freebsd/src.git
18 years agoMFS: 4.9-RELEASE should fetch the release packages with pkg_add -r.release/4.9.0
Murray Stokely [Mon, 27 Oct 2003 07:57:32 +0000 (27 07:57 +0000)]
MFS: 4.9-RELEASE should fetch the release packages with pkg_add -r.

18 years agoThe next release of FreeBSD will most likely be 4.9.1, according to
Bruce A. Mah [Mon, 27 Oct 2003 05:37:00 +0000 (27 05:37 +0000)]
The next release of FreeBSD will most likely be 4.9.1, according to
current RE plans.

Approved by: re (implicitly)
Requested by: scottl, murray

18 years agoBump __FreeBSD_version for FreeBSD 4.9.
Murray Stokely [Mon, 27 Oct 2003 02:20:56 +0000 (27 02:20 +0000)]
Bump __FreeBSD_version for FreeBSD 4.9.

18 years agoIdentify as 4.9-RELEASE.
Murray Stokely [Mon, 27 Oct 2003 02:18:05 +0000 (27 02:18 +0000)]
Identify as 4.9-RELEASE.

18 years agoMFS: rev. 1.35.2.35:
Max Khon [Sun, 26 Oct 2003 19:27:16 +0000 (26 19:27 +0000)]
MFS: rev. 1.35.2.35:
Fix setting PIO mode timings on AcerLabs Aladdin IV/V chipsets.

Approved by: re

18 years agoMFC: fix, improve & extend NFS support.
Dag-Erling Smørgrav [Thu, 23 Oct 2003 20:57:22 +0000 (23 20:57 +0000)]
MFC: fix, improve & extend NFS support.

Submitted by: hmp
Approved by: re (murray)

18 years agoBump version numbers in release documentation for upcoming
Bruce A. Mah [Thu, 23 Oct 2003 17:16:03 +0000 (23 17:16 +0000)]
Bump version numbers in release documentation for upcoming
4.9-RELEASE.

Approved by:  re (implicitly)

18 years agoVersion bump.
Murray Stokely [Thu, 23 Oct 2003 04:58:05 +0000 (23 04:58 +0000)]
Version bump.

18 years agoVersion bump. This supfile should track RELENG_4_9.
Murray Stokely [Thu, 23 Oct 2003 04:56:27 +0000 (23 04:56 +0000)]
Version bump.  This supfile should track RELENG_4_9.

18 years agoVersion bump -- 'make update' should check out RELENG_4_9 here.
Murray Stokely [Thu, 23 Oct 2003 04:53:45 +0000 (23 04:53 +0000)]
Version bump -- 'make update' should check out RELENG_4_9 here.

18 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_4_9'.
cvs2svn [Wed, 22 Oct 2003 20:36:28 +0000 (22 20:36 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_4_9'.

18 years agoAdd a note about the changes to the HTT support.
John Baldwin [Wed, 22 Oct 2003 20:36:27 +0000 (22 20:36 +0000)]
Add a note about the changes to the HTT support.

Approved by: re (scottl)

18 years agoMFC: Always enable hyperthreading to avoid lockups due to APIC interrupt
John Baldwin [Wed, 22 Oct 2003 20:15:10 +0000 (22 20:15 +0000)]
MFC: Always enable hyperthreading to avoid lockups due to APIC interrupt
routing problems.
- Enable cpu_idle_hlt for the SMP case by default since that significantly
  helps HTT performance.
- Provide a facility for disabling individual CPUs by refusing to schedule
  processes on them.
- Use the above to provide a sysctl for disabling or enabling all logical
  CPUs.  The default is to disable all logical CPUs.
- Remove the 'HTT' kernel option now that it is obsolete.

Submitted by:   ps
Approved by:    re (murray)

18 years agoMFC: Always enable hyperthreading to avoid lockups due to APIC interrupt
John Baldwin [Wed, 22 Oct 2003 19:04:09 +0000 (22 19:04 +0000)]
MFC: Always enable hyperthreading to avoid lockups due to APIC interrupt
routing problems.
- Enable cpu_idle_hlt for the SMP case by default since that significantly
  helps HTT performance.
- Provide a facility for disabling individual CPUs by refusing to schedule
  processes on them.
- Use the above to provide a sysctl for disabling or enabling all logical
  CPUs.  The default is to disable all logical CPUs.
- Remove the 'HTT' kernel option now that it is obsolete.

