]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoFix a bug that prevented the restoration of hard links to files that
John Polstra [Thu, 9 Jul 1998 03:57:26 +0000 (03:57 +0000)]
Fix a bug that prevented the restoration of hard links to files that
had the schg flag set.  Reported by Matthew Thyer <thyerm@camtech.net.au>.

26 years agoSlight optimisation. Don't do the same calculation twice.
Julian Elischer [Thu, 9 Jul 1998 02:00:02 +0000 (02:00 +0000)]
Slight optimisation. Don't do the same calculation twice.

26 years agoDon't update superblock if mounted readonly,
Julian Elischer [Wed, 8 Jul 1998 23:52:27 +0000 (23:52 +0000)]
Don't update superblock if mounted readonly,
also fixes some problems with softupdates on root.
More cleanups are needed here..
Submitted by: Luoqi Chen <luoqi@watermarkgroup.com>

26 years agonatd_interface can now be specified as either an IP address or
Jacques Vidrine [Wed, 8 Jul 1998 23:31:28 +0000 (23:31 +0000)]
natd_interface can now be specified as either an IP address or
an interface name, so describe that here.

26 years agoDetect user id 0 as a number instead of a string. String comparisons
Alexander Langer [Wed, 8 Jul 1998 22:42:08 +0000 (22:42 +0000)]
Detect user id 0 as a number instead of a string.  String comparisons
fail to detect 00.

PR: 7218
Submitted by: Michal Listos <mcl@Amnesiac.123.org>
Niall Smart <rotel@indigo.ie>

26 years agoOops, I broke netboot.
KATO Takenori [Wed, 8 Jul 1998 16:01:34 +0000 (16:01 +0000)]
Oops, I broke netboot.

Submitted by: Takahashi Yoshihiro <nyan@wyvern.cc.kogakuin.ac.jp>

26 years agoAdd lockf().
Dmitrij Tejblum [Wed, 8 Jul 1998 15:42:57 +0000 (15:42 +0000)]
Add lockf().

26 years agoAllow either an IP address or an interface to be specified in
Jacques Vidrine [Wed, 8 Jul 1998 15:40:53 +0000 (15:40 +0000)]
Allow either an IP address or an interface to be specified in
the rc.conf variable ``natd_interface''.  rc.network will
determine whether it is an IP address or an interface name,
and invoke natd with the -a or -n flag as appropriate.

PR: 6947
Reviewed by: jkh@FreeBSD.ORG

26 years agoMake this look more FreeBSDish.
Dmitrij Tejblum [Wed, 8 Jul 1998 15:38:06 +0000 (15:38 +0000)]
Make this look more FreeBSDish.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r37510,
Dmitrij Tejblum [Wed, 8 Jul 1998 15:28:56 +0000 (15:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r37510,
which included commits to RCS files with non-trunk default branches.

26 years agoAn implementation of lockf() function call.
Dmitrij Tejblum [Wed, 8 Jul 1998 15:28:56 +0000 (15:28 +0000)]
An implementation of lockf() function call.

Obtained from: NetBSD (author: Klaus Klein)

26 years agoDeclare lockf().
Dmitrij Tejblum [Wed, 8 Jul 1998 15:14:48 +0000 (15:14 +0000)]
Declare lockf().

26 years agoAdd XPG4.2.
Dmitrij Tejblum [Wed, 8 Jul 1998 15:05:05 +0000 (15:05 +0000)]
Add XPG4.2.

26 years agoUse not-so-new printf formats %r and/or %z instead of %n and/or %+x.
Bruce Evans [Wed, 8 Jul 1998 10:53:58 +0000 (10:53 +0000)]
Use not-so-new printf formats %r and/or %z instead of %n and/or %+x.

26 years agoFixed (un)sign extension bugs in %+n format. -4 became
Bruce Evans [Wed, 8 Jul 1998 10:41:32 +0000 (10:41 +0000)]
Fixed (un)sign extension bugs in %+n format.  -4 became
(long)(u_long)(u_int)-4 = 0x00000000fffffffc on machines with 32-bit
ints and 64-bit longs.

Restored %z format for printing signed hex.  %+x shouldn't have been
used since it is an error in userland.

