]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
27 years agoBetter fix for the previous overflow problem using the code from
Nate Williams [Thu, 5 Dec 1996 03:14:14 +0000 (03:14 +0000)]
Better fix for the previous overflow problem using the code from
ogetrlimit().

Suggested by: bde

27 years agoAdded Miguel Angel Sagreras <msagre@cactus.fi.uba.ar> for his ibcs2 patches.
Nate Williams [Thu, 5 Dec 1996 01:47:33 +0000 (01:47 +0000)]
Added Miguel Angel Sagreras <msagre@cactus.fi.uba.ar> for his ibcs2 patches.

27 years ago'ulimit' can overflow the int return value, so avoid the overflow by
Nate Williams [Thu, 5 Dec 1996 01:45:08 +0000 (01:45 +0000)]
'ulimit' can overflow the int return value, so avoid the overflow by
subtracting one if ulimit returns -1.

Submitted by: Miguel Angel Sagreras <msagre@cactus.fi.uba.ar>

27 years agoFixed an off by 1 error in unix2dostime(). The first day of each month
Bruce Evans [Wed, 4 Dec 1996 13:42:09 +0000 (13:42 +0000)]
Fixed an off by 1 error in unix2dostime().  The first day of each month
was converted to the last day of the previous month.  This bug was
introduced in the optimizations in rev.1.4.

27 years agoBump maximum username length to 16 characters. This brings us into
Jordan K. Hubbard [Wed, 4 Dec 1996 11:31:14 +0000 (11:31 +0000)]
Bump maximum username length to 16 characters.  This brings us into
line with BSD/OS and Linux's username limits, making transitioning from
either operating system a lot easier than it is now.  I'm currently
running with this change on my system, as are several others, and have
experienced no ill effects.

This is not for 2.2!  This needs to get shaken out longer term in 3.0.

Previously-approved-by: davidg
27 years agoFix syntax error loc->doc
Andrey A. Chernov [Wed, 4 Dec 1996 09:42:39 +0000 (09:42 +0000)]
Fix syntax error loc->doc

27 years agoMerge changes made to the en version (1.160 -> 1.161).
Masafumi Max NAKANE [Wed, 4 Dec 1996 09:04:42 +0000 (09:04 +0000)]
Merge changes made to the en version (1.160 -> 1.161).

27 years agoFirst cut at adding a little bit about the ports framework's dir structure.
David E. O'Brien [Wed, 4 Dec 1996 08:49:52 +0000 (08:49 +0000)]
First cut at adding a little bit about the ports framework's dir structure.

27 years agoAdded Pedro Giffuni to the contributor's section for his two ports:
David E. O'Brien [Wed, 4 Dec 1996 08:07:44 +0000 (08:07 +0000)]
Added Pedro Giffuni to the contributor's section for his two ports:
CIM      Simula Compiler using C
Isearch  A text search Engine from CNIDR  (to be commited later)

27 years agoOptimize by changing "sort | uniq" -> "sort -u". "make index" is now
Satoshi Asami [Wed, 4 Dec 1996 05:53:17 +0000 (05:53 +0000)]
Optimize by changing "sort | uniq" -> "sort -u".  "make index" is now
412 secs instead of 498 secs on my machine!

Submitted by: fenner

27 years agoSyncronize.
Satoshi Asami [Wed, 4 Dec 1996 04:36:59 +0000 (04:36 +0000)]
Syncronize.

Submitted by: The FreeBSD(98) Development Team

27 years agoReplace sbic driver (WD33C93 SCSI card driver) with new bs driver.
Satoshi Asami [Wed, 4 Dec 1996 04:32:52 +0000 (04:32 +0000)]
Replace sbic driver (WD33C93 SCSI card driver) with new bs driver.

Submitted by: The FreeBSD(98) Development Team
Obtained from: NetBSD/pc98 based on NetBSD 1.2

27 years agoSynchronize with the RELENG_2_2 branch.
Satoshi Asami [Wed, 4 Dec 1996 04:21:30 +0000 (04:21 +0000)]
Synchronize with the RELENG_2_2 branch.

Definite 2.2 candidate.

Submitted by: The FreeBSD(98) Development Team

27 years agoRemove a self referential xref, and add an xref for modload.
Mike Pritchard [Wed, 4 Dec 1996 00:00:20 +0000 (00:00 +0000)]
Remove a self referential xref, and add an xref for modload.

Closes PR# 2104.

27 years agoUse the predefined entities for a couple of mailing lists.
Mike Pritchard [Tue, 3 Dec 1996 23:40:11 +0000 (23:40 +0000)]
Use the predefined entities for a couple of mailing lists.

27 years agoRemove some garbage text from the man page.
Mike Pritchard [Tue, 3 Dec 1996 23:26:49 +0000 (23:26 +0000)]
Remove some garbage text from the man page.
Closes PR# 1971.

