]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoMinor cosmetic changes to minimise diffs for MFC.
Søren Schmidt [Thu, 7 Mar 2002 16:34:19 +0000 (16:34 +0000)]
Minor cosmetic changes to minimise diffs for MFC.

22 years agoFix a couble of bugs in the rebuild code, return errors properly.
Søren Schmidt [Thu, 7 Mar 2002 16:32:21 +0000 (16:32 +0000)]
Fix a couble of bugs in the rebuild code, return errors properly.

22 years agoFix build for OpenPAM. The directories needed tweeking.
Mark Murray [Thu, 7 Mar 2002 16:03:56 +0000 (16:03 +0000)]
Fix build for OpenPAM. The directories needed tweeking.

22 years agoAdd new errno ``ENOATTR''.
Brian Feldman [Thu, 7 Mar 2002 15:13:44 +0000 (15:13 +0000)]
Add new errno ``ENOATTR''.

22 years agoUpdate version string.
Jacques Vidrine [Thu, 7 Mar 2002 14:36:28 +0000 (14:36 +0000)]
Update version string.

22 years agoRevert/fix the most controversial parts of the last commit:
Mark Murray [Thu, 7 Mar 2002 14:11:41 +0000 (14:11 +0000)]
Revert/fix the most controversial parts of the last commit:

1) place __FBSDID() correctly IAW style(9).

2) revert (and slightly correct) a "complex" expression in an if()
   statement.

22 years ago1) Rev.1.35 of dd.c has a more serious regression. It backs out rev.1.31,
Mark Murray [Thu, 7 Mar 2002 14:00:33 +0000 (14:00 +0000)]
1) Rev.1.35 of dd.c has a more serious regression.  It backs out rev.1.31,
   thus breaking systems with unpolluted <sys/stat.h>'s.

2) Back out an initialisation of a variable in BSS.

Reported by: bde (1), many(2)

22 years agoUpdate FreeBSD SCM IDs.
David E. O'Brien [Thu, 7 Mar 2002 12:52:27 +0000 (12:52 +0000)]
Update FreeBSD SCM IDs.

22 years agoInstall the DSI and ISI trap handlers and their appropriate locations.
Benno Rice [Thu, 7 Mar 2002 12:22:44 +0000 (12:22 +0000)]
Install the DSI and ISI trap handlers and their appropriate locations.

22 years agoCopy the "implementation" of pmap_prefault from sparc64.
Benno Rice [Thu, 7 Mar 2002 12:22:08 +0000 (12:22 +0000)]
Copy the "implementation" of pmap_prefault from sparc64.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r91799,
Dag-Erling Smørgrav [Thu, 7 Mar 2002 12:11:34 +0000 (12:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r91799,
which included commits to RCS files with non-trunk default branches.

22 years agos/PAM_REINITIALISE_CRED/PAM_REINITIALIZE_CRED/
Dag-Erling Smørgrav [Thu, 7 Mar 2002 12:11:34 +0000 (12:11 +0000)]
s/PAM_REINITIALISE_CRED/PAM_REINITIALIZE_CRED/

Sponsored by: DARPA, NAI Labs

22 years agoThis file is not needed any more
Dag-Erling Smørgrav [Thu, 7 Mar 2002 12:03:50 +0000 (12:03 +0000)]
This file is not needed any more

22 years agoAdd missing #include <string.h>
Dag-Erling Smørgrav [Thu, 7 Mar 2002 11:59:04 +0000 (11:59 +0000)]
Add missing #include <string.h>

22 years agoThis commit was generated by cvs2svn to compensate for changes in r91796,
Dag-Erling Smørgrav [Thu, 7 Mar 2002 11:59:04 +0000 (11:59 +0000)]
This commit was generated by cvs2svn to compensate for changes in r91796,
which included commits to RCS files with non-trunk default branches.

22 years agoDarn. There is (now was) a signed/unsigned issue that resulted in a
Mark Murray [Thu, 7 Mar 2002 10:41:11 +0000 (10:41 +0000)]
Darn. There is (now was) a signed/unsigned issue that resulted in a
very long loop.

