]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoThe sequencer downloading code assumes that all jump
Scott Long [Thu, 5 Dec 2002 00:47:27 +0000 (00:47 +0000)]
The sequencer downloading code assumes that all jump
labels are acurate in relation to a fully compiled
sequencer program (all patches downloaded).  Correct
a few occurances of a relative jump across a macro
that ended up jumping us into the last instruction
of the macro.

Spproved by: re (bmah)

21 years agoForced commit:
Chris Costello [Thu, 5 Dec 2002 00:40:31 +0000 (00:40 +0000)]
Forced commit:
 Credit Hiten M. Pandya <hiten@angelica.unixdaemons.com> with reporting that
 suser_cred(9) should've been suser(9).

21 years agoDocument the following MAC policies:
Chris Costello [Thu, 5 Dec 2002 00:05:38 +0000 (00:05 +0000)]
Document the following MAC policies:
o Biba: A data integrity policy
o BSD Extended: Support for the firewall-like access controls (ugidfw(8))
o MLS: Multi-level security, a confidentiality policy

(These files originally lived in src/share/man/man9)

Approved by: re (blanket)
Sponsored by: DARPA, Network Associates Labs
Obtained from: TrustedBSD Project

21 years agoRemove an unused variable that snuck in the last commit.
Eric Anholt [Wed, 4 Dec 2002 23:39:05 +0000 (23:39 +0000)]
Remove an unused variable that snuck in the last commit.

Submitted by: Christian Brueffer <chris@unixpages.org>
Approved by: re (rwatson)

21 years agoNew release note: INDEX-5.
Bruce A. Mah [Wed, 4 Dec 2002 22:55:21 +0000 (22:55 +0000)]
New release note:  INDEX-5.

Modified release notes:  gcc-3.2.1 release, further clarification of
Perl stuff.

21 years agoLast minute fixes to ahc and ahd:
Scott Long [Wed, 4 Dec 2002 22:51:29 +0000 (22:51 +0000)]
Last minute fixes to ahc and ahd:

ahd_pci.c:
Retrieve the allow_memio hint from the resource manager to
determine whether or not to try PCI MEMIO.

aic79xx_osm.h:
aic7xxx_osm.h:
Don't wrongly abuse the callout_reset() interface when trying
to abuse timeouts generated from the CAM layer.  This fixes the
console freeze and lost timeout problem that many have reported,
especially on SMP systems.

aic79xx_pci.c
aic7xxx_pci.c
Rewrite the MEMIO test routine to prevent certain broken chipsets
from trying to burst multiple DWORDs to the registers.  Also make
the routine better detect byte merging by the host bridge and
deal with it.

aic79xx.reg:
Correct an incorrect register definition.

Approved by: re (rwatson, jhb)

21 years agoRemove a const that generates a warning: the const isn't strictly
Robert Watson [Wed, 4 Dec 2002 22:01:03 +0000 (22:01 +0000)]
Remove a const that generates a warning: the const isn't strictly
wrong, since we don't want the variable changed, but since we assign
it to variables that may also refer to other non-const strings,
warnings were generated that could break LINT.

Approved by: re
Spotted by: sam

21 years agoMerge from the English version and fix some mistranslations.
Hiroki Sato [Wed, 4 Dec 2002 21:39:38 +0000 (21:39 +0000)]
Merge from the English version and fix some mistranslations.

Reviewed by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Yoshihiko SARUMARU <mistral@imasy.or.jp>
References: [doc-jp-work 531]
Approved by: re (bmah)

21 years agoI really hate the GCC directory structure of gcc/gcc/gcc/<compiler files>
David E. O'Brien [Wed, 4 Dec 2002 21:31:04 +0000 (21:31 +0000)]
I really hate the GCC directory structure of gcc/gcc/gcc/<compiler files>

21 years agoConsistently mark std(in|out|err) with .Dv, because that's how they
Ruslan Ermilov [Wed, 4 Dec 2002 18:57:46 +0000 (18:57 +0000)]
Consistently mark std(in|out|err) with .Dv, because that's how they
are marked up in stdio(3), and because they are defined expressions
of type "FILE *".

Approved by: re

21 years agoHold the page queues lock around calls to pmap_remove().
Alan Cox [Wed, 4 Dec 2002 18:40:39 +0000 (18:40 +0000)]
Hold the page queues lock around calls to pmap_remove().

Approved by: re

21 years agoFix witness warning in DRM memory info sysctl by making a temporary copy of the
Eric Anholt [Wed, 4 Dec 2002 18:27:45 +0000 (18:27 +0000)]
Fix witness warning in DRM memory info sysctl by making a temporary copy of the
data under the lock and outputting it to the sysctl later.

