]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
27 years agoAdd new files in include/rpc.
wpaul [Wed, 28 May 1997 04:46:38 +0000 (04:46 +0000)]
Add new files in include/rpc.

27 years agoResolve conflicts.
wpaul [Wed, 28 May 1997 04:45:15 +0000 (04:45 +0000)]
Resolve conflicts.

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26209,
wpaul [Wed, 28 May 1997 04:42:53 +0000 (04:42 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26209,
which included commits to RCS files with non-trunk default branches.

27 years agoStep two: import Secure RPC changes to include/rpc.
wpaul [Wed, 28 May 1997 04:42:53 +0000 (04:42 +0000)]
Step two: import Secure RPC changes to include/rpc.

Reviewed by: Mark Murray

27 years agoResolve conflicts.
wpaul [Wed, 28 May 1997 04:38:30 +0000 (04:38 +0000)]
Resolve conflicts.

27 years agoHold onto your hats folks, here it comes: the first stage of the
wpaul [Wed, 28 May 1997 04:35:55 +0000 (04:35 +0000)]
Hold onto your hats folks, here it comes: the first stage of the
Secure RPC import I've been threatening.

This step adds some necessary protocol definition files and headers to
rpcsvc, including the ones needed for NIS+.

Reviewed by: Mark Murray

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26206,
wpaul [Wed, 28 May 1997 04:35:55 +0000 (04:35 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26206,
which included commits to RCS files with non-trunk default branches.

27 years agoDefine NPRIMES in terms of the number of elements in 'primes' (as opposed
alex [Wed, 28 May 1997 00:47:27 +0000 (00:47 +0000)]
Define NPRIMES in terms of the number of elements in 'primes' (as opposed
to hardcoding it).

27 years agoNuke the printing of the unredirect message unless bootverbose.
fsmp [Tue, 27 May 1997 19:28:10 +0000 (19:28 +0000)]
Nuke the printing of the unredirect message unless bootverbose.

27 years agoMinor cleanup of APIC_IO code.
fsmp [Tue, 27 May 1997 19:24:36 +0000 (19:24 +0000)]
Minor cleanup of APIC_IO code.

Submitted by: Stefan Esser <se@freebsd.org>

27 years agoAlways write out rc.conf, regardless of status.
jkh [Tue, 27 May 1997 18:56:03 +0000 (18:56 +0000)]
Always write out rc.conf, regardless of status.
PR: 3698

27 years agoFix problem introduced in the CMD640 workaround which could lead
se [Tue, 27 May 1997 18:28:08 +0000 (18:28 +0000)]
Fix problem introduced in the CMD640 workaround which could lead
to a NULL pointer dereference.

PR: kern/3696
Submitted by: Joel.Faedi@esial.u-nancy.fr

27 years agoCorrect SNAP server paths.
jkh [Tue, 27 May 1997 16:41:52 +0000 (16:41 +0000)]
Correct SNAP server paths.
Bring in CDE fixes.

27 years agoStaticize.
bde [Tue, 27 May 1997 16:26:39 +0000 (16:26 +0000)]
Staticize.

27 years agoTypo fix.
max [Tue, 27 May 1997 13:46:40 +0000 (13:46 +0000)]
Typo fix.
PR: 3693
Submitted by: Kazuo Horikawa <k-horik@yk.rim.or.jp>

27 years agoscprobe() referred to the psm_softc structure even after it was
yokota [Tue, 27 May 1997 11:55:41 +0000 (11:55 +0000)]
scprobe() referred to the psm_softc structure even after it was
freed ;-<

PR: kern/3694

27 years agoAn overhaul of style.9 to clear up some of the ambiguities. A number of
peter [Tue, 27 May 1997 10:00:08 +0000 (10:00 +0000)]
An overhaul of style.9 to clear up some of the ambiguities.  A number of
things are explicitly stated now rather than being implied by example.

