about these.

22 years agoAdded this makefile. This is not attached to the build yet. I often
Bruce Evans [Wed, 30 Jan 2002 09:27:03 +0000 (09:27 +0000)]
Added this makefile.  This is not attached to the build yet.  I often
install parts of /etc manually and it helps to have a makefile for
each subdir even if the main makefile doesn't invoke it.

22 years agoStop saying that "express" mode is for impatient people. It's
Jordan K. Hubbard [Wed, 30 Jan 2002 01:34:16 +0000 (01:34 +0000)]
Stop saying that "express" mode is for impatient people.  It's
really for impatient and EXPERT people who know sysinstall
backwards and forwards.

MFC after: 1 week

22 years agoBack out rev 1.78, which is incorrect now that the PAM modules have been
Dag-Erling Smørgrav [Tue, 29 Jan 2002 23:27:54 +0000 (23:27 +0000)]
Back out rev 1.78, which is incorrect now that the PAM modules have been
fixed to accept a NULL PAM_RHOST.

22 years agoAttempt to fixup select(2) and poll(2), this should fix some races with
Alfred Perlstein [Tue, 29 Jan 2002 22:54:19 +0000 (22:54 +0000)]
Attempt to fixup select(2) and poll(2), this should fix some races with
other threads as well as speed up the interfaces.

To fix the race and accomplish the speedup, remove selholddrop and
pollholddrop.  The entire concept is somewhat bogus because holding
the individual struct file pointers offers us no guarantees that
another thread context won't close it on us thereby removing our
access to our own reference.

Selholddrop and pollholddrop also would do multiple locks and unlocks
of mutexes _per-file_ in the fd arrays to be scanned, this needed to
be sped up.

Instead of using selholddrop and pollholddrop, simply hold the
filedesc lock over the selscan and pollscan functions.  This should
protect us against close(2)'s on the files as reduce the multiple
lock/unlock pairs per fd into a single lock over the filedesc.

22 years agoFix a signal 11 error that occurs if you try to use the 'T' option on
Murray Stokely [Tue, 29 Jan 2002 22:35:40 +0000 (22:35 +0000)]
Fix a signal 11 error that occurs if you try to use the 'T' option on
an existing FreeBSD partition.

Reported by: Brent Cook <busterb@mail.utexas.edu>

22 years agoForced commit. Please ignore the pppd-related log entry for revision
Crist J. Clark [Tue, 29 Jan 2002 22:16:50 +0000 (22:16 +0000)]
Forced commit. Please ignore the pppd-related log entry for revision
1.81. It was the product of a botched cvs(1) command line (and was also
a forced, null-delta commit).

22 years agoThe huge dependency lists of some of our packages has brought
Murray Stokely [Tue, 29 Jan 2002 21:41:08 +0000 (21:41 +0000)]
The huge dependency lists of some of our packages has brought
attention to the sub-optimal way that we deal with package
dependencies.  Traditionally, for each package in an INDEX that the
user wants to add, we check all of the dependencies first even if the
package is already installed.  With some GNOME packages, this can
cause package_extract to be called for 50 different dependencies when
we know the top level package is already installed.

The new behavior is to not check dependencies for packages that are
already installed.  This fixes a bug where sysinstall gets itself into
a CPU intensive loop when trying to install sawfish gnome with the
most recent ports/INDEX.  There is a bug somewhere in the ports INDEX,
but with over 6,400 ports we need to be a little more forgiving here.

22 years agoBy commit of usr.sbin/pccard/pccardd/cardd.c at Nov 29 (Dec
MIHIRA Sanpei Yoshiro [Tue, 29 Jan 2002 21:17:05 +0000 (21:17 +0000)]
By commit of usr.sbin/pccard/pccardd/cardd.c at Nov 29 (Dec
10 in -STABLE), pccardd's string comparison between
pccard.conf's entry and PC card's CIS tupple became strict
matching.

As influences of this commit, some PC cards don't work since
some /etc/default/pccard.conf's card identifiers entries are
incorrectly described.

