]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoOnly create directories that are really needed.
Ruslan Ermilov [Sun, 13 Apr 2003 14:34:17 +0000 (14:34 +0000)]
Only create directories that are really needed.

21 years agoOops. Back out a style "fix" in the previous commit. SRCS must be defined
Bruce Evans [Sun, 13 Apr 2003 14:21:56 +0000 (14:21 +0000)]
Oops.  Back out a style "fix" in the previous commit.  SRCS must be defined
although there is only one file in it because the file is misnamed main.c
so the default of isdntest.c doesn't work.

21 years agoSilence `make -s' (echo -> ${ECHO}).
Bruce Evans [Sun, 13 Apr 2003 14:16:06 +0000 (14:16 +0000)]
Silence `make -s' (echo -> ${ECHO}).

Fixed some style bugs.

21 years agoSilence `make -s' (echo -> ${ECHO}).
Bruce Evans [Sun, 13 Apr 2003 14:13:28 +0000 (14:13 +0000)]
Silence `make -s' (echo -> ${ECHO}).

21 years agoInstall bootstrap-tools into a separate subtree of ${WORLDTMP}.
Ruslan Ermilov [Sun, 13 Apr 2003 11:42:55 +0000 (11:42 +0000)]
Install bootstrap-tools into a separate subtree of ${WORLDTMP}.
This allows us to use them as early as possible while building
bootstrap-, build-, and cross-tools.  Some cleanups to follow.

This change resolves the gperf(1) bootstrapping issue (missing
-E option) in gnu/usr.bin/cc/cc1plus while in the cross-tools
stage when upgrading from 4.0-RELEASE.

21 years agoRemove the "run GEOM in userland harness" test stuff, it's been OBE.
Poul-Henning Kamp [Sun, 13 Apr 2003 09:11:47 +0000 (09:11 +0000)]
Remove the "run GEOM in userland harness" test stuff, it's been OBE.

21 years agoGive mdconfig a couple of seconds to deconfigure in case debugging is being
Poul-Henning Kamp [Sun, 13 Apr 2003 09:06:39 +0000 (09:06 +0000)]
Give mdconfig a couple of seconds to deconfigure in case debugging is being
printed to a serial console.

Reverse the order of args to diff to make it more logical.

21 years agoTime has run from the "run GEOM in userland" harness, and the new regression
Poul-Henning Kamp [Sun, 13 Apr 2003 09:02:06 +0000 (09:02 +0000)]
Time has run from the "run GEOM in userland" harness, and the new regression
test is built to test GEOM as running in the kernel.

This commit is basically "unifdef -D_KERNEL" to remove the mainly #include
related code to support the userland-harness.

21 years agoInclude <signal.h> instead of depending on namespace pollution in
Bruce Evans [Sun, 13 Apr 2003 08:47:30 +0000 (08:47 +0000)]
Include <signal.h> instead of depending on namespace pollution in
<sys/param.h>.  Include <sys/types.h> instead of of <sys/param.h>
so that further such dependencies don't develop.

21 years agoFixed some style bugs in rev.1.44.
Bruce Evans [Sun, 13 Apr 2003 08:27:35 +0000 (08:27 +0000)]
Fixed some style bugs in rev.1.44.

21 years agoDriver for Granch SBNI16 SHDSL modem
Max Khon [Sun, 13 Apr 2003 06:27:13 +0000 (06:27 +0000)]
Driver for Granch SBNI16 SHDSL modem

Submitted by: Denis I. Timofeev <timofeev@granch.ru>
MFC after: 1 week

21 years agoNo need to unlock if error detected before locking.
Jeffrey Hsu [Sun, 13 Apr 2003 06:21:02 +0000 (06:21 +0000)]
No need to unlock if error detected before locking.

Submitted by: harti

21 years agomanual page for Granch SBNI12-xx network interface driver
Max Khon [Sun, 13 Apr 2003 06:13:47 +0000 (06:13 +0000)]
manual page for Granch SBNI12-xx network interface driver

21 years agoUse vm_object_pip_wait() rather than reimplementing it.
Alan Cox [Sun, 13 Apr 2003 05:10:44 +0000 (05:10 +0000)]
Use vm_object_pip_wait() rather than reimplementing it.