Reported by: nnd@mail.nsk.ru (Nickolay Dudorov)

22 years agoMove tunable initialisation so it can get access to physmem.
Benno Rice [Thu, 7 Mar 2002 10:15:17 +0000 (10:15 +0000)]
Move tunable initialisation so it can get access to physmem.

22 years agoCalculate physmem.
Benno Rice [Thu, 7 Mar 2002 10:09:24 +0000 (10:09 +0000)]
Calculate physmem.

22 years agoLeave the `copyright' variable exposed, since the main purpose of it
Mike Barcroft [Thu, 7 Mar 2002 10:06:00 +0000 (10:06 +0000)]
Leave the `copyright' variable exposed, since the main purpose of it
is to put a copyright in the object file.

Submitted by: bde

22 years agoUpdate for the 6.9 firmware family (6xxx controllers), as well as for
Mike Smith [Thu, 7 Mar 2002 09:55:41 +0000 (09:55 +0000)]
Update for the 6.9 firmware family (6xxx controllers), as well as for
the upcoming 7.4 family (7xxx controllers).

 - improved error reporting and handling
 - more diagnostic output
 - add extra command packet definitions
 - merge sources again with -stable

22 years agoDon't use __FBSDID yet. Looks like most (all?) of the rest of the kernel
Warner Losh [Thu, 7 Mar 2002 08:12:40 +0000 (08:12 +0000)]
Don't use __FBSDID yet.  Looks like most (all?) of the rest of the kernel
doesn't do that.

22 years agoImplement hw.cardbus.debug and hw.cardbus.cis_debug to help debug some
Warner Losh [Thu, 7 Mar 2002 08:10:42 +0000 (08:10 +0000)]
Implement hw.cardbus.debug and hw.cardbus.cis_debug to help debug some
of the cardbus problems that people may start seeing.

22 years agoMake hw.pccard.debug and hw.pccard.cis_debug tunable/sysctl. Setting to 1
Warner Losh [Thu, 7 Mar 2002 08:03:53 +0000 (08:03 +0000)]
Make hw.pccard.debug and hw.pccard.cis_debug tunable/sysctl.  Setting to 1
will enable more verbose debugging output from the pccard system.

22 years agofix backwards spelling
David E. O'Brien [Thu, 7 Mar 2002 07:17:15 +0000 (07:17 +0000)]
fix backwards spelling

22 years agoImplement delivery of tlb shootdown ipis. This is currently more fine grained
Jake Burkholder [Thu, 7 Mar 2002 06:01:40 +0000 (06:01 +0000)]
Implement delivery of tlb shootdown ipis.  This is currently more fine grained
than the other implementations; we have complete control over the tlb, so we
only demap specific pages.  We take advantage of the ranged tlb flush api
to send one ipi for a range of pages, and due to the pm_active optimization
we rarely send ipis for demaps from user pmaps.

Remove now unused routines to load the tlb; this is only done once outside
of the tlb fault handlers.
Minor cleanups to the smp startup code.

This boots multi user with both cpus active on a dual ultra 60 and on a
dual ultra 2.

22 years agoModify the tlb demap API to take a pmap instead of a tlb context number.
Jake Burkholder [Thu, 7 Mar 2002 05:25:15 +0000 (05:25 +0000)]
Modify the tlb demap API to take a pmap instead of a tlb context number.
Due to allocating tlb contexts on the fly, we only ever need to demap the
primary context, non-primary contexts have already been implicitly flushed
by context switching.  All we really need to tell is if its a kernel demap
or not, and its easier just to compare against the kernel_pmap which is a
constant.

22 years agoImplement kthread context stealing. This is a bit of a misnomer because
Jake Burkholder [Thu, 7 Mar 2002 05:15:43 +0000 (05:15 +0000)]
Implement kthread context stealing.  This is a bit of a misnomer because
the context is not actually stolen, as it would be for i386.  Instead of
deactivating a user vmspace immediately when switching out, and recycling
its tlb context, wait until the next context switch to a different user
vmspace.  In this way we can switch from a user process to any number of
kernel threads and back to the same user process again, without losing any
of its mappings in the tlb that would not already be knocked by the automatic
replacement algorithm.  This is not expected to have a measurable performance
improvement on the machines we currently run on, but it sounds cool and makes
the sparc64 port SMPng buzz word compliant.

