]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
29 years agoChanged order of release of vnode/object to fix a problem where the vnode
David Greenman [Mon, 6 Feb 1995 02:20:40 +0000 (02:20 +0000)]
Changed order of release of vnode/object to fix a problem where the vnode
is freed with an old object still attached (subsequently causing a panic).
Fixes NFS server panic "object/pager mismatch".

Submitted by: John Dyson

29 years agoCorrect a bug in the P function: according to 7th edition UNIX (tm), the
Jean-Marc Zucconi [Mon, 6 Feb 1995 00:14:50 +0000 (00:14 +0000)]
Correct a bug in the P function: according to 7th edition UNIX (tm), the
newline must be part of the pattern space i.e. `echo a|sed -e P' must print
a
a
and not
aa
This is consistent with gnu sed, SunOS, Ultrix (and probably others!)

29 years agoFixed the annoing bug that caused writes to /dev/console to go to
Søren Schmidt [Sun, 5 Feb 1995 23:44:25 +0000 (23:44 +0000)]
Fixed the annoing bug that caused writes to /dev/console to go to
the active virtual terminal (Boy, do I have little hair left).
Misc other fixes, try to enable cursor earlier in the game.

29 years agoAdd the ports-base target and correct two typos (hosbase!=hostbase)
Gary Palmer [Sun, 5 Feb 1995 23:31:25 +0000 (23:31 +0000)]
Add the ports-base target and correct two typos (hosbase!=hostbase)

29 years agoWrote manual page for yppush and edited Makefile to install it.
Bill Paul [Sun, 5 Feb 1995 21:48:04 +0000 (21:48 +0000)]
Wrote manual page for yppush and edited Makefile to install it.
Added conditional definition for _PATH_YP in yppush.c in case _PATH_YP
isn't defined anywhere else.

29 years agoRemoved the name length limitations on the device-driver names entirely.
Poul-Henning Kamp [Sun, 5 Feb 1995 21:15:21 +0000 (21:15 +0000)]
Removed the name length limitations on the device-driver names entirely.
Now device-names can be more than 3 chars.

29 years agoAdded "panic" command to ddb, so we can do something sensible even if
Poul-Henning Kamp [Sun, 5 Feb 1995 21:09:04 +0000 (21:09 +0000)]
Added "panic" command to ddb, so we can do something sensible even if
we have no symbols.

29 years agoDocument the getenv(PWD) feature.
Poul-Henning Kamp [Sun, 5 Feb 1995 18:14:38 +0000 (18:14 +0000)]
Document the getenv(PWD) feature.

29 years agoAllow zero as value for certain arguments to indicate "take from disklabel".
Poul-Henning Kamp [Sun, 5 Feb 1995 18:03:37 +0000 (18:03 +0000)]
Allow zero as value for certain arguments to indicate "take from disklabel".

29 years agoPCVT userland utilities.
Jordan K. Hubbard [Sun, 5 Feb 1995 13:49:25 +0000 (13:49 +0000)]
PCVT userland utilities.
Submitted by: hm

29 years agoThis was wrong - PCVT and syscons don't share the same entrypoint
Jordan K. Hubbard [Sun, 5 Feb 1995 11:29:38 +0000 (11:29 +0000)]
This was wrong - PCVT and syscons don't share the same entrypoint
names.
Submitted by: mh

29 years agoAdd pcvt termcap entries (quite a lot of them - whew!)
Jordan K. Hubbard [Sun, 5 Feb 1995 11:17:03 +0000 (11:17 +0000)]
Add pcvt termcap entries (quite a lot of them - whew!)
Submitted by: hm

29 years agoAdd man page for pcvt console driver.
Jordan K. Hubbard [Sun, 5 Feb 1995 11:16:33 +0000 (11:16 +0000)]
Add man page for pcvt console driver.
Submitted by: hm

29 years agoAdd pcvt's header files.
Jordan K. Hubbard [Sun, 5 Feb 1995 11:02:45 +0000 (11:02 +0000)]
Add pcvt's header files.
Submitted by: hm

29 years agoBring pcvt console driver into the tree - now users can chose whichever
Jordan K. Hubbard [Sun, 5 Feb 1995 11:01:24 +0000 (11:01 +0000)]
Bring pcvt console driver into the tree - now users can chose whichever
console driver they like best (2 is ok, it was 4 that was a little insane
for awhile there! :-).
Submitted by: hm

29 years agoBring PCVT's config files into the tree.
Jordan K. Hubbard [Sun, 5 Feb 1995 10:56:47 +0000 (10:56 +0000)]
Bring PCVT's config files into the tree.
Submitted by: mh

29 years agoChange the defaults for newfs to disregard the geometry in the disklabel.
Poul-Henning Kamp [Sun, 5 Feb 1995 08:42:31 +0000 (08:42 +0000)]
Change the defaults for newfs to disregard the geometry in the disklabel.

