]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoBuild sendmail with STARTTLS support unless NO_OPENSSL is set.
gshapiro [Tue, 10 Oct 2000 18:15:41 +0000 (18:15 +0000)]
Build sendmail with STARTTLS support unless NO_OPENSSL is set.

23 years agoUse sendmail's version of vacation. It is command line and database
gshapiro [Tue, 10 Oct 2000 18:14:30 +0000 (18:14 +0000)]
Use sendmail's version of vacation.  It is command line and database
compatible with the old version but includes new functionality and bug fixes.

Since it is not part of the NO_SENDMAIL make.conf option, libsmdb and
libsmutil should always be built for vacation's sake.

PR: 15227

23 years agomail.local(8) is no longer installed as a set-user-id binary.
gshapiro [Tue, 10 Oct 2000 18:12:30 +0000 (18:12 +0000)]
mail.local(8) is no longer installed as a set-user-id binary.

Users may have to adjust their configuration to call mail.local as root
by adding the F=S flag to the local mailer.  Most probably already have this.

23 years agoDisable the SMTP EXPN and VRFY commands by default to prevent information
gshapiro [Tue, 10 Oct 2000 18:10:38 +0000 (18:10 +0000)]
Disable the SMTP EXPN and VRFY commands by default to prevent information
leakage.

Submitted by: Maxime Henrion <mux@qualys.com>

23 years agoInstall the sendmail cf building tools (contrib/sendmail/cf) in
gshapiro [Tue, 10 Oct 2000 18:08:19 +0000 (18:08 +0000)]
Install the sendmail cf building tools (contrib/sendmail/cf) in
/usr/share/sendmail/cf.

PR: 19790

23 years agofix conflicting types for ng_ing_rcvmsg() and ng_ing_rcvdata().
hm [Tue, 10 Oct 2000 15:22:00 +0000 (15:22 +0000)]
fix conflicting types for ng_ing_rcvmsg() and ng_ing_rcvdata().

23 years agoupdate with information about isdn4bsd upgrade
hm [Tue, 10 Oct 2000 15:11:45 +0000 (15:11 +0000)]
update with information about isdn4bsd upgrade

23 years ago* Add rudimentary DDB support (no kgdb, no backtrace, no single step).
dfr [Tue, 10 Oct 2000 14:57:10 +0000 (14:57 +0000)]
* Add rudimentary DDB support (no kgdb, no backtrace, no single step).
* Track recent changes to SWI code.
* Allocate RIDs for pmaps (untested).
* Implement assembler version of cpu_switch - its cleaner that way.

23 years agoupdate to include the new isdn4bsd drivers
hm [Tue, 10 Oct 2000 14:54:45 +0000 (14:54 +0000)]
update to include the new isdn4bsd drivers

23 years agoadd new version of ELECOM Air@Hawk/LD-WL11/PCC
sanpei [Tue, 10 Oct 2000 14:27:30 +0000 (14:27 +0000)]
add new version of ELECOM Air@Hawk/LD-WL11/PCC
(Yasuhiro Magara <gama@rose.zero.ad.jp> in [bsd-nomads:14588])

add "wi" cards into HARDWARE.TXT and RELNOTES.TXT

23 years agocorrect "device iwic0" to "device iwic"
hm [Tue, 10 Oct 2000 14:23:04 +0000 (14:23 +0000)]
correct "device iwic0" to "device iwic"

23 years agoA quote from the ia64 compiler:
dfr [Tue, 10 Oct 2000 14:18:50 +0000 (14:18 +0000)]
A quote from the ia64 compiler:
`char' is promoted to `int' when passed through `...'
(so you should pass `int' not `char' to `va_arg')

23 years agoupdate i4b version to 0.96.00
hm [Tue, 10 Oct 2000 14:16:55 +0000 (14:16 +0000)]
update i4b version to 0.96.00

23 years agoSet i4b version number to 0.96.00.
hm [Tue, 10 Oct 2000 14:14:20 +0000 (14:14 +0000)]
Set i4b version number to 0.96.00.
It is (nearly) identical to i4b-00.96.00-beta-101000.tar.gz

23 years agoDo not exit when no holiday file is found. Issue warning and continue.
hm [Tue, 10 Oct 2000 13:28:49 +0000 (13:28 +0000)]
Do not exit when no holiday file is found. Issue warning and continue.

