]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
27 years agoCatch up with the original.
Hiroyuki Hanai [Wed, 7 May 1997 04:45:54 +0000 (04:45 +0000)]
Catch up with the original.
Submitted by: tomo@maple.elcom.nitech.ac.jp (WATANABE tomoo)

27 years agoRestore unintentially lost backwards compat behavior of defaulting to
Peter Wemm [Wed, 7 May 1997 04:28:26 +0000 (04:28 +0000)]
Restore unintentially lost backwards compat behavior of defaulting to
family inet if not specified.  (eg: "ifconfig ed0" down would fail because
no family was specified, even though the up/down status is not per family)

Pointed out by: Wolfgang Helbig <helbig@MX.BA-Stuttgart.De>

27 years agoMake a lot of include-files self-contained. I excluded the patches changing
Eivind Eklund [Wed, 7 May 1997 02:27:18 +0000 (02:27 +0000)]
Make a lot of include-files self-contained.  I excluded the patches changing
int's to gid_t and uid_t - should I commit these, too?

Closes PR misc/2625.

Submitted by: Julian Assange <proff@iq.org>

27 years agoFixed missing __P() to complete the K&R support in this header.
Bruce Evans [Wed, 7 May 1997 00:58:23 +0000 (00:58 +0000)]
Fixed missing __P() to complete the K&R support in this header.

27 years agoAllow a '-' flag as a placeholder so that path_to_pid_file
Brian Somers [Tue, 6 May 1997 23:11:06 +0000 (23:11 +0000)]
Allow a '-' flag as a placeholder so that path_to_pid_file
can be specified without the flags field.
Fix bogus "trail" left in parse routine where the code jumps
past the end of the line and wanders into oblivion.

27 years agoForce user to config SMP kernel with "options APIC_IO".
Steve Passe [Tue, 6 May 1997 21:29:57 +0000 (21:29 +0000)]
Force user to config SMP kernel with "options APIC_IO".

Reviewed by: Peter Wemm <peter@spinner.DIALix.COM>

27 years agoPull up the IP header in ip_mloopback(). This makes sure that the
Bill Fenner [Tue, 6 May 1997 21:22:04 +0000 (21:22 +0000)]
Pull up the IP header in ip_mloopback().  This makes sure that the
operations on the header inside ip_mloopback() are performed on
a private copy instead of a shared cluster.

PR: kern/3410

27 years agoFix a long-standing bug with dump not treating errors correctly. They
Joerg Wunsch [Tue, 6 May 1997 20:52:28 +0000 (20:52 +0000)]
Fix a long-standing bug with dump not treating errors correctly.  They
were mishandled as an EOF, which became fatal if your first tape was
accidentally write-protected.

PR: bin/850 dump treats write-protect as an EOT...

27 years agoMake ident equal file name, ie SMP-GENERIC.
Steve Passe [Tue, 6 May 1997 20:40:52 +0000 (20:40 +0000)]
Make ident equal file name, ie SMP-GENERIC.

27 years agoA *little* more descriptive test for options.
Steve Passe [Tue, 6 May 1997 18:24:17 +0000 (18:24 +0000)]
A *little* more descriptive test for options.

27 years agoAdded a generic config file for SMP kernels.
Steve Passe [Tue, 6 May 1997 18:18:51 +0000 (18:18 +0000)]
Added a generic config file for SMP kernels.

27 years agoPR: none
Peter da Silva [Tue, 6 May 1997 16:21:18 +0000 (16:21 +0000)]
PR: none
Submitted by: pds

27 years agoFix a race condition that did, after all, exist.
Poul-Henning Kamp [Tue, 6 May 1997 15:19:38 +0000 (15:19 +0000)]
Fix a race condition that did, after all, exist.

Reviewed by: phk
Submitted by: dfr

27 years agoAdd SMPHDRS to CLEANFILES instead of replacing its current value.
Doug Rabson [Tue, 6 May 1997 14:29:56 +0000 (14:29 +0000)]
Add SMPHDRS to CLEANFILES instead of replacing its current value.

27 years agoFix FTP link to the statically linked FreeBSD executables
Wolfram Schneider [Tue, 6 May 1997 11:18:44 +0000 (11:18 +0000)]
Fix FTP link to the statically linked FreeBSD executables
freefall -> hub

27 years agoremoved the "#error ..." line preventing casual invokation of SMP_AUTOSTART.
Steve Passe [Tue, 6 May 1997 07:10:06 +0000 (07:10 +0000)]
removed the "#error ..." line preventing casual invokation of SMP_AUTOSTART.
autostart appears to be working now, at least on my dual P6.
I have no explanation why....