22 years agoUnconditionally limit maxproc so that it is not possible
Mike Silbersack [Thu, 7 Mar 2002 04:50:36 +0000 (04:50 +0000)]
Unconditionally limit maxproc so that it is not possible
to exhaust all kmaps.  The only reward for setting maxproc
to a value which will cause kmap exhaustion is a panic
during a forkbomb attack.

MFC after: 3 days

22 years agoInclude machine/smp.h.
Jake Burkholder [Thu, 7 Mar 2002 04:48:17 +0000 (04:48 +0000)]
Include machine/smp.h.

22 years agoAdd needed includes of machine/smp.h, remove nested include in sys/smp.h
Jake Burkholder [Thu, 7 Mar 2002 04:43:51 +0000 (04:43 +0000)]
Add needed includes of machine/smp.h, remove nested include in sys/smp.h
so that inlines in machine/smp.h can use variables declared in sys/smp.h.

22 years agoFix a bug in the vm_map_clean() procedure. msync()ing an area of memory
Matthew Dillon [Thu, 7 Mar 2002 03:54:56 +0000 (03:54 +0000)]
Fix a bug in the vm_map_clean() procedure.  msync()ing an area of memory
that has just been mapped MAP_ANON|MAP_NOSYNC and has not yet been accessed
will panic the machine.

MFC after: 1 day

22 years agoMFen.
Jun Kuriyama [Thu, 7 Mar 2002 03:53:15 +0000 (03:53 +0000)]
MFen.

22 years agoNow pam_alreadyloggedin lives in the ports.
Brian Feldman [Thu, 7 Mar 2002 02:23:19 +0000 (02:23 +0000)]
Now pam_alreadyloggedin lives in the ports.

22 years agoTwo style(9) fixes:
Warner Losh [Thu, 7 Mar 2002 00:11:42 +0000 (00:11 +0000)]
Two style(9) fixes:
- return(foo); (note parens)
- use __FBSDID()

22 years agoCheck for NULL on resource allocation. For the moment, punt, but we should
Warner Losh [Thu, 7 Mar 2002 00:05:26 +0000 (00:05 +0000)]
Check for NULL on resource allocation.  For the moment, punt, but we should
be smarter about a) cleanup and b) fallback.

22 years agoAdd entry for xterm-color
Luigi Rizzo [Wed, 6 Mar 2002 19:47:20 +0000 (19:47 +0000)]
Add entry for xterm-color

22 years agoSync with GENERIC. WITNESS_SKIPSPIN, and the ciss and iir devices.
Crist J. Clark [Wed, 6 Mar 2002 19:44:08 +0000 (19:44 +0000)]
Sync with GENERIC. WITNESS_SKIPSPIN, and the ciss and iir devices.

22 years agoProperly clear the status of a join operation if the joining thread is
Daniel Eischen [Wed, 6 Mar 2002 19:28:41 +0000 (19:28 +0000)]
Properly clear the status of a join operation if the joining thread is
canceled or the joinee is detached.

22 years agoTo quote BDE: "the k part of awk should understand the ctype macros and not
David E. O'Brien [Wed, 6 Mar 2002 18:37:26 +0000 (18:37 +0000)]
To quote BDE: "the k part of awk should understand the ctype macros and not
               need the -funsigned-char hack."

Also add needed CLEANSFILE.

22 years agoAdd the pam_alreadyloggedin(8) module, which allows for authentication
Brian Feldman [Wed, 6 Mar 2002 18:21:28 +0000 (18:21 +0000)]
Add the pam_alreadyloggedin(8) module, which allows for authentication
based on information that the user is already logged in.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoSimpler way to handling the names needed on the various platforms.
David E. O'Brien [Wed, 6 Mar 2002 18:06:39 +0000 (18:06 +0000)]
Simpler way to handling the names needed on the various platforms.

22 years agoAdd the "Avlab Technology, PCI IO 2S" card.
John Hay [Wed, 6 Mar 2002 17:37:27 +0000 (17:37 +0000)]
Add the "Avlab Technology, PCI IO 2S" card.