Reviewed by: scottl
Approved by: re

21 years agomdoc(7) police: markup overhaul.
Ruslan Ermilov [Wed, 4 Dec 2002 17:49:15 +0000 (17:49 +0000)]
mdoc(7) police: markup overhaul.

Approved by: re

21 years agoUpdate for Gcc 3.2.1 release.
David E. O'Brien [Wed, 4 Dec 2002 17:47:34 +0000 (17:47 +0000)]
Update for Gcc 3.2.1 release.
We also have more wide char support in libc now.

21 years agoGperf is usable again.
David E. O'Brien [Wed, 4 Dec 2002 17:34:21 +0000 (17:34 +0000)]
Gperf is usable again.

21 years agoMinimal change to cause pkg_version to examine the new INDEX-5 file
Kris Kennaway [Wed, 4 Dec 2002 17:22:48 +0000 (17:22 +0000)]
Minimal change to cause pkg_version to examine the new INDEX-5 file
instead of INDEX.  This should be done better since this file name will
depend on the FreeBSD major version.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107609,
David E. O'Brien [Wed, 4 Dec 2002 17:08:08 +0000 (17:08 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107609,
which included commits to RCS files with non-trunk default branches.

21 years agoGcc 3.2.1 release virgin vendor import. (19-Nov-2002)
David E. O'Brien [Wed, 4 Dec 2002 17:08:08 +0000 (17:08 +0000)]
Gcc 3.2.1 release virgin vendor import. (19-Nov-2002)

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107606,
David E. O'Brien [Wed, 4 Dec 2002 16:53:59 +0000 (16:53 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107606,
which included commits to RCS files with non-trunk default branches.

21 years agoGcc 3.2.1 release virgin vendor import. (19-Nov-2002)
David E. O'Brien [Wed, 4 Dec 2002 16:53:59 +0000 (16:53 +0000)]
Gcc 3.2.1 release virgin vendor import. (19-Nov-2002)

21 years agoUpdate HEAD with FSF branch 3.2.1 release files.
David E. O'Brien [Wed, 4 Dec 2002 16:35:55 +0000 (16:35 +0000)]
Update HEAD with FSF branch 3.2.1 release files.

21 years agoUse pure stock files.
David E. O'Brien [Wed, 4 Dec 2002 16:31:48 +0000 (16:31 +0000)]
Use pure stock files.

21 years agomdoc(7) police: markup overhaul.
Ruslan Ermilov [Wed, 4 Dec 2002 16:28:45 +0000 (16:28 +0000)]
mdoc(7) police: markup overhaul.

Approved by: re

21 years agoMerge rev 1.8 (printf format error fixes) into Gcc 3.2.1 release.
David E. O'Brien [Wed, 4 Dec 2002 16:13:54 +0000 (16:13 +0000)]
Merge rev 1.8 (printf format error fixes) into Gcc 3.2.1 release.

21 years agoGperf is usable again. We don't need this.
David E. O'Brien [Wed, 4 Dec 2002 16:13:07 +0000 (16:13 +0000)]
Gperf is usable again.  We don't need this.

21 years agoFix typo to sync with FSF sources.
David E. O'Brien [Wed, 4 Dec 2002 16:09:42 +0000 (16:09 +0000)]
Fix typo to sync with FSF sources.

21 years agoMerge revs 1.10 (-mno-align-long-strings) and 1.11 (LONG_TYPE_SIZE override)
David E. O'Brien [Wed, 4 Dec 2002 16:07:58 +0000 (16:07 +0000)]
Merge revs 1.10 (-mno-align-long-strings) and 1.11 (LONG_TYPE_SIZE override)
into Gcc 3.2.1 release.

21 years agoMerge rev 1.11 (-mno-align-long-strings) into Gcc 3.2.1 release.
David E. O'Brien [Wed, 4 Dec 2002 16:04:26 +0000 (16:04 +0000)]
Merge rev 1.11 (-mno-align-long-strings) into Gcc 3.2.1 release.

21 years agoUpdate for Gcc 3.2.1 release.
David E. O'Brien [Wed, 4 Dec 2002 16:01:15 +0000 (16:01 +0000)]
Update for Gcc 3.2.1 release.

21 years agoMerge revs 1.2 (-fformat-extensions), 1.7 (FORCE_OPTIMIZATION_DOWNGRADE knob),
David E. O'Brien [Wed, 4 Dec 2002 15:59:56 +0000 (15:59 +0000)]
Merge revs 1.2 (-fformat-extensions), 1.7 (FORCE_OPTIMIZATION_DOWNGRADE knob),
1.14 (-O/-O1 optimize alignment for time, not size) into Gcc 3.2.1 release.