27 years agoDescribe SNAPs and RELENGes as requested on USENET.
Jordan K. Hubbard [Tue, 6 May 1997 06:38:46 +0000 (06:38 +0000)]
Describe SNAPs and RELENGes as requested on USENET.

27 years agoSpecify the full path of units.lib in the FILES section.
Alexander Langer [Tue, 6 May 1997 03:31:14 +0000 (03:31 +0000)]
Specify the full path of units.lib in the FILES section.

27 years agoCreate the default rule with flags IP_FW_F_IN | IP_FW_F_OUT.
Alexander Langer [Tue, 6 May 1997 02:12:18 +0000 (02:12 +0000)]
Create the default rule with flags IP_FW_F_IN | IP_FW_F_OUT.
Closes PR#3100.

27 years agoBye bye CPLUSPLUSLIB hack! It's not needed any more.
John Polstra [Tue, 6 May 1997 00:49:36 +0000 (00:49 +0000)]
Bye bye CPLUSPLUSLIB hack!  It's not needed any more.

Don't merge this into -2.2 unless you understand the dependencies on
c++rt0, bsd.lib.mk, and gcc -shared.  I.e., let me do it.

27 years agoMake ppp exit on loss of carrier in -background mode.
Brian Somers [Mon, 5 May 1997 23:45:15 +0000 (23:45 +0000)]
Make ppp exit on loss of carrier in -background mode.

Submitted by: Jim Carroll <jim@carroll.com>

27 years agoCode to handle SMP/APIC_IO mapping of ISA INTs to APIC pins above IRQ15.
Steve Passe [Mon, 5 May 1997 22:56:37 +0000 (22:56 +0000)]
Code to handle SMP/APIC_IO mapping of ISA INTs to APIC pins above IRQ15.

 - doesn't break my system.
 - NOT yet verified on the affected motherboard.

Submitted by: "John S. Dyson" <toor@dyson.iquest.net>

27 years agoCode to handle SMP/APIC_IO mapping of ISA INTs to APIC pins above IRQ15.
Steve Passe [Mon, 5 May 1997 22:56:13 +0000 (22:56 +0000)]
Code to handle SMP/APIC_IO mapping of ISA INTs to APIC pins above IRQ15.

 - doesn't break my system.
 - NOT yet verified on the affected motherboard.

Stifle an annoying dma_start busy message for the sound cards.

Submitted by: "John S. Dyson" <toor@dyson.iquest.net>

27 years agouse frames instead of fields, and restore full meteor compatibility for
John-Mark Gurney [Mon, 5 May 1997 20:54:54 +0000 (20:54 +0000)]
use frames instead of fields, and restore full meteor compatibility for
generalized pixel support.

Submitted by: Amancio Hasty

27 years agoDo not kill -HUP syslog if running in user mode (-r), it allows
Andrey A. Chernov [Mon, 5 May 1997 15:00:15 +0000 (15:00 +0000)]
Do not kill -HUP syslog if running in user mode (-r), it allows
private user logs rotation without error diagnostics

27 years agoUse `MediaGX' instead of `Gx86'.
KATO Takenori [Mon, 5 May 1997 14:30:00 +0000 (14:30 +0000)]
Use `MediaGX' instead of `Gx86'.

27 years agoUse `M2' instead of `6x86 with MMX'. Cyrix seems to use `M2' officially.
KATO Takenori [Mon, 5 May 1997 14:11:09 +0000 (14:11 +0000)]
Use `M2' instead of `6x86 with MMX'.  Cyrix seems to use `M2' officially.

27 years agoDuplicate the TOC handling ioctls from the cd(4) driver. Duplicating
Joerg Wunsch [Mon, 5 May 1997 13:35:01 +0000 (13:35 +0000)]
Duplicate the TOC handling ioctls from the cd(4) driver.  Duplicating
the work is silly, and all this will be a moot point with Justin's CAM
driver, and there are still many ioctls missing -- but i need these
right now.  It's silly to need to move the CD-R back to the other
drive in order to determine the session start info when burning multi-
session CD-Rs.

27 years agoExtended copyright of PC-98 to 1997.
KATO Takenori [Mon, 5 May 1997 13:24:16 +0000 (13:24 +0000)]
Extended copyright of PC-98 to 1997.

