]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoFixes to the Aironet driver to clear up some WEP issues.
archie [Fri, 8 Dec 2000 19:00:10 +0000 (19:00 +0000)]
Fixes to the Aironet driver to clear up some WEP issues.

PR: 23097
Submitted by: Doug Ambrisko <ambrisko@whistle.com>

23 years agoArgh, revert the clobber changes. Since %ecx and %edx aren't call safe,
jhb [Fri, 8 Dec 2000 18:21:06 +0000 (18:21 +0000)]
Argh, revert the clobber changes.  Since %ecx and %edx aren't call safe,
calling the C functions mtx_enter_hard() and mtx_exit_hard() clobbers them.
Note that %eax is also not call safe, but it is already clobbered due to
cmpxchg.  However, now we are back to not compiling again, so these macros
are still left disabled for now.

23 years agostaticize.
phk [Fri, 8 Dec 2000 15:07:24 +0000 (15:07 +0000)]
staticize.

23 years agoRemove conflicts between unsinged char *s and int s.
non [Fri, 8 Dec 2000 13:50:20 +0000 (13:50 +0000)]
Remove conflicts between unsinged char *s and int s.
It was possible cause of kernel panic.

Pointed Out by: phk@FreeBSD.ORG

23 years agoThe drivers ncv, nsp and stg are already merged to -stable.
non [Fri, 8 Dec 2000 12:52:42 +0000 (12:52 +0000)]
The drivers ncv, nsp and stg are already merged to -stable.

23 years agomdoc(7) police: Once again, remove single-space hard sentence breaks.
ru [Fri, 8 Dec 2000 12:42:24 +0000 (12:42 +0000)]
mdoc(7) police: Once again, remove single-space hard sentence breaks.

23 years agostrunvis(3) and unvis(3) are the same files.
alex [Fri, 8 Dec 2000 12:35:06 +0000 (12:35 +0000)]
strunvis(3) and unvis(3) are the same files.

23 years agomdoc(7) police: once again, remove the hard sentence breaks.
ru [Fri, 8 Dec 2000 12:25:05 +0000 (12:25 +0000)]
mdoc(7) police: once again, remove the hard sentence breaks.

23 years agoThere are four types of encoding now, not three. Most of them use the
alex [Fri, 8 Dec 2000 12:17:53 +0000 (12:17 +0000)]
There are four types of encoding now, not three.  Most of them use the
backslash as a special char, but not all.

23 years agoRun weekly a little bit later to give more time for daily to complete
obrien [Fri, 8 Dec 2000 10:55:33 +0000 (10:55 +0000)]
Run weekly a little bit later to give more time for daily to complete
(since it starts later now).

23 years agoRemove a warning.
n_hibma [Fri, 8 Dec 2000 10:44:15 +0000 (10:44 +0000)]
Remove a warning.

Reported by: phk

23 years agoRemove unused declarations for spending and sdelayed, and remove unused
jake [Fri, 8 Dec 2000 10:39:00 +0000 (10:39 +0000)]
Remove unused declarations for spending and sdelayed, and remove unused
#defines for bits corresponding to pending interrupts.  Yay threads!

23 years agoWhitespace. Make the indentation for MPASS and MPASS2 consistent and
jake [Fri, 8 Dec 2000 09:51:13 +0000 (09:51 +0000)]
Whitespace.  Make the indentation for MPASS and MPASS2 consistent and
slightly more sane.  Make the arguments to the nop MPASS2 match those
of the functional one.  Change 4 spaces to a tab.  Don't indeent a
label so its easier to see.

23 years agoResolve conflicts from the 20001201 merge.
msmith [Fri, 8 Dec 2000 09:22:18 +0000 (09:22 +0000)]
Resolve conflicts from the 20001201 merge.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r69746,
msmith [Fri, 8 Dec 2000 09:20:40 +0000 (09:20 +0000)]
This commit was generated by cvs2svn to compensate for changes in r69746,
which included commits to RCS files with non-trunk default branches.