Submitted by: Lydon Nerenberg

27 years agoThe semget man page uses the incorect symbol for alter
Mike Pritchard [Tue, 3 Dec 1996 23:03:22 +0000 (23:03 +0000)]
The semget man page uses the incorect symbol for alter
access.  Closes PR# 1712.

Submitted by: Kent Vander Velden

27 years agoMerge the changes in the en version (1.159 -> 1.160).
Masafumi Max NAKANE [Tue, 3 Dec 1996 22:50:06 +0000 (22:50 +0000)]
Merge the changes in the en version (1.159 -> 1.160).

27 years agoAdded support for '-T' and '-B' (text and binary) file recognition by
Nate Williams [Tue, 3 Dec 1996 21:56:15 +0000 (21:56 +0000)]
Added support for '-T' and '-B' (text and binary) file recognition by
peeking inside of Chris Torek's stdio library internals.  This is
similar to the code used for other systems, but didn't work on CT's new
implementation.

Submitted by: Gary Kline <kline@tera.com>

27 years agoAdded my 'ddial' patches to user-PPP. The new mode tries it's darndest
Nate Williams [Tue, 3 Dec 1996 21:38:52 +0000 (21:38 +0000)]
Added my 'ddial' patches to user-PPP.  The new mode tries it's darndest
to keep the link up, so it re-dials whenever it detects the link go
down.  This is useful for 'dedicated' links who use PPP.

It's been used for over a year w/out problems at different sites.

27 years agoAdd support for detecting and hopefully using the passwd.adjunct.byname
Bill Paul [Tue, 3 Dec 1996 17:55:49 +0000 (17:55 +0000)]
Add support for detecting and hopefully using the passwd.adjunct.byname
NIS map which is present on SunOS NIS servers with the SunOS C2 security
hack^Woption installed. I'm convinced that the C2 security option restricts
access to the passwd.adjunct.byname map in the same way that I restrict
access to the master.passwd.{byname,buid} maps (checking for reserved ports),
which means that we should be able to handle passwd.adjunct.byname map
correctly.

If _havemaster() doesn't find a master.passwd.byname map, it will now
test for a passwd.adjunct.byname map before defaulting back to the
standard non-shadowed passwd.{byname,byuid} maps. If _pw_breakout_yp()
sees that the adjunct map was found and the password from the standard
maps starts with ##, it will try to grab the correct password field
from the adjunct map. As with the master.passwd maps, this only happens
if the caller is root, so the shadowing feature is preserved; non-root
users just get back ##username as the encrypted password.

Note that all we do is grab the second field from the passwd.adjunct.byname
entry, which is designated to be the real encrypted password. There are
other auditing fields in the entry but they aren't of much use to us.

Also switched back to using yp_order() to probe for the maps (instead
of yp_first()). The original problem with yp_order() was that it barfed
with NIS+ servers in YP compat mode since they don't support the
YPPROC_ORDER procedure. This condition is handled a bit more gracefully
in yplib now: we can detect the error and just punt on the probing.

27 years agoClear the busfree interrupt when one occurs, after a SELTO, or a bus reset.
Justin T. Gibbs [Tue, 3 Dec 1996 17:06:00 +0000 (17:06 +0000)]
Clear the busfree interrupt when one occurs, after a SELTO, or a bus reset.

27 years agoRe-enable reselections as the first thing after we see a bus free. We seem
Justin T. Gibbs [Tue, 3 Dec 1996 17:04:51 +0000 (17:04 +0000)]
Re-enable reselections as the first thing after we see a bus free.  We seem
to miss reselections from some devices and since the reselection response
timeout is only 200ns, enabling reselections too late may be the cause of our
problem.

27 years agoPR kern/2141:
Poul-Henning Kamp [Tue, 3 Dec 1996 16:08:00 +0000 (16:08 +0000)]
PR kern/2141:

        function ed_attach_NE2000_pci() in if_ed.c passes
        an uninitialized block of memory (got with malloc())
        to ed_attach. This prevents a proper initialization
        of the device descriptor and in my case causes a panic
        during the probe, while printing out device info.

Reviewed by: phk
Submitted by: Luigi Rizzo <luigi@iet.unipi.it>

27 years agoAdded FURUSAWA Kazuhisa for his port of xloadface.
David E. O'Brien [Tue, 3 Dec 1996 13:38:40 +0000 (13:38 +0000)]
Added FURUSAWA Kazuhisa for his port of xloadface.

27 years agoRemove file that we don't have any way of using. We don't have the other
Peter Wemm [Tue, 3 Dec 1996 06:21:49 +0000 (06:21 +0000)]
Remove file that we don't have any way of using.  We don't have the other
Makefiles for other OS's in our tree that this uses when not using bmake.