Obtained from: Quite a few people over the last few weeks
Reviewed by: core

27 years agosysconfig -> rc.conf
jkh [Tue, 27 May 1997 07:19:57 +0000 (07:19 +0000)]
sysconfig -> rc.conf
Submitted by: Igor Vinokurov <igor@cs.ibank.ru>

27 years agoYAMF22
jkh [Tue, 27 May 1997 07:17:12 +0000 (07:17 +0000)]
YAMF22

27 years agoprevent `struct proc' from being declared inside parameter list.
charnier [Tue, 27 May 1997 06:17:22 +0000 (06:17 +0000)]
prevent `struct proc' from being declared inside parameter list.
PR: kern/3548

27 years agoAdd support for APIC_IO to pci IRQ configuration.
fsmp [Tue, 27 May 1997 04:09:01 +0000 (04:09 +0000)]
Add support for APIC_IO to pci IRQ configuration.

The support for APIC_IO was lost in the new set of pci modules.  This patch
restores the ability to build SMP/APIC_IO kernels.

27 years agoUpdate Makefile to build tcpdump-3.3 .
fenner [Tue, 27 May 1997 02:21:28 +0000 (02:21 +0000)]
Update Makefile to build tcpdump-3.3 .

PR: 3371
Submitted by: Chris Timmons <skynyrd@opus.cts.cwu.edu>

27 years agoGet rid of the T2CHECK() macro, since it duplicates functionality
fenner [Tue, 27 May 1997 02:20:36 +0000 (02:20 +0000)]
Get rid of the T2CHECK() macro, since it duplicates functionality
available from TCHECK(), TCHECK2(), TTEST() and TTEST2().

27 years agoMerge tcpdump 3.3.
fenner [Tue, 27 May 1997 02:17:42 +0000 (02:17 +0000)]
Merge tcpdump 3.3.
The print_nfs.c changes are pretty extensive; this is partially because
 LBL did a lot of cleanup and partially because I removed lots of
 pointless changes away from the LBL style.

PR: 3371
mostly-Submitted by: Chris Timmons <skynyrd@opus.cts.cwu.edu>

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26180,
fenner [Tue, 27 May 1997 02:11:31 +0000 (02:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26180,
which included commits to RCS files with non-trunk default branches.

27 years agoVirgin import of LBL tcpdump v3.3
fenner [Tue, 27 May 1997 02:11:31 +0000 (02:11 +0000)]
Virgin import of LBL tcpdump v3.3

27 years agoBump minor version number.
fenner [Tue, 27 May 1997 00:08:01 +0000 (00:08 +0000)]
Bump minor version number.
Add -I${PCAP_DISTDIR}/bpf to CFLAGS since our <net/bpf.h> is out of date.
Remove -Wall and -Dlint.
DHAVE_ETHER_HOSTTON=1 -DHAVE_STRERROR=1 \
-DHAVE_SOCKADDR_SA_LEN=1 -DLBL_ALIGN=1

CFLAGS+=-I. -Dyylval=pcap_lval ${DEFS}

SHLIB_MAJOR=2
SHLIB_MINOR=2

#
# Magic to grab sources out of src/contrib
#
PCAP_DISTDIR?=${.CURDIR}/../../contrib/libpcap
CFLAGS+=-I${PCAP_DISTDIR} -I${PCAP_DISTDIR}/lbl -I${PCAP_DISTDIR}/bpf
.PATH: ${PCAP_DISTDIR}
.PATH: ${PCAP_DISTDIR}/bpf/net

beforeinstall:
.for i in pcap.h pcap-namedb.h
${INSTALL} -C -o ${BINOWN} -g ${BINGRP} -m 444 ${PCAP_DISTDIR}/$i \
${DESTDIR}/usr/include
.endfor

tokdefs.h grammar.c: grammar.y
${YACC} ${YACCFLAGS} -d ${PCAP_DISTDIR}/grammar.y
mv y.tab.c grammar.c
mv y.tab.h tok/home/ncvs/CVSROOT/commitcheck

27 years agoMerge libpcap 0.3
fenner [Tue, 27 May 1997 00:05:23 +0000 (00:05 +0000)]
Merge libpcap 0.3
Update FREEBSD-upgrade to mention bumping library version number.