23 years agoUpdate to the 20001201 ACPI CA snapshot.
msmith [Fri, 8 Dec 2000 09:20:40 +0000 (09:20 +0000)]
Update to the 20001201 ACPI CA snapshot.

23 years agoAdd macros MPASS3 and MPASS4, which take the file and line number
jake [Fri, 8 Dec 2000 09:17:16 +0000 (09:17 +0000)]
Add macros MPASS3 and MPASS4, which take the file and line number
as parameters.  Use them in the mutex inlines so that the file and
line numbers are those of the caller instead of always in this file.

23 years ago - Convert a lot of homebrew debugging output to use the ACPI CA debugging
msmith [Fri, 8 Dec 2000 09:16:20 +0000 (09:16 +0000)]
 - Convert a lot of homebrew debugging output to use the ACPI CA debugging
   infrastructure.  It's not perfect, but it's a lot better than what
   we've been using so far.  The following rules apply to this:
    o BSD component names should be capitalised
    o Layer names should be taken from the non-CA set for now.  We
      may elect to add some new BSD-specific layers later.

 - Make it possible to turn off selective debugging flags or layers
   by listing them in debug.acpi.layer or debug.acpi.level prefixed
   with !.

 - Fully implement support for avoiding nodes in the ACPI namespace.
   Nodes may be listed in the debug.acpi.avoid environment variable;
   these nodes and all their children will be ignored (although still
   scanned over) by ACPI functions which scan the namespace.  Multiple
   nodes can be specified, separated by whitespace.

 - Implement support for selectively disabling ACPI subsystem components
   via the debug.acpi.disable environment variable.  The following
   components can be disabled:
    o bus creation/scanning of the ACPI 'bus'
    o children attachment of children to the ACPI 'bus'
    o button the acpi_button control-method button driver
    o ec the acpi_ec embedded-controller driver
    o isa acpi replacement of PnP BIOS for ISA device discovery
    o lid the control-method lid switch driver
    o pci pci root-bus discovery
    o processor CPU power/speed management
    o thermal system temperature detection and control
    o timer ACPI timecounter
   Multiple components may be disabled by specifying their name(s)
   separated by whitespace.

 - Add support for ioctl registration.  ACPI subsystem components may
   register ioctl handlers with the /dev/acpi generic ioctl handler,
   allowing us to avoid the need for a multitude of /dev/acpi* control
   devices, etc.

23 years agoChange the calling conventions of the MTX_ENTER macro to match
jake [Fri, 8 Dec 2000 08:49:36 +0000 (08:49 +0000)]
Change the calling conventions of the MTX_ENTER macro to match
that of MTX_EXIT.  Don't assume that the reg parameter to MTX_ENTER
holds curproc, load it explicitly.  Put semi-colons at the end of
the macros to be more consistent and so its harder to forget them
when these change.

23 years agoKill some bogus "register" keywords.
phk [Fri, 8 Dec 2000 06:57:39 +0000 (06:57 +0000)]
Kill some bogus "register" keywords.
Go Ansi on the functions.

23 years agoWell, the previous commit wasn't entirely correct either. For now, just
jhb [Fri, 8 Dec 2000 05:03:34 +0000 (05:03 +0000)]
Well, the previous commit wasn't entirely correct either.  For now, just
disable the optimized mutex micro-operations for the non-I386_CPU case
and fall back to the C stubs that call the atomic_foo() inlines.

23 years agoWhen TFTP tries to open a file, it is expecting struct open_file
ps [Fri, 8 Dec 2000 05:02:12 +0000 (05:02 +0000)]
When TFTP tries to open a file, it is expecting struct open_file
member f_devdata to be a pointer to a socket number.  When currdev
is "pxe", that assumption is correct.  When currdev is "disk*", that
assumption is incorrect.

Submitted by: Jim Browne <jbrowne@jbrowne.com>

23 years agoAdd "empty" makefile. Cross-building fails without it.
marcel [Fri, 8 Dec 2000 04:08:09 +0000 (04:08 +0000)]
Add "empty" makefile. Cross-building fails without it.