We pretend we have one head with two megabyte worth of sectors per cylinder.

The code try to access another head in what it belives to the same
physical cylinder, because it belives that it would be faster than
waiting for the next free sector under this head to come around.

Most modern drives doesn't have a "classical" geometry, and thus
we end up fooling ourselves doing the above optimization.  With this
change we will fill a cylinder sequentially if we can, and thus get
much more mileage from the track-buffer/cache built into the drives.

As a result a lot of seeks to the next or previous track should be
avoided by this.

(My disk is a lot less noisy actually...)

You can still get the old behaviour, by specifying zero for the
numbers.

This will also solve the problem with newfs barfing at really big
drives.

Obtained from: adult advice from Kirk.

29 years agoUpdate kmem_malloc() call to new waitflag(s) interface.
Bruce Evans [Sun, 5 Feb 1995 07:08:27 +0000 (07:08 +0000)]
Update kmem_malloc() call to new waitflag(s) interface.

This might fix recent problems on thud and freefall.

29 years agoCollapsed _masterpw_breakout_yp() and _pw_breakout_yp() into a
Bill Paul [Sun, 5 Feb 1995 02:12:49 +0000 (02:12 +0000)]
Collapsed _masterpw_breakout_yp() and _pw_breakout_yp() into a
single function.

29 years agoUse absolute path to md5 program. It lives in /sbin and people are unlikely
Poul-Henning Kamp [Sat, 4 Feb 1995 22:56:28 +0000 (22:56 +0000)]
Use absolute path to md5 program.  It lives in /sbin and people are unlikely
to have $PATH to it.

29 years agoCreated manual page for ypserv and changed Makefile to install it.
Bill Paul [Sat, 4 Feb 1995 21:32:04 +0000 (21:32 +0000)]
Created manual page for ypserv and changed Makefile to install it.

Also tweaked server.c to support newer versions of tcpwrapper (log_tcp.h
is now tcpd.h and FROM_UNKNOWN changed to STRING_UNKNOWN).

29 years agoadded *.info and *.db to the default "ignore" list in cvs to avoid obviously
Poul-Henning Kamp [Sat, 4 Feb 1995 20:27:23 +0000 (20:27 +0000)]
added *.info and *.db to the default "ignore" list in cvs to avoid obviously
common mistakes.

29 years agoAdded final '\n'
Poul-Henning Kamp [Sat, 4 Feb 1995 20:02:01 +0000 (20:02 +0000)]
Added final '\n'

29 years agoAdded final '\n'.
Poul-Henning Kamp [Sat, 4 Feb 1995 19:59:58 +0000 (19:59 +0000)]
Added final '\n'.

29 years agoAdd missing trailing '\n'
Poul-Henning Kamp [Sat, 4 Feb 1995 19:56:39 +0000 (19:56 +0000)]
Add missing trailing '\n'

29 years agoSupport 32bit access to IDE disks, if it seems to work for the drive.
Poul-Henning Kamp [Sat, 4 Feb 1995 19:39:36 +0000 (19:39 +0000)]
Support 32bit access to IDE disks, if it seems to work for the drive.

You will normally have to have a VLB or other 32bit IDE "controller" for
this to work.

Depending on your setup, this may gain you 20-100 % speed from your disk.

Reviewed by: phk
Submitted by: vak@cronyx.ru

29 years agoA cute hack to speed up things by Keith: if getenv("PWD") is the same
Poul-Henning Kamp [Sat, 4 Feb 1995 19:29:22 +0000 (19:29 +0000)]
A cute hack to speed up things by Keith:  if getenv("PWD") is the same
inode as ".", then just return that.  I added a check so it must start with
a '/'.

Reviewed by: phk
Submitted by: bostic@cs.berkeley.edu (Keith Bostic)

29 years agoA couple of sensible changes from down under...
Poul-Henning Kamp [Sat, 4 Feb 1995 19:20:49 +0000 (19:20 +0000)]
A couple of sensible changes from down under...