21 years agoThe dpt, ex and idt modules are for i386 only.
Yoshihiro Takahashi [Sun, 13 Apr 2003 01:32:03 +0000 (01:32 +0000)]
The dpt, ex and idt modules are for i386 only.

21 years agoThe pcspeaker was for pc98.
Yoshihiro Takahashi [Sun, 13 Apr 2003 01:16:16 +0000 (01:16 +0000)]
The pcspeaker was for pc98.

21 years agoTeach gcc how to put unitialized data into BSS on FreeBSD.
Alexander Kabaev [Sun, 13 Apr 2003 01:04:01 +0000 (01:04 +0000)]
Teach gcc how to put unitialized data into BSS on FreeBSD.

21 years agoPermit vm_object_pip_add() and vm_object_pip_wakeup() on the kmem_object
Alan Cox [Sun, 13 Apr 2003 00:43:48 +0000 (00:43 +0000)]
Permit vm_object_pip_add() and vm_object_pip_wakeup() on the kmem_object
without Giant held.

21 years agoEliminate unnecessary gotos from kmem_malloc().
Alan Cox [Sun, 13 Apr 2003 00:23:42 +0000 (00:23 +0000)]
Eliminate unnecessary gotos from kmem_malloc().

21 years ago - Unbreak priority prop. for timeshare threads. Always place something on
Jeff Roberson [Sat, 12 Apr 2003 22:33:24 +0000 (22:33 +0000)]
 - Unbreak priority prop. for timeshare threads.  Always place something on
   the current queue if its priority is really elevated.  This needs more work
   as there are cases where a next queue kse could be holding up what would
   be a curr queue kse, and thus hurting interactivity.  Also, when a thread
   with an elevated priority has its priority lowered it should be placed
   back on the next queue.

21 years agoIf we hit access ahead of a spoil event, we should have negative
Poul-Henning Kamp [Sat, 12 Apr 2003 17:04:34 +0000 (17:04 +0000)]
If we hit access ahead of a spoil event, we should have negative
delta access-counts and proceed.

21 years agoNew release notes: MADV_PROTECT, vpd(4), DIRECTIO,
Bruce A. Mah [Sat, 12 Apr 2003 16:38:33 +0000 (16:38 +0000)]
New release notes:  MADV_PROTECT, vpd(4), DIRECTIO,
pcspeaker.ko/atspeaker.ko -> speaker.ko, pkg_info(1) -b, pkg_info(1)
-Q, removal of obsolete upgrade mechanisms.

21 years agoFix a bug which resulted in orphanization getting confused every now
Poul-Henning Kamp [Sat, 12 Apr 2003 16:36:19 +0000 (16:36 +0000)]
Fix a bug which resulted in orphanization getting confused every now
and then.

21 years agoModified release notes: GNOME 2.2.1, KDE 3.1.1a.
Bruce A. Mah [Sat, 12 Apr 2003 15:21:10 +0000 (15:21 +0000)]
Modified release notes:  GNOME 2.2.1, KDE 3.1.1a.

21 years agoWhitespace cleanup from the sect3->sect2 reorg.
Bruce A. Mah [Sat, 12 Apr 2003 15:18:52 +0000 (15:18 +0000)]
Whitespace cleanup from the sect3->sect2 reorg.

Translation teams can ignore this.

21 years agoAdd id attributes for all sect2 and sect3 sections.
Bruce A. Mah [Sat, 12 Apr 2003 15:13:26 +0000 (15:13 +0000)]
Add id attributes for all sect2 and sect3 sections.

21 years agoMake sure legacy stuff comes last. Some tools that use
Ruslan Ermilov [Sat, 12 Apr 2003 15:07:46 +0000 (15:07 +0000)]
Make sure legacy stuff comes last.  Some tools that use
legacy stuff (binutils) depend on this order.

For this to work, provide (and use) specialized versions
of bsd.prog.mk and bsd.lib.mk that include the standard
versions first, then augment CFLAGS, DPADD, LDADD, and
LDFLAGS as necessary, with the legacy stuff.