Submitted by: Terry Glanfield <Terry.Glanfield@program-products.co.uk>

22 years agoRewrite the section on downloading files to get all of the URLs in one
Bruce A. Mah [Wed, 6 Mar 2002 17:24:19 +0000 (17:24 +0000)]
Rewrite the section on downloading files to get all of the URLs in one
place, near the front of the document.

Also try to provide some introduction on the difference between CDROM
and network installs.

22 years agoNo functional change, but big code cleanup. WARNS, lint(1) and style(9).
Mark Murray [Wed, 6 Mar 2002 17:18:09 +0000 (17:18 +0000)]
No functional change, but big code cleanup. WARNS, lint(1) and style(9).

22 years agoPoint sparc64 &release.url; at a directory, rather than a specific
Bruce A. Mah [Wed, 6 Mar 2002 16:57:53 +0000 (16:57 +0000)]
Point sparc64 &release.url; at a directory, rather than a specific
file, to be consistent with other platforms.

22 years agoUnbreak the pam_krb5 build: cast a couple of const pointers
Peter Pentchev [Wed, 6 Mar 2002 16:49:02 +0000 (16:49 +0000)]
Unbreak the pam_krb5 build: cast a couple of const pointers
to normal char *.  A better fix might be some const'ifying
of the Heimdal code, but this will do to fix the build
for the present.

Approved by: des

22 years agoRename runq_find() to runq_findproc(), and hide it behind #ifdef DIAGNOSTIC,
Dag-Erling Smørgrav [Wed, 6 Mar 2002 15:34:07 +0000 (15:34 +0000)]
Rename runq_find() to runq_findproc(), and hide it behind #ifdef DIAGNOSTIC,
as it can have a severe impact on performance under high load, and the bug
it was meant to catch was fixed ages ago.

22 years agoRemove a bogus cast.
Dag-Erling Smørgrav [Wed, 6 Mar 2002 15:24:51 +0000 (15:24 +0000)]
Remove a bogus cast.

22 years agoYA patch I forgot to commit last night.
Dag-Erling Smørgrav [Wed, 6 Mar 2002 15:23:18 +0000 (15:23 +0000)]
YA patch I forgot to commit last night.

22 years agoAdd GETPID to the list of operations for which semctl(2) returns the value.
Maxim Konovalov [Wed, 6 Mar 2002 14:14:36 +0000 (14:14 +0000)]
Add GETPID to the list of operations for which semctl(2) returns the value.

Reviewed by: alfred, ru
Approved by: ru
MFC after: 3 days

22 years agoRemoved gprof4. 8-byte counters are now handled by gprof.
Bruce Evans [Wed, 6 Mar 2002 14:10:48 +0000 (14:10 +0000)]
Removed gprof4.  8-byte counters are now handled by gprof.

22 years agoBelatedly OpenPAMify. I forgot this patch in last night's megacommit.
Dag-Erling Smørgrav [Wed, 6 Mar 2002 12:46:56 +0000 (12:46 +0000)]
Belatedly OpenPAMify.  I forgot this patch in last night's megacommit.

Sponsored by: DARPA, NAI Labs

22 years agoFix a typo, unbreak the world.
Maxim Konovalov [Wed, 6 Mar 2002 12:28:51 +0000 (12:28 +0000)]
Fix a typo, unbreak the world.

Thanks to: mux
Approved by: ru

22 years agoAdd forgotten NOPROFILE that broke world.
Dag-Erling Smørgrav [Wed, 6 Mar 2002 12:11:05 +0000 (12:11 +0000)]
Add forgotten NOPROFILE that broke world.

22 years agoBoundary check an array holding tab stops.
Ruslan Ermilov [Wed, 6 Mar 2002 12:05:17 +0000 (12:05 +0000)]
Boundary check an array holding tab stops.

PR: bin/35590
Submitted by: "Tim J. Robbins" <tim@robbins.dropbear.id.au>

22 years agoDon't (blindly) truncate the unit number to 4 digits when formatting the
Bruce Evans [Wed, 6 Mar 2002 11:34:02 +0000 (11:34 +0000)]
Don't (blindly) truncate the unit number to 4 digits when formatting the
string returned by device_get_nameunit().