27 years agoRemove file no longer shipped with sendmail
Peter Wemm [Tue, 3 Dec 1996 06:20:18 +0000 (06:20 +0000)]
Remove file no longer shipped with sendmail

27 years agoRemove defunct files that are no longer shipped (and don't work with this
Peter Wemm [Tue, 3 Dec 1996 06:19:51 +0000 (06:19 +0000)]
Remove defunct files that are no longer shipped (and don't work with this
version of sendmail any more)

27 years agoBack out rev 1.6, it's in the official source now (a few lines up)
Peter Wemm [Tue, 3 Dec 1996 06:18:11 +0000 (06:18 +0000)]
Back out rev 1.6, it's in the official source now (a few lines up)

27 years agoMerge 8.8.3->8.8.4 changes onto our mainline where we've edited the
Peter Wemm [Tue, 3 Dec 1996 06:15:52 +0000 (06:15 +0000)]
Merge 8.8.3->8.8.4 changes onto our mainline where we've edited the
file at some point in the past.

Obtained from: Eric Allman <eric@sendmail.org>

27 years agoThis commit was generated by cvs2svn to compensate for changes in r20103,
Peter Wemm [Tue, 3 Dec 1996 06:07:35 +0000 (06:07 +0000)]
This commit was generated by cvs2svn to compensate for changes in r20103,
which included commits to RCS files with non-trunk default branches.

27 years agoImport sendmail 8.8.4. See RELEASE_NOTES for changes.
Peter Wemm [Tue, 3 Dec 1996 06:07:35 +0000 (06:07 +0000)]
Import sendmail 8.8.4.  See RELEASE_NOTES for changes.

27 years agoUpdate the bootblocks to use the new definitions that changed in the
Nate Williams [Tue, 3 Dec 1996 03:32:15 +0000 (03:32 +0000)]
Update the bootblocks to use the new definitions that changed in the
recent psm/syscons/kbdio code update.

Submitted by: Kazutaka YOKOTA <yokota@zodiac.mech.utsunomiya-u.ac.jp>

27 years agoAdd IP_OPTIONS and the multicast-related setsockopts to the
Bill Fenner [Tue, 3 Dec 1996 02:52:29 +0000 (02:52 +0000)]
Add IP_OPTIONS and the multicast-related setsockopts to the
list of IP setsockopts the Linux emulator recognizes.

Explicitly disallow IP_HDRINCL since Linux's handling of
raw output is different than BSD's.

Closes PR#kern/2111.

Submitted by: y-nakaga@ccs.mt.nec.co.jp (Yoshihisa NAKAGAWA)

27 years agoBack out the non-forking YPPROC_ALL stuff. Whatever drugs I was doing
Bill Paul [Tue, 3 Dec 1996 02:37:39 +0000 (02:37 +0000)]
Back out the non-forking YPPROC_ALL stuff. Whatever drugs I was doing
when I came up with this idea weren't strong enough to help me see it
through. If this was a self-contained application and I had complete
control over what data got sent through what socket and when, I might
be able to get everything to work right without blocking, but instead
I have RPC/XDR in between me and the socket layer, and they have their
own ideas about what to do.

Maybe one day I'll go totally mad and figure out the right way to do
this; in the meantime this mess goes on the back burner.

27 years agoFix math in SIGINFO printer.
Bill Fenner [Mon, 2 Dec 1996 21:42:11 +0000 (21:42 +0000)]
Fix math in SIGINFO printer.

27 years ago2 small changes:
Julian Elischer [Mon, 2 Dec 1996 21:07:33 +0000 (21:07 +0000)]
2 small changes:
1/ increase the tun MTU from 1500 to 1600 to allow it to be used with
packets formatted according to RFC1490 and RFC1717

2/ allow the tsleep() when reading, to be interruptable by signals
so that one can now do:
od -xc </dev/tun0
to dump packets for debugging without getting hung.

Passed on by: Archie@whistle.com (archie Cobbs)

Nice but not neccessary in 2.2

27 years agoupgrade STANDARDS from POSIX 1003.1-88 to 1003.1-90 using .St macro
Wolfram Schneider [Mon, 2 Dec 1996 20:03:58 +0000 (20:03 +0000)]
upgrade STANDARDS from POSIX 1003.1-88 to 1003.1-90 using .St macro
use ``is expected to conform to'' phrase, not ``conforms to''

Pointed out by: Bruce->NIST-PCTS

27 years agoSome imporvemnets to the vx driver.
Guido van Rooij [Mon, 2 Dec 1996 18:44:31 +0000 (18:44 +0000)]
Some imporvemnets to the vx driver.
1. 'connector_table' is shortened to 'conn_tab'.
2. More reliable connector change code.
3. Display message like "vx0: selected bnc. (link1)"
   when the connector changed by link[012].
4. Handle MII properly.
5. Potentially slightly better performance.
6. Fixed a silly typo.