23 years agoInital PowerPC loader build support.
obrien [Tue, 10 Oct 2000 13:22:41 +0000 (13:22 +0000)]
Inital PowerPC loader build support.

Submitted by: Benno Rice <benno@jeamland.net>

23 years agoFix the embeded VCS ID for FreeBSD vs. NetBSD.
obrien [Tue, 10 Oct 2000 13:15:26 +0000 (13:15 +0000)]
Fix the embeded VCS ID for FreeBSD vs. NetBSD.

23 years agoAdd a quirk entry for the USB Sony DSC drive.
n_hibma [Tue, 10 Oct 2000 13:12:11 +0000 (13:12 +0000)]
Add a quirk entry for the USB Sony DSC drive.

23 years agoArchitecture-specific setjump()/longjmp() bits for the PowerPC
obrien [Tue, 10 Oct 2000 13:11:47 +0000 (13:11 +0000)]
Architecture-specific setjump()/longjmp() bits for the PowerPC
needed by the loader.

Submitted by: Benno Rice <benno@jeamland.net>
Obtained from: NetBSD (Wolfgang Solfrank <ws@@tools.de>)

23 years agoprint out error reason if it was failed in assign_io
sanpei [Tue, 10 Oct 2000 13:01:18 +0000 (13:01 +0000)]
print out error reason if it was failed in assign_io

Obtained from: PAO3

23 years agoRemove unneccessary includes. (phk)
n_hibma [Tue, 10 Oct 2000 12:41:46 +0000 (12:41 +0000)]
Remove unneccessary includes. (phk)

23 years agoMark directories as directories, not as regular files.
des [Tue, 10 Oct 2000 12:32:44 +0000 (12:32 +0000)]
Mark directories as directories, not as regular files.

23 years agoRemove unneeded includes (phk)
n_hibma [Tue, 10 Oct 2000 10:56:53 +0000 (10:56 +0000)]
Remove unneeded includes (phk)

23 years agoAdd MAE0021 - Jetstream Int V.90 56k Voice Series 2.
tanimura [Tue, 10 Oct 2000 10:06:26 +0000 (10:06 +0000)]
Add MAE0021 - Jetstream Int V.90 56k Voice Series 2.

PR: i386/19920
Submitted by: Peter Ortner <port@iname.com>

23 years agoawi needs to access memory with 8bit,
onoe [Tue, 10 Oct 2000 09:50:25 +0000 (09:50 +0000)]
awi needs to access memory with 8bit,
but pccardd apparently maps memory with MDF_16BITS flag.
So memory mapped access is disabled and use IO port instead for now.

This fixes the problem for config index 0x01 in the pccard.conf
with the message: "awi0: failed to complete selftest (timeout)"

23 years agoWe want the FreeBSD ID as the RCS ID, not the NetBSD one.
obrien [Tue, 10 Oct 2000 08:57:30 +0000 (08:57 +0000)]
We want the FreeBSD ID as the RCS ID, not the NetBSD one.

23 years agoRemoved the FreeBSD ID I was forced to add in rev 1.9.
obrien [Tue, 10 Oct 2000 08:53:27 +0000 (08:53 +0000)]
Removed the FreeBSD ID I was forced to add in rev 1.9.

23 years agoMerged from sys/conf/{files.i386,options.i386} revisions 1.334 and
kato [Tue, 10 Oct 2000 08:12:15 +0000 (08:12 +0000)]
Merged from sys/conf/{files.i386,options.i386} revisions 1.334 and
1.140, respectively.

23 years agoFixed breakage of CFLAGS and misplacement of $FreeBSD$ in previous commit.
bde [Tue, 10 Oct 2000 06:29:33 +0000 (06:29 +0000)]
Fixed breakage of CFLAGS and misplacement of $FreeBSD$ in previous commit.

23 years agoRemove LIBRSAGLUE, add LIBSSL
gshapiro [Tue, 10 Oct 2000 06:17:51 +0000 (06:17 +0000)]
Remove LIBRSAGLUE, add LIBSSL

Reviewed by: kris

23 years agoAdded some handling code to work with SLI configs, and removed some unused
cokane [Tue, 10 Oct 2000 03:51:54 +0000 (03:51 +0000)]
Added some handling code to work with SLI configs, and removed some unused
older code that was phased out but not removed, heh.

23 years agoAdd an amusing spoonerism.
kris [Tue, 10 Oct 2000 01:55:08 +0000 (01:55 +0000)]
Add an amusing spoonerism.