27 years agoSynchornize with sys/i386/isa/clock.c revision upto 1.94.
KATO Takenori [Mon, 5 May 1997 13:13:47 +0000 (13:13 +0000)]
Synchornize with sys/i386/isa/clock.c revision upto 1.94.

27 years agoSynchronize with sys/i386/i386/trap.c revision 1.94.
KATO Takenori [Mon, 5 May 1997 13:12:52 +0000 (13:12 +0000)]
Synchronize with sys/i386/i386/trap.c revision 1.94.

27 years agoMake this C++ safe.
Jordan K. Hubbard [Mon, 5 May 1997 11:18:55 +0000 (11:18 +0000)]
Make this C++ safe.
Submitted by: Nadav Eiron <nadav@barcode.co.il>

27 years agoUpdate to list more changes in 3.0. Also point people at the donor's
Jordan K. Hubbard [Mon, 5 May 1997 11:09:33 +0000 (11:09 +0000)]
Update to list more changes in 3.0.  Also point people at the donor's
gallery rather than including an out-of-date list.

27 years agocorrect the order of the variables
Peter Wemm [Mon, 5 May 1997 09:34:33 +0000 (09:34 +0000)]
correct the order of the variables
use #ifdef where possible instead of #if defined

Submitted by: the KNF police, ie: bde :-)

27 years agoCorrect typo in NIS flags.
Jordan K. Hubbard [Mon, 5 May 1997 09:32:34 +0000 (09:32 +0000)]
Correct typo in NIS flags.
PR: 3503
Submitted by: Christoph Kukulies <kuku@blues.physik.rwth-aachen.de>

27 years agoDOH! Fix package-extractor looping problem by moving some code
Jordan K. Hubbard [Mon, 5 May 1997 08:38:12 +0000 (08:38 +0000)]
DOH!  Fix package-extractor looping problem by moving some code
to the *proper* part of the loop.

27 years agochange `an' to `a'...
John-Mark Gurney [Mon, 5 May 1997 07:43:50 +0000 (07:43 +0000)]
change `an' to `a'...

Submitted by: Josh Gilliam
Closes PR#3471

27 years agoAdd inetd_flags and way of passing ipfw a configuration file
Jordan K. Hubbard [Mon, 5 May 1997 07:08:31 +0000 (07:08 +0000)]
Add inetd_flags and way of passing ipfw a configuration file
(if firewall = "somefilename").

Fix typo fixes and URLs which were accidently nuked out of this
file (submitted by: soil@quick.net via PR#3501).

Submitted by: "Danny J. Zerkel" <dzerkel@phofarm.com>

27 years agoAdapt routing configuration glue to rc.conf format.
Jordan K. Hubbard [Mon, 5 May 1997 06:32:44 +0000 (06:32 +0000)]
Adapt routing configuration glue to rc.conf format.

27 years agoMention the removal of UPAGES, the return-to-userspace fork(), and the fast
Peter Wemm [Mon, 5 May 1997 05:58:33 +0000 (05:58 +0000)]
Mention the removal of UPAGES, the return-to-userspace fork(), and the fast
complete implementation of vfork().

27 years agoAdd in some more config file support -- add variables to shut up
Paul Traina [Mon, 5 May 1997 05:16:03 +0000 (05:16 +0000)]
Add in some more config file support -- add variables to shut up
sysinstall about the dangerously dedicated message, and other
variables to allow pre-configuring the distribution sets.

Still todo: add a variable to define an initial set of packages that
should be loaded.
Reviewed by: jkh

27 years agoMake sure that *fork() always returns with %edx == 1 in the
John Dyson [Mon, 5 May 1997 04:08:12 +0000 (04:08 +0000)]
Make sure that *fork() always returns with %edx == 1 in the
child.  This was sometimes not happening correctly during my
threads code work.

27 years agoDon't source $ENV unless this is an interactive shell.
Steve Price [Mon, 5 May 1997 03:45:17 +0000 (03:45 +0000)]
Don't source $ENV unless this is an interactive shell.

27 years agore-enabled: . /etc/rc.${arch}
Andreas Klemm [Sun, 4 May 1997 22:33:17 +0000 (22:33 +0000)]
re-enabled: . /etc/rc.${arch}
PR:
Reviewed by: bruce
Submitted by:
Obtained from:

27 years agoUse "gcc -shared" instead of "ld -Bshareable" for building shared
John Polstra [Sun, 4 May 1997 21:30:40 +0000 (21:30 +0000)]
Use "gcc -shared" instead of "ld -Bshareable" for building shared
libraries.  Remove the now-unneeded CPLUSPLUSLIB hack.  I will also
remove the CPLUSPLUSLIB definitions from the Makefiles that use it,
after the dust settles.