Submitted by: Naoki Hamada <nao@sbl.cl.nec.co.jp>

27 years agoSome imporvemnets to the vx driver.
Guido van Rooij [Mon, 2 Dec 1996 18:38:37 +0000 (18:38 +0000)]
Some imporvemnets to the vx driver.
1. 'connector_table' is shortened to 'conn_tab'.
2. More reliable connector change code.
3. Display message like "vx0: selected bnc. (link1)"
   when the connector changed by link[012].
4. Handle MII properly.
5. Potentially slightly better performance.
6. Fixed a silly typo.

Submitted by: Naoki Hamada <nao@sbl.cl.nec.co.jp>

27 years agoMerge changes from vendor branch.
Garrett Wollman [Mon, 2 Dec 1996 17:14:48 +0000 (17:14 +0000)]
Merge changes from vendor branch.

27 years agoThis commit was generated by cvs2svn to compensate for changes in r20091,
Garrett Wollman [Mon, 2 Dec 1996 16:59:56 +0000 (16:59 +0000)]
This commit was generated by cvs2svn to compensate for changes in r20091,
which included commits to RCS files with non-trunk default branches.

27 years ago1996 `L' release of timezone data files. Includes corrections for historical
Garrett Wollman [Mon, 2 Dec 1996 16:59:56 +0000 (16:59 +0000)]
1996 `L' release of timezone data files.  Includes corrections for historical
data in several locations.

Obtained from: ado@elsie.nci.nih.gov, ftp://elsie.nci.nih.gov/pub

27 years agoFix a typo in the man page (the "-n" flag was added to the synopsis in
Bill Fenner [Mon, 2 Dec 1996 16:21:20 +0000 (16:21 +0000)]
Fix a typo in the man page (the "-n" flag was added to the synopsis in
a comment).

27 years agoSome more cosmetic tweaks.
Masafumi Max NAKANE [Mon, 2 Dec 1996 14:21:34 +0000 (14:21 +0000)]
Some more cosmetic tweaks.

Submitted by: simokawa@jp.freebsd.org

27 years agoMerge the changes made to the en version (1.38 ->1.44).
Masafumi Max NAKANE [Mon, 2 Dec 1996 13:49:26 +0000 (13:49 +0000)]
Merge the changes made to the en version (1.38 ->1.44).

Cosmetic tweaks.

Submitted by: simokawa@jp.freebsd.org

27 years agoMerged change made to the en version (1.11 -> 1.12).
Masafumi Max NAKANE [Mon, 2 Dec 1996 13:18:03 +0000 (13:18 +0000)]
Merged change made to the en version (1.11 -> 1.12).

27 years agoNow that lprps is part of the ports collection, changed some statement about
Masafumi Max NAKANE [Mon, 2 Dec 1996 13:11:42 +0000 (13:11 +0000)]
Now that lprps is part of the ports collection, changed some statement about
lprps.

27 years agoI have added a '-n' flag to the watch(8) command. This option
Jordan K. Hubbard [Mon, 2 Dec 1996 12:32:46 +0000 (12:32 +0000)]
I have added a '-n' flag to the watch(8) command.  This option
disables the ability to interactively select a new tty.  I have also
removed a check for uid == 0 because it gets in the way of using suid
mode based access control.  Watch (8)is only runnable by root, so this
does not really change things much.

Closes PR#2131

Submitted-By: adrian@virginia.edu
27 years agoAllow baud rate of 230400.
Jordan K. Hubbard [Mon, 2 Dec 1996 12:12:19 +0000 (12:12 +0000)]
Allow baud rate of 230400.
Submitted-By: Andrew L.Davydov <davydov@ns.okbmei.msk.su>
27 years agoAdd fr_* family
Andrey A. Chernov [Mon, 2 Dec 1996 12:00:46 +0000 (12:00 +0000)]
Add fr_* family
Submitted by: roberto@keltia.freenix.fr (Ollivier Robert)

27 years agoMerge the change made to the en version (1.59 -> 1.61.)
Masafumi Max NAKANE [Mon, 2 Dec 1996 08:00:57 +0000 (08:00 +0000)]
Merge the change made to the en version (1.59 -> 1.61.)

Submitted by: Takashi Saeki <saeki@saskg.semicon.sony.co.jp>

27 years agoMerge the change to the en version (1.17 -> 1.18.)
Masafumi Max NAKANE [Mon, 2 Dec 1996 07:56:12 +0000 (07:56 +0000)]
Merge the change to the en version (1.17 -> 1.18.)

27 years agoAdded new files to SRCS.
Masafumi Max NAKANE [Mon, 2 Dec 1996 07:25:53 +0000 (07:25 +0000)]
Added new files to SRCS.