Tested on: 4.0-RELEASE

21 years agoSlight reorganization of the release notes: Make the contributed
Bruce A. Mah [Sat, 12 Apr 2003 15:07:04 +0000 (15:07 +0000)]
Slight reorganization of the release notes:  Make the contributed
software section and ports/packages infrastructure section be sect2
level sections (instead of sect3 under userland).  Combine the
kernel/contributed and userland/contributed entries into the new sect2
contributed section.

The main point of doing this is that the distinction between kernel
and userland is largely irrelevent for contributed software, and
doesn't apply well to things such as IPFilter and KAME, which have
both kernel and userland components.

21 years agoRemoved invasion into these makefiles by the "legacy" stuff;
Ruslan Ermilov [Sat, 12 Apr 2003 14:44:49 +0000 (14:44 +0000)]
Removed invasion into these makefiles by the "legacy" stuff;
I have a better fix in the works.

21 years agoAvoid emitting duplicate makefile entries.
Poul-Henning Kamp [Sat, 12 Apr 2003 14:02:03 +0000 (14:02 +0000)]
Avoid emitting duplicate makefile entries.

21 years agoFix a typo in the -O parsing code that caused a crash when the -O option
Tim J. Robbins [Sat, 12 Apr 2003 12:11:14 +0000 (12:11 +0000)]
Fix a typo in the -O parsing code that caused a crash when the -O option
is used without supplying "sowner:sgroup".

Obtained from: NetBSD (jdolecek)
PR: 37171

21 years agoDisplay residency and sleep times (re and sl fields) larger than 127 as 127.
Tim J. Robbins [Sat, 12 Apr 2003 10:39:56 +0000 (10:39 +0000)]
Display residency and sleep times (re and sl fields) larger than 127 as 127.
This is what the manual page says ps should do, and what OpenBSD and NetBSD do.
Based on a patch from Ken Stailey.

PR: 27433, 46232

21 years agoprotoypes for the dos_partition encoding functions.
Poul-Henning Kamp [Sat, 12 Apr 2003 10:07:12 +0000 (10:07 +0000)]
protoypes for the dos_partition encoding functions.

21 years agoOptions that go into homonymous headers shouldn't specify the header name.
Dag-Erling Smørgrav [Sat, 12 Apr 2003 09:40:37 +0000 (09:40 +0000)]
Options that go into homonymous headers shouldn't specify the header name.

Submitted by: Hiten Pandya <hiten@unixdaemons.com>

21 years agoRetire the experimental bio_taskqueue(), it was not quite as usable as
Poul-Henning Kamp [Sat, 12 Apr 2003 09:13:01 +0000 (09:13 +0000)]
Retire the experimental bio_taskqueue(), it was not quite as usable as
hoped.  It can be revived from here, should other drivers be able to
use it.

21 years agoRemove the unused ioctl routine.
Poul-Henning Kamp [Sat, 12 Apr 2003 08:44:54 +0000 (08:44 +0000)]
Remove the unused ioctl routine.

21 years agoRetire the "frontstuff" record keeping, it was no match for the
Poul-Henning Kamp [Sat, 12 Apr 2003 08:41:26 +0000 (08:41 +0000)]
Retire the "frontstuff" record keeping, it was no match for the
in-band meta-data of BSD labels and a more complex solution will be needed.

21 years agoMove the functions for encoding decoding struct dos_partition into
Poul-Henning Kamp [Sat, 12 Apr 2003 08:34:40 +0000 (08:34 +0000)]
Move the functions for encoding decoding struct dos_partition into
a separate .c file so they can be used from userland as well.

21 years agoTrust the code more than the comment(s) and correct some false
Robert Drehmel [Sat, 12 Apr 2003 07:36:44 +0000 (07:36 +0000)]
Trust the code more than the comment(s) and correct some false
statements about pointer data type sizes, which spread probably
by copy-and-paste.

21 years ago - Clean up some debug code left over from my earlier megacommit.
Jeff Roberson [Sat, 12 Apr 2003 07:28:36 +0000 (07:28 +0000)]
 - Clean up some debug code left over from my earlier megacommit.

21 years ago - We only care about the base priority. Ignore the SCHED_FIFO_BIT so that
Jeff Roberson [Sat, 12 Apr 2003 07:00:16 +0000 (07:00 +0000)]
 - We only care about the base priority.  Ignore the SCHED_FIFO_BIT so that
   we dont get confused.

