]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoWhen a GEOM (/dev-)device is closed and we find that I/O requests are
Poul-Henning Kamp [Fri, 9 May 2003 21:25:28 +0000 (21:25 +0000)]
When a GEOM (/dev-)device is closed and we find that I/O requests are
still outstanding, give them a chance to complete.

If after 10 seconds we still find outstanding I/O requests, complete
the close with a console warning that the system is likely to panic
later on.

This is a workaround for umount -f not quite doing the right thing.

Approved by:    re/scottl

21 years agoTurn off ppp (ICE) and pppd (cc objects to a varargs decl) on amd64.
Peter Wemm [Fri, 9 May 2003 21:06:41 +0000 (21:06 +0000)]
Turn off ppp (ICE) and pppd (cc objects to a varargs decl) on amd64.

21 years agoBefore reading an ascii label, initialize with defaults so that
Poul-Henning Kamp [Fri, 9 May 2003 20:26:17 +0000 (20:26 +0000)]
Before reading an ascii label, initialize with defaults so that
getasciipartspec() has a sectorisize in case it needs one.

Approved by: re/jhb

21 years agoRemove Giant from kern_sigsuspend() and osigsuspend() as these should now
John Baldwin [Fri, 9 May 2003 19:11:32 +0000 (19:11 +0000)]
Remove Giant from kern_sigsuspend() and osigsuspend() as these should now
be MP safe.

Approved by: re (scottl)

21 years agoSigh, this shows just how much one can be conditioned my the environment:
Poul-Henning Kamp [Fri, 9 May 2003 19:07:59 +0000 (19:07 +0000)]
Sigh, this shows just how much one can be conditioned my the environment:

Just because we for the last ten years have fought for every byte
in the boot code on i386, doesn't mean that other architectures could
not actually have space to spare there.

Remore debugging message.

21 years agoInclude the MXCSR initial values, based on the AMD docs. This file
Peter Wemm [Fri, 9 May 2003 18:28:05 +0000 (18:28 +0000)]
Include the MXCSR initial values, based on the AMD docs.  This file
should really be renamed to fpu.h and npx.c to fpu.c since its part of
the core architecture on amd64 systems, not an isa 'numeric processor
extension'.

21 years agoTurn syscons on now that it works, so that anybody trying to run this
Peter Wemm [Fri, 9 May 2003 18:26:06 +0000 (18:26 +0000)]
Turn syscons on now that it works, so that anybody trying to run this
can see something.  Probing for keyboard still works for auto serial
console mode.

21 years agoTrivial addition of __amd64__ to the ifdefs for platforms that use
Peter Wemm [Fri, 9 May 2003 18:24:40 +0000 (18:24 +0000)]
Trivial addition of __amd64__ to the ifdefs for platforms that use
i386-style vga console support.

Approved by:  re

21 years agoThe disc1 package set has overfilled the disc. Axe linux-netscape-navigator-4.8
Scott Long [Fri, 9 May 2003 09:41:18 +0000 (09:41 +0000)]
The disc1 package set has overfilled the disc.  Axe linux-netscape-navigator-4.8
to free up space.  linux-netscape-communicator-4.8 remains and is a superset of
the former for those who are stuck are the 1990's =-)

21 years agoDocument the last field of the procfs 'status' node, which either
Hiten Pandya [Fri, 9 May 2003 09:20:26 +0000 (09:20 +0000)]
Document the last field of the procfs 'status' node, which either
displays the 'hostname' of the jail, or a hyphen '-' to indicate
that the process is not jailed.

PR: docs/37470
Submitted by: Adrian Filipi-Martin <adrian@ubergeeks.com>
Approved and Reviewed by: des (mentor), re (bmah)

21 years agoAdd missing MLINK for arc4rand(9). It is documented in
Hiten Pandya [Fri, 9 May 2003 09:12:10 +0000 (09:12 +0000)]
Add missing MLINK for arc4rand(9).  It is documented in
random(9) and referenced by gbde(4).

PR: docs/51254
Submitted by: Simon L. Nielsen <simon@nitro.dk>
Approved and Reviewed by: des (mentor), re (bmah)

21 years agoTo reserve space for 65536 bits, allocate
Robert Drehmel [Fri, 9 May 2003 09:11:27 +0000 (09:11 +0000)]
To reserve space for 65536 bits, allocate
65536 / (sizeof(int) * CHAR_BITS) `int's instead of
65536 / (sizeof(int) * CHAR_BITS) bytes to avoid a possible
segmentation fault if ports above 16383 are specified via the
-p option on a platform with 4 byte wide ints.