Submitted by: ps
Approved by: re (murray)

18 years agoMerge the following from the English version:
Hiroki Sato [Wed, 22 Oct 2003 15:24:33 +0000 (22 15:24 +0000)]
Merge the following from the English version:

1.22.2.384 -> 1.22.2.385 relnotes/common/new.sgml

Approved by: re (bmah)

18 years agoMFC: Add support for the Intel ICH5 SATA controller in both legacy and
John Baldwin [Wed, 22 Oct 2003 14:43:52 +0000 (22 14:43 +0000)]
MFC: Add support for the Intel ICH5 SATA controller in both legacy and
enhanced mode.

Submitted by: Dan Strick <strick@covad.net>
Approved by: re (murray)

18 years agoMFC linux_misc.c 1.149, linux_uid16.c 1.12: Reject negative ngrp arguments
Tim J. Robbins [Wed, 22 Oct 2003 13:51:46 +0000 (22 13:51 +0000)]
MFC linux_misc.c 1.149, linux_uid16.c 1.12: Reject negative ngrp arguments
in linux_setgroups() and linux_setgroups16().

Approved by: re (murray)

18 years agoMFC rev. 1.86:
Thomas Quinot [Tue, 21 Oct 2003 22:26:11 +0000 (21 22:26 +0000)]
MFC rev. 1.86:
(cdcleanup,cdsysctlinit): Defend against calling sysctl_ctx_free on an
 uninitialized sysctl_ctx.

Approved by: re (murray)

18 years agoMFC rev. 1.160:
Thomas Quinot [Tue, 21 Oct 2003 22:18:19 +0000 (21 22:18 +0000)]
MFC rev. 1.160:
(dacleanup, dasysctlinit): Defend against calling sysctl_ctx_free on
 an uninitialized sysctl_ctx.

Approved by: re (murray)

18 years agoFix problems with slowdowns and hangs with the DRM on -stable. The condition
Eric Anholt [Sun, 19 Oct 2003 19:17:15 +0000 (19 19:17 +0000)]
Fix problems with slowdowns and hangs with the DRM on -stable.  The condition
checks for sleeping waiting for IRQs were not wrapped in spldrm().

PR: kern/57334
Submitted by: Shobaki sam. <progen@free.fr>
Approved by: re (murray)
Obtained from: DRI CVS (Keith Whitwell)

18 years agoMFC: Fix minor grammatical nits in atkbd(4) note and safe(4) note.
Bruce A. Mah [Sun, 19 Oct 2003 18:33:34 +0000 (19 18:33 +0000)]
MFC:  Fix minor grammatical nits in atkbd(4) note and safe(4) note.
Add cross-reference for newly-MFC-ed pae(4) manpage.

Approved by: re (implicitly)

18 years agoMFC:
Hiroki Sato [Sun, 19 Oct 2003 17:11:49 +0000 (19 17:11 +0000)]
MFC:
- hifn(4) 7955 and 7956 chipset support.
- ipfw(4) limit rule processing fixed.
- OpenSSL 0.9.7c.
- timezone database tzdata2003d.

Approved by: re (bmah)

18 years agoMFC of PAE(4) v. 1.6 without section about PAE kernel config file and
Simon L. B. Nielsen [Sun, 19 Oct 2003 08:12:15 +0000 (19 08:12 +0000)]
MFC of PAE(4) v. 1.6 without section about PAE kernel config file and
without reference to the bus_dma(9), both of which does not exist on
-STABLE.

Approved by: re (murray)

18 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_4'.
cvs2svn [Sun, 19 Oct 2003 00:03:14 +0000 (19 00:03 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_4'.

18 years agoMFC Rev 1.40 to fix use-after-free problems with dynamic rules.
Scott Long [Fri, 17 Oct 2003 11:01:03 +0000 (17 11:01 +0000)]
MFC Rev 1.40 to fix use-after-free problems with dynamic rules.

Submitted by: mckusick
Approved by: re (murray>

18 years agoMFC: rev 1.178 - 1.180: map page zero read/write in case DISABLE_PSE is
Peter Wemm [Fri, 17 Oct 2003 06:54:00 +0000 (17 06:54 +0000)]
MFC: rev 1.178 - 1.180: map page zero read/write in case DISABLE_PSE is
used (APM breaks with DISABLE_PSE) or if PSE is turned off for Pentium4.