Reviewed by: phk
Submitted by: Stephen McKay <syssgm@devetir.qld.gov.au>

29 years agofixed spelling error.
Poul-Henning Kamp [Sat, 4 Feb 1995 19:11:14 +0000 (19:11 +0000)]
fixed spelling error.

29 years agoDisable I/O port accesses in case the BIOS enabled them.
Stefan Eßer [Sat, 4 Feb 1995 14:02:44 +0000 (14:02 +0000)]
Disable I/O port accesses in case the BIOS enabled them.
Clear all output lines except ATN and ACK when waiting
for target phase change.

Submitted by: wolf (Wolfgang Stanglmeier)

29 years agoMoved "#include <pci/pcivar.h>" from "usr.sbin/lsdev/i386.c"
Stefan Eßer [Sat, 4 Feb 1995 12:47:19 +0000 (12:47 +0000)]
Moved "#include <pci/pcivar.h>" from "usr.sbin/lsdev/i386.c"
to "/sys/i386/include/devconf.h", where it really belongs.

29 years agoAdding ppp configuration files entry for distribution packages.
Atsushi Murai [Sat, 4 Feb 1995 11:16:47 +0000 (11:16 +0000)]
Adding ppp configuration files entry for distribution packages.

29 years agocomment localhost by default.
Dima Ruban [Sat, 4 Feb 1995 08:32:15 +0000 (08:32 +0000)]
comment localhost by default.

29 years agoAdd long-awaited (:) support for sophisticated dependency checking. We now
Satoshi Asami [Sat, 4 Feb 1995 05:49:26 +0000 (05:49 +0000)]
Add long-awaited (:) support for sophisticated dependency checking.  We now
have three variables:

EXEC_DEPENDS - A list of "prog:dir" pairs of other ports this
  package depends on.  "prog" is the name of an
  executable.  make will search your $PATH for it and go
  into "dir" to do a "make all install" if it's not found.
LIB_DEPENDS - A list of "lib:dir" pairs of other ports this package
  depends on.  "lib" is the name of a shared library.
  make will use "ldconfig -r" to search for the
  library.  Note that lib can be any regular expression,
  and you need two backslashes in front of dots (.) to
  supress its special meaning (e.g., use
  "foo\\.2\\.:${PORTSDIR}/utils/foo" to match "libfoo.2.*").
DEPENDS - A list of other ports this package depends on being
  made first.  Use this for things that don't fall into
  the above two categories.

DEPENDS behaves exactly like before, so old Makefiles will still work
the same.  The two variables are lists of pairs as described above.

For instance, if your program depends on unzip and libjpeg.5.*, use
the following definitions:

EXEC_DEPENDS= unzip:${PORTSDIR}/archivers/unzip
LIB_DEPENDS= jpeg\\.5\\.:${PORTSDIR}/graphics/jpeg

gmake:${PORTSDIR}/utils/gmake is automatically added to EXEC_DEPENDS
if USE_GMAKE is defined.

If NO_DEPENDS is defined, the list will just be printed out one by one.

29 years agoCreated yp_mkdb(8) manual page, modified Makefile to install it.
Bill Paul [Sat, 4 Feb 1995 04:26:20 +0000 (04:26 +0000)]
Created yp_mkdb(8) manual page, modified Makefile to install it.

29 years agoChanged some comments.
Bill Paul [Sat, 4 Feb 1995 00:13:21 +0000 (00:13 +0000)]
Changed some comments.

29 years agoAdded options for starting NIS client and server processes. All are
Bill Paul [Fri, 3 Feb 1995 23:54:07 +0000 (23:54 +0000)]
Added options for starting NIS client and server processes. All are
off by default.

29 years agoInclude <time.h> instead of <sys/time.h> to get CLK_TCK. Including
Bruce Evans [Fri, 3 Feb 1995 22:28:34 +0000 (22:28 +0000)]
Include <time.h> instead of <sys/time.h> to get CLK_TCK.  Including
<sys/time.h> works because <sys/time.h> includes <time.h> if KERNEL
is not defined, but is ugly.

29 years agoInclude <limits.h> to get the definition of INT_MAX - don't depend on
Bruce Evans [Fri, 3 Feb 1995 22:21:09 +0000 (22:21 +0000)]
Include <limits.h> to get the definition of INT_MAX - don't depend on
namespace pollution in <time.h>.