Submitted by: Benno Rice <benno@hiro.netizen.com.au>
Approved by: jkh

23 years agoCorrection to previous log entry: add newline between <sys/*.h> and <*.h>.
wollman [Tue, 10 Oct 2000 01:53:00 +0000 (01:53 +0000)]
Correction to previous log entry: add newline between <sys/*.h> and <*.h>.

23 years agoDon't depend on <sys/stat.h> bogusly including <sys/time.h> (and thereby
wollman [Tue, 10 Oct 2000 01:50:26 +0000 (01:50 +0000)]
Don't depend on <sys/stat.h> bogusly including <sys/time.h> (and thereby
<time.h>).

23 years agoFix bug in "read" command.
archie [Tue, 10 Oct 2000 01:34:20 +0000 (01:34 +0000)]
Fix bug in "read" command.

Submitted by: Dave Cornelius <dc@packetdesign.com>

23 years agoo Change TX_BUFFER_LEN from 512 to 2048.
lile [Tue, 10 Oct 2000 00:54:02 +0000 (00:54 +0000)]
o Change TX_BUFFER_LEN from 512 to 2048.

o Remove bogus "spurious interrupt" message.

o Ring buffer head and avail were incorrectly calculated.

o Fix fragment count.

o Fix ring entry for single station, default to 16Mbit.

o Don't complain about long frames.

23 years agoIf write() gives -1/ENOBUFS, keep the packet and sleep for 1/10th
brian [Mon, 9 Oct 2000 22:49:44 +0000 (22:49 +0000)]
If write() gives -1/ENOBUFS, keep the packet and sleep for 1/10th
of a second before attempting to write it again (unless there's
something else to do instead).

23 years agoCreate fd_sets big enough to handle getdtablesize() descriptors.
brian [Mon, 9 Oct 2000 21:18:23 +0000 (21:18 +0000)]
Create fd_sets big enough to handle getdtablesize() descriptors.

23 years agoFix typo: terrabytes->terabytes.
bmah [Mon, 9 Oct 2000 21:17:16 +0000 (21:17 +0000)]
Fix typo:  terrabytes->terabytes.

PR: docs/21837
Submitted by: Tony Finch <dot@dotat.at>

23 years agoo Move from Alfred Perstein's "exclusion" technique of handling special
chris [Mon, 9 Oct 2000 20:06:13 +0000 (20:06 +0000)]
o Move from Alfred Perstein's "exclusion" technique of handling special
  file types to requiring all file types to properly implement fo_stat.
  This makes any new file type additions much easier as this code no
  longer has to be modified to accomodate it.

o Instead of using curproc in fdesc_allocvp, pass a `struct proc' pointer as
  a new fifth parameter.

23 years agoo Sanity check was inverted, resulting in a possible spurious panic
rwatson [Mon, 9 Oct 2000 20:04:39 +0000 (20:04 +0000)]
o Sanity check was inverted, resulting in a possible spurious panic
  during unmount if extended attributes were in use.  Correct by removing
  an unneeded (and undesirable) '!'.

23 years agoClose a race condition that doesn't really exist in -current. When
gibbs [Mon, 9 Oct 2000 19:54:54 +0000 (19:54 +0000)]
Close a race condition that doesn't really exist in -current.  When
a resource shortage occurs, freeze our queue and then set the resource
shortage flag while the controller data structure is locked.  The old
code did these in the wrong order potentially allowing our interrupt
handler to release the queue and clear the flag before the freeze
ever occurred.

23 years agoFix broken const'ness in declaration of sha1_loop().
archie [Mon, 9 Oct 2000 18:49:14 +0000 (18:49 +0000)]
Fix broken const'ness in declaration of sha1_loop().

23 years agokernacc() only knows about mapped memory, not K0SEG addresses.
gallatin [Mon, 9 Oct 2000 18:47:49 +0000 (18:47 +0000)]
kernacc() only knows about mapped memory, not K0SEG addresses.
Before calling kernacc(), make sure that we're not calling it
with a K0SEG address.

This gets alphas booting with SMP_DEBUG & INVARIANTS options

approved by: jhb

23 years agoFix typo in NETGRAPH_INIT() macro.
archie [Mon, 9 Oct 2000 18:37:11 +0000 (18:37 +0000)]
Fix typo in NETGRAPH_INIT() macro.