Approved by: re (scottl)

18 years agoMFC: tzdata2003d (updates Canada, Brazil, New Zealand)
Garrett Wollman [Thu, 16 Oct 2003 15:02:34 +0000 (16 15:02 +0000)]
MFC: tzdata2003d (updates Canada, Brazil, New Zealand)

Approved by: re (murray)

18 years agoMerge the following from the English version:
Hideyuki KURASHINA [Wed, 15 Oct 2003 15:47:29 +0000 (15 15:47 +0000)]
Merge the following from the English version:

   1.1.2.106  ->  1.1.2.113 errata/article.sgml
   1.22.2.358 -> 1.22.2.384 relnotes/common/new.sgml
   1.1.2.7    ->    1.1.2.8 share/sgml/release.dsl

Approved by: re (murray)

18 years agoMFC: Note sk(4) support for SK-9521 V2.0 and 3COM 3C940.
Hideyuki KURASHINA [Wed, 15 Oct 2003 15:31:32 +0000 (15 15:31 +0000)]
MFC:  Note sk(4) support for SK-9521 V2.0 and 3COM 3C940.

Approved by: re (murray)

18 years agoMFC: Note added support for 3C940 and Marvell Semiconductor 88E1000*
Wilko Bulte [Tue, 14 Oct 2003 18:31:42 +0000 (14 18:31 +0000)]
MFC: Note added support for 3C940 and Marvell Semiconductor 88E1000*
     gigabit PHY

Submitted by:   Jung-uk Kim <jkim@niksun.com>
Based on: Nathan L. Binkert's OpenBSD patch for sk(4)
Approved by: re (jhb, murray)

18 years agoMFC: Add support for SK-9521 V2.0 and 3COM 3C940.
Wilko Bulte [Tue, 14 Oct 2003 18:22:42 +0000 (14 18:22 +0000)]
MFC: Add support for SK-9521 V2.0 and 3COM 3C940.

Based on: Nathan L. Binkert's patch for OpenBSD
Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com>
Approved by: re (jhb, murray)

18 years agoMFC: v1.152
John-Mark Gurney [Tue, 14 Oct 2003 05:01:04 +0000 (14 05:01 +0000)]
MFC: v1.152

Fix problem with EVFILT_WRITE filters on pipes

Approved by: re@
Obtained from: nCircle

18 years agoMFC: update number of ports in the ports collection dialog box.
Murray Stokely [Sun, 12 Oct 2003 21:37:15 +0000 (12 21:37 +0000)]
MFC: update number of ports in the ports collection dialog box.

18 years agoMFC: Allow a return of 0 from __sys_write() to exit the loop in libc_r's
Daniel Eischen [Sat, 11 Oct 2003 18:37:09 +0000 (11 18:37 +0000)]
MFC: Allow a return of 0 from __sys_write() to exit the loop in libc_r's
     wrapped version of write().

Approved by: re@

18 years agoFix (one of) 4.9-RELEASE release breaker, "Loading kernel modules while
Makoto Matsushita [Sat, 11 Oct 2003 08:10:33 +0000 (11 08:10 +0000)]
Fix (one of) 4.9-RELEASE release breaker, "Loading kernel modules while
installation phase sometimes causes an error" bug.  It also includes
"there's an error while loading firewire modules" bug.

Since the results of readdir(3)'s whims, either sbp.ko or if_fwe.ko is
picked up for load BEFORE firewire.ko.  Unfortunately module_path is
not properly set, kldload(2) cannot find kernel modules, firewire.ko,
which is required to load sbp.ko/if_fwe.ko.  This is why you'll see
"loading failed" message.

Actually these are mostly comes from following commits on the HEAD:
src/release/Makefile rev 1.721
src/usr.sbin/sysinstall/modules.c rev 1.5

Approved by: murray (as of re@, pre-approved)

18 years agoMFC 1.3: 7955/7956 support
Sam Leffler [Wed, 8 Oct 2003 23:57:50 +0000 (8 23:57 +0000)]
MFC 1.3: 7955/7956 support

Approved by: re (murray)

18 years agoMFC 7955/7956 symmetric crypto support
Sam Leffler [Wed, 8 Oct 2003 23:52:00 +0000 (8 23:52 +0000)]
MFC 7955/7956 symmetric crypto support

Note performance is currently suboptimal.