Prepared to nuke %n format by cloning it to %r.  %n shouldn't have
been used because it means something completely different in
userland.  Now %+r is equivalent to ddb's original %r, and %r is
equivalent to ddb's original %n.

Ignore '+' flag in combination with unsigned formats %{o,p,u,x}.

26 years agoFixed bogus type of valuep in struct db_variable. It was `int *' and
Bruce Evans [Wed, 8 Jul 1998 09:11:43 +0000 (09:11 +0000)]
Fixed bogus type of valuep in struct db_variable.  It was `int *' and
became `long *' for alpha, but should always have been `db_expr_t *'.
Fixed variable types to match.

26 years agoSync with sys/i386/isa/syscons.c revision 1.265.
KATO Takenori [Wed, 8 Jul 1998 08:54:38 +0000 (08:54 +0000)]
Sync with sys/i386/isa/syscons.c revision 1.265.

26 years agoSync with sys/i386/boot/netboot/Makefile revision 1.19.
KATO Takenori [Wed, 8 Jul 1998 08:53:43 +0000 (08:53 +0000)]
Sync with sys/i386/boot/netboot/Makefile revision 1.19.

26 years agoWhen not acting as a router (ipforwarding=0), silently discard source
David Greenman [Wed, 8 Jul 1998 08:49:51 +0000 (08:49 +0000)]
When not acting as a router (ipforwarding=0), silently discard source
routed packets that aren't destined for us, as required by RFC-1122.
PR: 7191

26 years agoFixed db_printf format errors.
Bruce Evans [Wed, 8 Jul 1998 06:43:57 +0000 (06:43 +0000)]
Fixed db_printf format errors.

26 years agoAdd support for run-time configuration of core file names. In a nutshell,
Sean Eric Fagan [Wed, 8 Jul 1998 06:38:39 +0000 (06:38 +0000)]
Add support for run-time configuration of core file names.  In a nutshell,
you can specify the corefile name by using:

sysctl -w kern.corefile="format"

where format is a pathname (relative or absolute -- default is "%N.core"),
with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats.

Reviewed by: Mike Smith, with strong requests by Julian :)

26 years agoFixed db_printf format errors (except for ones using broken extensions
Bruce Evans [Wed, 8 Jul 1998 06:27:22 +0000 (06:27 +0000)]
Fixed db_printf format errors (except for ones using broken extensions
(nonstandard %n and '+' with %x), and ones not found by -Wformat on
386's (some db_expr_t's are still printed as ints).

I decided not to change the arg type for %n from [unsigned] int to
register_t, since about half of the uses of %n are to print plain
ints and casting to [unsigned] long for %n is no harder than for %x.

26 years agoAdd a "-k" option, to specify that the lock file should be kept
John Polstra [Wed, 8 Jul 1998 05:29:05 +0000 (05:29 +0000)]
Add a "-k" option, to specify that the lock file should be kept
rather than removed.

Submitted by: Nick Barnes <Nick.Barnes@pobox.com>

26 years agoMerge changes from vendor branch.
Peter Wemm [Wed, 8 Jul 1998 01:24:37 +0000 (01:24 +0000)]
Merge changes from vendor branch.

26 years agoImport NetBSD-current version of if_de driver onto vendor branch.
Peter Wemm [Wed, 8 Jul 1998 01:16:05 +0000 (01:16 +0000)]
Import NetBSD-current version of if_de driver onto vendor branch.

26 years agoCatch a few corner cases where FreeBSD differs enough from BSD 4.4 to
Julian Elischer [Wed, 8 Jul 1998 01:04:33 +0000 (01:04 +0000)]
Catch a few corner cases where FreeBSD differs enough from BSD 4.4 to
confuse Soft updates..
Should solve several "dangling deps" panics.

26 years agoNew flag for internal stdio use to enable dynamic string allocation for
Peter Wemm [Wed, 8 Jul 1998 00:52:40 +0000 (00:52 +0000)]
New flag for internal stdio use to enable dynamic string allocation for
asprintf() implementation.