29 years agoChange CLK_TCK to CLOCKS_PER_SEC.
Bruce Evans [Fri, 3 Feb 1995 22:09:56 +0000 (22:09 +0000)]
Change CLK_TCK to CLOCKS_PER_SEC.

Add a missing apostrophe that suggests inverting the frequency to get
tick size.  It read better before because `CLK_TCK' suggests a tick
size although it is actually a frequency.

29 years agoFixed potential Makefile glitch that could arise if /var/yp doesn't exist yet.
Bill Paul [Fri, 3 Feb 1995 22:01:17 +0000 (22:01 +0000)]
Fixed potential Makefile glitch that could arise if /var/yp doesn't exist yet.

29 years agoChange CLK_TCK to CLOCKS_PER_SEC. (CLK_TCK is a deprecated POSIX feature
Bruce Evans [Fri, 3 Feb 1995 21:59:45 +0000 (21:59 +0000)]
Change CLK_TCK to CLOCKS_PER_SEC.  (CLK_TCK is a deprecated POSIX feature
and is not necessarily related to the ANSI CLOCKS_PER_SEC).

Parenthesize macro args.

29 years agoFix previous change to preserve const'ness.
Bruce Evans [Fri, 3 Feb 1995 21:54:03 +0000 (21:54 +0000)]
Fix previous change to preserve const'ness.

29 years agoDon't define CLK_TCK here.
Bruce Evans [Fri, 3 Feb 1995 21:49:18 +0000 (21:49 +0000)]
Don't define CLK_TCK here.

Uniformize idempotency ifdef.

29 years agoDefine CLOCKS_PER_SEC.
Bruce Evans [Fri, 3 Feb 1995 21:47:48 +0000 (21:47 +0000)]
Define CLOCKS_PER_SEC.

Define CLK_TCK only if _ANSI_SOURCE is not defined.

Don't include <machine/limits.h> to get the definition of CLK_TCK.
CLK_TCK should never have been defined there, and the inclusion
polluted the namespace.

29 years agoDon't depend on namespace pollution in <machine/limits.h> for the
Bruce Evans [Fri, 3 Feb 1995 21:31:35 +0000 (21:31 +0000)]
Don't depend on namespace pollution in <machine/limits.h> for the
definition of CLK_TCK.  Use _BSD_CLK_TCK_ instead of CLK_TCK so that
we don't have to include the user header <time.h>.

29 years agoDefine _BSD_CLK_TCK_ and _BSD_CLOCKS_PER_SEC_ to support correct
Bruce Evans [Fri, 3 Feb 1995 21:23:46 +0000 (21:23 +0000)]
Define _BSD_CLK_TCK_ and _BSD_CLOCKS_PER_SEC_ to support correct
definitions of CLK_TCK and CLOCKS_PER_SEC.

29 years agoAllow all characters except "real" control chars.
Søren Schmidt [Fri, 3 Feb 1995 21:09:44 +0000 (21:09 +0000)]
Allow all characters except "real" control chars.
Fix NULL reference when serial console (thanks bruce).

29 years agoInclude <limits.h> to get LONG_MAX - don't depend on namespace pollution
Bruce Evans [Fri, 3 Feb 1995 20:55:37 +0000 (20:55 +0000)]
Include <limits.h> to get LONG_MAX - don't depend on namespace pollution
in <time.h>.

29 years agoInclude ncr.c from where it is now.
Bruce Evans [Fri, 3 Feb 1995 20:49:10 +0000 (20:49 +0000)]
Include ncr.c from where it is now.

29 years agoAdd #include <pci/pcivar.h> for definition of PCI_EXTERNAL_LEN.
Stefan Eßer [Fri, 3 Feb 1995 20:26:50 +0000 (20:26 +0000)]
Add #include <pci/pcivar.h> for definition of PCI_EXTERNAL_LEN.

Reviewed by:
Submitted by:
Obtained from:

29 years agoInclude <sys/types.h> explicitly to get declaration of u_long for
Bruce Evans [Fri, 3 Feb 1995 20:25:59 +0000 (20:25 +0000)]
Include <sys/types.h> explicitly to get declaration of u_long for
<netinet/in.h> - don't depend on namespace pollution in <stdio.h>.

29 years agoHave version number track CSV version number. No other change.
Justin T. Gibbs [Fri, 3 Feb 1995 17:18:44 +0000 (17:18 +0000)]
Have version number track CSV version number.  No other change.