23 years agoAdd Linuxulator manpage. Derived from i386 version.
marcel [Fri, 8 Dec 2000 02:52:26 +0000 (02:52 +0000)]
Add Linuxulator manpage. Derived from i386 version.

23 years agoThe osfulator can be compiled-in now:
marcel [Fri, 8 Dec 2000 02:49:01 +0000 (02:49 +0000)]
The osfulator can be compiled-in now:
o  add kernel option to the synopsis.
o  remove "only as module" from bugs.

Reviewed by: sheldonh

23 years agoAdd comments to the proc structure to describe how each member will be
jhb [Fri, 8 Dec 2000 00:58:54 +0000 (00:58 +0000)]
Add comments to the proc structure to describe how each member will be
locked.  This list is subject to change, but hopefully many changes will
not have to be made.

23 years agoAdd missing #include.
archie [Fri, 8 Dec 2000 00:49:37 +0000 (00:49 +0000)]
Add missing #include.

23 years agoOnly call bwillwrite() for vnodes. Do not penalize devices or pipes.
dillon [Thu, 7 Dec 2000 23:45:57 +0000 (23:45 +0000)]
Only call bwillwrite() for vnodes.  Do not penalize devices or pipes.

23 years agoWhen I converted this driver, I neglected to deal with packet alignment.
wpaul [Thu, 7 Dec 2000 23:30:51 +0000 (23:30 +0000)]
When I converted this driver, I neglected to deal with packet alignment.
We must force payload alignment to a longword boundary to make the
alpha happy. This should stop the driver from trapping on the alpha
when the interface is ifconfig'ed (actually, when the first frame is
received).

23 years agoHide intrstate in the #ifdef where it belongs.
phk [Thu, 7 Dec 2000 22:38:22 +0000 (22:38 +0000)]
Hide intrstate in the #ifdef where it belongs.

23 years agoDon't use a private malloc(9) type for something this M_DEVBUFish.
phk [Thu, 7 Dec 2000 22:33:12 +0000 (22:33 +0000)]
Don't use a private malloc(9) type for something this M_DEVBUFish.

Noticed long time ago by: bde

23 years agoAdd needed include of <sys/devicestat.h>
phk [Thu, 7 Dec 2000 22:31:13 +0000 (22:31 +0000)]
Add needed include of <sys/devicestat.h>

23 years agoMove extern tsc_present outside function to quelch a warning.
phk [Thu, 7 Dec 2000 22:30:11 +0000 (22:30 +0000)]
Move extern tsc_present outside function to quelch a warning.

23 years agoAdd __ucmpdi2 prototype to quelch a warning.
phk [Thu, 7 Dec 2000 22:28:20 +0000 (22:28 +0000)]
Add __ucmpdi2 prototype to quelch a warning.

23 years agoAdd back some now needed #include <sys/systm.h>
phk [Thu, 7 Dec 2000 22:19:06 +0000 (22:19 +0000)]
Add back some now needed #include <sys/systm.h>

Fix various warnings while here anyway.

23 years agoBuild the sym driver as a kernel module.
alex [Thu, 7 Dec 2000 21:46:04 +0000 (21:46 +0000)]
Build the sym driver as a kernel module.

23 years agoDocument 'untested' status of ExtIO module on AS2100. Even when we
wilko [Thu, 7 Dec 2000 21:35:50 +0000 (21:35 +0000)]
Document 'untested' status of ExtIO module on AS2100. Even when we
don't know if these things were ever produced.

23 years agoDocument support for AlphaServer 2100A.
wilko [Thu, 7 Dec 2000 21:13:39 +0000 (21:13 +0000)]
Document support for AlphaServer 2100A.

Thanks to: gallatin & Compaq for open-sourcing bits of Tru64 UNIX
Tested by: wilko

23 years agoExplicitly document the fact that securelevel > 0 means that kernel modules
ben [Thu, 7 Dec 2000 21:09:22 +0000 (21:09 +0000)]
Explicitly document the fact that securelevel > 0 means that kernel modules
may not be (un)loaded.