21 years agoMerge revs 1.{2,15} (FREEBSD_NATIVE); 1.21 ("GCC_OPTIONS"); 1.{16,25,27}
David E. O'Brien [Wed, 4 Dec 2002 15:57:11 +0000 (15:57 +0000)]
Merge revs 1.{2,15} (FREEBSD_NATIVE); 1.21 ("GCC_OPTIONS"); 1.{16,25,27}
(cross compiling issues) into GCC 3.2.1 release.

Note revs 1.{5,13} (ELF and objformat support) are OBE with the a.out dike-out.

21 years agoMerge revs 1.2 (kernel printf framework) and 1.3 (suppress K&R main warnings)
David E. O'Brien [Wed, 4 Dec 2002 15:48:26 +0000 (15:48 +0000)]
Merge revs 1.2 (kernel printf framework) and 1.3 (suppress K&R main warnings)
into Gcc 3.2.1 release.

21 years agomdoc(7) police: overhaul.
Ruslan Ermilov [Wed, 4 Dec 2002 15:47:41 +0000 (15:47 +0000)]
mdoc(7) police: overhaul.

Approved by: re

21 years agoGcc 3.2.1 release virgin vendor import. (19-Nov-2002)
David E. O'Brien [Wed, 4 Dec 2002 15:42:16 +0000 (15:42 +0000)]
Gcc 3.2.1 release virgin vendor import. (19-Nov-2002)

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107590,
David E. O'Brien [Wed, 4 Dec 2002 15:42:16 +0000 (15:42 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107590,
which included commits to RCS files with non-trunk default branches.

21 years agoGcc 3.2.1 release virgin vendor import. (19-Nov-2002)
David E. O'Brien [Wed, 4 Dec 2002 15:15:07 +0000 (15:15 +0000)]
Gcc 3.2.1 release virgin vendor import. (19-Nov-2002)

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107588,
David E. O'Brien [Wed, 4 Dec 2002 15:15:07 +0000 (15:15 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107588,
which included commits to RCS files with non-trunk default branches.

21 years agoFixed a partition type for pc98 when create or change slices in the fdisk
Yoshihiro Takahashi [Wed, 4 Dec 2002 15:07:05 +0000 (15:07 +0000)]
Fixed a partition type for pc98 when create or change slices in the fdisk
editor.

Approved by: re (jhb)

21 years agoMFi386: revision 1.551.
Yoshihiro Takahashi [Wed, 4 Dec 2002 15:04:18 +0000 (15:04 +0000)]
MFi386: revision 1.551.

Approved by: re (jhb)

21 years agoChange the process title as soon as possible to mask information passed on
Dag-Erling Smørgrav [Wed, 4 Dec 2002 15:00:10 +0000 (15:00 +0000)]
Change the process title as soon as possible to mask information passed on
the command line by getty(8).  This is not a perfect fix, but drastically
reduces the window of exposure.

Approved by: re (rwatson)
MFC after: 1 week

21 years agomdoc(7) police: nit.
Ruslan Ermilov [Wed, 4 Dec 2002 14:46:33 +0000 (14:46 +0000)]
mdoc(7) police: nit.

Approved by: re

21 years agomdoc(7) police: bump document date for the C rewrite, nit.
Ruslan Ermilov [Wed, 4 Dec 2002 14:45:48 +0000 (14:45 +0000)]
mdoc(7) police: bump document date for the C rewrite, nit.

Approved by: re

21 years agomdoc(7) police: overhaul.
Ruslan Ermilov [Wed, 4 Dec 2002 14:44:20 +0000 (14:44 +0000)]
mdoc(7) police: overhaul.

Approved by: re

21 years agomdoc(7) police: s/nul character/null character/.
Ruslan Ermilov [Wed, 4 Dec 2002 14:42:49 +0000 (14:42 +0000)]
mdoc(7) police: s/nul character/null character/.

Approved by: re

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107579,
Dag-Erling Smørgrav [Wed, 4 Dec 2002 14:37:35 +0000 (14:37 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107579,
which included commits to RCS files with non-trunk default branches.

21 years agoAdd default cases to the switches (perforce change 21902)
Dag-Erling Smørgrav [Wed, 4 Dec 2002 14:37:35 +0000 (14:37 +0000)]
Add default cases to the switches (perforce change 21902)

21 years agoUse the correct value when writing the Day Of Week byte in the CMOS.
Poul-Henning Kamp [Wed, 4 Dec 2002 13:46:49 +0000 (13:46 +0000)]
Use the correct value when writing the Day Of Week byte in the CMOS.
The correct range is [1...7] with Sunday=1, but we have been writing
[0...6] with Sunday=0.