22 years agoSupport runtime specification of the history counter type by whatever
Bruce Evans [Wed, 6 Mar 2002 11:25:58 +0000 (11:25 +0000)]
Support runtime specification of the history counter type by whatever
generated the gmon data.  The support is currently limited to what is
easy to implement and/or needed:

    signedess: signed or insigned
    size: 8, 16, 32 or 64 bits
    format: a binary integer in gprof's format (gprof is not a cross-tool).

High-resolution kernel profiling uses signed 64-bit counters.  Normal
kernel profiling and user profiling use unsigned 16-bit counters but
should use 32-bit ones.

22 years agoLog:
Maxim Konovalov [Wed, 6 Mar 2002 11:20:13 +0000 (11:20 +0000)]
Log:
Remove eaccess(2) absence workaround. Add eaccess(2) checks for FILRD,
FILWR, FILEX and FILEXIST cases.

We cannot MFC this because there is no eaccess(2) in -stable yet.

PR: bin/35076
Reviewed by: ru
Approved by: ru

22 years agoMaximum semid is seminfo.semmni not seminfo.semmsl.
Maxim Konovalov [Wed, 6 Mar 2002 10:52:49 +0000 (10:52 +0000)]
Maximum semid is seminfo.semmni not seminfo.semmsl.

PR: kern/34979
Submitted by: James Gritton <jamie@gritton.org>
Reviewed by: alfred, ru
Approved by: ru
MFC after: 1 week

22 years agoFixed some misspellings of 2 as sizeof(UNIT) so that they won't break
Bruce Evans [Wed, 6 Mar 2002 09:47:36 +0000 (09:47 +0000)]
Fixed some misspellings of 2 as sizeof(UNIT) so that they won't break
things when sizeof(UNIT) becomes a runtime parameter.  The relevant 2
is the one in profil(2)'s scaling of pc's to bucket numbers:

   bucket = (pc - offset) / 2 * profil_scale / 65536

gprof(1) must duplicate this scaling, bug for bug compatibly, so it
must first do an integer division by 2 although this mainly makes
scales larger than 65536 useless.  sizeof(UNIT) was already wrong in
gprof4, but there were no problems because the fake profil scale is a
multiple of 2.

There are also some rounding bugs in the scaling, but these are only
problems if profil(2) is used directly to create unusual (and not
useful) scales.

22 years agoFixed compilation warnings.
Ruslan Ermilov [Wed, 6 Mar 2002 08:55:34 +0000 (08:55 +0000)]
Fixed compilation warnings.

22 years ago1-true-AWK has a build-tool target.
David E. O'Brien [Wed, 6 Mar 2002 08:24:32 +0000 (08:24 +0000)]
1-true-AWK has a build-tool target.

22 years agoMake it more obvious where to find RELNOTESng source files if someone went
Bruce A. Mah [Wed, 6 Mar 2002 07:59:18 +0000 (07:59 +0000)]
Make it more obvious where to find RELNOTESng source files if someone went
browsing in the old release notes location.

While I'm here, update README files to be closer to reality.

Requested by: obrien
MFC after: 1 day

22 years agoFirst round of markup/spelling/grammar fixes.
Bruce A. Mah [Wed, 6 Mar 2002 07:36:29 +0000 (07:36 +0000)]
First round of markup/spelling/grammar fixes.

22 years agoFixed a printf format error again. Rev.127 was clobbered in rev.1.128
Bruce Evans [Wed, 6 Mar 2002 06:33:33 +0000 (06:33 +0000)]
Fixed a printf format error again.  Rev.127 was clobbered in rev.1.128
by removing parentheses.  The main bug is in gcc: on machines with
64-bit longs and 64-bit long longs,

    (unsigned long long)rdp->total_sectors / ((1024L * 1024L) / DEV_BSIZE))

has type plain unsigned long instead of the correctly promoted type
unsigned long long, so it can not be printfed using %llu format.  Even
1ULL / 1L is mispromoted.  Anyway, casting the correct operand
automatically avoids the problem.  We do not want to to pessimize the
division; we just want to convert to a common maximal type for printing.