Approved by: re (murray)

18 years agoAvoid segfaulting if current working directory is unreadable.
Ruslan Ermilov [Wed, 8 Oct 2003 08:14:22 +0000 (8 08:14 +0000)]
Avoid segfaulting if current working directory is unreadable.
(HEAD is not subject to this bug.)

PR: bin/56087
Submitted by: Volker Stolz <stolz@i2.informatik.rwth-aachen.de>
Approved by: murray@re

18 years agoGo back to PIO mode for now; MMIO mode appears to cause problems on
Murray Stokely [Wed, 8 Oct 2003 06:01:57 +0000 (8 06:01 +0000)]
Go back to PIO mode for now; MMIO mode appears to cause problems on
some systems.

Submitted by: Mike Silbersack <silby@silby.com>

18 years agoThe test for whether or not this is a Pentium 4 for the purpose of
Jacques Vidrine [Wed, 8 Oct 2003 02:26:39 +0000 (8 02:26 +0000)]
The test for whether or not this is a Pentium 4 for the purpose of
disabling PSE on such processors was unnecessarily made twice.

But much worse, the second test was incorrect.

Symptoms reported by: Mike Tancsa <mike@sentex.net> and others
``This commit looks wrong.'': tegge
Pointy hat to: nectar

18 years agoAdd NO_PSE_HACK kernel option.
Jacques Vidrine [Tue, 7 Oct 2003 16:58:53 +0000 (7 16:58 +0000)]
Add NO_PSE_HACK kernel option.

Reminded by: jhb

18 years agoDisable PSE at boot time for Pentium 4 processors. On many such CPUs,
Jacques Vidrine [Tue, 7 Oct 2003 12:34:37 +0000 (7 12:34 +0000)]
Disable PSE at boot time for Pentium 4 processors.  On many such CPUs,
data corruption will result otherwise.  It seems that the bug is
aggravated by recent work related to PAE.

This change was originally based on rev 1.386, but now the test for
CPU type is correct and the display of the warning is postponed until
we have initialized the kernel message buffer (e.g. for dmesg).

peter@ believes that only certain revisions of the Pentium 4 processor
are affected.  However, we are unsure which so we choose to be safe by
disabling PSE on all Pentium 4s.

The test can be disabled entirely with NO_PSE_HACK.

Approved by: re

18 years agoRespect DESTDIR.
Ruslan Ermilov [Tue, 7 Oct 2003 06:59:19 +0000 (7 06:59 +0000)]
Respect DESTDIR.

PR: 54004
Submitted by: Sergey Akifyev <asa@gascom.ru>
Approved by: murray@re

18 years agoMFC: Add a warning about the safe(4) driver's experimental status.
Bruce A. Mah [Mon, 6 Oct 2003 19:11:17 +0000 (6 19:11 +0000)]
MFC:  Add a warning about the safe(4) driver's experimental status.

Submitted by: sam
Approved by: re (implicitly)

18 years agoMFC 1.81: "short cable" fix for DP8315
Sam Leffler [Mon, 6 Oct 2003 18:29:23 +0000 (6 18:29 +0000)]
MFC 1.81: "short cable" fix for DP8315

Approved by: re (murray)

18 years agoMFC 1.240: add IPSEC_FILTERGIF suport for FAST_IPSEC
Sam Leffler [Mon, 6 Oct 2003 18:25:59 +0000 (6 18:25 +0000)]
MFC 1.240: add IPSEC_FILTERGIF suport for FAST_IPSEC

Approved by: re (murray)

18 years agoMFC 1.2:
Sam Leffler [Mon, 6 Oct 2003 17:03:08 +0000 (6 17:03 +0000)]
MFC 1.2:

o add missing {}'s that to safe_dmamap_uniform that caused extraneous copies
  for partly-aligned operations through /dev/crypto (unlikely)
o add missing case in iov code that never showed up because of the above bug

Approved by: re (murray)

18 years agoNew errata: SA-03:15.
Bruce A. Mah [Mon, 6 Oct 2003 04:21:21 +0000 (6 04:21 +0000)]
New errata:  SA-03:15.

