]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoThis commit was generated by cvs2svn to compensate for changes in r117839,
Dag-Erling Smørgrav [Mon, 21 Jul 2003 19:54:17 +0000 (19:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117839,
which included commits to RCS files with non-trunk default branches.

21 years agoMFP4: revert constification of pam_message.msg (perforce change #34785)
Dag-Erling Smørgrav [Mon, 21 Jul 2003 19:54:17 +0000 (19:54 +0000)]
MFP4: revert constification of pam_message.msg (perforce change #34785)

21 years agoEnable eisa support on alpha.
Bernd Walter [Mon, 21 Jul 2003 19:15:25 +0000 (19:15 +0000)]
Enable eisa support on alpha.
We need eisa bridge support on some models, which now depends on having
eisa in the kernel.

21 years agoAdd a new macro __nonnull(x) to use the new GCC33 attribute which checks
Poul-Henning Kamp [Mon, 21 Jul 2003 18:35:51 +0000 (18:35 +0000)]
Add a new macro __nonnull(x) to use the new GCC33 attribute which checks
that an argument is not a NULL pointer.

Apply various obvious places.

I belive __printf*() implies __nonnull() so it is not needed on functions
already tagged that way.

21 years agoDocument uma_zone_set_max and its non-obvious behaviour.
Hartmut Brandt [Mon, 21 Jul 2003 14:20:58 +0000 (14:20 +0000)]
Document uma_zone_set_max and its non-obvious behaviour.

Reviewed by: bmilekic

21 years agoThis is a pseudo physical interface for the HARP ATM stack. When loaded
Hartmut Brandt [Mon, 21 Jul 2003 13:56:22 +0000 (13:56 +0000)]
This is a pseudo physical interface for the HARP ATM stack. When loaded
it attaches to all existing NATM network interfaces in the system
and creates a HARP physical interface for each of them. This allows
us to use the same set of ATM drivers for all ATM stuff. It is
possible to use the same interface for HARP, NATM and netgraph at the
same time.

21 years agoUse the clear function provided by the video adapter driver.
Yoshihiro Takahashi [Mon, 21 Jul 2003 13:04:54 +0000 (13:04 +0000)]
Use the clear function provided by the video adapter driver.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)

21 years agoSupported the gdc_clear function.
Yoshihiro Takahashi [Mon, 21 Jul 2003 13:00:09 +0000 (13:00 +0000)]
Supported the gdc_clear function.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)

21 years agoQuote from a Problem Report:
Maxim Konovalov [Mon, 21 Jul 2003 11:06:47 +0000 (11:06 +0000)]
Quote from a Problem Report:

The output format specifier for the round-trip time in ping6 should be
changed to %.3f instead of %g since %g doesn't accurately represent the
precision of the number being output. In particular, %g truncates trailing
zeroes. 0.01 ms does not mean the same thing as 0.010 ms. Although they
are numerically identical, they do not have the same precision.

PR: bin/52324, bin/52750
Submitted by: dg
MFC after: 1 week

21 years agoAdd my birthday to calendar
Kirill Ponomarev [Mon, 21 Jul 2003 10:43:37 +0000 (10:43 +0000)]
Add my birthday to calendar

Approved by: fjoe

21 years agoo Initialize do_pipe before command parsing.
Maxim Konovalov [Mon, 21 Jul 2003 09:56:05 +0000 (09:56 +0000)]
o Initialize do_pipe before command parsing.

PR: bin/54649
Submitted by: Andy Gilligan <andy@evo6.org>
MFC after: 3 days

21 years agoAlways start new sentences on a new line.
Hartmut Brandt [Mon, 21 Jul 2003 07:51:29 +0000 (07:51 +0000)]
Always start new sentences on a new line.

Submitted by: Anil Madhavapeddy <anil@recoil.org>

21 years agoadd monitor mode
Sam Leffler [Mon, 21 Jul 2003 02:49:42 +0000 (02:49 +0000)]
add monitor mode

21 years agoMissed one, remove one more.
David E. O'Brien [Mon, 21 Jul 2003 02:47:04 +0000 (02:47 +0000)]
Missed one, remove one more.