Use gcc's LIBRARY_PATH environment variable to handle DESTDIR,
instead of -L flags in LDDESTDIR.  LIBRARY_PATH is documented in
the gcc info pages.  It is better than using -L flags, because it
modifies the search for start-up files as well as for libraries.
A new variable LDDESTDIRENV is used to contain the normally-empty
LIBRARY_PATH environment setting.

LDDESTDIR is no longer set in <bsd.lib.mk>.  It is still honored for
the time being, because a couple of userland Makefiles still (wrongly)
set it.  These should be fixed, and LDDESTDIR should vanish.

Removed the commented-out "LDDESTDIR+=-nostdlib", because "gcc -shared"
doesn't link in any standard libraries anyway.

Removed the ".if defined(LDADD)" around the _EXTRADEPEND target for
shared libraries.  This target is always necessary now, because
c++rt0.o is linked into every shared library.

Don't merge this into -2.2 without first merging the support for
"gcc -shared".

27 years agoFrom the PR:
Eivind Eklund [Sun, 4 May 1997 20:17:54 +0000 (20:17 +0000)]
From the PR:
ipcs -T' shows system information about shared memory,
message queues and semaphores. But the manual description does
not mention semapores.

Now it does.
PR: docs/3489
Submitted by: k-horik@yk.rim.or.jp

27 years agoFix typo.
Eivind Eklund [Sun, 4 May 1997 20:04:59 +0000 (20:04 +0000)]
Fix typo.
PR: Closes PR docs/3488
Submitted by: k-horik@yk.rim.or.jp

27 years agoRemove the last reference to sup as a current method. We now want
Eivind Eklund [Sun, 4 May 1997 18:04:38 +0000 (18:04 +0000)]
Remove the last reference to sup as a current method.  We now want
cvsup-mirrors now, not sup-mirrors.

27 years agoMention new features:
Joerg Wunsch [Sun, 4 May 1997 16:26:53 +0000 (16:26 +0000)]
Mention new features:

. if_media support (obtained from Jason Thorpe's NetBSD implementation),
. multi-session ISO-9660 support

Move aic6[26]30 and floppy tapes out into an ``unmaintained'' subsection,
so people don't expect the same quality of support for them as for other
drivers.

Our ATAPI CD-ROM driver should no longer be considered ALPHA quality
now, i think.

27 years agoOops. The function cd9660_mountroot() is gone, but i've committed an
Joerg Wunsch [Sun, 4 May 1997 16:17:49 +0000 (16:17 +0000)]
Oops.  The function cd9660_mountroot() is gone, but i've committed an
even more bogus prototype for it in my previous commit.

27 years agoThis mega-commit brings the following:
Joerg Wunsch [Sun, 4 May 1997 15:24:23 +0000 (15:24 +0000)]
This mega-commit brings the following:

. It makes cd9660 root f/s working again.
. It makes CD9660 a new-style option.
. It adds support to mount an ISO9660 multi-session CD-ROM as the root
  filesystem (the last session actually, but that's what is expected
  behaviour).

Sigh.  The CDIOREADTOCENTRYS did a copyout() of its own, and thus has
been unusable for me for this work.  Too bad it didn't simply stuff
the max 100 entries into the struct ioc_read_toc_entry, but relied on
a user supplied data buffer instead. :-(  I now had to reinvent the
wheel, and created a CDIOREADTOCENTRY ioctl command that can be used
in a kernel context.

While doing this, i noticed the following bogosities in existing CD-ROM
drivers:

wcd: This driver is likely to be totally bogus when someone tries
two succeeding CDIOREADTOCENTRYS (or now CDIOREADTOCENTRY)
commands with requesting MSF format, since it apparently
operates on an internal table.

scd: This driver apparently returns just a single TOC entry only for
the CDIOREADTOCENTRYS command.

I have only been able to test the CDIOREADTOCENTRY command with the
cd(4) driver.  I hereby request the respective maintainers of the
other CD-ROM drivers to verify my code for their driver.  When it
comes to merging this CD-ROM multisession stuff into RELENG_2_2 i will
only consider drivers where i've got a confirmation that it actually
works.