Approved by: re (implicitly)

18 years agoMFC: SA-03:15, mount_msdos(8) -M.
Bruce A. Mah [Mon, 6 Oct 2003 04:15:56 +0000 (6 04:15 +0000)]
MFC:  SA-03:15, mount_msdos(8) -M.

Approved by: re (implicitly)

18 years agoDriver names for DRIVER_MODULE in current are 'ed' and 'ep', but in
Warner Losh [Mon, 6 Oct 2003 02:53:52 +0000 (6 02:53 +0000)]
Driver names for DRIVER_MODULE in current are 'ed' and 'ep', but in
-stable they need to be 'if_ed' and 'if_ep'.  ifconfig should be
modified, but this is a lower risk change this late in the release
cycle.

Approved by: re@

18 years agoBump release.manpath.freebsd to 4.8-stable.
Hiroki Sato [Sun, 5 Oct 2003 16:59:31 +0000 (5 16:59 +0000)]
Bump release.manpath.freebsd to 4.8-stable.

Approved by: re (bmah)

18 years agoMFC: SA-03:18.
Hiroki Sato [Sun, 5 Oct 2003 16:57:58 +0000 (5 16:57 +0000)]
MFC: SA-03:18.

Approved by: re (bmah)

18 years agoNew errata: SA-03:16.
Hiroki Sato [Sun, 5 Oct 2003 16:56:24 +0000 (5 16:56 +0000)]
New errata: SA-03:16.
MFC: SA-03:14, SA-03:17, SA-03:18.

Approved by: re (bmah)

18 years agoNew release note:
Hiroki Sato [Sun, 5 Oct 2003 08:24:06 +0000 (5 08:24 +0000)]
New release note:
SA-03:16.
MFC:
SA-03:14, SA-03:17.

Approved by: re (bmah)

18 years agoRevision 1.10.2.4 introduced a regression: the offset was ignored.
Jacques Vidrine [Sat, 4 Oct 2003 15:50:35 +0000 (4 15:50 +0000)]
Revision 1.10.2.4 introduced a regression: the offset was ignored.
This could cause a reader of /proc/*/regs to spin, because read(2)
would never return end-of-file.  In practice, this has much less
impact than I would expect.  Tools such as truss(1) make only a
single, fixed-size read (into a `struct reg').  (Similarly for
write(2).)

The cause was a sort of typo: `uiomove_frombuf' was spelled
`uiomove'.

Noticed by: David Rhodus <drhodus@catpa.com>
Approved by: re (implicitly)

18 years agoMFC: 1.48
Joe Marcus Clarke [Fri, 3 Oct 2003 16:50:17 +0000 (3 16:50 +0000)]
MFC: 1.48

Export PARALLEL_PACAKGE_BUILD to the environment as well to catch more
packages that should/should not be included on the first CD.  This is needed
for the recent KDE space-saving change.

Approved by: re (bmah)

18 years agoCorrect several integer underflows/overflows in linprocfs. See the
Jacques Vidrine [Fri, 3 Oct 2003 12:03:50 +0000 (3 12:03 +0000)]
Correct several integer underflows/overflows in linprocfs.  See the
log message for kern_subr.c 1.77 for more details.

18 years agoMFC OpenSSL 0.9.7c.
Jacques Vidrine [Fri, 3 Oct 2003 01:32:14 +0000 (3 01:32 +0000)]
MFC OpenSSL 0.9.7c.

Approved by: re

18 years agoMFC (the spirit of) procfs_dbregs.c 1.23, procfs_fpregs.c 1.29,
Jacques Vidrine [Thu, 2 Oct 2003 16:49:49 +0000 (2 16:49 +0000)]
MFC (the spirit of) procfs_dbregs.c 1.23, procfs_fpregs.c 1.29,
    pseudofs_vnops.c 1.42, kern_subr.c 1.77, uio.h 1.28:
Correct several integer underflows/overflows in procfs.

Approved by: re

18 years agoCorrect a case in readv(2) where the file descriptor reference count
Jacques Vidrine [Thu, 2 Oct 2003 15:08:01 +0000 (2 15:08 +0000)]
Correct a case in readv(2) where the file descriptor reference count
should have been decremented but was not.

Reported by: Joost Pol <joost@pine.nl>
Approved by: re

18 years agoAdd an include I forgot in the previous commit. Eeek.
Mike Silbersack [Thu, 2 Oct 2003 02:22:58 +0000 (2 02:22 +0000)]
Add an include I forgot in the previous commit.  Eeek.