The Soekris computers flagged the zero, zapped the date, so if you
rebooted your soekris on a sunday, it would come up with a wrong
date.

Bruce has a more extensive rework of this code, but we will stick with
the minimalist fix for now.

Spotted by: Soren Kristensen <soren@soekris.com>
Thanks to: Michael Sierchio <kudzu@tenebras.com>.
Confirmed by: bde
Approved by: re

21 years agortld support for PowerPC. Mostly obtained from NetBSD, with mods
Peter Grehan [Wed, 4 Dec 2002 07:32:20 +0000 (07:32 +0000)]
rtld support for PowerPC. Mostly obtained from NetBSD, with mods
for binutils 2.13

Reviewed by:  benno

Approved by:  re (blanket)

21 years agoMinimal libc for PowerPC.
Peter Grehan [Wed, 4 Dec 2002 07:25:14 +0000 (07:25 +0000)]
Minimal libc for PowerPC.

Reviewed by: benno

Approved by: re (blanket)

21 years agoReduce WARNS level to 3 in the non-crypto case to fix make release on sparc64.
Jake Burkholder [Wed, 4 Dec 2002 04:34:59 +0000 (04:34 +0000)]
Reduce WARNS level to 3 in the non-crypto case to fix make release on sparc64.
This may only be turning up now to changes in the cpp predefines for sparc64,
which may be turning on more strict alignment checking.

Approved by: re

21 years agoFix minor gramm-o.
Bruce A. Mah [Wed, 4 Dec 2002 04:17:28 +0000 (04:17 +0000)]
Fix minor gramm-o.

Submitted by: keramida
Approved by: re (implicitly)

21 years agoSigh. Another wrong version string. However, this one isn't used in the
Peter Wemm [Wed, 4 Dec 2002 00:07:15 +0000 (00:07 +0000)]
Sigh. Another wrong version string.  However, this one isn't used in the
build.

Approved by: re

21 years agoNew release notes: adduser(8)/rmuser(8) and the Great Perl Rewrite,
Bruce A. Mah [Tue, 3 Dec 2002 23:03:45 +0000 (23:03 +0000)]
New release notes:  adduser(8)/rmuser(8) and the Great Perl Rewrite,
dump(8) -L.

Modified release notes: DEVICE_POLLING for nge(4) [1], clarification
of some UFS2 items, Perl distribution in sysinstall.

Submitted by: Hiten Pandya <hiten@angelica.unixdaemons.com> [1]
Approved by: re (implicitly)

21 years agoReformulate how sysinstall handles file system options in the label
Robert Watson [Tue, 3 Dec 2002 22:25:47 +0000 (22:25 +0000)]
Reformulate how sysinstall handles file system options in the label
editor, in order to support specifying UFS2 as a newfs option.

(1) Support three different newfs types: NEWFS_UFS, NEWFS_MSDOS, and
    NEWFS_CUSTOM.  Don't mix up the arguments to them: you can't use
    soft updates on an msdos file system.

(2) Distinguish adding new arguments to the newfs command line from
    replacing it.  Permit the addition of new arguments by the user for
    NEWFS_UFS.  If we entirely replace the command line provided by
    sysinstall, call it NEWFS_CUSTOM.  'N' will now add additional
    arguments; 'Z' will opt to replace the newfs command line entirely,
    but will prompt the user with their current command line as a
    starting point.

(3) Construct the newfs command line dynamically based on the options
    provided by the user at label-time.  Right now, this means selecting
    UFS1 vs. UFS2, and the soft updates flag.  Drop in some variables
    to support ACLs and MAC Multilabel in the future also, but don't
    expose them now.

This provides sysinstall with the ability to do more "in band" editing
of the newfs command line, so we can provide more support for the user,
but doesn't sacrifice the ability to entirely specify the newfs command
line of the user is willing to give up on the cushiness factor.  It
also makes it easier for us to specify defaults in the future, and
define conditional behavior based on user configuration selections.
For now, we default to UFS1, and permit UFS2 to be used as the root
only on non-i386 systems.

While I was there, I dropped the default fragment and block sizes,
since newfs has much more sensible defaults now.

Reviewed by: jhb, marcel
Approved by: re
ia64 bits from: marcel

21 years agoOops. Use the correct version number. The number that was here was the
Peter Wemm [Tue, 3 Dec 2002 21:59:42 +0000 (21:59 +0000)]
Oops. Use the correct version number.  The number that was here was the
one that I originally used on freefall etc.