Obtained from: Todd Miller <Todd.Miller@courtesan.com> via OpenBSD

26 years agoUse a different hack for building libgcc2: `XCC= ${CC}' instead of
Bruce Evans [Wed, 8 Jul 1998 00:45:50 +0000 (00:45 +0000)]
Use a different hack for building libgcc2: `XCC= ${CC}' instead of
`XCC= <relative cc> -B<path to relative cc1> ...'.  This is equivalent
when cc and cc1, etc. have just been bootstrapped by `make world'.
The relative versions normally won't work if the target system is
not binary compatible.  Bootstrapping different versions of gcc
without going through `make world' is slightly more broken than
before.

Uniformized macro names (P1OBJS -> LIB1POBJS, etc.).

Don't give full paths to sources.

26 years agoReplace my original asprintf() and vasprintf() hacks with something
Peter Wemm [Wed, 8 Jul 1998 00:44:56 +0000 (00:44 +0000)]
Replace my original asprintf() and vasprintf() hacks with something
more cleanly integrated with stdio.  This should be faster and cleaner
since it doesn't memcpy() the data into a seperate buffer.  This lets
stdio allocate and manage the buffer and then hand it over to the user.

Obtained from: Todd Miller <Todd.Miller@courtesan.com> via OpenBSD

26 years agoUse the installed versions of cc, cc1 and cpp for building underscore.
Bruce Evans [Wed, 8 Jul 1998 00:14:15 +0000 (00:14 +0000)]
Use the installed versions of cc, cc1 and cpp for building underscore.
They have been bootstrapped by `make world' since long before the
hacks here were cloned from ../libgcc/Makefile.  The versions just
built in "../*" normally won't work if the target system is not binary
compatible.

Don't use OBJS to defeat `make depend'; just put generated sources in
SRCS.

Added temporary files to CLEANFILES.

26 years agoUse the installed version of perl for `make install'. It is now
Bruce Evans [Tue, 7 Jul 1998 23:44:43 +0000 (23:44 +0000)]
Use the installed version of perl for `make install'.  It is now
bootstrapped by `make world'.  The version just built in ".."
normally won't work if the target system is not binary compatible.

26 years agoAdd definitions for PT_LOPROC and PT_HIPROC.
John Polstra [Tue, 7 Jul 1998 23:32:57 +0000 (23:32 +0000)]
Add definitions for PT_LOPROC and PT_HIPROC.

Submitted by: Kapil Chowksey <kchowksey@hss.hns.com>

26 years agoUse the installed version of perl for `make depend'. It is now
Bruce Evans [Tue, 7 Jul 1998 23:32:39 +0000 (23:32 +0000)]
Use the installed version of perl for `make depend'.  It is now
bootstrapped by `make world'.  The version just built in ".."
normally won't work if the target system is not binary compatible.
The bootstrapped version has a better chance of working.

This makes the fixes and bugs in the previous 3 commits irrelevant.
Rev.1.11 was just wrong and rev.1.10 became unnecessary when
perl/perl was added to build-tools.  Don't expect to build perl/usub
without using `make world' or equivalent if you don't have perl
installed.

26 years agoUse the installed version of mklocale for `make all'. It is now
Bruce Evans [Tue, 7 Jul 1998 23:05:59 +0000 (23:05 +0000)]
Use the installed version of mklocale for `make all'.  It is now
bootstrapped by `make world'.  The version just built in ".."
normally won't work if the target system is not binary compatible.

Don't build or install anything if _BUILD_TOOLS is defined.  Then
we only want to build and install the mklocale binary, but the layout
of the mklocale tree forces recursing to mklocale/data for at least
the obj target even when _BUILD_TOOLS is defined.

26 years agoUse the installed version of colldef for `make all'. It is now
Bruce Evans [Tue, 7 Jul 1998 23:04:25 +0000 (23:04 +0000)]
Use the installed version of colldef for `make all'.  It is now
bootstrapped by `make world'.  The version just built in ".."
normally won't work if the target system is not binary compatible.

Don't build or install anything if _BUILD_TOOLS is defined.  Then
we only want to build and install the colldef binary, but the layout
of the colldef tree forces recursing to colldef/data for at least
the obj target even when _BUILD_TOOLS is defined.

26 years agoPrevious fix didn't make it past the Bruce filter. Clean up
Jordan K. Hubbard [Tue, 7 Jul 1998 22:20:50 +0000 (22:20 +0000)]
Previous fix didn't make it past the Bruce filter.  Clean up
non-KNF indentation and use err() properly.