18 years agoAdd the Dell PERC 4/Di to the list of supported devices for amr(4).
Simon L. B. Nielsen [Wed, 1 Oct 2003 20:36:57 +0000 (1 20:36 +0000)]
Add the Dell PERC 4/Di to the list of supported devices for amr(4).

Reported by: David Magda <dmagda@magda.ca>
Approved by: re (bmah)

18 years agoMFC rev 1.309 ; increase the # of kmap entries allowed.
Mike Silbersack [Wed, 1 Oct 2003 19:39:42 +0000 (1 19:39 +0000)]
MFC rev 1.309 ; increase the # of kmap entries allowed.

Approved by: re (murray)

18 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_4'.
cvs2svn [Wed, 1 Oct 2003 12:32:43 +0000 (1 12:32 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_4'.

18 years agoMFC: The knobs controlling ipfw2 and ip6fw were not protected when running
Bruce M Simpson [Wed, 1 Oct 2003 07:07:07 +0000 (1 07:07 +0000)]
MFC: The knobs controlling ipfw2 and ip6fw were not protected when running
at an elevated securelevel. Fix this behaviour and function as documented.

PR: kern/39396
Approved by: re (rwatson)

18 years agoFix a function prototype error.
Luoqi Chen [Wed, 1 Oct 2003 04:02:18 +0000 (1 04:02 +0000)]
Fix a function prototype error.

Approved by: re@

18 years agoFix a bug which was introduced in version 1.187.2.16 of
Andrew Gallatin [Wed, 1 Oct 2003 01:21:37 +0000 (1 01:21 +0000)]
Fix a bug which was introduced in version 1.187.2.16 of
sys/vm/vm_map.c and affects the unwiring memory:

The map entries are processed in a loop, checking to make sure the
entry is wired and asserting it has a wired count. However, another
loop was inserted more-or-less in the middle of the of the unwiring
path.  This loop picks up the "entry" loop variable from the first loop
without first setting it to start_entry.  Naturally, the second loop
is never entered and the pages backing the entries are never unwired.
This can lead to a leak of wired pages.

Reviewed by: alc
Approved by: re (murray)

18 years agoMFC rev 1.154
Mike Silbersack [Tue, 30 Sep 2003 18:23:22 +0000 (30 18:23 +0000)]
MFC rev 1.154

Approved by: re

18 years agoMFC: Fix a cut n paste typo.
Tom Rhodes [Mon, 29 Sep 2003 15:53:54 +0000 (29 15:53 +0000)]
MFC: Fix a cut n paste typo.

18 years agoUpdate for 4.9-RC.
Murray Stokely [Sun, 28 Sep 2003 22:56:57 +0000 (28 22:56 +0000)]
Update for 4.9-RC.

18 years agoWelcome to 4.9-RC.
Murray Stokely [Sun, 28 Sep 2003 22:55:45 +0000 (28 22:55 +0000)]
Welcome to 4.9-RC.
RC1 is imminent.

Approved by: re

18 years ago - Update to list of FTP sites.
Ken Smith [Sun, 28 Sep 2003 13:05:26 +0000 (28 13:05 +0000)]
- Update to list of FTP sites.

Approved by: murray

18 years agoSave and restore the interrupt mask in fork1() instead of using
Ian Dowse [Sun, 28 Sep 2003 11:08:31 +0000 (28 11:08 +0000)]
Save and restore the interrupt mask in fork1() instead of using
spl0() and hence assuming that the old interrupt mask was zero.
Although fork1() is normally called with a zero mask, one exception
to this is when it is called by kthread_create() before configure()
has unmasked interrupts for the first time. In that case it can
cause interrupt storm hangs during device configuration, because
interrupts can be delivered before the corresponding device drivers
are ready to receive them.

This should fix the boot-time USB hangs that have been reported.

Approved by: re (murray)

18 years agoMFC rev 1.55
Paul Saab [Fri, 26 Sep 2003 16:02:04 +0000 (26 16:02 +0000)]
MFC rev 1.55
By not setting No_CRC in the Mode Control Register, we must also
reduce the size of the packet by 4 bytes to remove the ethernet crc.