23 years agoBlow away the v_specmountpoint define, replacing it with what it was
eivind [Mon, 9 Oct 2000 17:31:39 +0000 (17:31 +0000)]
Blow away the v_specmountpoint define, replacing it with what it was
defined as (rdev->si_mountpoint)

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 16:01:36 +0000 (16:01 +0000)]
update to i4b version 0.95.04
This commit adds the ReleaseNotes file which contains a list of major i4b
changes happened since the last version comitted to FreeBSD (0.90.01).

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 15:41:15 +0000 (15:41 +0000)]
update to i4b version 0.95.04

23 years agoReduce buffer size from 64K to 4K.
sobomax [Mon, 9 Oct 2000 14:54:32 +0000 (14:54 +0000)]
Reduce buffer size from 64K to 4K.

Approved by: cg

23 years agoupdate to i4b version 0.95.04 (oops, forgot this one ..)
hm [Mon, 9 Oct 2000 14:30:50 +0000 (14:30 +0000)]
update to i4b version 0.95.04 (oops, forgot this one ..)

23 years agoremove MAINTAINER: i'll develop and maintain i4b from now on in the
hm [Mon, 9 Oct 2000 14:29:16 +0000 (14:29 +0000)]
remove MAINTAINER: i'll develop and maintain i4b from now on in the
FreeBSD-current tree

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 14:22:51 +0000 (14:22 +0000)]
update to i4b version 0.95.04

23 years agofix a typo & allow softclock to run on alphas.
gallatin [Mon, 9 Oct 2000 14:20:09 +0000 (14:20 +0000)]
fix a typo & allow softclock to run on alphas.

This gets my AS500 and UP1000 booting again.

23 years agoDon't make_dev() in bpfopen() unless we need to.
phk [Mon, 9 Oct 2000 14:19:09 +0000 (14:19 +0000)]
Don't make_dev() in bpfopen() unless we need to.

23 years agoDon't hold an extra reference to vnodes. Devfs vnodes are sufficiently
phk [Mon, 9 Oct 2000 14:18:07 +0000 (14:18 +0000)]
Don't hold an extra reference to vnodes.  Devfs vnodes are sufficiently
cheap to setup that it doesn't really matter that we recycle device
vnodes at kleenex speed.

Implement first cut try at killing cloned devices when they are
not needed anymore.  For now only the bpf driver is involved in
this experiment.  Cloned devices can set the SI_CHEAPCLONE flag
which allows us to destroy_dev() it when the vcount() drops to zero
and the vnode is reclaimed.  For now it's a requirement that the
driver doesn't keep persistent state from close to (re)open.

Some whitespace changes.

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 13:41:07 +0000 (13:41 +0000)]
update to i4b version 0.95.04

23 years agoremove the tina-dd driver fragment, it is unlikely that it will get finished.
hm [Mon, 9 Oct 2000 13:32:02 +0000 (13:32 +0000)]
remove the tina-dd driver fragment, it is unlikely that it will get finished.

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 13:29:00 +0000 (13:29 +0000)]
update to i4b version 0.95.04

23 years agoupdate to i4b version 0.95.04
hm [Mon, 9 Oct 2000 13:18:17 +0000 (13:18 +0000)]
update to i4b version 0.95.04

23 years agoCleanup of bitstring.h:
dwmalone [Mon, 9 Oct 2000 12:34:51 +0000 (12:34 +0000)]
Cleanup of bitstring.h:

1) Add FreeBSD: tag.
2) Add parenthesis around macro args.
3) Add parenthesis around macros which are expressions.
4) Add do { ... } while (0) around macros which are compound statements.
5) Sync bitstr_size and bit_alloc with neater versions from NetBSD.
6) Fix bit_ffs and bit_ffc so that they don't search off the end of the
bitstirng.
7) Try to avoid rightshifting signed ints.

I didn't take NetBSD's version directly as the macros are significantly
slower for long bitstrings. Bruce reviewed a previous version of
this patch.

PR: 21204
Submitted by: bob@immure.com
Reviewed by: bde

23 years agoNow I see the error of my ways.
dcs [Mon, 9 Oct 2000 11:29:40 +0000 (11:29 +0000)]
Now I see the error of my ways.

Previous revision of this file changed the "boot" commands to take
no arguments from the stack. This is only valid in the case where
a kernel has not been loaded. In that case, load_kernel_and_modules
will be called, which takes a list of arguments from the stack.