Reported and debugged by: Steve Kargl <sgk@troutmask.apl.washington.edu>

21 years agofix typo in the previous commit.
Hidetoshi Shimokawa [Sat, 12 Apr 2003 06:43:28 +0000 (06:43 +0000)]
fix typo in the previous commit.

21 years agoRename MBUF_FRAG_TEST to MBUF_STRESS_TEST as it will be extended
Mike Silbersack [Sat, 12 Apr 2003 06:11:46 +0000 (06:11 +0000)]
Rename MBUF_FRAG_TEST to MBUF_STRESS_TEST as it will be extended
to include more than just frag tests.

21 years agoAdd terse descriptions for NLSPATH and LC_*.
Tim J. Robbins [Sat, 12 Apr 2003 05:20:15 +0000 (05:20 +0000)]
Add terse descriptions for NLSPATH and LC_*.

21 years agoMention the environment variables that affect the execution of uniq.
Tim J. Robbins [Sat, 12 Apr 2003 04:17:14 +0000 (04:17 +0000)]
Mention the environment variables that affect the execution of uniq.
Of particular interest is the fact that LC_COLLATE affects how uniq
determines whether lines are equal. This was the subject of a fairly heated
debate a year or so ago, and it turns out that the current behaviour is
correct and that the standard contained an error.

Now that the standard has been corrected by Cor. 1-2002, refer to 1003.1-2001
instead of the 1992 edition in the Standards section.

21 years agoRefer to 1003.1-2001 in the Standards section now that TC1 allows the
Tim J. Robbins [Sat, 12 Apr 2003 03:49:28 +0000 (03:49 +0000)]
Refer to 1003.1-2001 in the Standards section now that TC1 allows the
traditional BSD behaviour (-n option) and FreeBSD extensions (\c).

21 years agoDocument the -L and -P options as being mutually exclusive.
Tim J. Robbins [Sat, 12 Apr 2003 03:22:54 +0000 (03:22 +0000)]
Document the -L and -P options as being mutually exclusive.

21 years agoDocument the -L and -P options to the cd and pwd commands as being
Tim J. Robbins [Sat, 12 Apr 2003 03:20:28 +0000 (03:20 +0000)]
Document the -L and -P options to the cd and pwd commands as being
mutually exclusive. The fact that the most recent one specified on the
command line is the one that takes effect is an implementation detail and
users should not rely on this.

21 years agoStyle fix.
David Xu [Sat, 12 Apr 2003 02:54:46 +0000 (02:54 +0000)]
Style fix.

21 years agoFix race between a process registering a NOTE_EXIT EVFILT_PROC event and
Kelly Yancey [Sat, 12 Apr 2003 01:57:04 +0000 (01:57 +0000)]
Fix race between a process registering a NOTE_EXIT EVFILT_PROC event and
the target process exiting which causes attempts to register the kevent
to randomly fail depending on whether the target runs to completion before
the parent can call kevent(2).  The bug actually effects EVFILT_PROC
events on any zombie process, but the most common manifestation is with
parents trying to monitor child processes.

MFC after: 2 weeks
Sponsored by: NTT Multimedia Communications Labs

21 years ago - Use the sched_nice() api instead of setting the nice value directly.
Jeff Roberson [Sat, 12 Apr 2003 01:05:19 +0000 (01:05 +0000)]
 - Use the sched_nice() api instead of setting the nice value directly.

Tested by: Steve Kargl <sgk@troutmask.apl.washington.edu>

21 years agoCheck SIG_HOLD action ealier to avoid missing test it in later code.
David Xu [Sat, 12 Apr 2003 00:38:47 +0000 (00:38 +0000)]
Check SIG_HOLD action ealier to avoid missing test it in later code.

21 years agoDefault the userland to 486 capabilities.
David E. O'Brien [Fri, 11 Apr 2003 21:28:17 +0000 (21:28 +0000)]
Default the userland to 486 capabilities.