29 years agoKeep track of when we actually are awaiting a reply to an SDTR or WDTR
Justin T. Gibbs [Fri, 3 Feb 1995 17:15:12 +0000 (17:15 +0000)]
Keep track of when we actually are awaiting a reply to an SDTR or WDTR
message instead of relying on the fact that we are scheduled to send them.
The old method worked 99.9% of the time, but someone reported some periferals
that did MSG_REJECT at odd times (sometimes before we could send an SDTR
or WDTR) that we would construe as the response to an SDTR or WDTR message.
This also removes a possible race condition where after a bus reset (the
result of a command time out not during intial probe time), we might queue
two commands both requesting SDTR, WDTR or both.

29 years ago#include<kernel.h> to quiet a compiler warning having to do with the
Justin T. Gibbs [Fri, 3 Feb 1995 17:08:17 +0000 (17:08 +0000)]
#include<kernel.h> to quiet a compiler warning having to do with the
DATA_SET (pcidevice_set, ahc_device) macro.  Maybe this will allow
the aic7870 to be probed again.  I don't have a PCI machine here
to test it.

29 years agoFix to h2ph "undefined function" bug
Ugen J.S. Antsilevich [Fri, 3 Feb 1995 15:16:03 +0000 (15:16 +0000)]
Fix to h2ph "undefined function" bug
i reported today earlier..tested and works OK..
( To those who want to experience bug try running aub
with old version of socket.ph and with new one or just any
perl script  "requiring " <sys/socket.ph> or <sys/cdefs.ph> )

29 years agoFix hanging while playing very short notes.
Andrey A. Chernov [Fri, 3 Feb 1995 10:19:38 +0000 (10:19 +0000)]
Fix hanging while playing very short notes.

29 years agoFixed bmap run-length brokeness.
David Greenman [Fri, 3 Feb 1995 06:46:28 +0000 (06:46 +0000)]
Fixed bmap run-length brokeness.
Use bmap run-length extension when doing clustered paging.

Submitted by: John Dyson

29 years agoAdded ${.CURDIR}s to afterinstall target.
Bill Paul [Fri, 3 Feb 1995 03:44:41 +0000 (03:44 +0000)]
Added ${.CURDIR}s to afterinstall target.

29 years agoPut ${.CURDIR}s in front of Makefile.yp and mknetid in the required places.
Bill Paul [Fri, 3 Feb 1995 03:41:38 +0000 (03:41 +0000)]
Put ${.CURDIR}s in front of Makefile.yp and mknetid in the required places.
Also took out uneeded BINDIR & BINMODE stuff.

29 years agoRemoved a pile of vfs_unbusy_pages()...both unnecessary and wrong - resulted
David Greenman [Fri, 3 Feb 1995 03:40:08 +0000 (03:40 +0000)]
Removed a pile of vfs_unbusy_pages()...both unnecessary and wrong - resulted
in serious system instability. Changed a B_INVAL to a B_NOCACHE so that
buffer data is properly disposed of.

Submitted by: John Dyson, Rick Macklin, and ohki@gssm.otsuka.tsukuba.ac.jp

29 years agoMake B_NOCACHE and B_INVAL buffers work correctly - throw away the data in
David Greenman [Fri, 3 Feb 1995 03:35:56 +0000 (03:35 +0000)]
Make B_NOCACHE and B_INVAL buffers work correctly - throw away the data in
the page cache.

Submitted by: John Dyson

29 years agoRemoved reference in comments to -o option of yppasswdd: -o option
Bill Paul [Fri, 3 Feb 1995 01:11:57 +0000 (01:11 +0000)]
Removed reference in comments to -o option of yppasswdd: -o option
no longer exists.

29 years agoFixed a rather serious bug that presents itself when FreeBSD is configured
Bill Paul [Fri, 3 Feb 1995 01:09:35 +0000 (01:09 +0000)]
Fixed a rather serious bug that presents itself when FreeBSD is configured
as an NIS client. The pw_breakout_yp routines that are used to populate the
_pw_passwd structire only do anything if the bits in the pw_fields member
_pw_passwd are cleared. Unfortunately, we can get into a state where
pw_fields has garbage in it right before the YP lookup functions are
called, which causes the breakout functions to screw up in a big way.
Here's how to duplicate the problem:

- Configure FreeBSD as an NIS client
- Log in as a user who's password database records reside only in
  the NIS passwd maps.
- Type ps -aux

Result: your processes appear to be owned by 'root' or 'deamon.'
/bin/ls can exhibit the same problem.