Suggested-by: Chris Timmons <skynyrd@opus.cts.cwu.edu>

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26175,
fenner [Tue, 27 May 1997 00:01:12 +0000 (00:01 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26175,
which included commits to RCS files with non-trunk default branches.

27 years agoVirgin import of libpcap 0.3
fenner [Tue, 27 May 1997 00:01:12 +0000 (00:01 +0000)]
Virgin import of libpcap 0.3

27 years agoYet another fix for configuration mechanism 1 register accesses:
se [Mon, 26 May 1997 21:52:41 +0000 (21:52 +0000)]
Yet another fix for configuration mechanism 1 register accesses:
Adjust the data port address by adding the two low order bits of
the register number. The address port takes only a word address
(i.e. ignores the two low order bits written to it).

27 years agoFix previous fix: The enable bit is bit 31 (0x8000000) and not bit 15.
se [Mon, 26 May 1997 21:25:24 +0000 (21:25 +0000)]
Fix previous fix: The enable bit is bit 31 (0x8000000) and not bit 15.

27 years agoSet enable bit when writing the configuration address in configuration
se [Mon, 26 May 1997 21:11:05 +0000 (21:11 +0000)]
Set enable bit when writing the configuration address in configuration
mode 1. Omission of this bit makes all config register accesses fail in
on recent chip sets ...

(The problem was reported and debug output provided by: Steve Passe)

27 years agoFix breakage from my last commit where mp_start() was missing from UP builds.
fsmp [Mon, 26 May 1997 18:40:45 +0000 (18:40 +0000)]
Fix breakage from my last commit where mp_start() was missing from UP builds.

27 years agoChanged inclusion of isa/icu.s to isa/ipl.s.
fsmp [Mon, 26 May 1997 18:00:34 +0000 (18:00 +0000)]
Changed inclusion of isa/icu.s to isa/ipl.s.
This is part of the breakup of UP/SMP specific INTerrupt code.

27 years agoSplit vector.s into UP and SMP specific files:
fsmp [Mon, 26 May 1997 17:58:27 +0000 (17:58 +0000)]
Split vector.s into UP and SMP specific files:
 - vector.s <- stub called by i386/exception.s
 - icu_vector.s <- UP
 - apic_vector.s <- SMP

Split icu.s into UP and SMP specific files:
 - ipl.s <- stub called by i386/exception.s (formerly icu.s)
 - icu_ipl.s <- UP
 - apic_ipl.s <- SMP

This was done in preparation for massive changes to the SMP INTerrupt
mechanisms.  More fine tuning, such as merging ipl.s into exception.s,
may be appropriate.

27 years agoThis file has been made obsolete by the new PCI code.
se [Mon, 26 May 1997 15:15:59 +0000 (15:15 +0000)]
This file has been made obsolete by the new PCI code.

27 years agoCompletely replace the PCI bus driver code to make it better reflect
se [Mon, 26 May 1997 15:08:43 +0000 (15:08 +0000)]
Completely replace the PCI bus driver code to make it better reflect
reality. There will be a new call interface, but for now the file
pci_compat.c (which is to be deleted, after all drivers are converted)
provides an emulation of the old PCI bus driver functions. The only
change that might be visible to drivers is, that the type pcici_t
(which had been meant to be just a handle, whose exact definition
should not be relied on), has been converted into a pcicfgregs* .

The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t
and has been converted to just call the PCI drivers functions to access
configuration space register, instead of inventing its own ...

This code is by no means complete, but assumed to be fully operational,
and brings the official code base more in line with my development code.

A new generic device descriptor data type has to be agreed on. The PCI
code will then use that data type to provide new functionality:

1) userconfig support
2) "wired" PCI devices
3) conflicts checking against ISA/EISA
4) maps will depend on the command register enable bits
5) PCI to Anything bridges can be defined as devices,
   and are probed like any "standard" PCI device.

The following features are currently missing, but will be added back,
soon:

1) unknown device probe message
2) suppression of "mirrored" devices caused by ancient, broken chip-sets

This code relies on generic shared interrupt support just commited to
kern_intr.c (plus the modifications of isa.c and isa_device.h).