26 years ago msgs -p (and maybe other flags, too) gives misleading error messages,
Jordan K. Hubbard [Tue, 7 Jul 1998 12:02:59 +0000 (12:02 +0000)]
    msgs -p (and maybe other flags, too) gives misleading error messages,
        especially on a new install, where /var/msgs/bounds doesn't exist.  I
        moved my bounds file out of the way to create this before and after
        on a quick 'n' dirty hack, which is probably the 23rd best way to do it,
        but it works:

PR: 6963
Submitted by: Matthew Fuller <fullermd@mortis.futuresouth.com>

26 years agoOops, don't build tools for building games, etc. when we're not building
Bruce Evans [Tue, 7 Jul 1998 09:59:48 +0000 (09:59 +0000)]
Oops, don't build tools for building games, etc. when we're not building
games, etc.

Define _BUILD_TOOLS in sub-makes for building tools.  This will be used
to avoid using uninstalled tools in colldef and mklocale.

26 years agoUse the installed versions of caesar and strfile. They are now
Bruce Evans [Tue, 7 Jul 1998 06:22:23 +0000 (06:22 +0000)]
Use the installed versions of caesar and strfile.  They are now
bootstrapped by `make world', and the installed versions are just
as likely to work as the installed version of cc if `make' is run
directly.

26 years agoBuild internal tools in build-tools so that they have some chance of
Bruce Evans [Tue, 7 Jul 1998 05:37:34 +0000 (05:37 +0000)]
Build internal tools in build-tools so that they have some chance of
working when the target system is not binary compatible.  Use various
hacks to work around minor problems in the source and binary tree
layouts:
- caesar and strfile are built normally (the source layout is good),
  then installed by copying them to ${WORLDTMP}/usr/bin (they are
  installed in ${WORLDTMP}/usr/games, but I don't want to put that
  in $PATH).
- colldef and mklocale are built and installed normally.  Messy and
  incomplete relative path searches for them and caesar and strfile
  can now go away.
- internal tools that aren't installed are now built and left lying
  around for the `make all' pass to use.  If the target system is
  not binary compatible, it is critical that these tools don't get
  rebuilt.  Cleaning of the obj tree before building the internal
  tools should ensure this.
- most internal tools are built using internal build-tools targets,
  but tn3270 is simpler for a change - it has all the tools in a
  separate tree, so they can be built using `make all'.

26 years agoChanged `#if defined(i386)' to `#ifdef __i386__'.
Bruce Evans [Tue, 7 Jul 1998 05:00:09 +0000 (05:00 +0000)]
Changed `#if defined(i386)' to `#ifdef __i386__'.

`#if defined(ONE_THING)' is a style bug, and i386 instead of __i386__
is a bug, since i386 is never defined when the kernel is compiled
by with the default flags (`gcc -ansi ...').  Here the bug disabled
the call to pmap_setvidram(), so ISA video memory was not mapped
WC on 686's.  The bug may have been masked by bugs in the committer's
version of gcc - `gcc -ansi' incorrectly defines i386 for gcc = the
version of egcs on the 2.2.6 cdrom.

26 years agoDon't depend on gcc's feature of casting lvalues.
Bruce Evans [Tue, 7 Jul 1998 04:36:23 +0000 (04:36 +0000)]
Don't depend on gcc's feature of casting lvalues.

26 years agoDon't assume that longs are 32 bits in struct dos_partition.
Bruce Evans [Tue, 7 Jul 1998 04:19:25 +0000 (04:19 +0000)]
Don't assume that longs are 32 bits in struct dos_partition.

26 years agoQuick fix for type mismatches which were fatal if longs aren't 32
Bruce Evans [Tue, 7 Jul 1998 04:08:44 +0000 (04:08 +0000)]
Quick fix for type mismatches which were fatal if longs aren't 32
bits.  We used a private, wrong, version of `struct dirent' to help
break getdirentries(), and we use a silly check that the size of this
struct is a power of 2 to help break mount() if getdirentries() would
not work.  This fix just changes the struct to match `struct dirent'
(except for the name length).

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 02:43:26 +0000 (02:43 +0000)]
Added a `build-tools' target for internal tools.