27 years agoFiles for sections recently added to the English handbook.
Masafumi Max NAKANE [Mon, 2 Dec 1996 07:23:47 +0000 (07:23 +0000)]
Files for sections recently added to the English handbook.

The translation for these files is in progress, so I only put the header
and leave the contents empty.

27 years agoDisplay multicast addresses if the "a" flag is used in combination
Bill Fenner [Mon, 2 Dec 1996 06:38:30 +0000 (06:38 +0000)]
Display multicast addresses if the "a" flag is used in combination
with "-i".  This was already in the man page but was not implemented.

Obtained from: Multicast release 3.5

27 years agoDon't run routed by default.
Jordan K. Hubbard [Mon, 2 Dec 1996 05:01:02 +0000 (05:01 +0000)]
Don't run routed by default.
Virtually-Demanded-At-Gunpoint-By: joerg
27 years agoMerge change in the en version (1.158->1.159) to jp version.
Masafumi Max NAKANE [Sun, 1 Dec 1996 21:11:27 +0000 (21:11 +0000)]
Merge change in the en version (1.158->1.159) to jp version.

27 years agoAdded Koichi Sato and Yoshiaki Suzuki for their extensive help on making
Masafumi Max NAKANE [Sun, 1 Dec 1996 21:04:12 +0000 (21:04 +0000)]
Added Koichi Sato and Yoshiaki Suzuki for their extensive help on making
port of CAP.

27 years agoMerge the change in the English version (1.39->1.40) into jp version.
Masafumi Max NAKANE [Sun, 1 Dec 1996 20:32:34 +0000 (20:32 +0000)]
Merge the change in the English version (1.39->1.40) into jp version.

27 years agoAlot of fixes from kazu:
Søren Schmidt [Sun, 1 Dec 1996 19:05:50 +0000 (19:05 +0000)]
Alot of fixes from kazu:

1. All the suggestions earlier made by Bruce: renaming some symbols,
stricter error checking, removing redundant code, etc.

2. The `psm' driver preserves the default counter resolution and
report rate, whatever they are after reset. (Based on reports and
suggestion from Nate and Rob Bolin).

3. The `psm' driver now does not check the so-called sync. bit in the
first byte of the data packet by default, so that the tapping feature
of ALPUS GlidePoint works (based on reports from Louis Mamakos). I
tested the code with ALPUS Desktop GlidePoint (M/N GP101) and found
no problem; tapping worked. It appears ALPUS produces several models
of GlidePoint. I hope the other models are OK too.

The check code can still be activated by defining the PSM_CHECKSYNC
option in the config file. (The bit checking slightly reduces, if not
completely eliminates, weird mouse behavior cased by unsynchronized
mouse data packets. It also helps us to detect if the mouse interrupt
can ever be lost. But, well, if there are devices which cannot be
supported this way...)

4. The `psm' driver does not include the protocol emulation code by
default. The code can still be compiled in if the PSM_EMULATION option
is specified in the config file. Louis Mamakos suggests the emulation
code is putting too much in the kernel, and `moused' works well.
I will think about this later and decide if the entire emulation
code should be removed.

5. And, of course, the fix in `scprobe()' from Bruce to cure the
UserConfig problem. My code in `kbdio.c' is slightly different from
his patch, but has the same effect. There still is a possibility that
`scprobe()' gets confused, if, for whatever reasons, the user holds
down a key for very long time during the boot process. But we cannot
cope with everything, can we?

Submitted by: Kazutaka YOKOTA (yokota@zodiac.mech.utsunomiya-u.ac.jp)

27 years agoSimplify the disklabel trickery, and thus avoid the dreaded
Joerg Wunsch [Sun, 1 Dec 1996 17:19:29 +0000 (17:19 +0000)]
Simplify the disklabel trickery, and thus avoid the dreaded
``Operation not supported by device''.

Requested by: jkh

27 years agoRemoved all references to b_cylinder (aka b_cylin). It was evil and
Bruce Evans [Sun, 1 Dec 1996 16:34:41 +0000 (16:34 +0000)]
Removed all references to b_cylinder (aka b_cylin).  It was evil and
hasn't been used for a year or two since disksort() started sorting
on b_pblkno.

27 years agoDon't allow empty pathnames. POSIX standard.
Bruce Evans [Sun, 1 Dec 1996 16:05:44 +0000 (16:05 +0000)]
Don't allow empty pathnames.  POSIX standard.

Most of the standard utilities that depended on (or were broken in
a different way by) the old behaviour of interpreting "" as "."
were fixed a year or two ago.  There is still a fairly harmless
bug in tar and a harmless bug in gzip.  Tar apparently replaces
"/" by "" when it strips leading slashes.