27 years agoAdd kern/kern_intr.c as a standard kernel component. This file contains
se [Mon, 26 May 1997 14:46:23 +0000 (14:46 +0000)]
Add kern/kern_intr.c as a standard kernel component. This file contains
the new interrupt registration code, which is meant to be architecture
indeppendent.

27 years agoAdd support for shared interrupts to the kernel. This code is meant
se [Mon, 26 May 1997 14:42:24 +0000 (14:42 +0000)]
Add support for shared interrupts to the kernel. This code is meant
be (eventually) architecture independent. It provides an emulation
of the ISA interrupt registration function register_intr(), but that
function does no longer manipulated the interrupt controller and
interrupt descriptor table, but calls the architecture dependent
function setup_icu() for that purpose.

After the ISA/EISA bus code has been modified to directly call the new
interrupt registartion functions (intr_create() and intr_connect()),
the emulation of register_intr() should be dropped.

The C level interrupt handler function should take a (void*) argument,
and the function pointer type (inthand2_t) should defined in some  other
place than isa_device.h.

This commit is a pre-requisite for the removal of the PCI specific shared
interrupt code.

Reviewed by:    dfr,bde

27 years agoAdd support for shared interrupts to the kernel. This code is meant
se [Mon, 26 May 1997 14:37:43 +0000 (14:37 +0000)]
Add support for shared interrupts to the kernel. This code is meant
be (eventually) architecture independent. It provides an emulation
of the ISA interrupt registration function register_intr(), but that
function does no longer manipulated the interrupt controller and
interrupt descriptor table, but calls the architecture dependent
function setup_icu() for that purpose.

After theISA/EISA bus code has been modified to directly call the new
interrupt registartion functions (intr_create() and intr_connect()),
the emulation of register_intr() should be dropped.

The C level interrupt handler function should take a (void*) argument,
and the function pointer type (inthand2_t) should defined in some  other
place than isa_device.h.

This commit is a pre-requisite for the removal of the PCI specific shared
interrupt code.

Reviewed by: dfr,bde

27 years agoAdded a test called 'LATE_START'.
fsmp [Mon, 26 May 1997 09:23:30 +0000 (09:23 +0000)]
Added a test called 'LATE_START'.

This is now the default, it delays most of the MP startup to the function
machdep.c:cpu_startup().  It should be possible to move the 2 functions
found there (mp_start() & mp_announce()) even further down the path once
we know exactly where that should be...

Help from: Peter Wemm <peter@spinner.dialix.com.au>

27 years agoLame typo fix here too.
jkh [Mon, 26 May 1997 05:33:16 +0000 (05:33 +0000)]
Lame typo fix here too.

27 years agomake beforeinstall in libalias from the includes target
peter [Mon, 26 May 1997 05:27:37 +0000 (05:27 +0000)]
make beforeinstall in libalias from the includes target

27 years agoOne of the new cvs's features is to be able to run a checker on the
peter [Mon, 26 May 1997 05:15:29 +0000 (05:15 +0000)]
One of the new cvs's features is to be able to run a checker on the
log messages after they've been entered.  This is more flexible than
using the editinfo script since it works for all log message types
and doesn't have to deal with trying to run the editor for the user.

The problem is that the verifymsg script can't modify the file like
editinfo can, which makes it useless for cleaning up the message (as is
needed for remote commits etc).  This change causes the verifymsg handler
to read back the message after the verify script has run and returned an
"OK" exit code.

27 years agoYAMF22
jkh [Mon, 26 May 1997 04:57:56 +0000 (04:57 +0000)]
YAMF22

27 years agoUninitialised inp variable in div_bind().
peter [Mon, 26 May 1997 03:33:48 +0000 (03:33 +0000)]
Uninitialised inp variable in div_bind().

Submitted by: Åge Røbekk <aagero@aage.priv.no>

27 years agoCorrected accidental breakage in the last commit.
yokota [Mon, 26 May 1997 01:02:41 +0000 (01:02 +0000)]
Corrected accidental breakage in the last commit.