Honor LDFLAGS for building internal tools.

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 02:30:45 +0000 (02:30 +0000)]
Added a `build-tools' target for internal tools.

Honor LDFLAGS for building internal tools.

26 years agoFinished previous fix - don't forget to add one dummy options header
Bruce Evans [Tue, 7 Jul 1998 02:27:09 +0000 (02:27 +0000)]
Finished previous fix - don't forget to add one dummy options header
to CLEANFILES.

Fixed lots of style bugs.

26 years agoAdded a `build-tools' target for internal tools. (linux_genassym may
Bruce Evans [Tue, 7 Jul 1998 02:04:20 +0000 (02:04 +0000)]
Added a `build-tools' target for internal tools.  (linux_genassym may
not actually work for cross compiling, but that is another problem.)

Honor LDFLAGS for building internal tools.  (Tools should normally
be built static to avoid problems with picking up target shared
libraries.  bsd.kmod doesn't set -static yet, and has some problems
with `LDFLAGS=-static ...' in the environment.)

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 01:51:18 +0000 (01:51 +0000)]
Added a `build-tools' target for internal tools.

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 01:43:54 +0000 (01:43 +0000)]
Added a `build-tools' target for internal tools.

Honor LDFLAGS for building internal tools.

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 01:41:58 +0000 (01:41 +0000)]
Added a `build-tools' target for internal tools.

Honor LDFLAGS for building internal tools.

Always build intermediate object files explicitly so that binaries don't
change when they are rebuilt.

Fixed some style bugs.

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 01:25:37 +0000 (01:25 +0000)]
Added a `build-tools' target for internal tools.

Restored a variant of explicit rule for `setup', with modifications to
always build setup.o so that `setup' doesn't change every time it is
rebuilt because it has a temporary file name in it.

26 years agoAdded a `build-tools' target for internal tools.
Bruce Evans [Tue, 7 Jul 1998 01:06:58 +0000 (01:06 +0000)]
Added a `build-tools' target for internal tools.

Removed explicit dependencies of foo.o on foo.c.  These were mainly
placeholders for comments about missing dependencies of tools objects
on headers.  This problem needs to be handled more generally.

26 years agoFixed type mismatches which were fatal when sizeof(long) > sizeof(int).
Bruce Evans [Mon, 6 Jul 1998 22:08:00 +0000 (22:08 +0000)]
Fixed type mismatches which were fatal when sizeof(long) > sizeof(int).

26 years agoFixed classic sign extension bug `-(long)sizeof(foo)'.
Bruce Evans [Mon, 6 Jul 1998 21:45:12 +0000 (21:45 +0000)]
Fixed classic sign extension bug `-(long)sizeof(foo)'.

Fixed the type of the string table size variable to match its use
(assuming that int32_t is 4 bytes and other unportable things).
bfd uses `unsigned char string_chars[BYTES_IN_WORD]', where
BYTES_IN_WORD can be 4, 8, or perhaps even 2 or 3, but it is
assumed to be precisely 4 bytes here.

Fixed printf format errors (don't assume that n_value in struct
n_list has type u_long, since it should have size BYTES_IN_WORD
and longs may be longer than words).

26 years agoFixed printf format errors.
Bruce Evans [Mon, 6 Jul 1998 21:01:54 +0000 (21:01 +0000)]
Fixed printf format errors.

26 years agoFixed type mismatches and style bugs in rev.1.30.
Bruce Evans [Mon, 6 Jul 1998 20:52:08 +0000 (20:52 +0000)]
Fixed type mismatches and style bugs in rev.1.30.

26 years agoDeclared version_string() properly (1970's style).
Bruce Evans [Mon, 6 Jul 1998 20:40:57 +0000 (20:40 +0000)]
Declared version_string() properly (1970's style).

26 years agoFixed printf format errors.
Bruce Evans [Mon, 6 Jul 1998 20:28:08 +0000 (20:28 +0000)]
Fixed printf format errors.

26 years agoRemoved bogus #ifdef INET - <net/if_slvar.h> no longer depends on it.
Bruce Evans [Mon, 6 Jul 1998 20:09:36 +0000 (20:09 +0000)]
Removed bogus #ifdef INET - <net/if_slvar.h> no longer depends on it.