  - Lexar Media compact flash
  - IO DATA CBIDE2 in 16 bit mode
  - TOSHIBA Portable 24X Speed CD-ROM Drive PA2673UJ
  - Hewlett Packard M820e (CD-writer)

Update these card configs.

PR: 33815
Obtained from: [bsd-nomads:16128]

22 years ago*sigh* Misfire. Backout previous commit, as it was meant for
Jacques Vidrine [Tue, 29 Jan 2002 17:40:45 +0000 (17:40 +0000)]
*sigh* Misfire.  Backout previous commit, as it was meant for
another branch.

22 years agoBackout 1.120, EINVAL isn't a proper error return when the passed fd is
Alfred Perlstein [Tue, 29 Jan 2002 17:12:10 +0000 (17:12 +0000)]
Backout 1.120, EINVAL isn't a proper error return when the passed fd is
negative, the 'pointer' referred to by the manpage is actually the
struct file's f_offset field.

Pointed out by: bde

22 years agoFreeBSD 4.4 doesn't have __FBSDID. Do it the old-fashioned way instead.
Jacques Vidrine [Tue, 29 Jan 2002 17:11:28 +0000 (17:11 +0000)]
FreeBSD 4.4 doesn't have __FBSDID.  Do it the old-fashioned way instead.

Reported by: Steven Farmer <steve@geoenergycorp.com>
Pointy hat to: nectar

22 years agoMove cardaddr to memory structure.
Warner Losh [Tue, 29 Jan 2002 16:57:50 +0000 (16:57 +0000)]
Move cardaddr to memory structure.

Forgotten by: imp
Reminded by: Dave Cornejo

22 years agoTidy up gecos field for `bin'.
Ruslan Ermilov [Tue, 29 Jan 2002 14:00:03 +0000 (14:00 +0000)]
Tidy up gecos field for `bin'.

22 years ago(null delta)
Sheldon Hearn [Tue, 29 Jan 2002 13:01:26 +0000 (13:01 +0000)]
(null delta)

The previous delta uncommented kerberos-adm, not kserver-adm.

22 years agoUncomment kserver-adm, which is IANA-sanctioned and has no apparent
Sheldon Hearn [Tue, 29 Jan 2002 12:28:51 +0000 (12:28 +0000)]
Uncomment kserver-adm, which is IANA-sanctioned and has no apparent
conflicts.

PR: conf/34316
Submitted by: Sean Chittenden <sean@chittenden.org>
MFC after: 2 weeks

22 years agoCorrect function's description.
Chris D. Faulhaber [Tue, 29 Jan 2002 12:18:45 +0000 (12:18 +0000)]
Correct function's description.

Obtained from: TrustedBSD Project

22 years agoAdd EDIMAX ethernet card for NEWCARD from NetBSD
Warner Losh [Tue, 29 Jan 2002 07:08:56 +0000 (07:08 +0000)]
Add EDIMAX ethernet card for NEWCARD from NetBSD

22 years agoModule for exca. Eventually, this will be shared between pcic and pccbb.
Warner Losh [Tue, 29 Jan 2002 06:53:32 +0000 (06:53 +0000)]
Module for exca.  Eventually, this will be shared between pcic and pccbb.

22 years agopccbb needs exca now.
Warner Losh [Tue, 29 Jan 2002 06:51:09 +0000 (06:51 +0000)]
pccbb needs exca now.

22 years agoMigrate towards using the new exca module for 16bit stuff. This is a
Warner Losh [Tue, 29 Jan 2002 06:50:18 +0000 (06:50 +0000)]
Migrate towards using the new exca module for 16bit stuff.  This is a
WIP, but works for me.

Also do some minor code factoring and code cleanup while I'm here.

22 years agoFirst draft of common code between pccbb and pcic drivers for talking
Warner Losh [Tue, 29 Jan 2002 06:48:38 +0000 (06:48 +0000)]
First draft of common code between pccbb and pcic drivers for talking
to ExCA register sets.  These registers exist in both ISA and PCI
devices in a couple different ways, and this will provide a common
base for future building.  This code is a rehash of the pccbb 16-bit
code, which was a rehash of the pcic code, which was a rehash of the
netbsd i82365 code.  More hashing to come.