27 years agoNow I can even execute "df" on my diskless :-)
Poul-Henning Kamp [Sun, 4 May 1997 15:04:49 +0000 (15:04 +0000)]
Now I can even execute "df" on my diskless :-)

27 years agodon't #ifdef out reference to i586_ctr_freq.
Peter Wemm [Sun, 4 May 1997 14:28:22 +0000 (14:28 +0000)]
don't #ifdef out reference to i586_ctr_freq.

27 years agoDon't remove i586_ctr_freq from scope, leave it defined as zero. This
Peter Wemm [Sun, 4 May 1997 14:25:00 +0000 (14:25 +0000)]
Don't remove i586_ctr_freq from scope, leave it defined as zero.  This
simplifies some assumptions and stops some code compile problems.

This should fix the compile hiccup in PR#3491, but smp kernel profiling
isn't likely to be fixed by this.

27 years agoFixed sourcing of /etc/etc.${arch}/rc.${arch} on startup.
Andreas Klemm [Sun, 4 May 1997 12:50:33 +0000 (12:50 +0000)]
Fixed sourcing of /etc/etc.${arch}/rc.${arch} on startup.
PR:
Reviewed by:
Submitted by:
Obtained from:

27 years agoYikes! I thought I updated these references to sup ages ago.. Hmmmm!
Jordan K. Hubbard [Sun, 4 May 1997 11:57:51 +0000 (11:57 +0000)]
Yikes!  I thought I updated these references to sup ages ago..  Hmmmm!

27 years ago1. Add a {pointer, v_id} pair to the vnode to store the reference to the
Poul-Henning Kamp [Sun, 4 May 1997 09:17:38 +0000 (09:17 +0000)]
1.  Add a {pointer, v_id} pair to the vnode to store the reference to the
    ".." vnode.  This is cheaper storagewise than keeping it in the
    namecache, and it makes more sense since it's a 1:1 mapping.

2.  Also handle the case of "." more intelligently rather than stuff
    the namecache with pointless entries.

3.  Add two lists to the vnode and hang namecache entries which go from
    or to this vnode.  When cleaning a vnode, delete all namecache
    entries it invalidates.

4.  Never reuse namecache enties, malloc new ones when we need it, free
    old ones when they die.  No longer a hard limit on how many we can
    have.

5.  Remove the upper limit on namelength of namecache entries.

6.  Make a global list for negative namecache entries, limit their number
    to a sysctl'able (debug.ncnegfactor) fraction of the total namecache.
    Currently the default fraction is 1/16th.  (Suggestions for better
    default wanted!)

7.  Assign v_id correctly in the face of 32bit rollover.

8.  Remove the LRU list for namecache entries, not needed.  Remove the
    #ifdef NCH_STATISTICS stuff, it's not needed either.

9.  Use the vnode freelist as a true LRU list, also for namecache accesses.

10. Reuse vnodes more aggresively but also more selectively, if we can't
    reuse, malloc a new one.  There is no longer a hard limit on their
    number, they grow to the point where we don't reuse potentially
    usable vnodes.  A vnode will not get recycled if still has pages in
    core or if it is the source of namecache entries (Yes, this does
    indeed work :-)  "." and ".." are not namecache entries any longer...)

11. Do not overload the v_id field in namecache entries with whiteout
    information, use a char sized flags field instead, so we can get
    rid of the vpid and v_id fields from the namecache struct.  Since
    we're linked to the vnodes and purged when they're cleaned, we don't
    have to check the v_id any more.

12. NFS knew about the limitation on name length in the namecache, it
    shouldn't and doesn't now.

Bugs:
        The namecache statistics no longer includes the hits for ".."
        and "." hits.

Performance impact:
        Generally in the +/- 0.5% for "normal" workstations, but
        I hope this will allow the system to be selftuning over a
        bigger range of "special" applications.  The case where
        RAM is available but unused for cache because we don't have
        any vnodes should be gone.

Future work:
        Straighten out the namecache statistics.

        "desiredvnodes" is still used to (bogusly ?) size hash
        tables in the filesystems.

        I have still to find a way to safely free unused vnodes
        back so their number can shrink when not needed.

        There is a few uses of the v_id field left in the filesystems,
        scheduled for demolition at a later time.

        Maybe a one slot cache for unused namecache entries should
        be implemented to decrease the malloc/free frequency.

27 years agoUpdate our history a little (whoops - I thought I'd committed
Jordan K. Hubbard [Sun, 4 May 1997 08:11:24 +0000 (08:11 +0000)]
Update our history a little (whoops - I thought I'd committed
this days ago!).