PR: 23350
Submitted by: Gordon Tetlow <gordont@bluemtn.net>

23 years agoRemove unneeded include of <pci.h>
alex [Thu, 7 Dec 2000 20:58:16 +0000 (20:58 +0000)]
Remove unneeded include of <pci.h>

Approved by: groudier

23 years agoFix bug where minimal dist selection didn't check off the right
jkh [Thu, 7 Dec 2000 20:04:35 +0000 (20:04 +0000)]
Fix bug where minimal dist selection didn't check off the right
menu item.

Noticed by: Dan "Sheepdip" Langille <dan@freebsddiary.org>

23 years agoRetire this ancient version of rdist.
obrien [Thu, 7 Dec 2000 19:24:55 +0000 (19:24 +0000)]
Retire this ancient version of rdist.

23 years agoput localdir first
obrien [Thu, 7 Dec 2000 16:35:21 +0000 (16:35 +0000)]
put localdir first

23 years agoWhitespace-only to sync with -stable.
ru [Thu, 7 Dec 2000 15:09:48 +0000 (15:09 +0000)]
Whitespace-only to sync with -stable.

23 years agoMFS: Silence compilation warnings.
ru [Thu, 7 Dec 2000 14:59:11 +0000 (14:59 +0000)]
MFS: Silence compilation warnings.

23 years agoCreate a pmtimer device instance for GENERIC and NEWCARD kernels by default.
iwasaki [Thu, 7 Dec 2000 14:27:02 +0000 (14:27 +0000)]
Create a pmtimer device instance for GENERIC and NEWCARD kernels by default.

Submitted by: Masayuki FUKUI <fukui@sonic.nm.fujitsu.co.jp>

23 years agoAdd a PCI vendor/device database, used by pciconf (and available for use
msmith [Thu, 7 Dec 2000 10:56:29 +0000 (10:56 +0000)]
Add a PCI vendor/device database, used by pciconf (and available for use
by other tools as well).

Note that omissions and corrections for this file should be resolved
via http://www.yourvote.com/pci, as this is the master source for this
database, rather than by editing this file directly.

23 years agoUpdate to improve handling of verbose PCI vendor/device information.
msmith [Thu, 7 Dec 2000 10:52:59 +0000 (10:52 +0000)]
Update to improve handling of verbose PCI vendor/device information.

 - Read the database from /usr/share/misc (or wherever else we're pointed)
   rather than compiling it in.
 - Decode the class/subclass fields if requested.
 - Print things in a slightly longer but more readable format.

23 years agoAdd Isochronus transfer mode support required by
roger [Thu, 7 Dec 2000 10:28:25 +0000 (10:28 +0000)]
Add Isochronus transfer mode support required by
USB WebCams, using a patch from Peter Housel.

With this change ugen, and with Peter's 'vid' program
in ports/graphics/vid, we can capture single images from USB Cameras
using the OmniVision OV511 chipset (including some models of the
Creative WebCam 3)

NetBSD merged in Peter's patch to their ugen.c file
several months ago, so this brings us back in line.

Submitted by: Peter Housel <housel@acm.org>
http://members.home.com/housel/
Approved by: Nick Hibma

23 years agoNon functional change.
roger [Thu, 7 Dec 2000 10:13:18 +0000 (10:13 +0000)]
Non functional change.
Change a few indentations to tabs.
Change the functions to use ANSI sytle parameters.

This lowers the diffs between our copy of ugen.c and NetBSD's copy

Approved by: Nick Hibma

23 years agoAdd support for compiling ray driver.
imp [Thu, 7 Dec 2000 03:16:21 +0000 (03:16 +0000)]
Add support for compiling ray driver.

23 years agoFix broken register restraints that needlessly clobbered registers %ecx
jhb [Thu, 7 Dec 2000 02:23:16 +0000 (02:23 +0000)]
Fix broken register restraints that needlessly clobbered registers %ecx
and %edx resulting in gcc not having enough registers left to work with.

23 years agoConvert from spl -> mutex.
archie [Thu, 7 Dec 2000 02:09:39 +0000 (02:09 +0000)]
Convert from spl -> mutex.