26 years agoActivate newfs_msdos.
Robert Nordier [Mon, 6 Jul 1998 20:05:41 +0000 (20:05 +0000)]
Activate newfs_msdos.

26 years agoAdd newfs_msdos: extensive newfs-style support for creating FAT12,
Robert Nordier [Mon, 6 Jul 1998 20:01:34 +0000 (20:01 +0000)]
Add newfs_msdos: extensive newfs-style support for creating FAT12,
FAT16, and FAT32 file systems in all their various horriblenesses.

Approved-in-concept by: joerg

26 years agoioctl() request args are unsigned longs, so don't attempt to store
Bruce Evans [Mon, 6 Jul 1998 19:54:39 +0000 (19:54 +0000)]
ioctl() request args are unsigned longs, so don't attempt to store
them as ints.  Among other bugs, doing so at best caused benign
overflow followed by fatal sign extension on machines with 32-bit
ints and 64-bit longs.

26 years agoIntroduce (fairly hacky) workaround for odd TCP behavior with application
Bill Fenner [Mon, 6 Jul 1998 19:27:14 +0000 (19:27 +0000)]
Introduce (fairly hacky) workaround for odd TCP behavior with application
 writes of size (100,208]+N*MCLBYTES.

The bug:
 sosend() hands each mbuf off to the protocol output routine as soon as it
 has copied it, in the hopes of increasing parallelism (see
  http://www.kohala.com/~rstevens/vanj.88jul20.txt ). This works well for
 TCP as long as the first mbuf handed off is at least the MSS.  However,
 when doing small writes (between MHLEN and MINCLSIZE), the transaction is
 split into 2 small MBUF's and each is individually handed off to TCP.
 TCP assumes that the first small mbuf is the whole transaction, so sends
 a small packet.  When the second small mbuf arrives, Nagle prevents TCP
 from sending it so it must wait for a (potentially delayed) ACK.  This
 sends throughput down the toilet.

The workaround:
 Set the "atomic" flag when we're doing small writes.  The "atomic" flag
 has two meanings:
 1. Copy all of the data into a chain of mbufs before handing off to the
    protocol.
 2. Leave room for a datagram header in said mbuf chain.
 TCP wants the first but doesn't want the second.  However, the second
 simply results in some memory wastage (but is why the workaround is a
 hack and not a fix).

The real fix:
 The real fix for this problem is to introduce something like a "requested
 transfer size" variable in the socket->protocol interface.  sosend()
 would then accumulate an mbuf chain until it exceeded the "requested
 transfer size".  TCP could set it to the TCP MSS (note that the
 current interface causes strange TCP behaviors when the MSS > MCLBYTES;
 nobody notices because MCLBYTES > ethernet's MTU).

26 years agoRestored rev.1.11, which I somehow clobbered in rev.1.12.
Bruce Evans [Mon, 6 Jul 1998 19:11:35 +0000 (19:11 +0000)]
Restored rev.1.11, which I somehow clobbered in rev.1.12.

26 years agoAttempt to load serial eeprom contents in both 93c46 and 93c56/66 mode before
Justin T. Gibbs [Mon, 6 Jul 1998 18:38:57 +0000 (18:38 +0000)]
Attempt to load serial eeprom contents in both 93c46 and 93c56/66 mode before
giving up.

PR: 6966

26 years ago"vender" -> "vendor"
Eivind Eklund [Mon, 6 Jul 1998 16:10:06 +0000 (16:10 +0000)]
"vender" -> "vendor"

Requested by: Aage Røbekk <aagero@aage.priv.no>

26 years agoShorten line.
Eivind Eklund [Mon, 6 Jul 1998 16:07:43 +0000 (16:07 +0000)]
Shorten line.

Prodded (long time ago) by: bde

26 years agoSync with sys/i386/isa/fd.c and wd.c revisions 1.115 and 1.170,
KATO Takenori [Mon, 6 Jul 1998 10:09:42 +0000 (10:09 +0000)]
Sync with sys/i386/isa/fd.c and wd.c revisions 1.115 and 1.170,
respectively.

26 years agooops ended comment before the comment ended..
Julian Elischer [Mon, 6 Jul 1998 09:10:56 +0000 (09:10 +0000)]
oops ended comment before the comment ended..