22 years agoDefault to UTC on sparc64 also.
David E. O'Brien [Wed, 6 Mar 2002 06:18:21 +0000 (06:18 +0000)]
Default to UTC on sparc64 also.

22 years agoAdd a carriage return to each function declaration, and make use of ID tag
Eric Melville [Wed, 6 Mar 2002 03:26:11 +0000 (03:26 +0000)]
Add a carriage return to each function declaration, and make use of ID tag
macros.

MFC after: 1 day

22 years agoAdd a sequential iteration optimization to vm_object_page_clean(). This
Matthew Dillon [Wed, 6 Mar 2002 02:42:56 +0000 (02:42 +0000)]
Add a sequential iteration optimization to vm_object_page_clean().  This
moderately improves msync's and VM object flushing for objects containing
randomly dirtied pages (fsync(), msync(), filesystem update daemon),
and improves cpu use for small-ranged sequential msync()s in the face of
very large mmap()ings from O(N) to O(1) as might be performed by a database.

A sysctl, vm.msync_flush_flag, has been added and defaults to 3 (the two
committed optimizations are turned on by default).  0 will turn off both
optimizations.

This code has already been tested under stable and is one in a series of
memq / vp->v_dirtyblkhd / fsync optimizations to remove O(N^2) restart
conditions that will be coming down the pipe.

MFC after: 3 days

22 years agoCreate a manual to document the new kernel byte order conversion
Mike Barcroft [Wed, 6 Mar 2002 01:53:35 +0000 (01:53 +0000)]
Create a manual to document the new kernel byte order conversion
functions.

Reviewed by: tmm

22 years agocleanup readability syntax prior to ongoing b_resid work commits.
Matthew Dillon [Wed, 6 Mar 2002 00:44:30 +0000 (00:44 +0000)]
cleanup readability syntax prior to ongoing b_resid work commits.

MFC after: 1 day

22 years agoAdd installation instructions for the FreeBSD/sparc64 port. I have very
Thomas Moestl [Tue, 5 Mar 2002 23:22:08 +0000 (23:22 +0000)]
Add installation instructions for the FreeBSD/sparc64 port. I have very
likely introduced language and markup bugs, but it's a start.

Makefile, initial docbook conversion & cleanups by: murray

22 years agoNew release notes: CPU_ATHLON_SSE_HACK, ufm driver (+ MFC),
Bruce A. Mah [Tue, 5 Mar 2002 22:39:51 +0000 (22:39 +0000)]
New release notes:  CPU_ATHLON_SSE_HACK, ufm driver (+ MFC),
awk is BWK awk for sparc64, ctags(1) works for files with C++ comments
(+ MFC), dump(8) -s, last(1) -y, lint update/import from NetBSD.

22 years agoSwitch to OpenPAM. Bump library version. Modules are now versioned, so
Dag-Erling Smørgrav [Tue, 5 Mar 2002 21:56:25 +0000 (21:56 +0000)]
Switch to OpenPAM.  Bump library version.  Modules are now versioned, so
applications linked with Linux-PAM will still work.
Remove pam_get_pass(); OpenPAM has pam_get_authtok().
Remove pam_prompt(); OpenPAM has pam_{,v}{error,info,prompt}().
Remove pam_set_item(3) man page as OpenPAM has its own.

Sponsored by: DARPA, NAI Labs

22 years ago- use des_ecb3_encrypt().
Hajimu UMEMOTO [Tue, 5 Mar 2002 21:12:41 +0000 (21:12 +0000)]
- use des_ecb3_encrypt().
- style: added spaces after /* and before */

Obtained from: KAME
MFC after: 2 weeks

22 years agoOops, now, encription and decription are separate function.
Hajimu UMEMOTO [Tue, 5 Mar 2002 20:32:49 +0000 (20:32 +0000)]
Oops, now, encription and decription are separate function.

MFC after: 2 weeks

22 years agoOfficial timeout routines for netgraph nodes that know how to
Julian Elischer [Tue, 5 Mar 2002 20:26:20 +0000 (20:26 +0000)]
Official timeout routines for netgraph nodes that know how to
use (and abuse) the node locking system.