21 years ago - Call sched_exit_{kse,thread} and sched_fork{kse,thread} so that thr works
Jeff Roberson [Fri, 11 Apr 2003 19:24:37 +0000 (19:24 +0000)]
 - Call sched_exit_{kse,thread} and sched_fork{kse,thread} so that thr works
   with ULE.  This was not strictly required by sched_4bsd.

21 years ago - Add sched_exit_*
Jeff Roberson [Fri, 11 Apr 2003 19:24:00 +0000 (19:24 +0000)]
 - Add sched_exit_*
 - Call sched_exit_kse() from sched_exit() instead of implementing it here.

21 years ago - Only select kseqs with more than one kse to steal. The running kse
Jeff Roberson [Fri, 11 Apr 2003 18:40:34 +0000 (18:40 +0000)]
 - Only select kseqs with more than one kse to steal.  The running kse
   is reflected in the load now and you can't very well migrate that.

21 years ago - When migrating a kse from one kseq to the next actually insert it onto
Jeff Roberson [Fri, 11 Apr 2003 18:37:34 +0000 (18:37 +0000)]
 - When migrating a kse from one kseq to the next actually insert it onto
   the second kseq's run queue so that it is referenced by the kse when
   it is switched out.
 - Spell ksq_rslices properly.

Reported by: Ian Freislich <ianf@za.uu.net>

21 years agolibbinutils.a needs basename(3); when the latter is provided
Ruslan Ermilov [Fri, 11 Apr 2003 18:02:10 +0000 (18:02 +0000)]
libbinutils.a needs basename(3); when the latter is provided
by -legacy, liblegacy.a should come last in the LDADD list.

21 years agolibc_gen/basename.c depends on include/libgen.h.
Ruslan Ermilov [Fri, 11 Apr 2003 17:58:17 +0000 (17:58 +0000)]
libc_gen/basename.c depends on include/libgen.h.

21 years agoAdd new sysctl MIB (hw.acpi.supported_sleep_state) to indicate
Mitsuru IWASAKI [Fri, 11 Apr 2003 16:53:56 +0000 (16:53 +0000)]
Add new sysctl MIB (hw.acpi.supported_sleep_state) to indicate
the list of supported sleep state.
This should help people understand what following message means.

acpi0: AcpiGetSleepTypeData failed - AE_NOT_FOUND

MFC after: 3 days

21 years agoClarify NO_SWAPPING description.
Mike Barcroft [Fri, 11 Apr 2003 14:48:13 +0000 (14:48 +0000)]
Clarify NO_SWAPPING description.

21 years agoCopy %gs from current CPU not from a stale PCB backup.
David Xu [Fri, 11 Apr 2003 14:47:34 +0000 (14:47 +0000)]
Copy %gs from current CPU not from a stale PCB backup.

21 years agoset_user_ldt_rv() should check same proc not thread,
David Xu [Fri, 11 Apr 2003 14:45:07 +0000 (14:45 +0000)]
set_user_ldt_rv() should check same proc not thread,
this commit fixes an user LDT smp rendezvous bug.

21 years agoReturn "/boot/kernel/kernel" instead of "/kernel" if the sysctl() call
Tim J. Robbins [Fri, 11 Apr 2003 13:54:28 +0000 (13:54 +0000)]
Return "/boot/kernel/kernel" instead of "/kernel" if the sysctl() call
fails. The documentation was incorrectly updated in getbootfile.3 rev. 1.10.

21 years agoStyle(9) a comment added in previous commit.
Andrew Gallatin [Fri, 11 Apr 2003 13:30:32 +0000 (13:30 +0000)]
Style(9) a comment added in previous commit.

Pointed out by: bde

21 years ago- Call ether_ifdetach() before shutting down the hardware.
Maxime Henrion [Fri, 11 Apr 2003 12:52:52 +0000 (12:52 +0000)]
- Call ether_ifdetach() before shutting down the hardware.
- Remove a useless device_is_alive() check.
- Disable interrupts if bus_child_present() so that this
  check is more useful.

This fixes the hangs I was seeing when unloading the fxp driver.

Suggestions from: hsu, njl

21 years agoThe data in an sf_buf should not be modified by the mbuf system. Mark
Alan Cox [Fri, 11 Apr 2003 07:02:36 +0000 (07:02 +0000)]
The data in an sf_buf should not be modified by the mbuf system.  Mark
the mbuf as read only.