Pointed out by: Steve Passe, smp@csn.net

27 years agoDe-couple ppp from libalias. If libalias isn't there, the
brian [Mon, 26 May 1997 00:44:10 +0000 (00:44 +0000)]
De-couple ppp from libalias.  If libalias isn't there, the
alias commands simply won't work.  Only root may specify the
location of the alias lib (otherwise, it's hard-coded).

Make logprintf silently fail if LogOpen hasn't been called.

Suggested by: eivind

27 years agoInclude <alias.h>, not "alias.h"
brian [Sun, 25 May 1997 19:49:44 +0000 (19:49 +0000)]
Include <alias.h>, not "alias.h"

27 years agoThis commit adds support to ypbind(8) for binding to non-local servers.
wpaul [Sun, 25 May 1997 19:49:33 +0000 (19:49 +0000)]
This commit adds support to ypbind(8) for binding to non-local servers.

The standard SunOS ypbind(8) (and, until now, the FreeBSD ypbind)
only selects servers based on whether or not they respond to clnt_broadcast().
Ypbind(8) broadcasts to the YPPROC_DOMAIN_NONACK procedure and waits
for answers; whichever server answers first is the one ypbind uses
for the local client binding.

This mechanism fails when binding across subnets is desired. In order
for a client on one subnet to bind to a server on another subnet, the
gateway(s) between the client and server must be configured to forward
broadcasts. If this is not possible, then a slave server must be
installed on the remote subnet. If this is also not possible, you
have to force the client to bind to the remote server with ypset(8).

Unfortunately, this last option is less than ideal. If the remote
server becomes unavailable, ypbind(8) will lose its binding and
revert to its broadcast-based search behavior. Even if there are
other servers available, or even if the original server comes back
up, ypbind(8) will not be able to create a new binding since all
the servers are on remote subnets where its broadcasts won't be heard.
If the administrator isn't around to run ypset(8) again, the system
is hosed.

In some Linux NIS implementations, there exists a yp.conf file where
you can explicitly specify a server address and avoid the use of
ypbind altogether. This is not desireable since it removes the
possibility of binding to an alternate server in the event that the
one specified in yp.conf crashes.

Some people have mentioned to me how they though the 'restricted mode'
operation (using the -S flag) could be used as a solution for this
problem since it allows one to specify a list of servers. In fact,
this is not the case: the -S flag just tells ypbind(8) that when it
listens for replies to its broadcasts, it should only honor them if
the replying hosts appear in the specified restricted list.

This behavior has now been changed. If you use the -m flag in conjunction
with the -S flag, ypbind(8) will use a 'many-cast' instead of a broadcast
for choosing a server. In many-cast mode, ypbind(8) will transmit directly
to the YPPROC_DOMAIN_NONACK procedure of all the servers specified in
the restricted mode list and then wait for a reply. As with the broadcast
method, whichever server from the list answers first is used for the
local binding. All other behavior is the same: ypbind(8) continues
to ping its bound server every 60 seconds to insure it's still alive
and will many-cast again if the server fails to respond. The code used
to achieve this is in yp_ping.c; it includes a couple of modified RPC
library routines.

Note that it is not possible to use this mechanism without using
the restricted list since we need to know the addresses of the available
NIS servers ahead of time in order to transmit to them.

Most-recently-requested by: Tom Samplonius

27 years agoTypo: .SH --> .Sh.
max [Sun, 25 May 1997 19:11:26 +0000 (19:11 +0000)]
Typo: .SH --> .Sh.
PR: 3683
Submitted by: Kazuo Horikawa <k-horik@yk.rim.or.jp>

27 years agoTypo fix.
max [Sun, 25 May 1997 18:47:33 +0000 (18:47 +0000)]
Typo fix.
PR: 3684
Submitted by: Kazuo Horikawa <k-horik@yk.rim.or.jp>

27 years agoRemoved global hwisrs, replacing it with vec (formerly static to icu.s).
fsmp [Sun, 25 May 1997 16:58:03 +0000 (16:58 +0000)]
Removed global hwisrs, replacing it with vec (formerly static to icu.s).