Approved by: re

21 years agoOops, forgot this cosmetic tweak to the perl menu item in my previous
John Baldwin [Tue, 3 Dec 2002 21:02:26 +0000 (21:02 +0000)]
Oops, forgot this cosmetic tweak to the perl menu item in my previous
commit.

Approved by: re (this is what they actually reviewed)

21 years agoAdd support for the PC98 platform to the ATA driver.
Søren Schmidt [Tue, 3 Dec 2002 20:20:44 +0000 (20:20 +0000)]
Add support for the PC98 platform to the ATA driver.
This mostly consists of functionality to serialize accesses to
the two ATA channels (which can also be used to "fix" certain
PCI based controllers).
Add support for Acard controllers.
Enable the ATA driver in PC98 GENERIC, and add device hints.
Update man page with latest support.

The PC98 core team has kindly provided me with a PC98
machine that made this all possible, thanks to all that
contributed to that effort, without that this would
probably newer have been possible..

Approved by: re@

21 years ago- Convert the installation of the X package to using installPackage()
John Baldwin [Tue, 3 Dec 2002 19:42:30 +0000 (19:42 +0000)]
- Convert the installation of the X package to using installPackage()
  rather than installX11package().
- Add a perl psuedo-dist that installs the perl package.  The perl
  distribution is selected by default when a User distribution set is
  selected.  It is not selected when a Minimal distribution set is
  selected.  The perl distribution may be toggled manually in the
  custom menu just as other distributions.

Approved by: re

21 years agoRename installX11package() to installPackage() and tweak it so that it
John Baldwin [Tue, 3 Dec 2002 19:36:12 +0000 (19:36 +0000)]
Rename installX11package() to installPackage() and tweak it so that it
can be used to install any arbitrary package.

Approved by: re

21 years agoAdd the `L' option to dump to notify it that it is dumping a
Kirk McKusick [Tue, 3 Dec 2002 18:21:09 +0000 (18:21 +0000)]
Add the `L' option to dump to notify it that it is dumping a
live filesystem. To obtain a consistent dump image, dump takes
a snapshot of the filesystem and then does a dump of the snapshot.
The snapshot is removed when the dump is complete.

Also add an operator warning that the `L' option should be used
if dump is run on a live filesystem without the `L' option being
specified. The alternative would be to silently use a snapshot
any time that a live filesystem is dumped, but this change in
dump semantics seemed too drastic at this time.

Sponsored by:   DARPA & NAI Labs.
Approved by: re

21 years agoHave to use bread() rather than UFS_BALLOC() when obtaining a
Kirk McKusick [Tue, 3 Dec 2002 18:19:27 +0000 (18:19 +0000)]
Have to use bread() rather than UFS_BALLOC() when obtaining a
previously allocated block as the previous use of the block may
have fallen out of the cache. Failure to reread its contents cause
zeroed results to be written instead of the proper contents.
Conversely, when the block is going to be entirely filled in, it
is not necessary reread the old contents.

Sponsored by:   DARPA & NAI Labs.
Approved by: re

21 years ago"i386" is probably a better long name for the i386 platform, as
Bruce A. Mah [Tue, 3 Dec 2002 18:00:30 +0000 (18:00 +0000)]
"i386" is probably a better long name for the i386 platform, as
opposed to IA-32.

Pointed out by: Andrew I Baznikin <dikiy@scn.ru>, jhb
Approved by: re (implicitly)

21 years agoMinor, cosmetic change to buildworld output to make it reflect
Bruce A. Mah [Tue, 3 Dec 2002 17:49:05 +0000 (17:49 +0000)]
Minor, cosmetic change to buildworld output to make it reflect
rev. 1.270 more accurately.

Reviewed by: ru
Approved by: re (jhb)

21 years agoUpdate the second copy of libstand to deal with the new ufs2 superblock
Jake Burkholder [Tue, 3 Dec 2002 16:25:29 +0000 (16:25 +0000)]
Update the second copy of libstand to deal with the new ufs2 superblock
format.  The one in sys/boot/libstand is not connected to the build.

Approved by: re

21 years agoFix a bug in the fdisk editor that partially masked the chunk deletion
John Baldwin [Tue, 3 Dec 2002 16:09:55 +0000 (16:09 +0000)]
Fix a bug in the fdisk editor that partially masked the chunk deletion
bug fixed yesterday.  New slices created in the fdisk editor and slices
whose sub-type is changed are of type 'mbr' if their sub-type is not a
magic type, not type 'unknown'.