21 years agooverride the tag. usb_block_allocmem allocates a new tag, which when we
John-Mark Gurney [Mon, 21 Jul 2003 02:41:01 +0000 (02:41 +0000)]
override the tag.  usb_block_allocmem allocates a new tag, which when we
go looking for free fragments won't match.  Since we never free this, we
can "throw away" the tag.  This is very dirty, and needs to be reimplemented
properly, but fixes performance problems with uhci.

Also assert that when we overlay a structure on some space, that the
space is large enough for the structure.

21 years agoAlways deliver synchronous signal to UTS for SA threads.
David Xu [Mon, 21 Jul 2003 00:26:52 +0000 (00:26 +0000)]
Always deliver synchronous signal to UTS for SA threads.

21 years agoTest correct macro for "without crypto" option(s).
Mark Murray [Sun, 20 Jul 2003 23:29:46 +0000 (23:29 +0000)]
Test correct macro for "without crypto" option(s).

21 years agotrack changes to 802.11 code:
Sam Leffler [Sun, 20 Jul 2003 21:38:20 +0000 (21:38 +0000)]
track changes to 802.11 code:

o override new_state method per new model
o use ieee80211_state_name instead of private copy

21 years agoo change ieee80211_new_state handling to use a proper method that drivers
Sam Leffler [Sun, 20 Jul 2003 21:36:08 +0000 (21:36 +0000)]
o change ieee80211_new_state handling to use a proper method that drivers
  override in their sub-class; this eliminates the hack of interpreting the
  EINPROGRESS return value to mean "don't do any of the normal work"
o correct active scanning so the first channel is only scanned once and so
  per-channel passive mode is properly honored
o expose 802.11 FSM state names so every driver doesn't keep a private copy
o eliminate node parameter to ieee80211_begin_scan; it was not being used

21 years agoDo not call acpi_MatchHid() for all probe cases since it accesses the
Nate Lawson [Sun, 20 Jul 2003 21:12:21 +0000 (21:12 +0000)]
Do not call acpi_MatchHid() for all probe cases since it accesses the
namespace.  To compensate for it only being used in the !ECDT case, use
a more robust approach to indicate a device was probed via ECDT by setting
the private ivar to be &acpi_ec_devclass.  Without the acpi_MatchHid() call
now, it might have been possible for a non-EC device to have had its magic
match our previous flag.

Pointed out by: takawata

21 years agoClose a race condition by passing status retrieved via a non-SCI call
Nate Lawson [Sun, 20 Jul 2003 21:11:32 +0000 (21:11 +0000)]
Close a race condition by passing status retrieved via a non-SCI call
to EcGpeQueryHandler on to any waiting threads through the softc.  Similar
behavior was in the original version.

Also:
* Merge EcQuery into EcGpeQueryHandler to simplify locking
* Hold EcLock from the initial read of the CSR down to the wakeup or
  until after the query command has been processed.
* ec_gpebit only needs to be a UINT8

21 years agoAdd my birthday to the calendar.
Simon L. B. Nielsen [Sun, 20 Jul 2003 20:31:18 +0000 (20:31 +0000)]
Add my birthday to the calendar.

Approved by: ceri (mentor)

21 years agoAdd missing arguments to _amd64_restore_context() when called from
Daniel Eischen [Sun, 20 Jul 2003 12:41:38 +0000 (12:41 +0000)]
Add missing arguments to _amd64_restore_context() when called from
THR_SETCONTEXT().

21 years agoThe `mk' target is long dead.
Ruslan Ermilov [Sun, 20 Jul 2003 12:38:29 +0000 (12:38 +0000)]
The `mk' target is long dead.

21 years agoThis code isn't WARNS=6 clean in the standard (crypto) case
Ruslan Ermilov [Sun, 20 Jul 2003 12:26:24 +0000 (12:26 +0000)]
This code isn't WARNS=6 clean in the standard (crypto) case
due to bugs in OpenSSL headers.  I was testing in the wrong
environmement: standalone build without crypto/ sources.

21 years agoPass getvfsbyname() the address of a struct xvfsconf instead of
Tim J. Robbins [Sun, 20 Jul 2003 11:27:54 +0000 (11:27 +0000)]
Pass getvfsbyname() the address of a struct xvfsconf instead of
struct vfsconf. This silences a warning, but could also prevent
stack corruption problems if xvfsconf ever became larger than vfsconf.