27 years agoMade the array vec[] a global.
fsmp [Sun, 25 May 1997 16:55:26 +0000 (16:55 +0000)]
Made the array vec[] a global.
This allows the APIC code to reorder the vectors at runtime.

27 years agoFixed up ssigala@globalnet.it's "Jumping Daemon" screen saver for < v3.0.
obrien [Sun, 25 May 1997 16:06:29 +0000 (16:06 +0000)]
Fixed up ssigala@globalnet.it's "Jumping Daemon" screen saver for < v3.0.

27 years agoAdd lines for new files in ipfilter
darrenr [Sun, 25 May 1997 15:56:18 +0000 (15:56 +0000)]
Add lines for new files in ipfilter

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26124,
darrenr [Sun, 25 May 1997 15:50:46 +0000 (15:50 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26124,
which included commits to RCS files with non-trunk default branches.

27 years agoImport version 3.2alpha7
darrenr [Sun, 25 May 1997 15:50:46 +0000 (15:50 +0000)]
Import version 3.2alpha7

27 years agoImport version 3.2alpha7
darrenr [Sun, 25 May 1997 15:50:46 +0000 (15:50 +0000)]
Import version 3.2alpha7

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26121,
darrenr [Sun, 25 May 1997 15:49:17 +0000 (15:49 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26121,
which included commits to RCS files with non-trunk default branches.

27 years agoImport version 3.2alpha7
darrenr [Sun, 25 May 1997 15:49:17 +0000 (15:49 +0000)]
Import version 3.2alpha7
(they're currently in src/contrib/ipfilter/ipfilter/ by mistake, if someone
 from core would like to delete that directory with three files as I'm not
 meant to do that :)

27 years agoThis commit was generated by cvs2svn to compensate for changes in r26119,
darrenr [Sun, 25 May 1997 15:45:04 +0000 (15:45 +0000)]
This commit was generated by cvs2svn to compensate for changes in r26119,
which included commits to RCS files with non-trunk default branches.

27 years agoImport version 3.2alpha7
darrenr [Sun, 25 May 1997 15:45:04 +0000 (15:45 +0000)]
Import version 3.2alpha7

27 years agoAdd an entry for the Epson scanners since some of them repsond to more
joerg [Sun, 25 May 1997 14:20:28 +0000 (14:20 +0000)]
Add an entry for the Epson scanners since some of them repsond to more
than one LUN:

Submitted by: simokawa@sat.t.u-tokyo.ac.jp (Hidetoshi Shimokawa)

27 years agoAdd a `strip' target which remove the ^H from plain text files.
wosch [Sun, 25 May 1997 12:43:06 +0000 (12:43 +0000)]
Add a `strip' target which remove the ^H from plain text files.

27 years agoRemove some quite bogus code that tried to find subnet names but instead
phk [Sun, 25 May 1997 08:36:20 +0000 (08:36 +0000)]
Remove some quite bogus code that tried to find subnet names but instead
spammed named with bogus queries.

27 years agouncomment wl again now that it compiles...
peter [Sun, 25 May 1997 07:18:22 +0000 (07:18 +0000)]
uncomment wl again now that it compiles...

27 years agoMake the multicast code compile under -current. This has not been reviewed
peter [Sun, 25 May 1997 07:17:02 +0000 (07:17 +0000)]
Make the multicast code compile under -current.  This has not been reviewed
or tested yet.  Mike may want to make some adjustments.

27 years agoConnect the ipdivert div_usrreqs struct to the ip proto switch table
peter [Sun, 25 May 1997 06:09:23 +0000 (06:09 +0000)]
Connect the ipdivert div_usrreqs struct to the ip proto switch table

27 years agoFix warnings (from LINT). Missing static prototype, missing vm includes
peter [Sun, 25 May 1997 04:57:11 +0000 (04:57 +0000)]
Fix warnings (from LINT).  Missing static prototype, missing vm includes
for vnode_pager_setsize().