The reason this happens:

- When ps(1) needs to match a username to a UID, it calls getpwuid().

- root is in the local password file, so getpwuid() calls  __hashpw()
  and __hashpw() populates the _pw_passwd struct, including the pw_fields
  member. This happens before NIS lookups take place because, by coincidence,
  ps(1) tends to display processes owned by root before it happens upon
  a proccess owned by you.

- When your UID comes up, __hashpw() fails to find your entry in the
  local password database, so it bails out, BUT THE BITS IN THE pw_fields
  STRUCTURE OF _pw_passwd ARE NEVER CLEARED AND STILL CONTAIN INFORMATION
  FROM THE PREVIOUS CALL TO __hash_pw()!!

- If we have NIS enabled, the NIS lookup functions are called.

- The pw_breakout_yp routines see that the pw_fields bits are set and
  decline to place the data retrieved from the NIS passwd maps into the
  _pw_passwd structure.

- getpwuid() returns the results of the last __hashpw() lookup instead
  of the valid NIS data.

- Hijinxs ensue when user_from_uid() caches this bogus information and
  starts handing out the wrong usernames.

AAAARRRRRRRRRGGGGGGHHHHHHHHHH!!!

*Please* don't tell me I'm the only person to have noticed this.

Fixed by having __hashpw() check the state of pw_fields just before
bailing out on a failed lookup and clearing away any leftover garbage.
What a fun way to spend an afternoon.

29 years agoGive the user a chance to specify all the network setup parameters again
Jordan K. Hubbard [Fri, 3 Feb 1995 00:42:27 +0000 (00:42 +0000)]
Give the user a chance to specify all the network setup parameters again

29 years agoWhoops, actually DO the last config!
Jordan K. Hubbard [Thu, 2 Feb 1995 23:39:44 +0000 (23:39 +0000)]
Whoops, actually DO the last config!

29 years agoGet ready for the XFree86 3.1.1 distribution (next snapshot),
Jordan K. Hubbard [Thu, 2 Feb 1995 23:13:26 +0000 (23:13 +0000)]
Get ready for the XFree86 3.1.1 distribution (next snapshot),
compat1x -> compat1xdist

29 years agoPut the release number in the first banner - makes it easier to quickly
Jordan K. Hubbard [Thu, 2 Feb 1995 23:09:30 +0000 (23:09 +0000)]
Put the release number in the first banner - makes it easier to quickly
tell if I've muffed it.

29 years agoRemoved misleading copyright note. This file contains just
Stefan Eßer [Thu, 2 Feb 1995 22:01:40 +0000 (22:01 +0000)]
Removed misleading copyright note. This file contains just
definitions taken from the PCI specs. Part of them were typed
in by Wolfgang Stanglmeier, the (at that time unneeded) rest
by Charles Hannum (thanks !).

29 years agoFix some last boners in the setup script.
Jordan K. Hubbard [Thu, 2 Feb 1995 20:46:40 +0000 (20:46 +0000)]
Fix some last boners in the setup script.

29 years agoLook for files to install first in the object directory, then in the
Bruce Evans [Thu, 2 Feb 1995 17:44:02 +0000 (17:44 +0000)]
Look for files to install first in the object directory, then in the
source directory.

Don't create target directories at install time.  mtree already handles
them correctly.

29 years agoCleaned up and updated the yppasswdd man page to reflect FreeBSD-specific
Bill Paul [Thu, 2 Feb 1995 17:25:58 +0000 (17:25 +0000)]
Cleaned up and updated the yppasswdd man page to reflect FreeBSD-specific
changes and new options. (more new man pages and an NIS tutorial to follow)

29 years agoDisable disconnect after drive timeouts, since these
Stefan Eßer [Thu, 2 Feb 1995 15:50:57 +0000 (15:50 +0000)]
Disable disconnect after drive timeouts, since these
may be the result of reselect following too fast for
the driver to notice. Not the final solution, but the
problem has been seen only with very few devices.

Reviewed by: se
Submitted by: wolf (Wolfgang Stanglmeier)

29 years agoWhy not give user a chance to start it, too?
Jordan K. Hubbard [Thu, 2 Feb 1995 14:30:36 +0000 (14:30 +0000)]
Why not give user a chance to start it, too?