PR: 53863
Submitted by: Lukas Ertl

21 years agoMake this code WARNS=6 clean again (after GCC 3.1.1 import).
Ruslan Ermilov [Sun, 20 Jul 2003 10:24:09 +0000 (10:24 +0000)]
Make this code WARNS=6 clean again (after GCC 3.1.1 import).

Submitted by: Marius Strobl <marius@alchemy.franken.de>

21 years agoNow that we have the stubs for alpha and we can build it
Mike Makonnen [Sun, 20 Jul 2003 01:34:40 +0000 (01:34 +0000)]
Now that we have the stubs for alpha and we can build it
on that platform, invert the test for the platforms on
which libthr is built. Amd64 and powerpc are the only
platforms excluded.

Compile tested on: amd64, alpha

21 years agoClarify the ACPI shutdown messages.
Nate Lawson [Sun, 20 Jul 2003 00:52:57 +0000 (00:52 +0000)]
Clarify the ACPI shutdown messages.

21 years agoAdd ECDT (ACPI 2.0) support. This allows the EC to be enabled before the
Nate Lawson [Sun, 20 Jul 2003 00:48:38 +0000 (00:48 +0000)]
Add ECDT (ACPI 2.0) support.  This allows the EC to be enabled before the
namespace has been evaluated.  Machines with ACPI 2.0 expect this behavior
and have AML which calls EC functions early in the boot process.  If the
ECDT is not available, fall back to original probe behavior.

Other minor changes:
* Add GPE bit and GLK usage to the device announcement
* Always use the global lock in the ECDT case, but potentially downgrade to
  not using it if _GLK is 0 once the namespace is available.  This is
  announced with "Changing GLK from 1 to 0"
* Remove the acpi_object_list definitions which were earlier deprecated

Ideas from: takawata

21 years agoTake advantage of the use of file system IDs to simplify umount(8)
Ian Dowse [Sun, 20 Jul 2003 00:11:27 +0000 (00:11 +0000)]
Take advantage of the use of file system IDs to simplify umount(8)
and make it work more reliably in a number of cases that have
traditionally been troublesome. The new behaviour is:
 1) If the filesystem can be determined by the fsid or device,
    or uniquely identified by the mountpoint, then just go ahead
    and call unmount(2) using the file system ID.
 2) Otherwise use fstatfs(2) to resolve the path into a file system
    ID (checking with stat(2) that it is a filesystem root directory).

Case 2 can potentially block if an NFS server is down, but it can
always be avoided by using an unambiguous specification. It handles
all the hard cases such as symlinks and mismatches between the mount
list and reality. For example, if a filesystem was mounted as /mnt
inside a chroot, it will show up in the mount list as /mnt, but now
you can unmount it from outside the chroot with "umount /chroot_path/mnt".

21 years agoUse an up-to-date make binary in the upgrade_checks target if
Ruslan Ermilov [Sat, 19 Jul 2003 23:17:08 +0000 (23:17 +0000)]
Use an up-to-date make binary in the upgrade_checks target if
one is already available.  This avoids sometimes unnecessary
step of attempting to rebuild the make binary again which may
fail at all if, for example, one has removed his /usr/include
before doing an installworld (to keep /usr/include tidied up).

Pointed out by: kris, marcel

21 years agoTrim /rescue.
David E. O'Brien [Sat, 19 Jul 2003 20:40:37 +0000 (20:40 +0000)]
Trim /rescue.

Discussed on: freebsd-arch@

21 years agoDisabling multicast on vlan interface caused kernel panic.
Hajimu UMEMOTO [Sat, 19 Jul 2003 16:47:16 +0000 (16:47 +0000)]
Disabling multicast on vlan interface caused kernel panic.

PR: kern/40723
Submitted by: Hideki ONO <ono@kame.net>
MFC after: 1 week

21 years agoThe MD framework for libthr on alpha
Mike Makonnen [Sat, 19 Jul 2003 15:57:52 +0000 (15:57 +0000)]
The MD framework for libthr on alpha

21 years agoAdd description about tagged queuing.
Hidetoshi Shimokawa [Sat, 19 Jul 2003 14:46:59 +0000 (14:46 +0000)]
Add description about tagged queuing.