23 years agoPartially re-write T2 chipset support based on Tru64 platform support
gallatin [Thu, 7 Dec 2000 01:06:19 +0000 (01:06 +0000)]
Partially re-write T2 chipset support based on Tru64 platform support
files which Compaq open-sourced (with a BSD license).

This commit adds support for proper PCI interrupt mapping and much
better support for swizzling between "standard" isa IRQs and the stdio
irqs used by the t2.  This also adds enabling/disabling/eoi support
for AlphaServer 2100A machines.  The 2100A (or lynx) interrupt
hardware is is very different (and much nicer) than the 2100.
Previously, only AS2100 and AS2000 machines worked.

This commits also lays the groundwork for supporting ExtIO modules.
These modules are essentially a second hose.  This work is left
unfinished pending testing on real hardware.  Wilko tells me that
ExtIO modules are quite rare, and may not actually exist in the wild.

Obtained from: Tru64
Tested by: wilko

23 years agoReally fix phys_pager:
alfred [Wed, 6 Dec 2000 21:52:23 +0000 (21:52 +0000)]
Really fix phys_pager:

Backout the previous delta (rev 1.4), it didn't make any difference.

If the requested handle is NULL then don't add it to the list of
objects, to be found by handle.

The problem is that when asking for a NULL handle you are implying
you want a new object.  Because objects with NULL handles were
being added to the list, any further requests for phys backed
objects with NULL handles would return a reference to the initial
NULL handle object after finding it on the list.

Basically one couldn't have more than one phys backed object without
a handle in the entire system without this fix.  If you did more
than one shared memory allocation using the phys pager it would
give you your initial allocation again.

23 years agoAdd necessary bwillwrite() in writev() entry point.
dillon [Wed, 6 Dec 2000 20:55:09 +0000 (20:55 +0000)]
Add necessary bwillwrite() in writev() entry point.

Deal with excessive dirty buffers when msync() syncs non-contiguous
dirty buffers by checking for the case in UFS *before* checking for
clusterability.

23 years agoDo not duplicate our efforts checking FreeBSD version. For a new version,
ru [Wed, 6 Dec 2000 18:31:21 +0000 (18:31 +0000)]
Do not duplicate our efforts checking FreeBSD version.  For a new version,
simply define the empty string before the .Fx macro definition, like this:

.ds Fx*4.3

23 years agoInstall the stock (as supplied by groff(1) distribution) tmac.an
ru [Wed, 6 Dec 2000 17:02:33 +0000 (17:02 +0000)]
Install the stock (as supplied by groff(1) distribution) tmac.an
and tmac.andoc files.  The man(1), catman(1) and bsd.man.mk have
been modified to call groff(1) with -mandoc argument.

23 years agoBackout rev 1.2 (emulate `:o' and `/o' letters using 'o'),
ru [Wed, 6 Dec 2000 12:45:10 +0000 (12:45 +0000)]
Backout rev 1.2 (emulate `:o' and `/o' letters using 'o'),
since they are already emulated in tmac.tty-char.

23 years agoBuild and install groff(1) texinfo(1) documentation.
ru [Wed, 6 Dec 2000 12:17:01 +0000 (12:17 +0000)]
Build and install groff(1) texinfo(1) documentation.

23 years agoGroff 1.16.1 implements .psbb request internally in troff(1)
ru [Wed, 6 Dec 2000 11:56:34 +0000 (11:56 +0000)]
Groff 1.16.1 implements .psbb request internally in troff(1)
rather than as a macro using the .sy request, so we may drop
the unsafe (-U) mode.

23 years agoUpgrade to groff 1.16.1.
ru [Wed, 6 Dec 2000 11:38:34 +0000 (11:38 +0000)]
Upgrade to groff 1.16.1.

23 years agoHmm, font `L' is still used in old Sun docs.
ru [Wed, 6 Dec 2000 11:25:33 +0000 (11:25 +0000)]
Hmm, font `L' is still used in old Sun docs.