MFC after: 1 week

22 years agoThree p_ucred -> td_ucred's missed in jhb's earlier pass; all appear to
Robert Watson [Tue, 5 Mar 2002 19:45:45 +0000 (19:45 +0000)]
Three p_ucred -> td_ucred's missed in jhb's earlier pass; all appear to
be safe.

22 years agoOn the sparc64 platform we install this as "gawk" as Gawk is currently
David E. O'Brien [Tue, 5 Mar 2002 19:40:25 +0000 (19:40 +0000)]
On the sparc64 platform we install this as "gawk" as Gawk is currently
borked on the sparc64.  We are using BWK's AWK at the moment.

22 years agoOn the sparc64 platform we install the `1 true AWK' as "awk". On all other
David E. O'Brien [Tue, 5 Mar 2002 19:37:12 +0000 (19:37 +0000)]
On the sparc64 platform we install the `1 true AWK' as "awk".  On all other
platforms we install as "nawk".

At the moment Gawk is borked on sparc64, however BWK AWK works fine.

22 years agoAdd the `1 true AWK' back to the mix.
David E. O'Brien [Tue, 5 Mar 2002 19:36:08 +0000 (19:36 +0000)]
Add the `1 true AWK' back to the mix.

22 years agoThe change from td->td_proc->p_ucred to td->td_ucred has shortened some
Robert Watson [Tue, 5 Mar 2002 19:31:25 +0000 (19:31 +0000)]
The change from td->td_proc->p_ucred to td->td_ucred has shortened some
lines: more agressively line wrap under those circumstances.

22 years agoEnvironmental variable was not being passed to a subshell as intended.
Crist J. Clark [Tue, 5 Mar 2002 19:13:05 +0000 (19:13 +0000)]
Environmental variable was not being passed to a subshell as intended.

PR: bin/35558
Submitted by: Nicolas Rachinsky <list@rachinsky.de>

22 years ago- Use td_ucred for jail checks.
John Baldwin [Tue, 5 Mar 2002 18:57:36 +0000 (18:57 +0000)]
- Use td_ucred for jail checks.
- Move jail checks and some other checks involving constants and stack
  variables out from under Giant.  This isn't perfectly safe atm because
  jail_sysvipc_allowed is read w/o a lock meaning that its value could be
  stale.  This global variable will soon become a per-jail flag, however,
  at which time it will either not need a lock or will use the prison lock.

22 years agoReserve a mount flag, MNT_MULTILABEL, used by the MAC subsystem and
Robert Watson [Tue, 5 Mar 2002 18:48:15 +0000 (18:48 +0000)]
Reserve a mount flag, MNT_MULTILABEL, used by the MAC subsystem and
individual filesystems to determine whether they should operate in
"file system as a single object" mode, or "file system as a set of objects
with individual labels" mode.  Note: in the trustedbsd_mac branch,
this is refered to as "MNT_MULTILEVEL", but the two mean the same thing.
MNT_MULTILABEL is more suggestive of a flexible policy system than one
providing purely hierarchal policies.  The need for a reserved flag will
go away once nmount() is done.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoMerge reservation of two SI_SUB constants for the MAC policy framework
Robert Watson [Tue, 5 Mar 2002 18:44:11 +0000 (18:44 +0000)]
Merge reservation of two SI_SUB constants for the MAC policy framework
and for individual MAC policies.  The framework event initializes the
access control subsystem; the policy event allows policies to register
themselves.  The gap in between is for all the things we'll think of
later.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years ago* Move bswlist declaration and initialization from kern/vfs_bio.c to
Eivind Eklund [Tue, 5 Mar 2002 18:20:58 +0000 (18:20 +0000)]
* Move bswlist declaration and initialization from kern/vfs_bio.c to
  vm/vm_pager.c, which is the only place it is used.
* Make the QUEUE_* definitions and bufqueues local to vfs_bio.c.
* constify buf_wmesg.