27 years agoFix some warnings (missing prototypes, wrong "generic" args etc)
peter [Sun, 25 May 1997 04:50:02 +0000 (04:50 +0000)]
Fix some warnings (missing prototypes, wrong "generic" args etc)
umapfs uses one of nullfs's functions...

27 years agoFix(?) two volatile cast warnings. The buffer is volatile within the
peter [Sun, 25 May 1997 04:13:27 +0000 (04:13 +0000)]
Fix(?) two volatile cast warnings.  The buffer is volatile within the
function so as to avoid compiler side effects, but functions calling it
shouldn't be affected (gcc shouldn't cache memory scope past a function
call)

27 years agomatch the #include opt_ file with options.i386 (it has opt_wavelan.h)
peter [Sun, 25 May 1997 04:04:20 +0000 (04:04 +0000)]
match the #include opt_ file with options.i386 (it has opt_wavelan.h)

27 years agoBroke up parse_mp_table() into 2 passes:
fsmp [Sun, 25 May 1997 02:49:03 +0000 (02:49 +0000)]
Broke up parse_mp_table() into 2 passes:
 - The 1st (preparse_mp_table()) counts the number of cpus, busses, etc. and
   records the LOCAL and IO APIC addresses.
 - The 2nd pass (parse_mp_table()) does the actual parsing of info and recording
   into the incore MP table.

This will allow us to defer the 2nd pass untill malloc() & private pages
are available (but thats for another day!).

27 years agoMinor cleanup.
fsmp [Sun, 25 May 1997 02:43:42 +0000 (02:43 +0000)]
Minor cleanup.
Comment about expanded use of NAPICIDS.
Change default NINTR from 64 to 48.

27 years agoWhen no last file specified for % command user barf(error diagnostic)
ache [Sat, 24 May 1997 21:45:55 +0000 (21:45 +0000)]
When no last file specified for % command user barf(error diagnostic)
instead of segmentation fault

27 years agoMake sh(1) less aware of the bit fields returned by wait by
steve [Sat, 24 May 1997 21:04:55 +0000 (21:04 +0000)]
Make sh(1) less aware of the bit fields returned by wait by
using the WIF* macros in sys/wait.h.

PR: bin/3668
Submitted by: dholland@eecs.harvard.edu

27 years agoAdd rules for building doc back into releases, conditionalized on NODOC.
jkh [Sat, 24 May 1997 20:15:13 +0000 (20:15 +0000)]
Add rules for building doc back into releases, conditionalized on NODOC.
Also clean up some of the comments at the top.  It's 12 o'clock, and
that is all.

27 years agoDelay mp_start() till after the msgbuf is mapped. We really want to delay
fsmp [Sat, 24 May 1997 18:59:02 +0000 (18:59 +0000)]
Delay mp_start() till after the msgbuf is mapped.  We really want to delay
it till even later but tss setup prevents that right now...

27 years agoNow that panic() is properly printing messages for early SMP panics all
fsmp [Sat, 24 May 1997 18:48:53 +0000 (18:48 +0000)]
Now that panic() is properly printing messages for early SMP panics all
the 'printf("..."); panic("\n")' sections are returned to 'panic("...")'.

27 years agoMove the printing of "cpu#%d" to AFTER the general panic argument string.
fsmp [Sat, 24 May 1997 18:35:44 +0000 (18:35 +0000)]
Move the printing of "cpu#%d" to AFTER the general panic argument string.
When a panic occurs early in the SMP boot process 'cpunumber()' hangs,
causing the panic string to be lost.  Now the system appears to hang
in 'breakpoint()', but at least the user sees the panic string before the
hang.

27 years agoFix the reconnect option, and add an explanation to vars.h
brian [Sat, 24 May 1997 17:32:42 +0000 (17:32 +0000)]
Fix the reconnect option, and add an explanation to vars.h

27 years agoAttempt to convert the ip_divert code to use the new-style protocol request
peter [Sat, 24 May 1997 17:23:11 +0000 (17:23 +0000)]
Attempt to convert the ip_divert code to use the new-style protocol request
switch.  I needed 'LINT' to compile for other reasons so I kinda got the
blood on my hands.  Note: I don't know how to test this, I don't know if
it works correctly.