22 years agoMinor sorting orders
Warner Losh [Tue, 29 Jan 2002 06:15:07 +0000 (06:15 +0000)]
Minor sorting orders

22 years agoSync to 1.27 of pccarddevs
Warner Losh [Tue, 29 Jan 2002 06:10:15 +0000 (06:10 +0000)]
Sync to 1.27 of pccarddevs

22 years agoMFNetBSD:
Warner Losh [Tue, 29 Jan 2002 06:09:39 +0000 (06:09 +0000)]
MFNetBSD:
1.159: sorting (not relevant to us, since I'd done thie before)
1.158: joda; Ericsson PRISM2 WIRELEASS LAN CARD.
1.157: aymeric Edimax Technology ethernet card

22 years agoHave SIOCGIFCONF return all (if any) AF_INET addresses for the
Marcel Moolenaar [Tue, 29 Jan 2002 06:00:11 +0000 (06:00 +0000)]
Have SIOCGIFCONF return all (if any) AF_INET addresses for the
interfaces we encounter. In Linux, all addresses are returned for
which gifconf handlers are installed. This boils down to AF_DECnet
and AF_INET. We care mostly about AF_INET for now. Adding additional
families is simple enough.

Returning the addresses is important for RPC clients to function
properly. Andrew found in some reference code that the logic that
handles the retransmission looks for an interface that's up and has
an AF_INET address. This obviously failed as we didn't return any
addresses at all.

Note also that with this change we don't return interfaces that don't
have AF_INET addresses, whereas before we returned any interface
present in the system. This is in line with what Linux does (modulo
interfaces with only AF_DECnet addresses of course :-)

Reported by: "Andrew Atrens" <atrens@nortelnetworks.com>
MFC after: 1 week

22 years agoAdd Linksys Instant Wireless WPC11 v2.5
Warner Losh [Tue, 29 Jan 2002 05:15:56 +0000 (05:15 +0000)]
Add Linksys Instant Wireless WPC11 v2.5

Submitted by: eliedtke@apogeetelecom.com

22 years agoGrr. The last commit contained update conflicts. It just goes to
Greg Lehey [Tue, 29 Jan 2002 02:49:15 +0000 (02:49 +0000)]
Grr.  The last commit contained update conflicts.  It just goes to
show that you should check the code even for a forced commit.

Spotted by: dd

22 years agoForced commit. The previous log message was garbled and should have read:
Greg Lehey [Tue, 29 Jan 2002 02:17:42 +0000 (02:17 +0000)]
Forced commit.  The previous log message was garbled and should have read:

vinumioctl: Fix a double fault caused by calling setjmp() without
            holding the config lock.  The call was in fact
            superfluous.

Dump analysis by:  Thomas Moestl <tmoestl@gmx.net>
Forced commit requested by:  obrien, kris

22 years agoForced commit. The previous commit log was garbled and didn't even
Greg Lehey [Tue, 29 Jan 2002 02:14:51 +0000 (02:14 +0000)]
Forced commit.  The previous commit log was garbled and didn't even
apply to this file.  The correct message is:

throw_rude_remark: Make sure we're holding the config lock before
                   proceeding.  There's no reason to assume that this
                   has ever happened, but the alternative might be a
                   double fault.

22 years agoBump __FreeBSD_version for sendfile fix to properly send 'nbytes' of
Alfred Perlstein [Tue, 29 Jan 2002 01:52:30 +0000 (01:52 +0000)]
Bump __FreeBSD_version for sendfile fix to properly send 'nbytes' of
file when headers are supplied (version 1.104 of src/sys/kern/uipc_syscalls.c)

22 years agoPut a complete set of pppd(8) sample configuration files in
Crist J. Clark [Tue, 29 Jan 2002 01:10:47 +0000 (01:10 +0000)]
Put a complete set of pppd(8) sample configuration files in
/usr/share/examples/pppd.

Remove the out-of-place pppd(8) configuration files in etc/ppp,
ppp.shells.sample and ppp.deny.

Make the appropriate changes to the build process, etc/Makefile and
etc/mtree/BSD.usr.mtree, so it all works.