21 years agoMFC: atkbd(4) bug fix.
Hiroki Sato [Sat, 19 Jul 2003 14:38:03 +0000 (14:38 +0000)]
MFC:  atkbd(4) bug fix.

21 years agoTurn a KASSERT back into an EINVAL return value. So, next time someone
Mike Makonnen [Sat, 19 Jul 2003 11:32:48 +0000 (11:32 +0000)]
Turn a KASSERT back into an EINVAL return value. So, next time someone
comes across it, it will turn into a core dump in userland instead of
a kernel panic. I had also inverted the sense of the test, so

Double pointy hat to: mtm

21 years agoMerge the following from the English version:
Hiroki Sato [Sat, 19 Jul 2003 08:36:38 +0000 (08:36 +0000)]
Merge the following from the English version:

1.583 -> 1.592 relnotes/common/new.sgml

21 years agomake sockstat not print wierd addresses on not connected unix domain sockets
John-Mark Gurney [Sat, 19 Jul 2003 06:23:56 +0000 (06:23 +0000)]
make sockstat not print wierd addresses on not connected unix domain sockets

Pointed out by: rwatson
Reviewed by: peter

21 years agoAdd new supported cards
Warner Losh [Sat, 19 Jul 2003 06:19:43 +0000 (06:19 +0000)]
Add new supported cards

21 years agoThree fixes:
Mike Silbersack [Sat, 19 Jul 2003 06:03:48 +0000 (06:03 +0000)]
Three fixes:

- Make m_prepend use m_gethdr instead of m_get where
  appropriate

- Make m_copym use m_gethdr instead of m_get where
  appropriate

- Add a call to m_fixhdr in m_defrag; m_defrag can't
  deal with corrupted pkthdr.len counts.

MFC after: 3 days

21 years agoAdd support for FA-511; Submitted by: Kenneth P. Stox; Pr 42858
Warner Losh [Sat, 19 Jul 2003 06:01:15 +0000 (06:01 +0000)]
Add support for FA-511; Submitted by: Kenneth P. Stox; Pr 42858

21 years agoAdd Addtron AWA-100 wireless PCI card
Warner Losh [Sat, 19 Jul 2003 05:53:57 +0000 (05:53 +0000)]
Add Addtron AWA-100 wireless PCI card

Submitted by: Robin Reagan
Pr: 37526

21 years agoAdd Linksys WCF12: from Scott Lambert
Warner Losh [Sat, 19 Jul 2003 05:51:44 +0000 (05:51 +0000)]
Add Linksys WCF12: from Scott Lambert

21 years agoSync to 1.60
Warner Losh [Sat, 19 Jul 2003 05:51:10 +0000 (05:51 +0000)]
Sync to 1.60

21 years agoMinor fix to the MBUF_STRESS_TEST code so that it keeps
Mike Silbersack [Sat, 19 Jul 2003 05:50:32 +0000 (05:50 +0000)]
Minor fix to the MBUF_STRESS_TEST code so that it keeps
pkthdr.len consistant at all times.  (Some debugging
code I'm working on is tripped otherwise.)

MFC after: 3 days

21 years agoAdd Linksys compact flash wireless card
Warner Losh [Sat, 19 Jul 2003 05:49:17 +0000 (05:49 +0000)]
Add Linksys compact flash wireless card

Submitted by: Scott Lambert
PR: 53881

21 years agoAdd support for DFE-670TXD to OLDCARD
Warner Losh [Sat, 19 Jul 2003 05:40:06 +0000 (05:40 +0000)]
Add support for DFE-670TXD to OLDCARD

Submitted by: David Wolfskill
PR: 53356

21 years agoAdd support for BUFFALO LPC3-CLX 10/100Base-T PC-Card
Warner Losh [Sat, 19 Jul 2003 05:36:09 +0000 (05:36 +0000)]
Add support for BUFFALO LPC3-CLX 10/100Base-T PC-Card

PR: 47786
Submitted by: IHA, genta -san

21 years agosync to 1.59
Warner Losh [Sat, 19 Jul 2003 05:33:54 +0000 (05:33 +0000)]
sync to 1.59

21 years agoAdd Buffalo LPC3_CLX.
Warner Losh [Sat, 19 Jul 2003 05:32:07 +0000 (05:32 +0000)]
Add Buffalo LPC3_CLX.