27 years agoThe wavelan driver doesn't even compile!
peter [Sat, 24 May 1997 12:23:44 +0000 (12:23 +0000)]
The wavelan driver doesn't even compile!

27 years agoMake firewall comment a little more explanatory.
jkh [Sat, 24 May 1997 11:31:56 +0000 (11:31 +0000)]
Make firewall comment a little more explanatory.

27 years agoAllow local configuration to override this file.
jkh [Sat, 24 May 1997 11:29:59 +0000 (11:29 +0000)]
Allow local configuration to override this file.
Submitted by: Russell Vincent <rv@groa.uct.ac.za>

27 years agoFixes skeleton directory test-it-is-a-directory logic.
davidn [Sat, 24 May 1997 10:41:49 +0000 (10:41 +0000)]
Fixes skeleton directory test-it-is-a-directory logic.
PR: 3666
Reviewed by:
Submitted by: iaint@css.tuu.utas.edu.au
Obtained from:

27 years agoAdd -lalias for ppp now.
jkh [Sat, 24 May 1997 09:03:28 +0000 (09:03 +0000)]
Add -lalias for ppp now.

27 years agoBootstrap aid from older releases.. 'make world' can't build xinstall
peter [Sat, 24 May 1997 05:39:19 +0000 (05:39 +0000)]
Bootstrap aid from older releases..  'make world' can't build xinstall
until the includes are installed, and it doesn't install the includes until
xinstall is built (the previous xinstall binary may not have -C).  As a
bandaid, define MAP_FAILED ourselves if <sys/mman.h> didn't.  Perhaps it
would be safer to simply disable mmap if all the prerequisites are not
there.

27 years agoFlip the daemon at the edge of the screen.
yokota [Sat, 24 May 1997 01:44:39 +0000 (01:44 +0000)]
Flip the daemon at the edge of the screen.
Submitted by: Sandro Sigala, Chris Shenton

27 years agosubmitted by: archie@whistle.com
julian [Fri, 23 May 1997 22:33:16 +0000 (22:33 +0000)]
submitted by: archie@whistle.com

Don't search for interface addresses matching interface "NULL"
it's likely to cause a page fault..
this can be triggered by the ipfw code rejecting a locally generated
packet (e.g. you decide to make some network unreachable by local users)

27 years agoMove libmd after all libraries to keep natural libraries order
ache [Fri, 23 May 1997 21:18:00 +0000 (21:18 +0000)]
Move libmd after all libraries to keep natural libraries order

27 years agoFix a couple of log()'s that came out with the wrong (default)
joerg [Fri, 23 May 1997 20:40:15 +0000 (20:40 +0000)]
Fix a couple of log()'s that came out with the wrong (default)
log level, as opposed to LOG_DEBUG.

27 years agoMerge from 2.2 - XiG support updates.
jkh [Fri, 23 May 1997 18:54:29 +0000 (18:54 +0000)]
Merge from 2.2 - XiG support updates.

27 years agoAdd "-assert pure-text" to the link line. The warnings about RRS
dfr [Fri, 23 May 1997 17:50:35 +0000 (17:50 +0000)]
Add "-assert pure-text" to the link line.  The warnings about RRS
relocations in text sections are now dependant on this flag.

27 years agoUse realpath on the source as well as the destination.
dfr [Fri, 23 May 1997 17:07:40 +0000 (17:07 +0000)]
Use realpath on the source as well as the destination.

27 years agoTypo in previous change.
dfr [Fri, 23 May 1997 17:07:00 +0000 (17:07 +0000)]
Typo in previous change.

27 years agoFix support for the Hitachi CDR7730 IDE cdrom drive.
sos [Fri, 23 May 1997 16:22:04 +0000 (16:22 +0000)]
Fix support for the Hitachi CDR7730 IDE cdrom drive.
Patch from pr1730.
Tested on our IDE cdrom drive farm...

27 years agoforgot to mention that the vms source was pruned
peter [Fri, 23 May 1997 14:55:31 +0000 (14:55 +0000)]
forgot to mention that the vms source was pruned