Reviewed by: gallatin

21 years ago - Add a SYSCTL node for the ule scheduler.
Jeff Roberson [Fri, 11 Apr 2003 03:47:14 +0000 (03:47 +0000)]
 - Add a SYSCTL node for the ule scheduler.
 - Allow user adjustable min and max time slices (suggested by hiten).
 - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a
   process is interactive or not much more quickly.
 - Place a process on the current run queue if it is interactive or if it is
   running at an interrupt thread priority due to priority prop.
 - Use the 'current' timeshare queue for interrupt threads, realtime threads,
   and idle threads that are running at higher priority due to priority prop.
   This fixes problems where priorities would have been elevated but we would
   not check the timeshare run queue until other lower priority tasks were
   no longer runnable.
 - Keep an array of loads indexed by the priority class as well as a global
   load.
 - Keep an bucket of nice values with a count of the number of kses currently
   runnable with that nice value.
 - Keep track of the minimum nice value of any running thread.
 - Remove the unused short term sleep accounting.  I was attempting to use
   this for load balancing but it didn't work out.
 - Define a kseq_print() for use with debugging.
 - Add KTR debugging at useful places so we can easily debug slice and
   priority assignment.
 - Decouple the runq assignment from the kseq assignment.  kseq_add now keeps
   track of statistics.  This is done so that the nice and load is still
   tracked for the currently running process.  Previously if a niced process
   was added while a non nice process was running the niced process would
   still get a slice since it was not aware of the unnice process.
 - Make adjustments for the sched api changes.

21 years ago - Catch up with sched api changes.
Jeff Roberson [Fri, 11 Apr 2003 03:39:48 +0000 (03:39 +0000)]
 - Catch up with sched api changes.

21 years ago - Adjust sched hooks for fork and exec to take processes as arguments instead
Jeff Roberson [Fri, 11 Apr 2003 03:39:07 +0000 (03:39 +0000)]
 - Adjust sched hooks for fork and exec to take processes as arguments instead
   of ksegs since they primarily operation on processes.
 - KSEs take ticks so pass the kse through sched_clock().
 - Add a sched_class() routine that adjusts a ksegrp pri class.
 - Define a sched_fork_{kse,thread,ksegrp} and sched_exit_{kse,thread,ksegrp}
   that will be used to tell the scheduler about new instances of these
   structures within the same process.  These will be used by THR and KSE.
 - Change sched_4bsd to reflect this API update.

21 years agofxp(4) works on sparc64.
Bruce A. Mah [Fri, 11 Apr 2003 03:00:08 +0000 (03:00 +0000)]
fxp(4) works on sparc64.

21 years agoMention DARPA and Network Associate Laboratories.
Jake Burkholder [Fri, 11 Apr 2003 02:53:49 +0000 (02:53 +0000)]
Mention DARPA and Network Associate Laboratories.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoPut the MAKEDEV item in correct alphabetical order.
Bruce A. Mah [Fri, 11 Apr 2003 00:51:15 +0000 (00:51 +0000)]
Put the MAKEDEV item in correct alphabetical order.

21 years agoNew release notes: PAE, diskinfo(8), jail(8) -i, jexec(8), jls(8),
Bruce A. Mah [Fri, 11 Apr 2003 00:50:33 +0000 (00:50 +0000)]
New release notes:  PAE, diskinfo(8), jail(8) -i, jexec(8), jls(8),
killall(1) -j, mount(8)/umount(8) -F, rtld(1) dynamic mapping and
libmap.conf(5), 1:1 threading.

Modified release notes:  Use wlan(4) manpage hyperlink.

21 years agoI deserve a big pointy hat for having missed all those references
Maxime Henrion [Thu, 10 Apr 2003 23:50:06 +0000 (23:50 +0000)]
I deserve a big pointy hat for having missed all those references
to bus_dmasync_op_t in my last commit.