Approved by: re

21 years agoSince OpenSSH drops privileges before calling pam_open_session(3),
Dag-Erling Smørgrav [Tue, 3 Dec 2002 15:48:11 +0000 (15:48 +0000)]
Since OpenSSH drops privileges before calling pam_open_session(3),
pam_lastlog(8) can't possibly work, so let OpenSSH handle lastlog.

Approved by: re (rwatson)

21 years agoThe default CD drive is /dev/cd0, not /dev/cd0c.
Dag-Erling Smørgrav [Tue, 3 Dec 2002 15:41:30 +0000 (15:41 +0000)]
The default CD drive is /dev/cd0, not /dev/cd0c.

Reviewed by: jhb
Approved by: re (rwatson)

21 years agoLicense tweak: remove clause 3 per permission of NAI.
Robert Watson [Tue, 3 Dec 2002 15:26:01 +0000 (15:26 +0000)]
License tweak: remove clause 3 per permission of NAI.

Approved by: re
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoMdoc markup and language fixes.
Robert Watson [Tue, 3 Dec 2002 15:19:15 +0000 (15:19 +0000)]
Mdoc markup and language fixes.

Submitted by: ru
Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoHook up a sample LOMAC labeling policy. Unlike the old LOMAC module,
Robert Watson [Tue, 3 Dec 2002 15:16:10 +0000 (15:16 +0000)]
Hook up a sample LOMAC labeling policy.  Unlike the old LOMAC module,
the file system initial labeling policy exists in userland, and is
fed into setfsmac(1).  This is based on the old LOMAC PLM.

Approved by: re
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoUnhook the old LOMAC module, now replaced with mac_lomac.
Robert Watson [Tue, 3 Dec 2002 15:12:36 +0000 (15:12 +0000)]
Unhook the old LOMAC module, now replaced with mac_lomac.

Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoDon't install old LOMAC include files; do install new mac_lomac
Robert Watson [Tue, 3 Dec 2002 15:05:32 +0000 (15:05 +0000)]
Don't install old LOMAC include files; do install new mac_lomac
include files.

Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoProperly account for prefetchable memory when a request is being made.
Warner Losh [Tue, 3 Dec 2002 08:34:20 +0000 (08:34 +0000)]
Properly account for prefetchable memory when a request is being made.
We allow the request to go through if it matches either a prefetchable
or a non-prefetchable part of the bridge.  We do not check to make
sure it is the right kind of memory because most drivers to not yet
properly set RF_PREFETCHABLE (only cardbus seems to do so, and I'm not
entirely sure it does it right).  RF_PREFETCHABLE was invented for
cardbus, so hasn't been properly documented yet.

This is still overridable by hw.pci.allow_unsupported_io_ranges, but
the need for that is greatly reduced, especially for the nvida driver.

Approved by: re
Reviewed by: jhb and many testers
Submitted by: Matt Emmerton (although this has been reworked somewhat)

21 years agoReplace the perl versions of adduser and rmuser with shell script versions.
Scott Long [Tue, 3 Dec 2002 05:41:09 +0000 (05:41 +0000)]
Replace the perl versions of adduser and rmuser with shell script versions.

Submitted by: Mike Makonnen <mtm@identd.net>
Approved by: re

21 years agoCorrect the estimated block count calculated by dump to account
Kirk McKusick [Tue, 3 Dec 2002 05:12:53 +0000 (05:12 +0000)]
Correct the estimated block count calculated by dump to account
for the minimal amount of space used by a snapshot.

Sponsored by:   DARPA & NAI Labs.

21 years agoProperly handle UFS2 sparsely allocated inodes. The UFS2 filesystem
Kirk McKusick [Tue, 3 Dec 2002 05:10:07 +0000 (05:10 +0000)]
Properly handle UFS2 sparsely allocated inodes. The UFS2 filesystem
only preallocates a small number of inodes. The dump program tries
to scan through all the allocated inodes on a filesystem which
causes bad behavior if they have never been allocated. Thus dump
must calculate the set of inodes that have actually been allocated
and scan only those inodes.

Sponsored by:   DARPA & NAI Labs.

21 years agoAdd spaces around FBSD_ENDFILE_SPEC as it is used in string concatenation.
David E. O'Brien [Tue, 3 Dec 2002 04:31:19 +0000 (04:31 +0000)]
Add spaces around FBSD_ENDFILE_SPEC as it is used in string concatenation.

Approved by: re(bmah)

21 years agoAdd lib/libfetch and usr.bin/fetch to FIXCRYPTO since they need the crypto
Dag-Erling Smørgrav [Tue, 3 Dec 2002 04:02:13 +0000 (04:02 +0000)]
Add lib/libfetch and usr.bin/fetch to FIXCRYPTO since they need the crypto
libraries for SSL support.