Approved by: re (bmah)
Reported by: Marco Wertejuk <wertejuk@mwcis.com>

21 years agoComment out the entire nomatch entry. The partially commented out
Warner Losh [Fri, 9 May 2003 05:58:22 +0000 (05:58 +0000)]
Comment out the entire nomatch entry.  The partially commented out
entry causes problems :-(

This effectively backs out the commit the re@ allowed me to commit.

21 years agoRedefine M_FREELIST to be 0x8000; 0x4000 conflicted with two other
Mike Silbersack [Fri, 9 May 2003 02:15:52 +0000 (02:15 +0000)]
Redefine M_FREELIST to be 0x8000; 0x4000 conflicted with two other
uses of m_flags in the kernel.  (A future commit will move all
private m_flags users here so they're obvious without a great
deal of searching.)

This should fix the mbuf double-free panics those using ppp or
ipfw reset rules have been seeing since the double-free detection
code went in.

21 years agoGive the kmem object's mutex a unique name, instead of "vm object",
Alan Cox [Fri, 9 May 2003 02:13:23 +0000 (02:13 +0000)]
Give the kmem object's mutex a unique name, instead of "vm object",
to avoid false reports of lock-order reversal with a system map mutex.

Approved by: re (jhb)

21 years agoWhitespace cleanup of previous commit. Translators can ignore this.
Bruce A. Mah [Thu, 8 May 2003 22:31:00 +0000 (22:31 +0000)]
Whitespace cleanup of previous commit.  Translators can ignore this.

Approved by: re (implicitly)

21 years agoNew release notes: New newsyslog(8) features.
Bruce A. Mah [Thu, 8 May 2003 22:29:28 +0000 (22:29 +0000)]
New release notes:  New newsyslog(8) features.

Submitted by: gad
Approved by: re (implicitly)

21 years agoTry to reflect the current state of disklabel (removed) and the
Bruce A. Mah [Thu, 8 May 2003 21:09:39 +0000 (21:09 +0000)]
Try to reflect the current state of disklabel (removed) and the
recently-activated bsdlabel(8).

21 years agoRename MAC_MAX_POLICIES to MAC_MAX_SLOTS, since the variables and
Robert Watson [Thu, 8 May 2003 19:49:42 +0000 (19:49 +0000)]
Rename MAC_MAX_POLICIES to MAC_MAX_SLOTS, since the variables and
constants in question refer to the number of label slots, not the
maximum number of policies that may be loaded.  This should reduce
confusion regarding an element in the MAC sysctl MIB, as well as
make it more clear what the affect of changing the compile-time
constants is.

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

21 years agoNote that -n and \c are implementation-defined as per TC1 to POSIX 2001.
Jens Schweikhardt [Thu, 8 May 2003 17:47:28 +0000 (17:47 +0000)]
Note that -n and \c are implementation-defined as per TC1 to POSIX 2001.

Approved by: re@ (bmah)
MFC after: 3 days

21 years agoDescribe the pattern filenames must be in before they are added to the hints
Tom Rhodes [Thu, 8 May 2003 17:25:41 +0000 (17:25 +0000)]
Describe the pattern filenames must be in before they are added to the hints
file.

PR: 46271
Approved by: re (blanket)

21 years agoAcquire Giant at the start of the raid rebuild kthreads.
John Baldwin [Thu, 8 May 2003 16:38:14 +0000 (16:38 +0000)]
Acquire Giant at the start of the raid rebuild kthreads.

Reported by: Masachika ISHIZUKA <ishizuka@ish.org>
Reviewed by: sos
Approved by: re (bmah)

21 years agoI goofed in rev 1.59 (a.out support axing) and missed getting FUNCTION_PROFILER
David E. O'Brien [Thu, 8 May 2003 16:02:22 +0000 (16:02 +0000)]
I goofed in rev 1.59 (a.out support axing) and missed getting FUNCTION_PROFILER
fully clean.  This resulted in inserted garbage into the produced assembly code
when the gcc -pg and -fPIC options are used together.

PR: i386/50598
Submitted by: Don Lewis <truckman@freebsd.org>

21 years agoAdd a comment describing why it's important for the values in this
David Schultz [Thu, 8 May 2003 13:50:44 +0000 (13:50 +0000)]
Add a comment describing why it's important for the values in this
file to be correct, and how to generate them automatically.