21 years ago- Clean up the fxp_release() and fxp_detach() functions.
Maxime Henrion [Thu, 10 Apr 2003 23:15:12 +0000 (23:15 +0000)]
- Clean up the fxp_release() and fxp_detach() functions.
- Be sure to teardown the interrupt first so that "kldunload if_fxp"
  doesn't panic the box.  It's now deadlocking rather than crashing,
  which isn't really better, but I'm unsure this is fxp(4)'s fault.
- Change a bus_dmamap_sync() call to also do a BUS_DMASYNC_PREREAD
  now that we can pass several operations.

21 years agoConvert the SMP_TSC kernel option into a loader tunable. Also enable
Dag-Erling Smørgrav [Thu, 10 Apr 2003 23:07:24 +0000 (23:07 +0000)]
Convert the SMP_TSC kernel option into a loader tunable.  Also enable
the TSC timecounter on single-CPU systems even when they are running
an SMP kernel.

21 years agoChange the operation parameter of bus_dmamap_sync() from an
Maxime Henrion [Thu, 10 Apr 2003 23:03:33 +0000 (23:03 +0000)]
Change the operation parameter of bus_dmamap_sync() from an
enum to an int and redefine the BUS_DMASYNC_* constants as
flags.  This allows us to specify several operations in one
call to bus_dmamap_sync() as in NetBSD.

21 years agoQuote authname and authkey. Some ISPs allow (or even require)
Giorgos Keramidas [Thu, 10 Apr 2003 21:33:02 +0000 (21:33 +0000)]
Quote authname and authkey.  Some ISPs allow (or even require)
non-alphanumeric characters in these and this will be a hint to the
users that quotes can and should be used in such cases.

PR:             docs/42292
Submitted by:   Matthias Andree <matthias.andree@web.de>
MFC after: 1 week

21 years agoRemove a potential panic condition introduced by reduced TCP wait
Robert Watson [Thu, 10 Apr 2003 20:33:10 +0000 (20:33 +0000)]
Remove a potential panic condition introduced by reduced TCP wait
state.  Those changed attempted to work around the changed invariant
that inp->in_socket was sometimes now NULL, but the logic wasn't
quite right, meaning that inp->in_socket would be dereferenced by
cr_canseesocket() if security.bsd.see_other_uids, jail, or MAC
were in use.  Attempt to clarify and correct the logic.

Note: the work-around originally introduced with the reduced TCP
wait state handling to use cr_cansee() instead of cr_canseesocket()
in this case isn't really right, although it "Does the right thing"
for most of the cases in the base system.  We'll need to address
this at some point in the future.

Pointed out by: dcs
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoEnable loadable modules to be unloaded on alphas with shared isa
Andrew Gallatin [Thu, 10 Apr 2003 20:32:29 +0000 (20:32 +0000)]
Enable loadable modules to be unloaded on alphas with shared isa
interrupts  by only disabling the interrupt in hardware if
the handler being removed is the only handler.

21 years ago- Kill the pv_flags member of the alpha mdpage since it stop being used
John Baldwin [Thu, 10 Apr 2003 18:42:06 +0000 (18:42 +0000)]
- Kill the pv_flags member of the alpha mdpage since it stop being used
  in rev 1.61 of pmap.c.
- Now that pmap_page_is_free() is empty and since it is just a hack for
  the Alpha pmap, remove it.

21 years agoCheck M_TRYWAIT as a flag, not a constant, in MBTOM() mbuf allocator
Robert Watson [Thu, 10 Apr 2003 17:52:10 +0000 (17:52 +0000)]
Check M_TRYWAIT as a flag, not a constant, in MBTOM() mbuf allocator
to malloc flag converter.

21 years agoCatch up with the kernel. Move the current cpu indicator to the thread.
Julian Elischer [Thu, 10 Apr 2003 17:41:40 +0000 (17:41 +0000)]
Catch up with the kernel. Move the current cpu indicator to the thread.

21 years agoMove the _oncpu entry from the KSE to the thread.
Julian Elischer [Thu, 10 Apr 2003 17:35:44 +0000 (17:35 +0000)]
Move the _oncpu entry from the KSE to the thread.
The entry in the KSE still exists but it's purpose will change a bit
when we add the ability to lock a KSE to a cpu.

21 years agoPrint real memory/avail memory on startup like other platforms. Hide
Jake Burkholder [Thu, 10 Apr 2003 17:18:52 +0000 (17:18 +0000)]
Print real memory/avail memory on startup like other platforms.  Hide
printing the model under bootverbose.