When a kernel is presently loaded, though, the list of arguments must
be passed to the boot command, which was the behaviour before the last
revision.

Fix things for both cases.

Noticed by: S-Max and others on that chat room

23 years agoFixed include files to use sys/{cons,fb,kb}io.h instead of
kato [Mon, 9 Oct 2000 11:07:18 +0000 (11:07 +0000)]
Fixed include files to use sys/{cons,fb,kb}io.h instead of
machine/console.h.

23 years ago.. fsck wrappers aftercommit #1: I don't know how these files got lost,
adrian [Mon, 9 Oct 2000 10:26:15 +0000 (10:26 +0000)]
.. fsck wrappers aftercommit #1: I don't know how these files got lost,
but they did. Oops.

23 years agoReviewed by: rwatson, bp
adrian [Mon, 9 Oct 2000 10:23:31 +0000 (10:23 +0000)]
Reviewed by: rwatson, bp
Approved by: rwatson
Obtained from: NetBSD source tree

Second part of the fsck wrappers commit. This commit enables the new fsck
code (removing the fsck/* code and replacing it with the netbsd fsck
wrapper code), and enabling some FFS-based utilities to compile.

Details:

* quotacheck, fsdb required modification to use the fsck_ffs/ code rather
  than fsck/ . This might change later since quotacheck requires preen.c
  which should exist in fsck/ rather than fsck_ffs/

* src/Makefile has fsck_ffs added to it so it it built as part of the tree
  now

* share/doc/smm/03.fsck/ uses the SMM.doc/ stuff from fsck_ffs, not fsck.

I've tested this, and it shouldn't require any changes on your machine.
The fsck wrapper reads /etc/fsck and is command-line-compatible enough
to not require rc changes (well, most changes unless you want to do
anything nifty by specifying the fs types explicityly, read the man page
if you want further details on what it can do.)

This now allows us to support multiple filesystem types during bootup.

23 years agoReviewed by: rwatson, bp
adrian [Mon, 9 Oct 2000 10:22:56 +0000 (10:22 +0000)]
Reviewed by: rwatson, bp
Approved by: rwatson
Obtained from: NetbSD source tree

Second part of the fsck wrappers commit. This commit enables the new fsck
code (removing the fsck/* code and replacing it with the netbsd fsck
wrapper code), and enabling some FFS-based utilities to compile.

Details:

* quotacheck, fsdb required modification to use the fsck_ffs/ code rather
  than fsck/ . This might change later since quotacheck requires preen.c
  which should exist in fsck/ rather than fsck_ffs/

* src/Makefile has fsck_ffs added to it so it it built as part of the tree
  now

* share/doc/smm/03.fsck/ uses the SMM.doc/ stuff from fsck_ffs, not fsck.

I've tested this, and it shouldn't require any changes on your machine.
The fsck wrapper reads /etc/fsck and is command-line-compatible enough
to not require rc changes (well, most changes unless you want to do
anything nifty by specifying the fs types explicityly, read the man page
if you want further details on what it can do.)

This now allows us to support multiple filesystem types during bootup.

23 years ago.. forgot a ; at the end of the $FreeBSD$ string. How did I actually forget
adrian [Mon, 9 Oct 2000 09:42:51 +0000 (09:42 +0000)]
.. forgot a ; at the end of the $FreeBSD$ string. How did I actually forget
this? :-)

It builds again.

23 years ago- simplify option flags
sumikawa [Mon, 9 Oct 2000 09:28:49 +0000 (09:28 +0000)]
- simplify option flags
- do not print header when using -c
- linklocal address are not cleared when using -c

Obtained from: KAME

23 years agoSome missed magic in the fsck wrapper commit which is required by other
adrian [Mon, 9 Oct 2000 09:21:04 +0000 (09:21 +0000)]
Some missed magic in the fsck wrapper commit which is required by other
utilities which use bits of fsck_ffs - namely quotacheck and fsdb.
In depth, utilities.c contains blockcheck() which is needed by both,
but also a slew of routines which require bits of the FFS code to be
compiled in. This breaks the fs-specific and non-fs-specific code
up into two files (well, blockcheck() is the only routine in utilities.c,
that'll change later) which makes building fsck_ffs, quotacheck and
fsdb work yet again.

(You won't find commits to fsdb and quotacheck here before I haven't
committed the post-fsck-wrappers version of them yet.)