23 years agoThe PAM support for our OpenSSH is sponsored by Enitel ASA.
green [Wed, 6 Dec 2000 10:57:57 +0000 (10:57 +0000)]
The PAM support for our OpenSSH is sponsored by Enitel ASA.

PAM support by: Eivind Eklund <eivind@FreeBSD.org>

23 years agoThese groff(1) documents require sed(1) parsing.
ru [Wed, 6 Dec 2000 10:32:40 +0000 (10:32 +0000)]
These groff(1) documents require sed(1) parsing.

23 years agoReduce the diffs with the stock (FSF) version.
ru [Wed, 6 Dec 2000 09:32:10 +0000 (09:32 +0000)]
Reduce the diffs with the stock (FSF) version.

23 years agoSomewhere along the line, I misunderstood the whole FTP_PASSIVE_MODE debate
des [Wed, 6 Dec 2000 09:23:27 +0000 (09:23 +0000)]
Somewhere along the line, I misunderstood the whole FTP_PASSIVE_MODE debate
and had libfetch selecting passive mode even when FTP_PASSIVE_MODE was not
set at all, which is really quite surprising unless you know about it. So
change it to the agreed default behaviour of selecting passive mode if
FTP_PASSIVE_MODE is set, but not "no".

23 years agoBackout rev 1.2 (workaround for gcc 2.60 expression evaluation
ru [Wed, 6 Dec 2000 09:21:36 +0000 (09:21 +0000)]
Backout rev 1.2 (workaround for gcc 2.60 expression evaluation
bug) as later versions of gcc have apparently this bug fixed.

23 years agoFont `L' is not needed for BSD docs, but some use font `CW'.
ru [Wed, 6 Dec 2000 09:01:07 +0000 (09:01 +0000)]
Font `L' is not needed for BSD docs, but some use font `CW'.

23 years agoUse stock (FSF) version of this file.
ru [Wed, 6 Dec 2000 08:49:05 +0000 (08:49 +0000)]
Use stock (FSF) version of this file.

23 years agoFix bogon in previous commit, whereby the contents of SUBDIR were lost
sheldonh [Wed, 6 Dec 2000 08:45:03 +0000 (08:45 +0000)]
Fix bogon in previous commit, whereby the contents of SUBDIR were lost
in the assignment of man4.alpha!

This prevented the man4.i386 pages from being installed.

Reported by: Masayuki FUKUI <fukui@sonic.nm.fujitsu.co.jp>

23 years agoFixed device files building procedure. This makes
ru [Wed, 6 Dec 2000 08:35:46 +0000 (08:35 +0000)]
Fixed device files building procedure.  This makes
it possible to ``make all install'' in one pass.

23 years agoUntangle vfsinit() a bit. Use seperate sysinit functions rather than
peter [Wed, 6 Dec 2000 07:09:08 +0000 (07:09 +0000)]
Untangle vfsinit() a bit.  Use seperate sysinit functions rather than
having a super-function calling bits all over the place.

23 years agoSimplify this a bit so that it doesn't have to generate silly redundant
peter [Wed, 6 Dec 2000 06:59:38 +0000 (06:59 +0000)]
Simplify this a bit so that it doesn't have to generate silly redundant
__P() prototypes when an ansi-style static inline is a prototype already.
Since vnode_if.[ch] are generated on the fly, there are no CVS diffs to
mess up.

23 years agoBack this out, we apparently have the ipfw(4).
ru [Wed, 6 Dec 2000 06:50:24 +0000 (06:50 +0000)]
Back this out, we apparently have the ipfw(4).

23 years agoipfw(4) -> ipfw(8).
ru [Wed, 6 Dec 2000 06:48:28 +0000 (06:48 +0000)]
ipfw(4) -> ipfw(8).

23 years agoAdd back displaying the `world' start time when displaying the stop time.
obrien [Wed, 6 Dec 2000 04:35:45 +0000 (04:35 +0000)]
Add back displaying the `world' start time when displaying the stop time.
However, don't prefix it with '>>>', so those greping for times will
not see it.