27 years agoFix a problem with the new buffer_map management code. Additionally,
John Dyson [Sun, 1 Dec 1996 15:46:40 +0000 (15:46 +0000)]
Fix a problem with the new buffer_map management code.  Additionally,
decrease the size of buffer_map to approx 2/3 of what it used to be
(buffer_map can be smaller now.)  The original commit of these changes
increased the size of buffer_map to the point where the system would
not boot on large systems -- now large systems with large caches will
have even less problems than before.

27 years agoupdate E-Mail address of Andrew Webster
Wolfram Schneider [Sun, 1 Dec 1996 13:49:59 +0000 (13:49 +0000)]
update E-Mail address of Andrew Webster

27 years agoReflect change to the English version (1.13 -> 1.14).
Masafumi Max NAKANE [Sun, 1 Dec 1996 12:38:41 +0000 (12:38 +0000)]
Reflect change to the English version (1.13 -> 1.14).

27 years agoMove configResolv() to *after* the chroot(). This is what was causing
Jordan K. Hubbard [Sun, 1 Dec 1996 12:27:36 +0000 (12:27 +0000)]
Move configResolv() to *after* the chroot().  This is what was causing
NFS installation to fail the first time.  This will go into 2.2.

27 years agoSync the Original revision: line to the English version in -current (1.7->1.8).
Masafumi Max NAKANE [Sun, 1 Dec 1996 11:46:09 +0000 (11:46 +0000)]
Sync the Original revision: line to the English version in -current (1.7->1.8).
(No need to change the content.)

27 years agoMerge the change to the English version (1.157 -> 1.158).
Masafumi Max NAKANE [Sun, 1 Dec 1996 11:39:07 +0000 (11:39 +0000)]
Merge the change to the English version (1.157 -> 1.158).

27 years agoThis update adds the support for != 512 byte sector SCSI devices to
Søren Schmidt [Sun, 1 Dec 1996 11:25:38 +0000 (11:25 +0000)]
This update adds the support for != 512 byte sector SCSI devices to
the sd & od drivers. There is also slight changes to fdisk & newfs
in order to comply with different sectorsizes.
Currently sectors of size 512, 1024 & 2048 are supported, the only
restriction beeing in fdisk, which hunts for the sectorsize of
the device.
This is based on patches to od.c and the other system files by
John Gumb & Barry Scott, minor changes and the sd.c patches by
me.
There also exist some patches for the msdos filesys code, but I
havn't been able to test those (yet).

John Gumb (john@talisker.demon.co.uk)
Barry Scott (barry@scottb.demon.co.uk)

27 years agoCosmetic code cleanup from Matt's latest driver.
Rodney W. Grimes [Sun, 1 Dec 1996 06:01:00 +0000 (06:01 +0000)]
Cosmetic code cleanup from Matt's latest driver.

a)  Removal of private typedefs tulip_uint*_t, use standard u_int_*_t.

b)  Change [Dd][Cc]21.4. to just 21.4., seems Dec has done this to all
    of the drivers for all OS's.  (Did they get in trouble with someone?)
    [The few that remain can either not be eliminated, or are waiting for
    additional driver functional changes that will remove them.]

c)  Move some code from dc21040.h into the driver, later a whole block of that
    code and more will move to devar.h, but for now this makes it easier
    to study diffs.

d)  Add a big bold comment to the README.de file about it not reflecting
    reality anymore.

Note that these are all cosmetic changes and should be no functional
change in the driver whatsoever.  If _anyone_ spots a problem introduced
by this please let me know ASAP!

27 years agoCorrect some spelling errors.
Mike Pritchard [Sun, 1 Dec 1996 00:25:06 +0000 (00:25 +0000)]
Correct some spelling errors.

27 years agoUpdate getdomainname(3) to reflect that it operates on
Mike Pritchard [Sun, 1 Dec 1996 00:10:28 +0000 (00:10 +0000)]
Update getdomainname(3) to reflect that it operates on
domain names and not host names.

Pointed out by:  bde
Obtained from:  NetBSD

27 years agoFix some spelling errors.
Mike Pritchard [Sat, 30 Nov 1996 23:51:49 +0000 (23:51 +0000)]
Fix some spelling errors.

27 years agoFix some spelling errors.
Mike Pritchard [Sat, 30 Nov 1996 23:35:43 +0000 (23:35 +0000)]
Fix some spelling errors.

27 years agoFix an incorrect man page cross reference.
Mike Pritchard [Sat, 30 Nov 1996 23:24:20 +0000 (23:24 +0000)]
Fix an incorrect man page cross reference.