29 years agoMake this a bit more aggressive.. :-)
Jordan K. Hubbard [Thu, 2 Feb 1995 14:21:46 +0000 (14:21 +0000)]
Make this a bit more aggressive.. :-)

29 years agoHere's the _real_ XFree86 3.1.1 installation script. I'm ready.. :-)
Jordan K. Hubbard [Thu, 2 Feb 1995 14:18:04 +0000 (14:18 +0000)]
Here's the _real_ XFree86 3.1.1 installation script.  I'm ready.. :-)

29 years agoReapplied all of Stefan's changes. What a mess - the files were modified
David Greenman [Thu, 2 Feb 1995 13:12:18 +0000 (13:12 +0000)]
Reapplied all of Stefan's changes. What a mess - the files were modified
and moved at the same time. This made it *very* difficult to fix the
revision log lossage that happend when the files were moved. SIGH.

29 years agoFixed up include paths after copying these in the repository.
David Greenman [Thu, 2 Feb 1995 12:36:19 +0000 (12:36 +0000)]
Fixed up include paths after copying these in the repository.

29 years agoA better clean rule.
Jordan K. Hubbard [Thu, 2 Feb 1995 12:29:03 +0000 (12:29 +0000)]
A better clean rule.

29 years agoswap_pager.c:
David Greenman [Thu, 2 Feb 1995 09:09:15 +0000 (09:09 +0000)]
swap_pager.c:
Fixed long standing bug in freeing swap space during object collapses.
Fixed 'out of space' messages from printing out too often.
Modified to use new kmem_malloc() calling convention.
Implemented an additional stat in the swap pager struct to count the
amount of space allocated to that pager. This may be removed at some
point in the future.
Minimized unnecessary wakeups.

vm_fault.c:
Don't try to collect fault stats on 'swapped' processes - there aren't
any upages to store the stats in.
Changed read-ahead policy (again!).

vm_glue.c:
Be sure to gain a reference to the process's map before swapping.
Be sure to lose it when done.

kern_malloc.c:
Added the ability to specify if allocations are at interrupt time or
are 'safe'; this affects what types of pages can be allocated.

vm_map.c:
Fixed a variety of map lock problems; there's still a lurking bug that
will eventually bite.

vm_object.c:
Explicitly initialize the object fields rather than bzeroing the struct.
Eliminated the 'rcollapse' code and folded it's functionality into the
"real" collapse routine.
Moved an object_unlock() so that the backing_object is protected in
the qcollapse routine.
Make sure nobody fools with the backing_object when we're destroying it.
Added some diagnostic code which can be called from the debugger that
looks through all the internal objects and makes certain that they
all belong to someone.

vm_page.c:
Fixed a rather serious logic bug that would result in random system
crashes. Changed pagedaemon wakeup policy (again!).

vm_pageout.c:
Removed unnecessary page rotations on the inactive queue.
Changed the number of pages to explicitly free to just free_reserved
level.

Submitted by: John Dyson

29 years agoCalling semantics for kmem_malloc() have been changed...and the third
David Greenman [Thu, 2 Feb 1995 08:49:45 +0000 (08:49 +0000)]
Calling semantics for kmem_malloc() have been changed...and the third
argument is now more than just a single flag.

Submitted by: John Dyson

29 years agoCalling semantics for kmem_malloc() have been changed...and the third
David Greenman [Thu, 2 Feb 1995 08:49:08 +0000 (08:49 +0000)]
Calling semantics for kmem_malloc() have been changed...and the third
argument is now more than just a single flag. (kern_malloc.c)
Used new M_KERNEL value for socket allocations that previous were
"M_NOWAIT". Note that this will change when we clean up the M_ namespace
mess.

Submitted by: John Dyson

29 years agoMostly cosmetic changes. Use KERNBASE instead of UPT_MAX_ADDRESS in
David Greenman [Thu, 2 Feb 1995 08:45:14 +0000 (08:45 +0000)]
Mostly cosmetic changes. Use KERNBASE instead of UPT_MAX_ADDRESS in
some comparisons as it is more correct (we want the kernel page tables
included).
Reorganized some of the expressions for efficiency.
Fixed the new pmap_prefault() routine - it would sometimes pick up the
wrong page if the page in the shadow was present but the page in object
was paged out. The routine remains unused and commented out, however.
Explicitly free zero reference count page tables (rather than waiting
for the pagedaemon to do it).

Submitted by: John Dyson

29 years agoBump the date to February 2nd.
Jordan K. Hubbard [Thu, 2 Feb 1995 08:32:11 +0000 (08:32 +0000)]
Bump the date to February 2nd.