21 years agoAdd NTT-ME SS-LAN CARD MN128. This card entry has been
Takeshi Shibagaki [Thu, 10 Apr 2003 17:16:01 +0000 (17:16 +0000)]
Add NTT-ME SS-LAN CARD MN128. This card entry has been
committed into NEWCARD.

21 years agoNEC PK-WL001 is Lucent, not Prism-II.
Takeshi Shibagaki [Thu, 10 Apr 2003 17:04:03 +0000 (17:04 +0000)]
NEC PK-WL001 is Lucent, not Prism-II.

21 years agoSpell 55 with two '5's
Søren Schmidt [Thu, 10 Apr 2003 12:56:33 +0000 (12:56 +0000)]
Spell 55 with two '5's

21 years agoAdd a few more older SiS ID's.
Søren Schmidt [Thu, 10 Apr 2003 12:24:39 +0000 (12:24 +0000)]
Add a few more older SiS ID's.

21 years agoBrucify.
Dag-Erling Smørgrav [Thu, 10 Apr 2003 10:26:18 +0000 (10:26 +0000)]
Brucify.

21 years agoForced commit to fill in the fields below for the last commit. Also
Hartmut Brandt [Thu, 10 Apr 2003 10:18:56 +0000 (10:18 +0000)]
Forced commit to fill in the fields below for the last commit. Also
bump the date of the man page.

PR: docs/24797
Reviewed by: bde, jake, jhb

21 years agoWhen called with s == NULL, behave as if wc == L'\0' as required by the
Tim J. Robbins [Thu, 10 Apr 2003 09:20:38 +0000 (09:20 +0000)]
When called with s == NULL, behave as if wc == L'\0' as required by the
standard.

21 years agoAdd quirk for LOGITECH WMRPAD.
Matthew N. Dodd [Thu, 10 Apr 2003 08:25:09 +0000 (08:25 +0000)]
Add quirk for LOGITECH WMRPAD.

Submitted by: Mathew Kanner <mat@cnd.mcgill.ca>

21 years ago- Add LOGITECH WMRPAD, MOTOROLA SB410, and SANDISK SDDR05.
Matthew N. Dodd [Thu, 10 Apr 2003 08:24:33 +0000 (08:24 +0000)]
- Add LOGITECH WMRPAD, MOTOROLA SB410, and SANDISK SDDR05.
- Regenerate.

21 years agoFix debounce. I forgot these bits in my last commit.
Matthew N. Dodd [Thu, 10 Apr 2003 08:06:56 +0000 (08:06 +0000)]
Fix debounce.  I forgot these bits in my last commit.

Submitted by:  Mathew Kanner <mat@cnd.mcgill.ca>

21 years agoFix interrupt race.
Warner Losh [Thu, 10 Apr 2003 07:55:55 +0000 (07:55 +0000)]
Fix interrupt race.

From NetBSD

21 years agoSMC 8041 is the same thing as LINKSYS COMBO_ECARD, so no extra entry
Warner Losh [Thu, 10 Apr 2003 07:35:19 +0000 (07:35 +0000)]
SMC 8041 is the same thing as LINKSYS COMBO_ECARD, so no extra entry
is required.  NetBSD has one because it checks for the mac address
match as well wanting to give its own string in the description.
Since we do neither, we don't need a separate entry.

# I suspect that a few of the COREGA cards might fall into that category
# as well, but since I don't have access to any of them it is hard to know
# for sure.

21 years ago- fixed byte order of route lifetimes in route info opptions.
SUZUKI Shinsuke [Thu, 10 Apr 2003 07:31:34 +0000 (07:31 +0000)]
- fixed byte order of route lifetimes in route info opptions.
- corrected wording in a log message.

Obtained from: KAME
MFC after: 1 week

21 years agoAdd Billionton CFLT10N and SMC8041 (although I'm not 100% sure the latter is
Warner Losh [Thu, 10 Apr 2003 07:24:18 +0000 (07:24 +0000)]
Add Billionton CFLT10N and SMC8041 (although I'm not 100% sure the latter is
needed).  Remove duplicate kingston kne2 entry.

Info obtained from NetBSD