23 years agoString buffer overflow cleanup.
kris [Mon, 9 Oct 2000 09:07:45 +0000 (09:07 +0000)]
String buffer overflow cleanup.

Reviewed by: green, alex

23 years agoReviewed by: rwatson, bp
adrian [Mon, 9 Oct 2000 08:26:35 +0000 (08:26 +0000)]
Reviewed by: rwatson, bp
Approved by: rwatson
Obtained from: NetBSD-current source tree

The beginnings of the fsck wrappers stuff from NetBSD. This particular commit
brings a newly repo-copied sbin/fsck_ffs/ (from sbin/fsck/) into fsck wrappers
mode.

A quick overview (the code reflects this):

* Documentation changed to reflect fsck_ffs instead of fsck
* Simply acts on a single filesystem, doesn't try to do any multiple filesystem
  magic - this is done by the fsck wrappers now

And then specific to fsck_ffs:

* link to /sbin/fsck_4.2bsd and /sbin/fsck_ufs. This is because right now
  the filesystem is of type ufs not ffs, and that during autodetection the
  labeltype rather than the VFS type is used - this is because when doing
  an autodetection of filesystem type in the fsck wrapper program, it does
  not have any link between label type (4.2bsd, vinum, etc) and VFS string.

Note that this shouldn't break a build since the required buildworld Makefile
magic and import of the fsck wrapper code into src/sbin/fsck/ will happen
in a seperate commit.

23 years agoInitiate deorbit burn sequence for <machine/mouse.h>.
phk [Mon, 9 Oct 2000 08:08:36 +0000 (08:08 +0000)]
Initiate deorbit burn sequence for <machine/mouse.h>.

Replace all in-tree uses with <sys/mouse.h> which repo-copied a few
moments ago from src/sys/i386/include/mouse.h by peter.
This is also the appropriate fix for exo-tree sources.

Put warnings in <machine/mouse.h> to discourage use.
November 15th 2000 the warnings will be converted to errors.
January 15th 2001 the <machine/mouse.h> files will be removed.

23 years agoA couple of negative options was not commented out in NOTES/LINT. This
phk [Mon, 9 Oct 2000 07:29:41 +0000 (07:29 +0000)]
A couple of negative options was not commented out in NOTES/LINT.  This
obscured a #include bug in syscons.

23 years agoString buffer safety cleanup. I don't think any of these were exploitable
kris [Mon, 9 Oct 2000 06:08:00 +0000 (06:08 +0000)]
String buffer safety cleanup. I don't think any of these were exploitable
remotely, but they would be if e.g. it happened to call the logging
function using a DNS hostname.

Also replace random() by arc4random() - only one of these is arguably
required since it's directly used in the protocol, but we might as
well replace both to avoid using two different PRNGs.

Reviewed by: green, alex

23 years agoDon't overflow in command-line args
kris [Mon, 9 Oct 2000 05:44:39 +0000 (05:44 +0000)]
Don't overflow in command-line args

Submitted by: Mike Heffner <mheffner@vt.edu>

23 years agoUnremoved used include of <machine/ipl.h>. Removing it in rev.1.95
bde [Mon, 9 Oct 2000 05:39:03 +0000 (05:39 +0000)]
Unremoved used include of <machine/ipl.h>.  Removing it in rev.1.95
significantly pessimized syscalls by arranging to do null rescheduling
on return from every syscall.  (AST_RESCHED was not defined, and the
mask ~AST_RESCHED gets replaced by the useless mask ~0.  This bug has
been fixed before, in rev.1.92.)

23 years agoAdd :L and :U variable modifiers. These convert the variable's value to
will [Mon, 9 Oct 2000 04:53:36 +0000 (04:53 +0000)]
Add :L and :U variable modifiers.  These convert the variable's value to
all-lower or all-upper case characters, respectively.  These were added
to further reduce differences between NetBSD/OpenBSD and FreeBSD make(1)
to propagate OpenPackages.

PR: 19959
Submitted by: Gaspar Chilingarav <nm@web.am>

23 years agoAdd :C/// variable modifier, which is similar to :S/// in functionality
will [Mon, 9 Oct 2000 04:31:43 +0000 (04:31 +0000)]
Add :C/// variable modifier, which is similar to :S/// in functionality
but allows for regex.  I removed NO_REGEX since no one could give a reason
to have it, and since we don't use make in bootstrap tools, it's not needed.