PR: 47786
Submitted by: IHA, genta -san

21 years agoOverride libc function raise(), in threading mode, raise() will
David Xu [Sat, 19 Jul 2003 05:25:49 +0000 (05:25 +0000)]
Override libc function raise(), in threading mode, raise() will
send signal to current thread.

Reviewed by: deischen

21 years agoMake raise and _raise as weak symbols, so they can be overriden by
David Xu [Sat, 19 Jul 2003 05:22:56 +0000 (05:22 +0000)]
Make raise and _raise as weak symbols, so they can be overriden by
thread library.

Reviewed by: deischen

21 years agoAdd some very beta amd64 bits. These will also need some tweaking.
Daniel Eischen [Sat, 19 Jul 2003 04:44:21 +0000 (04:44 +0000)]
Add some very beta amd64 bits.  These will also need some tweaking.

21 years agoAdd amd64 versions of makecontext() and signalcontext() needed
Daniel Eischen [Sat, 19 Jul 2003 04:41:08 +0000 (04:41 +0000)]
Add amd64 versions of makecontext() and signalcontext() needed
for libkse (makecontext() is also needed for libthr).
These probably will need some tweaking.

21 years agoRewrite to reflect slight change in semantics for C99, and note a bug
Garrett Wollman [Sat, 19 Jul 2003 03:19:59 +0000 (03:19 +0000)]
Rewrite to reflect slight change in semantics for C99, and note a bug
in the standard.  Defer to gettimeofday(2) for error indications.

21 years agoC99 compliance: time() always sets its return value in both places
Garrett Wollman [Sat, 19 Jul 2003 02:53:46 +0000 (02:53 +0000)]
C99 compliance: time() always sets its return value in both places
(if present), even on error.

Pointed out by: Wojtek Lerch, on the Austin Group mailing-list

21 years agoAdd mutex for routing entries.
Jeffrey Hsu [Sat, 19 Jul 2003 00:21:13 +0000 (00:21 +0000)]
Add mutex for routing entries.

Reviewed by: bmilekic, silby

21 years agoAdjust for new telnet situation (single telnet source tree).
Mark Murray [Sat, 19 Jul 2003 00:17:36 +0000 (00:17 +0000)]
Adjust for new telnet situation (single telnet source tree).

21 years agoRevert to using yp_order() to probe for master.paswd.by* maps and
Bill Paul [Fri, 18 Jul 2003 23:51:15 +0000 (23:51 +0000)]
Revert to using yp_order() to probe for master.paswd.by* maps and
don't probe the server at all for passwd.by* maps. This fixes
interoperability with the Services For UNIX NIS server (which is
really a front end to Captive^WActiveDirectory). This server
incorrectly returns success for all YPPROC_MASTER requests,
even for maps that don't exist, which makes it impossible to
(ab)use it to probe for the existence of the master.passwd.by*
maps.

This is a little kludgey, but basically restores the original
behavior of getpwent.c as it is in -stable, and works around both
the lack of YPPROC_ORDER on NIS+ servers as well as the broken
YPPROC_MASTER on Services For UNIX servers.

21 years agoFix a busdma bogon:
Bill Paul [Fri, 18 Jul 2003 22:11:47 +0000 (22:11 +0000)]
Fix a busdma bogon:

Some of the calls to bus_dmamap_sync() were syncing the DMA descriptor
ring maps using the mbuf tag, when they should have been using the
descriptor ring tag instead.

21 years agoFix a printf format warning I introduced.
Poul-Henning Kamp [Fri, 18 Jul 2003 22:11:17 +0000 (22:11 +0000)]
Fix a printf format warning I introduced.
Use the macro max number of swap devices rather than cache the constant
in a variable.
Avoid a (now) pointless variable.

21 years agoReport the I/O base address in hex to be consistent with its
Diomidis Spinellis [Fri, 18 Jul 2003 21:53:42 +0000 (21:53 +0000)]
Report the I/O base address in hex to be consistent with its
specification format.