23 years agoThis is kind of a nasty hack, but it appears to solve the Compaq DL360
peter [Wed, 6 Dec 2000 03:47:14 +0000 (03:47 +0000)]
This is kind of a nasty hack, but it appears to solve the Compaq DL360
SMP problem.  Compaq, in their infinite wisdom, forgot to put the IO apic
intpin #0 connection to the 8259 PIC into the mptable.  This hack is to
look and see if intpin #0 has *no* table entry and adds a fake ExtInt
entry for the remap routines to use.  isa/clock.c will still test the
interrupts.  This entry is only ever used on an already broken system.

23 years agoPass RFSTOPPED to fork1() in kthread_create() to avoid a race condition
jhb [Wed, 6 Dec 2000 03:45:15 +0000 (03:45 +0000)]
Pass RFSTOPPED to fork1() in kthread_create() to avoid a race condition
where fork1() could put the process on the run queue where it could be
snatched up by another CPU before kthread_create() had set the proper
fork handler.  Instead, we put the new kthread on the runqueue after its
fork handler has been sent.

Noticed by: jake
Looked over by: peter

23 years agoCleanup XXXdir functions to eliminate global hash table of
deischen [Wed, 6 Dec 2000 03:15:49 +0000 (03:15 +0000)]
Cleanup XXXdir functions to eliminate global hash table of
telldir positions.  This will allow (future) locking on a
per-DIR basis (for MT-safety).  For now, this change does
the following:

  o Remove the hash table from telldir.c.  Recode to use queue
    macros.

  o Remove 'const' from 'telldir(const DIR *)'.

  o Remove 'register' variables as suggested in a recent
    thread.

No response from: -current

23 years agoAdd a LIST_HEAD to DIR for a queue of telldir positions.
deischen [Wed, 6 Dec 2000 03:14:28 +0000 (03:14 +0000)]
Add a LIST_HEAD to DIR for a queue of telldir positions.
Also add a location count (used as the magic for telldir)
to DIR.  A future change will also add a mutex/lock.

23 years ago- Add in PROC_LOCK() and PROC_UNLOCK() macros. For now these do simple
jhb [Wed, 6 Dec 2000 02:01:56 +0000 (02:01 +0000)]
- Add in PROC_LOCK() and PROC_UNLOCK() macros.  For now these do simple
  mutex operations.  In the future they may call functions that verify
  correct locking order between processes in the INVARIANTS case.
- Lock the process in PHOLD() and PREL().

23 years agoAdd appropriate defines to use snprintf/vsnprintf instead of
jedgar [Wed, 6 Dec 2000 01:49:08 +0000 (01:49 +0000)]
Add appropriate defines to use snprintf/vsnprintf instead of
sprintf/vsprintf.

Approved by: peter

23 years agoProtect accesses to member of struct proc with the proc lock.
jhb [Wed, 6 Dec 2000 01:45:20 +0000 (01:45 +0000)]
Protect accesses to member of struct proc with the proc lock.

23 years agoMove io_apic_{read,write} from apic_ipl.s (where they do not belong) into
peter [Wed, 6 Dec 2000 01:04:02 +0000 (01:04 +0000)]
Move io_apic_{read,write} from apic_ipl.s (where they do not belong) into
mpapic.c.  This gives us the benefit of C type checking.  These functions
are not called in any critical paths and are not used by the interrupt
routines.

23 years agoRestore a sense of cleanly supporting multiple platforms. That is,
mjacob [Wed, 6 Dec 2000 00:52:28 +0000 (00:52 +0000)]
Restore a sense of cleanly supporting multiple platforms. That is,
place the LOCKing macros within the areas within if_wxvar.h that
is set aside for them. Put any platform specific data also in those
areas.

For ease of maintenance purposes, merge in the OpenBSD version codebase here.

23 years agoMove $FreeBSD id up to top of file for multi-OS ease of support reasons.
mjacob [Wed, 6 Dec 2000 00:50:53 +0000 (00:50 +0000)]
Move $FreeBSD id up to top of file for multi-OS ease of support reasons.