Approved by: re (rwatson)

21 years agoAvoid recursive acquisition of the page queues lock in pmap_unuse_pt().
Alan Cox [Tue, 3 Dec 2002 04:00:42 +0000 (04:00 +0000)]
Avoid recursive acquisition of the page queues lock in pmap_unuse_pt().

Approved by: re

21 years agoImplement DIOCGDINFO for volumes. newfs will no longer build a file
Greg Lehey [Mon, 2 Dec 2002 23:56:59 +0000 (23:56 +0000)]
Implement DIOCGDINFO for volumes.  newfs will no longer build a file
system on a volume without a disk label.

Approved by: re (rwatson)

21 years agoDon't blindly call make_dev when configuring an object: the device
Greg Lehey [Mon, 2 Dec 2002 23:53:32 +0000 (23:53 +0000)]
Don't blindly call make_dev when configuring an object: the device
could already exist, and this triggers a booby trap panic in make_dev.

remove_plex_entry: Don't remove the stripe mutex here, it gets done in
free_plex.

Approved by: re (rwatson)

21 years agoRelax partition overlap check to allow Vinum partitions to overlap
Greg Lehey [Mon, 2 Dec 2002 23:49:59 +0000 (23:49 +0000)]
Relax partition overlap check to allow Vinum partitions to overlap
other partitiosns.  This is necessary when migrating conventional
partitions to Vinum and was broken by recent more stringent overlap
checks.  This is arguably the wrong way to do it.  A better method
would be to have the loader understand a subset of Vinum partitioning
and allow an install directly to Vinum, but until then, this is the
best we have.

Reviewed by: jhb
Approved by: re (rwatson)

21 years agoBreak up a bunch of crazy if statements to use a case statement instead
John Baldwin [Mon, 2 Dec 2002 21:42:29 +0000 (21:42 +0000)]
Break up a bunch of crazy if statements to use a case statement instead
to be cleaner.  Also, when deleting a chunk, try to find the mother chunk
as a whole chunk by default if this isn't a BSD partition or a unused or
whole chunk.  Before we just did this for FreeBSD and FAT slices, which
means that other chunk types such as EFI and mbr (mbr is used for slices
that don't have their own chunk type).

Submitted by: nyan (mostly)
Approved by: re

21 years agoUpdated release notes: Binutils 2.13.1-20021011, CVS
Bruce A. Mah [Mon, 2 Dec 2002 21:22:43 +0000 (21:22 +0000)]
Updated release notes:  Binutils 2.13.1-20021011, CVS
1.11.2.1-20021201.

MFCs noted:  gre(4), Heimdal Kerberos 0.5.1.

21 years agoMake a CR/LF sequence measure up to its name: the order of the CR and LF
Peter Pentchev [Mon, 2 Dec 2002 20:38:12 +0000 (20:38 +0000)]
Make a CR/LF sequence measure up to its name: the order of the CR and LF
characters was reversed, resulting in some network appliances, including
at least some NAS devices from Ascend, not recognizing our finger(1)
request.

PR: 45914
Submitted by: J R Matthews <jrm@delta-e.com.au>
Approved by: re (rwatson)
MFC after: 2 weeks

21 years agoReplace the remaining strcpy() instances with strlcpy(), fixing a segfault
Peter Pentchev [Mon, 2 Dec 2002 20:29:08 +0000 (20:29 +0000)]
Replace the remaining strcpy() instances with strlcpy(), fixing a segfault
when parsing a malformed command-line parameter.
Rearrange a risky usage of sprintf() in a loop.

Reported by: phrail@division7.us via the vuln-dev mailing list
Approved by: re (rwatson)

21 years agoUse the hotspot code to prevent people from overwriting their disklabel
Poul-Henning Kamp [Mon, 2 Dec 2002 20:23:54 +0000 (20:23 +0000)]
Use the hotspot code to prevent people from overwriting their disklabel
with stuff which would ruin the day for any open parititons.

Approved by: re

21 years agoia64 specific.
Marcel Moolenaar [Mon, 2 Dec 2002 20:15:16 +0000 (20:15 +0000)]
ia64 specific.
o  Mount the EFI file system as msdosfs and not ufs as it's a FAT
   file system. Introduce Mount_msdos() for this to go side-by-side
   with Mount().
o  Also, since mounting is performed as a command (which means it's
   queued, sorted, lost, found and executed), we cannot create a
   directory on the file system by calling mkdir. We must make sure
   the mkdir happens after the mount. Introduce Mkdir_command() to
   allow mkdir operations to be queued, sorted, lost, found and
   executed as well.