Approved by: re

18 years agoMFC:
Scott Long [Fri, 26 Sep 2003 15:24:14 +0000 (26 15:24 +0000)]
MFC:
aic79xx_pci.c:
aic7xxx_pci.c:
When performing our register test, be careful
to avoid resetting the chip when pausing the
controller.  The test reads the HCNTRL register
and then writes it back with the PAUSE bit
explicitly set.  If the last write to the controller
before our probe is to reset it, the CHIPRST
bit will still be set, so we must mask it off
before the PAUSE operation.  On some chip versions,
we cannot access registers for a few 100us after
a reset, so this inadvertant reset was causing PCI
errors to occur on the read to check for paused
status.

Approved by: re (murray)

18 years ago - Updated list of FTP sites.
Ken Smith [Fri, 26 Sep 2003 12:07:43 +0000 (26 12:07 +0000)]
- Updated list of FTP sites.

Approved by: murray

18 years agoMFC: 1.17, 1.18, 1.24, 1.28.
Ruslan Ermilov [Fri, 26 Sep 2003 07:55:18 +0000 (26 07:55 +0000)]
MFC: 1.17, 1.18, 1.24, 1.28.

Urged by: Sergey Benner <sigurd@sunbay.com>
Approved by: re (blanket)

18 years agoMFC:
Wilko Bulte [Thu, 25 Sep 2003 21:18:47 +0000 (25 21:18 +0000)]
MFC:
Clarify SRM bootability on Alpha for DAC960.

Approved by: re (bmah) / blanket

18 years agoMFC BUS_DMA_ZERO support for busdma memory allocations, and M_ZERO
Mike Silbersack [Thu, 25 Sep 2003 18:55:27 +0000 (25 18:55 +0000)]
MFC BUS_DMA_ZERO support for busdma memory allocations, and M_ZERO
support for contigmalloc, which is required for BUS_DMA_ZERO support.

Approved by:    re (scottl)

18 years agoMFC some of the security safety belts from sendmail 8.12.10 for
Gregory Neil Shapiro [Thu, 25 Sep 2003 04:53:38 +0000 (25 04:53 +0000)]
MFC some of the security safety belts from sendmail 8.12.10 for
4.9-RELEASE.  A full MFC of 8.12.10 will occur after 4.9-RELEASE.

Approved by: re (murray)

18 years agoFix a logic error in the check to see if arplookup() should free the route.
Bruce M Simpson [Wed, 24 Sep 2003 21:51:13 +0000 (24 21:51 +0000)]
Fix a logic error in the check to see if arplookup() should free the route.

Noticed by: Mike Hogsett
Reviewed by: ru

18 years agoUpdate version string.
Dag-Erling Smørgrav [Wed, 24 Sep 2003 19:28:35 +0000 (24 19:28 +0000)]
Update version string.

18 years agoMFC: (1.13) remove bogus calls to xfree().
Dag-Erling Smørgrav [Wed, 24 Sep 2003 19:16:50 +0000 (24 19:16 +0000)]
MFC: (1.13) remove bogus calls to xfree().

18 years agoMFC: (1.12) resp is a pointer to an array of structs.
Dag-Erling Smørgrav [Wed, 24 Sep 2003 18:32:22 +0000 (24 18:32 +0000)]
MFC: (1.12) resp is a pointer to an array of structs.

18 years agoMFC: (1.7) return the correct error value when a null query fails.
Dag-Erling Smørgrav [Wed, 24 Sep 2003 18:32:12 +0000 (24 18:32 +0000)]
MFC: (1.7) return the correct error value when a null query fails.

18 years agoProtect crfree() with splhigh; without this, there appears to be
Mike Silbersack [Tue, 23 Sep 2003 18:22:13 +0000 (23 18:22 +0000)]
Protect crfree() with splhigh; without this, there appears to be
a crfree race which can occur on SMP systems.

Submitted by: Ed Maste <emaste@sandvine.com>
Approved by: re

18 years agoFix a bug in arplookup(), whereby a hostile party on a locally
Bruce M Simpson [Tue, 23 Sep 2003 16:54:39 +0000 (23 16:54 +0000)]
Fix a bug in arplookup(), whereby a hostile party on a locally
attached network could exhaust kernel memory, and cause a system
panic, by sending a flood of spoofed ARP requests.