23 years agoEnable the same path deduction as for ftp install (great for snapshot tests)
jkh [Wed, 6 Dec 2000 00:47:54 +0000 (00:47 +0000)]
Enable the same path deduction as for ftp install (great for snapshot tests)
Allow user to change proxy server
Allow user to change ftp path

Submitted by: pho

23 years agoGC unused assembler function apic_eoi()
peter [Wed, 6 Dec 2000 00:38:04 +0000 (00:38 +0000)]
GC unused assembler function apic_eoi()

23 years agoAdd in #include of <sys/lock.h> since it was axed from <sys/proc.h>.
jhb [Wed, 6 Dec 2000 00:33:58 +0000 (00:33 +0000)]
Add in #include of <sys/lock.h> since it was axed from <sys/proc.h>.

Noticed by: Wesley Morgan <morganw@chemikals.org>
Pointy hat to: me

23 years agoAdd forgotten SYSCALL_MODULE_HELPER() for msgsys() syscall.
alfred [Tue, 5 Dec 2000 23:05:45 +0000 (23:05 +0000)]
Add forgotten SYSCALL_MODULE_HELPER() for msgsys() syscall.

Discovered by: Valentin Chopov <valentin@valcho.net>

23 years agoneed to adjust allocation size to properly deal with non PAGE_SIZE
alfred [Tue, 5 Dec 2000 22:22:24 +0000 (22:22 +0000)]
need to adjust allocation size to properly deal with non PAGE_SIZE
allocations, specifically with allocations < PAGE_SIZE when the code
doesn't work properly

23 years agoChange the spelling of `.' to `.' from .OBJDIR since `.' really is where
obrien [Tue, 5 Dec 2000 22:10:43 +0000 (22:10 +0000)]
Change the spelling of `.' to `.' from .OBJDIR since `.' really is where
generated files land.  Also give precedence to generated files.

23 years agoMove the ipfw(4) xref to the description of LOG_SECURITY instead of
jhb [Tue, 5 Dec 2000 20:30:28 +0000 (20:30 +0000)]
Move the ipfw(4) xref to the description of LOG_SECURITY instead of
LOG_UUCP.

PR: docs/23302
Submitted by: cshumway

23 years ago1. Several style cleanups:
jhb [Tue, 5 Dec 2000 20:23:43 +0000 (20:23 +0000)]
1. Several style cleanups:
   - Whitespace fixes.
   - Comment fixes (mostly capitalization and trailing periods).
   - Reorderings to put variables all in the same place, function prototypes
     sorted alphabetically, etc.
2. Removed unused and #ifdef'd out members from struct ithd.
3. Changed ESTCPULIM() to use an explicit (PRIO_MAX - PRIO_MIN) rather than
   PRIO_TOTAL.
4. Remove <sys/lock.h> #include and resort #include's.

Submitted by: bde (1, 3, 4)

23 years agoNote that Symbios is the new owner of the NCR SCSI chip designs. And
wilko [Tue, 5 Dec 2000 20:21:18 +0000 (20:21 +0000)]
Note that Symbios is the new owner of the NCR SCSI chip designs. And
that the document references to Symbios are ment to include NCR-stamped
chips as well.

Suggested by: obrien

23 years agogroff does not use MANDEPEND anymore.
ru [Tue, 5 Dec 2000 19:56:50 +0000 (19:56 +0000)]
groff does not use MANDEPEND anymore.

23 years agoAdded directories needed for Groff 1.16.1 distribution.
ru [Tue, 5 Dec 2000 19:18:45 +0000 (19:18 +0000)]
Added directories needed for Groff 1.16.1 distribution.

23 years agoUpgrade to version 1.16.1.
ru [Tue, 5 Dec 2000 19:15:36 +0000 (19:15 +0000)]
Upgrade to version 1.16.1.

23 years agoResolve conflicts after v1.16.1 import.
ru [Tue, 5 Dec 2000 19:06:41 +0000 (19:06 +0000)]
Resolve conflicts after v1.16.1 import.