Approved by: schweikh (mentor)
MFC after: 2 weeks

21 years agoRemove a lock held across casuptr() that snuck in last commit.
Mike Makonnen [Fri, 18 Jul 2003 21:26:45 +0000 (21:26 +0000)]
Remove a lock held across casuptr() that snuck in last commit.

21 years agoNew release notes: top(1) -H, mount(8)/umount(8) filesystem ID support.
Hiroki Sato [Fri, 18 Jul 2003 21:09:55 +0000 (21:09 +0000)]
New release notes:  top(1) -H, mount(8)/umount(8) filesystem ID support.

21 years agoMove the decision on whether to unset the contested
Mike Makonnen [Fri, 18 Jul 2003 17:58:37 +0000 (17:58 +0000)]
Move the decision on whether to unset the contested
bit or not from lock to unlock time.

Suggested by: jhb

21 years agoWhen mount(8) is invoked with the `-v' flag, display the filesystem
Ian Dowse [Fri, 18 Jul 2003 17:43:13 +0000 (17:43 +0000)]
When mount(8) is invoked with the `-v' flag, display the filesystem
ID for each file system in addition to the normal information.

In umount(8), accept filesystem IDs as well as the usual device and
path names. This makes it possible to unambiguously specify which
file system is to be unmounted even when two or more file systems
share the same device and mountpoint names (e.g. NFS mounts from
the same export into different chroots).

Suggested by: Dan Nelson <dnelson@allantgroup.com>

21 years agoFix a number of occurences of calling uma_zalloc() with neither
Hartmut Brandt [Fri, 18 Jul 2003 16:36:41 +0000 (16:36 +0000)]
Fix a number of occurences of calling uma_zalloc() with neither
M_WAITOK nor M_NOWAIT.

21 years agoReference PR for IntelligentStick quirk entry.
Thomas Quinot [Fri, 18 Jul 2003 16:26:36 +0000 (16:26 +0000)]
Reference PR for IntelligentStick quirk entry.

PR: kern/53005

21 years agoPrototype defttymode() instead of just declaring it.
Yaroslav Tykhiy [Fri, 18 Jul 2003 16:25:55 +0000 (16:25 +0000)]
Prototype defttymode() instead of just declaring it.

21 years agoAdd a comment above rip_ctloutput() documenting that the privilege
Robert Watson [Fri, 18 Jul 2003 16:10:36 +0000 (16:10 +0000)]
Add a comment above rip_ctloutput() documenting that the privilege
check for raw IP system management operations is often (although
not always) implicit due to the namespacing of raw IP sockets.  I.e.,
you have to have privilege to get a raw IP socket, so much of the
management code sitting on raw IP sockets assumes that any requests
on the socket should be granted privilege.

Obtained from: TrustedBSD Project
Product of: France

21 years agoWhen INVARIANTS is defined make sure that uma_zalloc_arg (and hence
Hartmut Brandt [Fri, 18 Jul 2003 16:04:36 +0000 (16:04 +0000)]
When INVARIANTS is defined make sure that uma_zalloc_arg (and hence
uma_zalloc) is called with exactly one of either M_WAITOK or M_NOWAIT and
that it is called with neither M_TRYWAIT or M_DONTWAIT. Print a warning
if anything is wrong. Default to M_WAITOK of no flag is given. This is the
same test as in malloc(9).

21 years agoWhitespace after keywords per style(9).
Garrett Wollman [Fri, 18 Jul 2003 16:04:32 +0000 (16:04 +0000)]
Whitespace after keywords per style(9).

21 years agoWhen reporting an error internalizing an ACL string, print out the
Robert Watson [Fri, 18 Jul 2003 16:00:26 +0000 (16:00 +0000)]
When reporting an error internalizing an ACL string, print out the
ACL that generated the error, rather than the function, which is
more user-friendly.

Obtained from: TrustedBSD Project
Product of: France

21 years agoCorrect a couple of slips: add a missing article at one point
Yaroslav Tykhiy [Fri, 18 Jul 2003 14:50:10 +0000 (14:50 +0000)]
Correct a couple of slips: add a missing article at one point
and the plaintext word "option" after "</option>" at another.