Approved by: re (jhb, rwatson)

21 years agoAdd a simplified version of the hot-spot code to enable us to protect
Poul-Henning Kamp [Mon, 2 Dec 2002 19:59:25 +0000 (19:59 +0000)]
Add a simplified version of the hot-spot code to enable us to protect
in-band disklabels from in-band vandalism.

Approve by: re

21 years agoAlign the FPU state in the ucontext and sigcontext to 16 bytes
Daniel Eischen [Mon, 2 Dec 2002 19:58:55 +0000 (19:58 +0000)]
Align the FPU state in the ucontext and sigcontext to 16 bytes
to accomodate the new SSE/XMM floating point save/restore
instructions.

This commit is mostly from bde and includes some style nits.

Approved by: re (jhb)

21 years agoCorrectly calculate the initial number of fragments in a filesystem
Kirk McKusick [Mon, 2 Dec 2002 19:31:53 +0000 (19:31 +0000)]
Correctly calculate the initial number of fragments in a filesystem
so that fsck does not complain with `SUMMARY BLK COUNT(S) WRONG IN
SUPERBLK' the first time it is run on a new filesystem.

Reported by: Poul-Henning Kamp <phk@freebsd.org>
Sponsored by:   DARPA & NAI Labs.

21 years agoVerify that alternate superblocks have a correct magic number before
Kirk McKusick [Mon, 2 Dec 2002 19:30:37 +0000 (19:30 +0000)]
Verify that alternate superblocks have a correct magic number before
trying to use them. Set a minimum value for numdirs when using an
alternate superblock to avoid spurious numdirs == 0 error. Calculate
new fields when using an alternate superblock from a UFS1 filesystem
to avoid segment faulting.

Sponsored by:   DARPA & NAI Labs.

21 years agoRemove a workaround for a binutils bug that was fixed in the recent
Thomas Moestl [Mon, 2 Dec 2002 18:56:30 +0000 (18:56 +0000)]
Remove a workaround for a binutils bug that was fixed in the recent
import, as it breaks the relocation kernel modules built with the new
binutils.
Note that this, together with the binutils import, marks a kernel module
flag day on sparc64: modules built with the old binutils will not work
with new kernels and vice versa. Mismatches will result in panics.

Approved by: re

21 years agoIf /usr/bin/make fails to pass the regression tests, it might not
Ruslan Ermilov [Mon, 2 Dec 2002 14:31:21 +0000 (14:31 +0000)]
If /usr/bin/make fails to pass the regression tests, it might not
be suitable for building.  The old action here was to upgrade the
make in /usr/bin.  This is both bogus and does not work under non
root.  So, rather than overwriting the existing /usr/bin/make, we
now install it into a safe location, and use it from there.

Reviewed by: imp, marcel
Approved by: re (rwatson)

21 years agoFor installkernel, use the fresh tools (if we've built them with
Ruslan Ermilov [Mon, 2 Dec 2002 14:03:40 +0000 (14:03 +0000)]
For installkernel, use the fresh tools (if we've built them with
buildworld).

Reviewed by: imp, marcel
Approved by: re (rwatson)

21 years agoCast argument to (long), missed in prior pass.
Robert Watson [Mon, 2 Dec 2002 13:26:16 +0000 (13:26 +0000)]
Cast argument to (long), missed in prior pass.

Approved by: re
Submitted by: marcel

21 years agoFix 32-bit/64-bit bug in format string.
Robert Watson [Mon, 2 Dec 2002 13:25:01 +0000 (13:25 +0000)]
Fix 32-bit/64-bit bug in format string.

Approved by: re
Submitted by: marcel
Pointy hat to: green

21 years agoRemove the ancient PEI cruft that is useless.
David E. O'Brien [Mon, 2 Dec 2002 10:50:38 +0000 (10:50 +0000)]
Remove the ancient PEI cruft that is useless.

21 years agoWe can handle the bfd_elf32_i386_vec vector at the same time as the
David E. O'Brien [Mon, 2 Dec 2002 10:01:18 +0000 (10:01 +0000)]
We can handle the bfd_elf32_i386_vec vector at the same time as the
bfd_elf32_i386_freebsd_vec.  This may help in intra-branch building.

21 years agoWe can handle the bfd_elf64_alpha_vec vector at the same time as the
David E. O'Brien [Mon, 2 Dec 2002 10:00:44 +0000 (10:00 +0000)]
We can handle the bfd_elf64_alpha_vec vector at the same time as the
bfd_elf64_alpha_freebsd_vec.  This may help in intra-branch building.