27 years agoImplement a new totally dynamic (up to MAXPHYS) buffer kva allocation
John Dyson [Sat, 30 Nov 1996 22:41:49 +0000 (22:41 +0000)]
Implement a new totally dynamic (up to MAXPHYS) buffer kva allocation
scheme.  Additionally, add the capability for checking for unexpected
kernel page faults.  The maximum amount of kva space for buffers hasn't
been decreased from where it is, but it will now be possible to do so.

This scheme manages the kva space similar to the buffers themselves.  If
there isn't enough kva space because of usage or fragementation, buffers
will be reclaimed until a buffer allocation is successful.  This scheme
should be very resistant to fragmentation problems until/if the LFS code
is fixed and uses the bogus buffer locking scheme -- but a 'fixed' LFS
is not likely to use such a scheme.

Now there should be NO problem allocating buffers up to MAXPHYS.

27 years agoThis commit changes the YPPROC_ALL procecdure so that it handles requests
Bill Paul [Sat, 30 Nov 1996 22:38:44 +0000 (22:38 +0000)]
This commit changes the YPPROC_ALL procecdure so that it handles requests
_without_ using fork().

The problem with YPPROC_ALL is that it transmits an entire map through
a TCP pipe as the result of a single RPC call. First of all, this requires
certain hackery in the XDR filter. Second, if the map being sent is
large, the server can end up spending lots of time in the XDR filter
sending to just the one client, while requests for other clients will
go unanswered.

My original solution for this was to fork() the request into a child
process which terminates after the map has been transmitted (or the
transfer is interrupted due to an error). This leaves the parent free
to handle other requests. But this solution is kind of lame: fork()
is relatively expensive, and we have to keep a cap on the number of
child processes to keep from swamping the system.

What we do now is grab control of the service transport handle and XDR
handle from the RPC library and send the records one at a time ourselves
instead of letting the RPC library do it. We send a record, then go
back to the svc_run() loop and select() on the socket. If select() says
we can still write data, we send the next record. Then we call
svc_getreqset() and handle other RPCs and loop around again. This way,
we can handle other RPCs between records.

We manage multiple YPPROC_ALL requests using a circular queue. When a
request is done, we dequeue it and destroy the handle. We also tag
each request with a ttl which is decremented whevever we run the queue
and a handle isn't serviced. This lets us nuke requests that have sat
idle for too long (if we didn't do this, we might run out of socket
descriptors.)

Now all I have to do is come up with an async resolver, and ypserv
won't need to fork() at all. :)

Note: these changes should not go into 2.2 unless they get a very
throrough shakedown before the final cutoff date.

27 years agoTruncate the file when opening it with write intent. Otherwise,
Joerg Wunsch [Sat, 30 Nov 1996 20:59:32 +0000 (20:59 +0000)]
Truncate the file when opening it with write intent.  Otherwise,
there's a good chance that garbage will remain at the end.

Closes PR # bin/2112: tftpd doesn't truncate ...

Reviewed by: fenner

27 years agoCorrecting Mark Huizer's address.
Chuck Robey [Sat, 30 Nov 1996 18:05:48 +0000 (18:05 +0000)]
Correcting Mark Huizer's address.
Obtained from: Mark, of course.

27 years agoMade the synopsis in the man page conform to the style guide.
Bruce Evans [Sat, 30 Nov 1996 16:12:46 +0000 (16:12 +0000)]
Made the synopsis in the man page conform to the style guide.

Made the usage message conform to the style guide.

Don't use the implementation variable `__progname'.

27 years agoFixed input of BREAKs when IGNPAR is set and IGNBRK is not set. BREAKs
Bruce Evans [Sat, 30 Nov 1996 15:52:56 +0000 (15:52 +0000)]
Fixed input of BREAKs when IGNPAR is set and IGNBRK is not set.  BREAKs
are always together with Framing Errors and they were incorrectly
treated as FE's and discarded.

Reorganized the BREAK/FE/PE tests.

Found by: NIST-PCTS

27 years agoReset h/w fifos (if any) in siostop(). Now ttyflush() works fairly well
Bruce Evans [Sat, 30 Nov 1996 15:29:31 +0000 (15:29 +0000)]
Reset h/w fifos (if any) in siostop().  Now ttyflush() works fairly well
with sio devices (not perfectly, since there is no way to flush the tx
holding register on 8250-16450's.  I'm not sure if resetting the fifos
flushes the tx shift register).

Reminded by: NIST-PCTS

27 years agoFixed sloppy clearing of TS_BUSY. Don't clear it until the transmitter
Bruce Evans [Sat, 30 Nov 1996 15:19:19 +0000 (15:19 +0000)]
Fixed sloppy clearing of TS_BUSY.  Don't clear it until the transmitter
is completely empty.  There is no interrupt for output completion, so
poll for it every 10 ms after output is nearly complete.  Now ttywait()
works right.