21 years agoClean up include files.
Hidetoshi Shimokawa [Fri, 18 Jul 2003 14:31:16 +0000 (14:31 +0000)]
Clean up include files.

21 years agoNew release note: arp(8) -i.
Yaroslav Tykhiy [Fri, 18 Jul 2003 14:29:35 +0000 (14:29 +0000)]
New release note: arp(8) -i.

21 years agoRemove extern declaration of ps_showallprocs. The definition is already gone.
Tim J. Robbins [Fri, 18 Jul 2003 13:51:15 +0000 (13:51 +0000)]
Remove extern declaration of ps_showallprocs. The definition is already gone.

PR: 54604
Submitted by: Pawel Jakub Dawidek

21 years agoImplement a new option: -i, which will allow to limit
Yaroslav Tykhiy [Fri, 18 Jul 2003 13:48:06 +0000 (13:48 +0000)]
Implement a new option: -i, which will allow to limit
the scope of operation to the ARP entries on a particular
interface.  It should be useful on machines with numerous
network interfaces, e.g., on inter-VLAN routers.

PR: bin/54151
Submitted by: Dmitry Morozovsky <marck at rinet.ru>
Discussed on: -net
MFC after: 2 weeks

21 years agoBig fixup of the makefiles. Sort out the dependancies so that "make"
Mark Murray [Fri, 18 Jul 2003 13:21:58 +0000 (13:21 +0000)]
Big fixup of the makefiles. Sort out the dependancies so that "make"
without "make depend" works, "make -j N" works, and lists of source
files are made vertical to reduce future diffs.

21 years agoWLAN related fixes:
Hiten Pandya [Fri, 18 Jul 2003 11:40:41 +0000 (11:40 +0000)]
WLAN related fixes:

- Add some wlan(4) related cross-references.
- Add some "device" lines to SYNOPSIS
- Correct grammar/sentencing in ath_hal(4) manual page

PR: docs/54114
Submitted by:  Lukas Ertl <l.ertl@univie.ac.at>
Approved by: des (mentor)

21 years agoDon't call print_pdu() when we are not debugging. This would result
Hartmut Brandt [Fri, 18 Jul 2003 11:17:04 +0000 (11:17 +0000)]
Don't call print_pdu() when we are not debugging. This would result
in calling fprintf() with a NULL fp. Strange enough this didn't result
in cores in stable, but results in cores now.

MFC after: 2 weeks

21 years agoIf a proposed swap device exceeds the 8G artificial limit which out
Poul-Henning Kamp [Fri, 18 Jul 2003 11:01:23 +0000 (11:01 +0000)]
If a proposed swap device exceeds the 8G artificial limit which out
radix-tree code imposes, truncate the device instead of rejecting it.

21 years agoMove the implementation of the vmspace_swap_count() (used only in
Poul-Henning Kamp [Fri, 18 Jul 2003 10:47:58 +0000 (10:47 +0000)]
Move the implementation of the vmspace_swap_count() (used only in
the "toss the largest process" emergency handling) from vm_map.c to
swap_pager.c.

The quantity calculated depends strongly on the internals of the
swap_pager and by moving it, we no longer need to expose the
internal metrics of the swap_pager to the world.

21 years agoAdd a new function swap_pager_status() which reports the total size of the
Poul-Henning Kamp [Fri, 18 Jul 2003 10:26:09 +0000 (10:26 +0000)]
Add a new function swap_pager_status() which reports the total size of the
paging space and how much of it is in use (in pages).

Use this interface from the Linuxolator instead of groping around in the
internals of the swap_pager.

21 years agoMerge swap_pager.c and vm_swap.c into swap_pager.c, the separation
Poul-Henning Kamp [Fri, 18 Jul 2003 10:02:44 +0000 (10:02 +0000)]
Merge swap_pager.c and vm_swap.c into swap_pager.c, the separation
is not natural and needlessly exposes a lot of dirty laundry.

Move private interfaces between the two from swap_pager.h to swap_pager.c
and staticize as much as possible.

No functional change.

21 years agoCorrect the device identifiers for the ProATM cards.
Hartmut Brandt [Fri, 18 Jul 2003 08:40:45 +0000 (08:40 +0000)]
Correct the device identifiers for the ProATM cards.