29 years agoClean these up a bunch more - gee I'm glad I do these snapshots, they uncover
Jordan K. Hubbard [Thu, 2 Feb 1995 08:31:38 +0000 (08:31 +0000)]
Clean these up a bunch more - gee I'm glad I do these snapshots, they uncover
*so many problems*.. Too bad the snaps themselves don't get made as a
consequence.. :-)

Bump the date forward Yet Another Day.  David has some changes he wants to
see rolled in, so I guess we're down for Feb 2nd as the snap date!

29 years agoCreate directory symlinks neccesary for locale because mtree
Andrey A. Chernov [Thu, 2 Feb 1995 08:03:24 +0000 (08:03 +0000)]
Create directory symlinks neccesary for locale because mtree
can't create symlinks.
Suggested by Rod.

29 years agoMkdir() now takes two flags.
Jordan K. Hubbard [Thu, 2 Feb 1995 05:49:06 +0000 (05:49 +0000)]
Mkdir() now takes two flags.

29 years agoThis was misleading - you select OK to continue, not Cancel! ;-)
Jordan K. Hubbard [Thu, 2 Feb 1995 05:44:14 +0000 (05:44 +0000)]
This was misleading - you select OK to continue, not Cancel! ;-)

29 years agoSince the pci directory has moved, this required updating.
Jordan K. Hubbard [Thu, 2 Feb 1995 05:42:47 +0000 (05:42 +0000)]
Since the pci directory has moved, this required updating.

29 years agoMake dirs for msdos fs's.
Poul-Henning Kamp [Thu, 2 Feb 1995 05:35:36 +0000 (05:35 +0000)]
Make dirs for msdos fs's.

29 years agoWe no longer need a welcome telling us to take a floppy out - it's already
Jordan K. Hubbard [Thu, 2 Feb 1995 04:57:05 +0000 (04:57 +0000)]
We no longer need a welcome telling us to take a floppy out - it's already
out.

29 years agoCreate an obj symlink when necessary.
Jordan K. Hubbard [Thu, 2 Feb 1995 04:53:30 +0000 (04:53 +0000)]
Create an obj symlink when necessary.
Submitted by: "Remy CARD" <Remy.Card@masi.ibp.fr>

29 years agoClean up the documentation in light of a few recent changes.
Jordan K. Hubbard [Thu, 2 Feb 1995 04:13:22 +0000 (04:13 +0000)]
Clean up the documentation in light of a few recent changes.

29 years agoWe don't need to go back to the first screen again - optimize this.
Jordan K. Hubbard [Thu, 2 Feb 1995 04:08:13 +0000 (04:08 +0000)]
We don't need to go back to the first screen again - optimize this.

29 years agoArgh! I left the setup script off the floppy. Time to re-roll it.
Jordan K. Hubbard [Thu, 2 Feb 1995 04:07:51 +0000 (04:07 +0000)]
Argh!  I left the setup script off the floppy.  Time to re-roll it.

29 years agoFixed problem passing received packets to bpf.
Paul Richards [Thu, 2 Feb 1995 04:07:45 +0000 (04:07 +0000)]
Fixed problem passing received packets to bpf.

Implemented all the devconf code properly.

29 years agoCosmetic fix.
Jordan K. Hubbard [Thu, 2 Feb 1995 03:36:51 +0000 (03:36 +0000)]
Cosmetic fix.
Submitted by: Christoph Kukulies <kuku@gilberto.physik.rwth-aachen.de>

29 years agoPPP does not find config files (Wrong path name not a /etc/ppp)
Atsushi Murai [Thu, 2 Feb 1995 01:54:27 +0000 (01:54 +0000)]
PPP does not find config files (Wrong path name not a /etc/ppp)
Submitted by: Michael Reifenberger <root@rz-wb.fh-sw.de>

29 years agocleaned up and modified slightly to reflect changes
Bill Paul [Wed, 1 Feb 1995 23:47:04 +0000 (23:47 +0000)]
cleaned up and modified slightly to reflect changes

29 years agoCleaned up Makefile a little, added man page links from ypchsh.1 ypchfn.1
Bill Paul [Wed, 1 Feb 1995 23:46:20 +0000 (23:46 +0000)]
Cleaned up Makefile a little, added man page links from ypchsh.1 ypchfn.1
and ypchpass.1 to yppasswd.1. Also fixed a typo: ypshfn != ypchfn.