Reminded by: NIST-PCTS

27 years agoThrow away input if CREAD is not set. POSIX requires no characters to be
Bruce Evans [Sat, 30 Nov 1996 15:03:05 +0000 (15:03 +0000)]
Throw away input if CREAD is not set.  POSIX requires no characters to be
received in this case even if the hardware doesn't have a CREAD bit.

Found by: NIST-PCTS

27 years agoCleaned up CONSPEED changes. `comdefaultrate' gives the default
Bruce Evans [Sat, 30 Nov 1996 14:51:04 +0000 (14:51 +0000)]
Cleaned up CONSPEED changes.  `comdefaultrate' gives the default
speed for the "com" console, not for general purpose "com" ports,
so there was no need to split it into comdefaultrate and condefaultrate.

27 years agoReenabled i586-optimized copyin/out.
Bruce Evans [Sat, 30 Nov 1996 14:08:44 +0000 (14:08 +0000)]
Reenabled i586-optimized copyin/out.

Should be in 2.2.  Don't put it there for a while.

27 years agoadd flag to allow only anonymous ftp logins
Torsten Blum [Sat, 30 Nov 1996 12:00:25 +0000 (12:00 +0000)]
add flag to allow only anonymous ftp logins

Reviewed by: pst

27 years agoMerge from RELENG_2_2 (duh). See rev 1.227.2.4 for details.
Satoshi Asami [Sat, 30 Nov 1996 10:31:50 +0000 (10:31 +0000)]
Merge from RELENG_2_2 (duh).  See rev 1.227.2.4 for details.

27 years agoAdd the Quantum DLT drives. All of them grok SC_MORE_LUS, so the tape
Joerg Wunsch [Sat, 30 Nov 1996 07:39:37 +0000 (07:39 +0000)]
Add the Quantum DLT drives.  All of them grok SC_MORE_LUS, so the tape
library devices will properly come up with all LUNs detected.

Submitted by: dawes@Physics.usyd.edu.au (David Dawes)

27 years agoAdd en_GB
Andrey A. Chernov [Sat, 30 Nov 1996 05:35:49 +0000 (05:35 +0000)]
Add en_GB
Submitted by: mark@linus.demon.co.uk (Mark Valentine)

27 years agoWe don't offer the LaTeX or ASCII version of the handbook on the FTP sites
Masafumi Max NAKANE [Sat, 30 Nov 1996 01:00:11 +0000 (01:00 +0000)]
We don't offer the LaTeX or ASCII version of the handbook on the FTP sites
anymore.  Plain text (latin1, PostScript and HTML are there.

Pointed-Out by: Mitsuru IWASAKI <iwasaki@pc.jaring.my>

27 years agorename obsolete 'TZ=US/Pacific' example to 'TZ=America/Los_Angeles'
Wolfram Schneider [Sat, 30 Nov 1996 00:29:51 +0000 (00:29 +0000)]
rename obsolete 'TZ=US/Pacific' example to 'TZ=America/Los_Angeles'

close PR #2125

27 years agoChange boolean nature of "router" variable since it's not a boolean
Jordan K. Hubbard [Fri, 29 Nov 1996 23:52:20 +0000 (23:52 +0000)]
Change boolean nature of "router" variable since it's not a boolean
anymore.
Noticed-By: joerg
27 years agoAdd PT locale
Andrey A. Chernov [Fri, 29 Nov 1996 21:01:29 +0000 (21:01 +0000)]
Add PT locale
Submitted by:  Paulo Menezes <paulo@isr.uc.pt>

27 years agoSyntax tweak.
John Fieber [Fri, 29 Nov 1996 20:04:11 +0000 (20:04 +0000)]
Syntax tweak.
The breakage is actually a subtle bug in the DTD, but fixing it may
have some other ugly implications that need to be explored first.

27 years agoAdd da_DK timedef.
Poul-Henning Kamp [Fri, 29 Nov 1996 19:28:27 +0000 (19:28 +0000)]
Add da_DK timedef.

27 years agoCheck for error return from uiomove to prevent looping endlessly in
David Greenman [Fri, 29 Nov 1996 19:03:42 +0000 (19:03 +0000)]
Check for error return from uiomove to prevent looping endlessly in
soreceive(). Closes PR#2114.

Submitted by: wpaul

27 years agoFixed sigaction() for SIGKILL and SIGSTOP. Reading the old action now
Bruce Evans [Fri, 29 Nov 1996 18:01:55 +0000 (18:01 +0000)]
Fixed sigaction() for SIGKILL and SIGSTOP.  Reading the old action now
succeeds.  Writing an action now succeeds iff the handler isn't changed.
(POSIX allows attempts to change the handler to be ignored or cause an
error.  Changing other parts of the action is allowed (except attempts
to mask unmaskable signals are silently ignored as usual).)

Found by: NIST-PCTS