21 years agoSince checkmntlist() and getmntentry() return a struct statfs that
Ian Dowse [Fri, 18 Jul 2003 08:01:10 +0000 (08:01 +0000)]
Since checkmntlist() and getmntentry() return a struct statfs that
includes the filesystem type name, remove the "type" output parameter.

21 years agoLook for logs in the directory where we place the output.
Dag-Erling Smørgrav [Fri, 18 Jul 2003 07:43:43 +0000 (07:43 +0000)]
Look for logs in the directory where we place the output.

21 years agoFix taint bug in previous commit.
Dag-Erling Smørgrav [Fri, 18 Jul 2003 07:29:00 +0000 (07:29 +0000)]
Fix taint bug in previous commit.

21 years agoGenerate index.html in the same directory.
Dag-Erling Smørgrav [Fri, 18 Jul 2003 07:24:13 +0000 (07:24 +0000)]
Generate index.html in the same directory.

21 years agoAdd some debug messages.
Hidetoshi Shimokawa [Fri, 18 Jul 2003 05:28:30 +0000 (05:28 +0000)]
Add some debug messages.

21 years agoCleanup thread accounting. Don't reset a threads timeslice
Daniel Eischen [Fri, 18 Jul 2003 02:46:55 +0000 (02:46 +0000)]
Cleanup thread accounting.  Don't reset a threads timeslice
when it blocks; it only gets reset when it yields.

Properly set a thread's default stack guardsize.

Reviewed by: davidxu

21 years agoAdd a preemption point when a mutex or condition variable is
Daniel Eischen [Fri, 18 Jul 2003 02:46:30 +0000 (02:46 +0000)]
Add a preemption point when a mutex or condition variable is
handed-off/signaled to a higher priority thread.  Note that when
there are idle KSEs that could run the higher priority thread,
we still add the preemption point because it seems to take the
kernel a while to schedule an idle KSE.  The drawbacks are that
threads will be swapped more often between CPUs (KSEs) and
that there will be an extra userland context switch (the idle
KSE is still woken and will probably resume the preempted
thread).  We'll revisit this if and when idle CPU/KSE wakeup
times improve.

Inspired by: Petri Helenius <pete@he.iki.fi>
Reviewed by: davidxu

21 years agoClean up KSE specific data (KSD) macros a bit.
Daniel Eischen [Fri, 18 Jul 2003 02:45:56 +0000 (02:45 +0000)]
Clean up KSE specific data (KSD) macros a bit.

Reviewed by: davidxu

21 years agoWhen the file system to unmount is specified by device name instead
Ian Dowse [Fri, 18 Jul 2003 01:10:16 +0000 (01:10 +0000)]
When the file system to unmount is specified by device name instead
of by mount point, umount had to take care not to unmount the wrong
file system if another file system was covering the requested one.
Now that the file system to unmount is specified to the kernel using
the filesystem ID, this confusion cannot occur, so remove the code
that checked for it.

21 years agoObey NOCRYPT
David E. O'Brien [Fri, 18 Jul 2003 01:07:18 +0000 (01:07 +0000)]
Obey NOCRYPT

Submitted by: Skip Ford <skip.ford@verizon.net>

21 years agoMake the 80x60, 132x25, 132x43, 132x50, and 132x60 VESA text modes
Robert Drehmel [Fri, 18 Jul 2003 00:06:10 +0000 (00:06 +0000)]
Make the 80x60, 132x25, 132x43, 132x50, and 132x60 VESA text modes
work when using a graphics chipset which identifies itself as
`VIA CLE266', used in some VIA EPIA boards.  Two values need to be
patched in the VESA mode information structure: the widths of the modes
mentioned above are encoded in a format which was unknown to the VESA
module (and to my copy of the VBE spec.) whereas the window memory
segment values seem to be just incorrect.

I tested this on a VIA EPIA-M9000 and -M10000.

21 years agoChanges to allow top to decide whether or not to show multiple threads per
Julian Elischer [Thu, 17 Jul 2003 23:56:40 +0000 (23:56 +0000)]
Changes to allow top to decide whether or not to show multiple threads per
process. Option -H enables it and it is toggled at the interactive
screen by 'H'.

Submitted by: Jung-uk Kim <jkim@niksun.com>