27 years agoCommit hooks for ifmedia support. It's optional in the Makefile, and
Peter Wemm [Sun, 4 May 1997 06:27:45 +0000 (06:27 +0000)]
Commit hooks for ifmedia support.  It's optional in the Makefile, and
can be trivially disabled.

27 years agoBring in the ifmedia components of NetBSD's ifconfig as a seperate file.
Peter Wemm [Sun, 4 May 1997 06:22:27 +0000 (06:22 +0000)]
Bring in the ifmedia components of NetBSD's ifconfig as a seperate file.

Obtained from: Jason Thorpe via NetBSD

27 years agoSecond try at cleanups and fixes (without if_media stuff for the moment)
Peter Wemm [Sun, 4 May 1997 06:14:47 +0000 (06:14 +0000)]
Second try at cleanups and fixes (without if_media stuff for the moment)
- parse command options using getopt for consistancy
- sanitise the command parsing so that it's less like spaghetti
- implement a "-l" option (idea from NetBSD - just list names)
- attempt to clean up the sysctl parsing loop some more.  It still needs
  to be taken out the back and shot though.
- cut down on global usage, but there's a lot more scope for this.
- make usage string a bit closer to reality (it was missing lots of things)

Unfortunately, I did this for the second time but with the memory of
the NetBSD version still recently in my mind.  It's hard to redo simple
changes or getopt stuff without making it look like what you've been
working with a few hours ago.

27 years agoforce null commit to skip over rev 1.25 so there's less cvs trouble
Peter Wemm [Sun, 4 May 1997 06:00:27 +0000 (06:00 +0000)]
force null commit to skip over rev 1.25 so there's less cvs trouble

27 years agoUpdate the URL for the CTM mirror at the University of Trier.
John Polstra [Sun, 4 May 1997 05:55:56 +0000 (05:55 +0000)]
Update the URL for the CTM mirror at the University of Trier.

Submitted by: Sascha Blank <blank@fox.uni-trier.de>

27 years agoUse newly introduced pid file field now for ppp
Andrey A. Chernov [Sun, 4 May 1997 02:41:29 +0000 (02:41 +0000)]
Use newly introduced pid file field now for ppp

27 years agoLog each ppp line to separate /var/log/ppp.tunX.log instead mixing of
Andrey A. Chernov [Sun, 4 May 1997 02:39:04 +0000 (02:39 +0000)]
Log each ppp line to separate /var/log/ppp.tunX.log instead mixing of
all lines into single /var/log/ppp.log

27 years agoFinish off and activate the smp_active sysctl handler..
Peter Wemm [Sun, 4 May 1997 02:08:09 +0000 (02:08 +0000)]
Finish off and activate the smp_active sysctl handler..

27 years ago1) Add yet one optional field: pid file to send SIGHUP to if log rotated
Andrey A. Chernov [Sun, 4 May 1997 01:53:53 +0000 (01:53 +0000)]
1) Add yet one optional field: pid file to send SIGHUP to if log rotated
(good thing for apache f.e.) This change is backward compatible with old
newsyslog.conf files.

2) Do not compress log if SIGHUP sending failed for some reason
(f.e. pid file deleted). Newcoming messages will be lost
otherwise.

3) Misc cleanup while I am here.

27 years agoUpdate some of these release notes.
Jordan K. Hubbard [Sun, 4 May 1997 00:26:56 +0000 (00:26 +0000)]
Update some of these release notes.

27 years agoGrab some of the NetBSD text for describing the new options and the
Peter Wemm [Sat, 3 May 1997 23:03:56 +0000 (23:03 +0000)]
Grab some of the NetBSD text for describing the new options and the
media commands.

27 years agoAdd the Archive 4586NP tape changer.
Joerg Wunsch [Sat, 3 May 1997 22:23:13 +0000 (22:23 +0000)]
Add the Archive 4586NP tape changer.

PR: kern/3481
Submitted by: bruce@zuhause.mn.org (Bruce Albrecht)

27 years agoNobody ever seemed to be interested in reviewing these changes, and i
Joerg Wunsch [Sat, 3 May 1997 22:17:43 +0000 (22:17 +0000)]
Nobody ever seemed to be interested in reviewing these changes, and i
found that my syslogd is now running them for several months...

Add an option to syslogd to restrict the IP addresses that are allowed
to log to this syslogd.  It's too late to develop the inter-syslogd
communications protocol mentioned in the BUGS section, some 10 years
too late.  Thus, restricting the IP address range is about the most
effective change we can do if we want to allow incoming syslog
messages at all.