26 years agoBring back some slight cleanups from 2.2
Julian Elischer [Mon, 6 Jul 1998 09:06:58 +0000 (09:06 +0000)]
Bring back some slight cleanups from 2.2

26 years agoCorrect use of .Nm. Spelling. Add rcsid and remove unused #includes.
Philippe Charnier [Mon, 6 Jul 1998 07:19:27 +0000 (07:19 +0000)]
Correct use of .Nm. Spelling. Add rcsid and remove unused #includes.

26 years ago-Wall.
Philippe Charnier [Mon, 6 Jul 1998 07:17:27 +0000 (07:17 +0000)]
-Wall.

26 years agoSync usage string and man page. Correct use of .Nm. Spelling. Remove
Philippe Charnier [Mon, 6 Jul 1998 07:15:53 +0000 (07:15 +0000)]
Sync usage string and man page. Correct use of .Nm. Spelling. Remove
unused #includes.

26 years agoCorrect use of .Nm.
Philippe Charnier [Mon, 6 Jul 1998 07:13:31 +0000 (07:13 +0000)]
Correct use of .Nm.

26 years agoRemove unused #includes. Spelling. Add rcsid. Do not dot terminate err()
Philippe Charnier [Mon, 6 Jul 1998 07:12:38 +0000 (07:12 +0000)]
Remove unused #includes. Spelling. Add rcsid. Do not dot terminate err()
strings.

26 years agoAdd rcsid. Remove unused #includes. Spelling.
Philippe Charnier [Mon, 6 Jul 1998 07:09:13 +0000 (07:09 +0000)]
Add rcsid. Remove unused #includes. Spelling.

26 years agoAdd rcsid, remove unused #includes, spelling.
Philippe Charnier [Mon, 6 Jul 1998 07:07:41 +0000 (07:07 +0000)]
Add rcsid, remove unused #includes, spelling.

26 years agoRemove unused #include. Add usage().
Philippe Charnier [Mon, 6 Jul 1998 07:06:15 +0000 (07:06 +0000)]
Remove unused #include. Add usage().

26 years agoCorrect use of .Nm. Spelling. Add rcsid, remove unused #includes, use err(3).
Philippe Charnier [Mon, 6 Jul 1998 07:04:50 +0000 (07:04 +0000)]
Correct use of .Nm. Spelling. Add rcsid, remove unused #includes, use err(3).

26 years agoAdd rcsid. remove unused #includes. Getopt returns -1 not EOF. Add usage().
Philippe Charnier [Mon, 6 Jul 1998 07:02:26 +0000 (07:02 +0000)]
Add rcsid. remove unused #includes. Getopt returns -1 not EOF. Add usage().
Cosmetics in err(3) string.

26 years agoLowercase 1st char. in err(3) strings.
Philippe Charnier [Mon, 6 Jul 1998 06:59:52 +0000 (06:59 +0000)]
Lowercase 1st char. in err(3) strings.

26 years agoremoce unused #include.
Philippe Charnier [Mon, 6 Jul 1998 06:58:32 +0000 (06:58 +0000)]
remoce unused #include.

26 years agoCorrect .Nm use. Add rcsid. Use min for minutes instead of mn.
Philippe Charnier [Mon, 6 Jul 1998 06:56:08 +0000 (06:56 +0000)]
Correct .Nm use. Add rcsid. Use min for minutes instead of mn.

26 years agoUpdate man page to allow adding address_family when -l is used (this is a no op
Philippe Charnier [Mon, 6 Jul 1998 06:53:43 +0000 (06:53 +0000)]
Update man page to allow adding address_family when -l is used (this is a no op
for now). Correct use of .Nm. Short usage string (see man page for full list).
Spelling.  Use err(3).

26 years agoCorrect use of .Nm, use .Bx Free for FreeBSD. Add rcsid, remove unused
Philippe Charnier [Mon, 6 Jul 1998 06:44:36 +0000 (06:44 +0000)]
Correct use of .Nm, use .Bx Free for FreeBSD. Add rcsid, remove unused
#includes. Spelling. Use err(3) and add usage().