Approved by: security-officer, jake (mentor)
Reported by: Apple Product Security <product-security@apple.com>

18 years agoMFC: rev 1.54, Do not set the No_CRC bit in the Mode Control Register.
Paul Saab [Tue, 23 Sep 2003 02:34:49 +0000 (23 02:34 +0000)]
MFC: rev 1.54, Do not set the No_CRC bit in the Mode Control Register.

18 years agoMFC: 1.60: Better documentation of -redirect_port port range syntax.
Ruslan Ermilov [Mon, 22 Sep 2003 18:59:45 +0000 (22 18:59 +0000)]
MFC: 1.60: Better documentation of -redirect_port port range syntax.

18 years agoMFC: r1.41 - disc1 is over 680megabytes right now, so remove
Murray Stokely [Mon, 22 Sep 2003 07:19:04 +0000 (22 07:19 +0000)]
MFC: r1.41 - disc1 is over 680megabytes right now, so remove
linux-netscape-navigator for now to free up space (-communicator is a
superset and is still included).

Approved by: re (murray)

18 years agoMFC: Fix name of no-reset device (ulpn -> unlpt) and fix style.
Stefan Eßer [Sun, 21 Sep 2003 20:17:54 +0000 (21 20:17 +0000)]
MFC: Fix name of no-reset device (ulpn -> unlpt) and fix style.
This file is now identical to HEAD, except for CVS IDs.

Approved by: re

18 years agoPut x11-wm/sawfish2 and irc/xchat on CD 1 instead of sawfish and xchat.
Joe Marcus Clarke [Sun, 21 Sep 2003 19:07:23 +0000 (21 19:07 +0000)]
Put x11-wm/sawfish2 and irc/xchat on CD 1 instead of sawfish and xchat.
Sawfish is gone, and xchat2 is more widely used these days since it is the
current stable release.

Approved by: re (murray)

18 years agoPrevent vmstat -m totals from overflowing by using a 64-bit type. Likely
Bruce M Simpson [Sat, 20 Sep 2003 19:10:01 +0000 (20 19:10 +0000)]
Prevent vmstat -m totals from overflowing by using a 64-bit type. Likely
to be needed for systems using PAE.

PR: bin/55124
Reviewed by: bde
Approved by: re, jake

18 years agoFix a minor glitch in conditional text.
Bruce A. Mah [Sat, 20 Sep 2003 00:20:26 +0000 (20 00:20 +0000)]
Fix a minor glitch in conditional text.

Approved by: re (implicitly)

18 years agoNew errata: SA-03:13.
Bruce A. Mah [Sat, 20 Sep 2003 00:19:42 +0000 (20 00:19 +0000)]
New errata:  SA-03:13.

Approved by: re (implicitly)

18 years agoMFC: SA-03:13, GNOME 2.4, KDE 3.1.4.
Bruce A. Mah [Sat, 20 Sep 2003 00:03:01 +0000 (20 00:03 +0000)]
MFC:  SA-03:13, GNOME 2.4, KDE 3.1.4.

18 years agoInclude a patch number in the version header in case 8.12.10 isn't
Gregory Neil Shapiro [Fri, 19 Sep 2003 21:19:36 +0000 (19 21:19 +0000)]
Include a patch number in the version header in case 8.12.10 isn't
MFC'ed before 4.9-RELEASE.  8.12.10 will be imported into -CURRENT
this weekend.

Approved by: re (murray)

18 years agoMFC: Preserve the state of %fs and %gs when sending a signal. NVidia
Daniel Eischen [Fri, 19 Sep 2003 18:35:36 +0000 (19 18:35 +0000)]
MFC: Preserve the state of %fs and %gs when sending a signal.  NVidia
     drivers (for one) seem to need this.  This has existed in -current
     for some time since both libthr and libkse also require this
     behavior.

Requested by: kde@
Approved by: re (jhb, murray)

18 years agoMFC up to 1.96
Doug Barton [Fri, 19 Sep 2003 06:55:23 +0000 (19 06:55 +0000)]
MFC up to 1.96

Approved by: re (bmah, murray)

18 years agoMFC rev. 1.22:
Thomas Quinot [Thu, 18 Sep 2003 20:15:55 +0000 (18 20:15 +0000)]
MFC rev. 1.22:
(atapi_action, case XPT_PATH_INQ): Handle properly the case of
CAM_TARGET_WILDCARD (target_id_t is an unsigned integer type).

Approved by: re (murray)