IMHO, we should encourage the system administrators to use this option,
and thus provide a knob in /etc/rc.* for it, defaulting to -a 127.0.0.1/32
(just as a hint about the usage).

Please state opinions about whether to merge this change into 2.2 or
not (i've got it running on RELENG_2_2 anyway).

27 years agoFix the umount problems for DEVFS.
Joerg Wunsch [Sat, 3 May 1997 21:19:54 +0000 (21:19 +0000)]
Fix the umount problems for DEVFS.

PR: 3276 & 3469 (the fixes), 2738, 2033 (reports)
Submitted by: Dmitrij Tejblum <dima@tejblum.dnttm.rssi.ru>

27 years agoadd SIOC{S,G}IFMEDIA ioctl support
Peter Wemm [Sat, 3 May 1997 21:07:13 +0000 (21:07 +0000)]
add SIOC{S,G}IFMEDIA ioctl support

27 years agoadd net/if_media.c
Peter Wemm [Sat, 3 May 1997 21:06:00 +0000 (21:06 +0000)]
add net/if_media.c

27 years agodefine SIOC{SG}IFMEDIA ioctls
Peter Wemm [Sat, 3 May 1997 21:05:03 +0000 (21:05 +0000)]
define SIOC{SG}IFMEDIA ioctls

Obtained from: NetBSD

27 years agoMake it compile on FreeBSD, add $Id$
Peter Wemm [Sat, 3 May 1997 20:58:10 +0000 (20:58 +0000)]
Make it compile on FreeBSD, add $Id$

27 years agoThis commit was generated by cvs2svn to compensate for changes in r25428,
Peter Wemm [Sat, 3 May 1997 20:54:36 +0000 (20:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r25428,
which included commits to RCS files with non-trunk default branches.

27 years agoBackend support for if_media interface ioctl's. This is for supporting
Peter Wemm [Sat, 3 May 1997 20:54:36 +0000 (20:54 +0000)]
Backend support for if_media interface ioctl's.  This is for supporting
the many-mode, many-media-option type ethernet cards, in particular, the
new if_de.

Obtained from: NetBSD  (Their implementation of BSDI's design)

27 years agoChange group ownership of /var/mail to mail and permission
Jordan K. Hubbard [Sat, 3 May 1997 20:15:15 +0000 (20:15 +0000)]
Change group ownership of /var/mail to mail and permission
to 0775.

This does *not* instantly make any program which "ensures"
mail spool consistency by creating lock files safe in any way
since other tools, like mail.local, will be using flock() semantics
and any such lock file will simply be ignored.  It does, however,
allow a lot of things which are currently suid root in order to create
such bogus lockfiles to, at least, be bogus at a much lower level of
privilege (and this is good).  Ultimately, of course, everybody should
just use flock.

27 years agosynchronize with bsd.doc.mk (long outstanding fix)
Andreas Klemm [Sat, 3 May 1997 19:58:39 +0000 (19:58 +0000)]
synchronize with bsd.doc.mk (long outstanding fix)
PR: closes docs/3479

27 years agodisabled checks for smp_active == 0.
Steve Passe [Sat, 3 May 1997 19:24:16 +0000 (19:24 +0000)]
disabled checks for smp_active == 0.
this was wasting precious cycles for no apparent (to me) reason.
it is currently bracketed by BOTHER_TO_CHECK, define to restore old behaviour.

27 years agocode to allow range checking on smp_active.
Steve Passe [Sat, 3 May 1997 18:24:25 +0000 (18:24 +0000)]
code to allow range checking on smp_active.
disabled by default, not sure its ready for prime time.

Submitted by:  Peter Wemm <peter@spinner.DIALix.COM>

27 years agoadded declaration for get_isa_apic_mask().
Steve Passe [Sat, 3 May 1997 18:05:31 +0000 (18:05 +0000)]
added declaration for get_isa_apic_mask().

Submitted by: "John S. Dyson" <toor@dyson.iquest.net>

27 years agoimproved io_apic_setup().
Steve Passe [Sat, 3 May 1997 18:01:56 +0000 (18:01 +0000)]
improved io_apic_setup().
deals with motherboards that map ISA IRQs to APIC IRQS above 15.

Submitted by: "John S. Dyson" <toor@dyson.iquest.net>

27 years agonew function to turn an APIC pin# into an INT mask.
Steve Passe [Sat, 3 May 1997 17:42:01 +0000 (17:42 +0000)]
new function to turn an APIC pin# into an INT mask.
added missing APIC_IO define.