26 years agoAdd the ability to suspend as well as hibernate to the system. This
Warner Losh [Mon, 6 Jul 1998 06:29:07 +0000 (06:29 +0000)]
Add the ability to suspend as well as hibernate to the system.  This
is the kernel part of my commits, the userlevel stuff will be done in
a separate commit.  Add the ability to suspend as well as hibernate to
syscons.  Create a new virtual key like hibernate for suspend.  Update
apm_bios.h to define more apm bios goodies.

26 years agoDon't expect the new code to be used without the right option file being
Julian Elischer [Mon, 6 Jul 1998 05:04:33 +0000 (05:04 +0000)]
Don't expect the new code to be used without the right option file being
included.

26 years agoFix braino in switching to TAILQ macro.
Julian Elischer [Mon, 6 Jul 1998 05:00:53 +0000 (05:00 +0000)]
Fix braino in switching to TAILQ macro.

26 years agoForgot to register the IPFIREWALL_FORWARD option.
Julian Elischer [Mon, 6 Jul 1998 04:55:22 +0000 (04:55 +0000)]
Forgot to register the IPFIREWALL_FORWARD option.

26 years agoDocument `-I' and `-R' options to `pkg_info'.
Joseph Koshy [Mon, 6 Jul 1998 04:39:20 +0000 (04:39 +0000)]
Document `-I' and `-R' options to `pkg_info'.

PR: 7155

26 years agoSupport for IPFW based transparent forwarding.
Julian Elischer [Mon, 6 Jul 1998 03:20:19 +0000 (03:20 +0000)]
Support for IPFW based transparent forwarding.
Any packet that can be matched by a ipfw rule can be redirected
transparently to another port or machine. Redirection to another port
mostly makes sense with tcp, where a session can be set up
between a proxy and an unsuspecting client. Redirection to another machine
requires that the other machine also be expecting to receive the forwarded
packets, as their headers will not have been modified.

/sbin/ipfw must be recompiled!!!

Reviewed by: Peter Wemm <peter@freebsd.org>
Submitted by: Chrisy Luke <chrisy@flix.net>

26 years agoDEVFS completely bypasses the cdevsw and bdevsw tables now.
Julian Elischer [Sun, 5 Jul 1998 23:10:22 +0000 (23:10 +0000)]
DEVFS completely bypasses the cdevsw and bdevsw tables now.
Each devfs node has (and has had fro a while) a pointer directly to
the correct cdefsw entry so just use it instead of doing the lookup.

There are several other places in the kernel that still use the tables
however, so they can't go away yet..

26 years agofix braino from yesterdays' megacommit
Julian Elischer [Sun, 5 Jul 1998 20:33:18 +0000 (20:33 +0000)]
fix braino from yesterdays' megacommit

Not sure of the result of it..
(may or may not effect anything) but it's fixed now.
(found by: comparing what cvsup sent back to me with what I tested..)

26 years agoAdd support for kernel gdb.
Doug Rabson [Sun, 5 Jul 1998 12:24:18 +0000 (12:24 +0000)]
Add support for kernel gdb.

26 years agoUse byte/word access in badaddr if relevant.
Doug Rabson [Sun, 5 Jul 1998 12:22:56 +0000 (12:22 +0000)]
Use byte/word access in badaddr if relevant.

26 years agoSupport channel B as well as channel A.
Doug Rabson [Sun, 5 Jul 1998 12:16:55 +0000 (12:16 +0000)]
Support channel B as well as channel A.

26 years agoAdd basic support for 2117x pci chipsets. Currently only pyxis (21174)
Doug Rabson [Sun, 5 Jul 1998 12:16:15 +0000 (12:16 +0000)]
Add basic support for 2117x pci chipsets.  Currently only pyxis (21174)
is supported.  Older chipsets will be easy to support later but right now,
I just want to boot my 164LX scratch machine :-).

26 years agoAdd macros for byte/word sized load and store instructions.
Doug Rabson [Sun, 5 Jul 1998 12:14:15 +0000 (12:14 +0000)]
Add macros for byte/word sized load and store instructions.

26 years agoAdd declaration of the NetBSD/alpha bootinfo.
Doug Rabson [Sun, 5 Jul 1998 12:13:18 +0000 (12:13 +0000)]
Add declaration of the NetBSD/alpha bootinfo.