This is mostly added to synch up with NetBSD/OpenBSD so as to eliminate
roadblocks in the OpenPackages project (see http://www.openpackages.org/).
It's also quite useful, and costs us only about 3 kilobytes of space.

PR: 21605
Submitted by: Hubert Feyrer <hubertf@NetBSD.org>
Reviewed by: silence on -arch
Obtained from: NetBSD

23 years agorm miniperl && cd /usr/ports && cvs update -PAd now recommended
imp [Mon, 9 Oct 2000 03:56:39 +0000 (03:56 +0000)]
rm miniperl && cd /usr/ports && cvs update -PAd now recommended

23 years agoo Record the pccard_function in the ivar.
imp [Mon, 9 Oct 2000 03:40:15 +0000 (03:40 +0000)]
o Record the pccard_function in the ivar.
o Use a macro function to get the ivar for the child rather than the casting
  directly.  This should be a little safer and easier to read.

23 years agoClean up error recovery code:
gibbs [Mon, 9 Oct 2000 01:46:01 +0000 (01:46 +0000)]
Clean up error recovery code:

aic7xxx.c:
In target mode, reset the TQINPOS on every restart of the sequencer.
In the past we did this only during a bus reset, but there are other
reasons the sequencer might be reset.

In ahc_clear_critical_section(), disable pausing chip interrupts while
we step the sequencer out of a critical section.  This avoids the
possibility of getting a pausing interrupt (unexpected bus free,
bus reset, etc.) that would prevent the sequencer from stepping.

Send the correct async notifications in the case of a BDR or bus reset.

In ahc_loadseq(), correct the calculation of our critical sections.
In some cases, the sections would be larger than needed.

aic7xxx.h:
Remove an unused SCB flag.

aic7xxx.seq:
MK_MESSAGE is cleared by the kernel, there is no need to waste
a sequencer instruction clearing it.

aic7xxx_freebsd.c:
Go through the host message loop instead of issuing a single
byte message directly in the ahc_timeout() case where we
are currently on the bus to the device.  The effect is the same,
but this way we get a nice printf saying that an expected BDR
was delivered instead of an unexpected bus free.

If we are requeuing an SCB for an error recovery action, be sure
to set the DISCONNECTED flag in the in-core version of the SCB.
This ensures that, in the SCB-paging case, the sequencer will
still recognize the reselection as valid even if the version
of the SCB with this flag set was never previously paged out
to system memory.  In the non-paging case, set the MK_MESSAGE
flag in SCB_CONTROL directly.

aic7xxx_pci.c:
Enable the Memeory Write and Invalidate bug workaround for
all aic7880 chips with revs < 1.  This bug is rarely triggered
in FreeBSD as most transfers end on cache-aligned boundaries,
but a recheck of my references indicates that these chips
are affected.

23 years agoOnly attach "legacy" PCI busses if none have been attached via any other
msmith [Mon, 9 Oct 2000 00:44:41 +0000 (00:44 +0000)]
Only attach "legacy" PCI busses if none have been attached via any other
method.

23 years agoValidate the PCI bus number that we fetch from our parent, since there's no
msmith [Mon, 9 Oct 2000 00:43:45 +0000 (00:43 +0000)]
Validate the PCI bus number that we fetch from our parent, since there's no
guarantee that everything attached to *it* is a PCI bus.

23 years agoChange the way that eventhandler lists are processed so that an event
msmith [Mon, 9 Oct 2000 00:41:29 +0000 (00:41 +0000)]
Change the way that eventhandler lists are processed so that an event
handler can safely remove itself from a list while being run.

(Note that it is not safe to remove anything else from the same list,
 as this may still cause corruption in the case where the removed
 item is next on the list.)

23 years agoResolve the inconsistency between "the number of resources of a particular
msmith [Mon, 9 Oct 2000 00:40:17 +0000 (00:40 +0000)]
Resolve the inconsistency between "the number of resources of a particular
kind we can manage in a set of configurations" and "the number of resources
of a particular kind that can be programmed into an ISA PnP adapter".

Submitted by: Motomichi Matsuzaki <mzaki@e-mail.ne.jp>
Submitted by: Hirokazu WATANABE <gwna@geocities.co.jp>

23 years agoInitiate deorbit burn sequence for <machine/console.h>.
phk [Sun, 8 Oct 2000 21:34:00 +0000 (21:34 +0000)]
Initiate deorbit burn sequence for <machine/console.h>.