Submitted by: "John S. Dyson" <toor@dyson.iquest.net>

27 years agoRecognize, but don't actually do much with a TI PCI1130 PcCard/CardBus bridge.
Poul-Henning Kamp [Sat, 3 May 1997 13:57:26 +0000 (13:57 +0000)]
Recognize, but don't actually do much with a TI PCI1130 PcCard/CardBus bridge.

27 years agoInitialize PCI/CardBus bridges.
Poul-Henning Kamp [Sat, 3 May 1997 13:52:29 +0000 (13:52 +0000)]
Initialize PCI/CardBus bridges.

Tested on: HP Omnibook 800 / TI PCI1130
Reviewed by: se

27 years agoMake nfs roots (diskless) functional again. It may still not be correct,
Poul-Henning Kamp [Sat, 3 May 1997 13:42:50 +0000 (13:42 +0000)]
Make nfs roots (diskless) functional again.  It may still not be correct,
but it is functional.

27 years agoSpelling/typo fix.
Daniel O'Callaghan [Sat, 3 May 1997 12:14:21 +0000 (12:14 +0000)]
Spelling/typo fix.

27 years agoUpdate the etc world from RELENG_2_2 which is now more up-to-date
Jordan K. Hubbard [Sat, 3 May 1997 11:22:17 +0000 (11:22 +0000)]
Update the etc world from RELENG_2_2 which is now more up-to-date
(gotta get myself -current again, this is a drag).

Also-fixes-problems-noted-by: Wolfgang Helbig & Joerg Wunsch
27 years agoBring in the Starter files for the contrib-crypto dir.
Mark Murray [Sat, 3 May 1997 09:16:07 +0000 (09:16 +0000)]
Bring in the Starter files for the contrib-crypto dir.

I am not going to commit anything to this area for a few days.
This is because
1) I want everyone to be DARN sure there is no export of crypto
   that may get our USA friends it trouble.
2) I have been asked by the folk developing KTH-eBones to hold off
   for their new release.

Worked with: rkw, jdp
CVS:
CVS:

27 years agobettter -> cleaner.
Jordan K. Hubbard [Sat, 3 May 1997 04:46:42 +0000 (04:46 +0000)]
bettter -> cleaner.

27 years agoNuke makefiles that duplicate those in libc now that libc is libc_r
John Birrell [Sat, 3 May 1997 04:05:57 +0000 (04:05 +0000)]
Nuke makefiles that duplicate those in libc now that libc is libc_r
aware. This simplifies things for libc_r.

27 years agoUse libc makefiles from now on instead of duplicating them in libc_r.
John Birrell [Sat, 3 May 1997 03:57:21 +0000 (03:57 +0000)]
Use libc makefiles from now on instead of duplicating them in libc_r.

Added Id strings too. It's useful to know who last made a change.

27 years agoChanged all paths to be relative to src/lib instead of src/lib/libc
John Birrell [Sat, 3 May 1997 03:50:06 +0000 (03:50 +0000)]
Changed all paths to be relative to src/lib instead of src/lib/libc
so that all these makefiles can be used to build libc_r too.

Added .if ${LIB} == "c" tests to restrict man page builds to libc
to avoid needlessly building them with libc_r too.

Split libc Makefile into Makefile and Makefile.inc to allow the
libc_r Makefile to include Makefile.inc too.

27 years agoLet gcc know about libc_r. Use -pthread to link against libc_r instead
John Birrell [Sat, 3 May 1997 02:58:54 +0000 (02:58 +0000)]
Let gcc know about libc_r. Use -pthread to link against libc_r instead
of libc.

27 years agoChange default build to include libc_r, but allow disadvantaged souls
John Birrell [Sat, 3 May 1997 02:51:59 +0000 (02:51 +0000)]
Change default build to include libc_r, but allow disadvantaged souls
to turn off the build.

27 years agoThis stub has not been required by libc for a long time. Nuke it.
John Birrell [Sat, 3 May 1997 02:47:52 +0000 (02:47 +0000)]
This stub has not been required by libc for a long time. Nuke it.

27 years agoFixed panic message in union_lock(): union_link --> union_lock.
KATO Takenori [Sat, 3 May 1997 01:55:19 +0000 (01:55 +0000)]
Fixed panic message in union_lock(): union_link --> union_lock.

27 years agoMinor typo.
Masafumi Max NAKANE [Fri, 2 May 1997 22:08:10 +0000 (22:08 +0000)]
Minor typo.
PR: 3472
Submitted by: Josh Gilliam <soil@quick.net>