22 years agoUse revoke_and_destroy_dev() instead of destroy_dev() when removing /dev/net
Brian Feldman [Tue, 5 Mar 2002 17:50:35 +0000 (17:50 +0000)]
Use revoke_and_destroy_dev() instead of destroy_dev() when removing /dev/net
pseudo-devices when an interface goes away.  Otherwise, an open /dev/net/foo0
when the interface is removed can cause a crash.

Not objected to by: jlemon

22 years agoChange wmesg to const char * instead of char *
Eivind Eklund [Tue, 5 Mar 2002 17:45:12 +0000 (17:45 +0000)]
Change wmesg to const char * instead of char *

22 years agoWhen multiple threads call atexit at the same time, some operations must
Tor Egge [Tue, 5 Mar 2002 17:34:37 +0000 (17:34 +0000)]
When multiple threads call atexit at the same time, some operations must
be serialized.  A mutex is used to protect the critical regions.

sbrk() and brk() are not thread safe.  Replace use of sbrk() with
a call to malloc to avoid race when one thread calls atexit
while another thread calls malloc.

Reviewed by: deischen

22 years agoChange every occurrence of "bootblock" into "boot1" after
Robert Drehmel [Tue, 5 Mar 2002 16:39:36 +0000 (16:39 +0000)]
Change every occurrence of "bootblock" into "boot1" after
the repo-move from "src/sys/boot/sparc64/bootblock".

22 years agoAdd roaming and microwave oven support. Some minor formatting nits.
Warner Losh [Tue, 5 Mar 2002 16:18:14 +0000 (16:18 +0000)]
Add roaming and microwave oven support.  Some minor formatting nits.
Some buggy firmware workarounds.  Fix some endian bugs.

These reduce the diffs from NetBSD, but NetBSD does have more changes since
my last manual merge.

22 years agoPart II: update various mechanically generated files to allow for new
Robert Watson [Tue, 5 Mar 2002 16:13:01 +0000 (16:13 +0000)]
Part II: update various mechanically generated files to allow for new
system call number allocations.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoRemove __P to reduce diffs with NetBSD and my private tree.
Warner Losh [Tue, 5 Mar 2002 16:12:41 +0000 (16:12 +0000)]
Remove __P to reduce diffs with NetBSD and my private tree.

22 years agoReserve system call numbers for the MAC framework. This will prevent
Robert Watson [Tue, 5 Mar 2002 16:11:11 +0000 (16:11 +0000)]
Reserve system call numbers for the MAC framework.  This will prevent
people working on the MAC tree from getting toasted whenever system call
numbers are allocated in the main tree (for example, for KSE :-).
Calls allocated: __mac_{get,set}_proc, __mac_{get,set}_{fd,file}().

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoMerge a little from NetBSD.
Warner Losh [Tue, 5 Mar 2002 16:06:54 +0000 (16:06 +0000)]
Merge a little from NetBSD.

22 years agoDocument all functions, global and static variables, and sysctls.
Eivind Eklund [Tue, 5 Mar 2002 15:38:49 +0000 (15:38 +0000)]
Document all functions, global and static variables, and sysctls.
Includes some minor whitespace changes, and re-ordering to be able to document
properly (e.g, grouping of variables and the SYSCTL macro calls for them, where
the documentation has been added.)

Reviewed by: phk (but all errors are mine)

22 years agoFix a warning.
Robert Drehmel [Tue, 5 Mar 2002 15:19:33 +0000 (15:19 +0000)]
Fix a warning.

22 years agoFix off-by-one error.
Jacques Vidrine [Tue, 5 Mar 2002 14:27:19 +0000 (14:27 +0000)]
Fix off-by-one error.

Obtained from: OpenBSD

22 years agoMove sysctl debug.sizeof into sys/sysctl.h where it belongs.
Poul-Henning Kamp [Tue, 5 Mar 2002 13:56:58 +0000 (13:56 +0000)]
Move sysctl debug.sizeof into sys/sysctl.h where it belongs.

22 years agoVendor import of OpenPAM Celandine.
Dag-Erling Smørgrav [Tue, 5 Mar 2002 13:49:56 +0000 (13:49 +0000)]
Vendor import of OpenPAM Celandine.