Replace all in-tree uses with necessary subset of <sys/{fb,kb,cons}io.h>.
This is also the appropriate fix for exo-tree sources.

Put warnings in <machine/console.h> to discourage use.
November 15th 2000 the warnings will be converted to errors.
January 15th 2001 the <machine/console.h> files will be removed.

23 years agoAdd HP PhotoSmart C500 id.
n_hibma [Sun, 8 Oct 2000 21:31:39 +0000 (21:31 +0000)]
Add HP PhotoSmart C500 id.

PR: kern/21841
Submitted by: josh@zipperup.org

23 years agoAdd copyright notices. Other systems have been barrowing our /etc files
obrien [Sun, 8 Oct 2000 19:20:36 +0000 (19:20 +0000)]
Add copyright notices.  Other systems have been barrowing our /etc files
w/o giving any credit.

23 years agoUse <space> not <tab> before VCS ID.
obrien [Sun, 8 Oct 2000 19:15:49 +0000 (19:15 +0000)]
Use <space> not <tab> before VCS ID.

23 years agoDuh, it is mtree(8) that creates boot/kernel for us as refered to in
obrien [Sun, 8 Oct 2000 18:45:01 +0000 (18:45 +0000)]
Duh, it is mtree(8) that creates boot/kernel for us as refered to in
rev 1.572.  Of course if you are manually running the various Makefile
targets, it may or may not get run.

23 years agoDon't install manpages.
obrien [Sun, 8 Oct 2000 16:56:04 +0000 (16:56 +0000)]
Don't install manpages.
They are being moved elsewhere, and they are causing problems being here.

23 years agoUnbreak detection of breakage in cy driver.
bde [Sun, 8 Oct 2000 14:46:04 +0000 (14:46 +0000)]
Unbreak detection of breakage in cy driver.

23 years agoMade this file compile again after move of stuff from <machine.ipl.h>
bde [Sun, 8 Oct 2000 14:41:13 +0000 (14:41 +0000)]
Made this file compile again after move of stuff from <machine.ipl.h>
to <sys/ipl.h>.

23 years agoUse schedsofttty() again (SMPng casualty with intentionally wrong fix
bde [Sun, 8 Oct 2000 14:37:39 +0000 (14:37 +0000)]
Use schedsofttty() again (SMPng casualty with intentionally wrong fix
in rev.1.101).

Made this file compile again after move of stuff from <machine.ipl.h>
to <sys/ipl.h>.

23 years agoUse schedsofttty() again so that siopoll() gets scheduled as designed
bde [Sun, 8 Oct 2000 14:13:31 +0000 (14:13 +0000)]
Use schedsofttty() again so that siopoll() gets scheduled as designed
(SMPng casualty in rev.1.308 with wrong fix in rev.1.310).

23 years agoKASSERT that mbuf external reference counts are > 0 before we remove
dwmalone [Sun, 8 Oct 2000 09:52:50 +0000 (09:52 +0000)]
KASSERT that mbuf external reference counts are > 0 before we remove
a reference. (Agreed to by Bosko some time ago).

23 years agoFixed printf format errors which corrupted /var/db/mounttab on i386's with
bde [Sun, 8 Oct 2000 09:24:45 +0000 (09:24 +0000)]
Fixed printf format errors which corrupted /var/db/mounttab on i386's with
64-bit longs (don't assume that time_t is long; assume that time_t's are
representable as longs).

23 years agoUse the correct timestamp for the -C -p case when the comparison fails.
bde [Sun, 8 Oct 2000 09:17:56 +0000 (09:17 +0000)]
Use the correct timestamp for the -C -p case when the comparison fails.
A garbage timestamp was used for at least installing /dev/null with -C -p
when the target doesn't already exist.

23 years agoEnable use of poll()
kris [Sun, 8 Oct 2000 09:09:05 +0000 (09:09 +0000)]
Enable use of poll()

23 years agoDon't overflow our fd_set. This is not a full sync with KAME because there
kris [Sun, 8 Oct 2000 08:02:35 +0000 (08:02 +0000)]
Don't overflow our fd_set. This is not a full sync with KAME because there
are a whole lot of other changes which may not be suitable for us.

Obtained from: KAME

23 years agoDon't write past the end of our fd_set
kris [Sun, 8 Oct 2000 07:58:32 +0000 (07:58 +0000)]
Don't write past the end of our fd_set

Obtained from: OpenBSD, via KAME