Caused much pain and suffering for: peter

21 years agoOops. Turn T_PAGEFLT back into an interrupt gate. It is *critical*
Peter Wemm [Thu, 8 May 2003 08:25:51 +0000 (08:25 +0000)]
Oops.  Turn T_PAGEFLT back into an interrupt gate.  It is *critical*
that interrupts be disabled and remain disabled until %cr2 is read.
Otherwise we can preempt and another process can fault, and by the
time we read %cr2, we see a different processes fault address.  This
Greatly Confuses vm_fault() (to say the least).  The i386 port has
got this marked as a bug workaround for a Cyrix CPU, which is what
lead me astray.  Its actually necessary for preemption, regardless
of whether Cyrix cpus had a bug or not.

21 years agoSIG_SETMASK is 3, not 1. Sigh.
Peter Wemm [Thu, 8 May 2003 07:41:24 +0000 (07:41 +0000)]
SIG_SETMASK is 3, not 1.  Sigh.

21 years agoSet the binutils arch for as/nm correctly on the amd64 platform.
Peter Wemm [Thu, 8 May 2003 06:39:31 +0000 (06:39 +0000)]
Set the binutils arch for as/nm correctly on the amd64 platform.
As far as binutils is concerned, the amd64 platform is still called
"x86-64"/"x86_64".  Setting things from ${MACHINE_ARCH} breaks that.

Approved by:  re (scottl)

21 years agoExclude rtld-elf for amd64. More porting is still needed.
Peter Wemm [Thu, 8 May 2003 06:37:12 +0000 (06:37 +0000)]
Exclude rtld-elf for amd64.  More porting is still needed.

Approved by:  re (scottl)

21 years agoExclude sys/boot for amd64. There are still toolchain issues to deal
Peter Wemm [Thu, 8 May 2003 06:35:39 +0000 (06:35 +0000)]
Exclude sys/boot for amd64.  There are still toolchain issues to deal
with.  In theory, gcc -m32 should work, but for now, do not tempt fate.

Approved by: re (scottl)

21 years agoAdd amd64 settings.
Peter Wemm [Thu, 8 May 2003 06:33:49 +0000 (06:33 +0000)]
Add amd64 settings.

Approved by:  re (scottl)

21 years agoExclude gprof and truss for amd64 too.
Peter Wemm [Thu, 8 May 2003 06:33:07 +0000 (06:33 +0000)]
Exclude gprof and truss for amd64 too.

Approved by:  re (scottl)

21 years agoFix an embarresing transcription error from i386 to amd64. Put the arguments
Peter Wemm [Thu, 8 May 2003 06:25:03 +0000 (06:25 +0000)]
Fix an embarresing transcription error from i386 to amd64.  Put the arguments
to sigprocmask(2) int the correct order. *blush*.
For sigsetjmp(), match up the pushq/popq in the non-savemask case.

21 years agorelease.9 has become the biggest festering pile of spaghetti s(*^% code.
David E. O'Brien [Thu, 8 May 2003 03:25:17 +0000 (03:25 +0000)]
release.9 has become the biggest festering pile of spaghetti s(*^% code.
Duplicate the target for Alpha so I can start hacking and slashing bits out.

Approved by: re(blanket to make Alpha build)

21 years agoTidy up modf.S and make it actually work. It wasn't extracting
Peter Wemm [Thu, 8 May 2003 03:19:37 +0000 (03:19 +0000)]
Tidy up modf.S and make it actually work.  It wasn't extracting
the value out of ST(0) before copying it to %xmm0.  Also remove bogus stack
frame and work in the red zone.

21 years agoRethink the way we count module references. Simply following
Alexander Kabaev [Thu, 8 May 2003 01:31:36 +0000 (01:31 +0000)]
Rethink the way we count module references. Simply following
DT_NEEDED links is not flexible enough for cases where dynamically
loaded modules form a dependency cycle.

This should fix an infinite recursion problem encountered by Yahoo.

Approved by: re (jhb)

21 years agoioctl VINUM_READCONFIG: Don't lock configuration here. vinum_scandisk
Greg Lehey [Thu, 8 May 2003 00:36:20 +0000 (00:36 +0000)]
ioctl VINUM_READCONFIG: Don't lock configuration here.  vinum_scandisk
            needs to do it anyway to handle the startup case.  This is
            part of a fix for the recently reported hangs.

Approved by:  re (scottl)

21 years agovinum_read: Don't lock configuration from userland. If vinum(8)
Greg Lehey [Thu, 8 May 2003 00:33:57 +0000 (00:33 +0000)]
vinum_read: Don't lock configuration from userland.  If vinum(8)
            crashes, the config remains locked and causes all
            subsequent start or read attempts to fail.  This is part
            of a fix for the recently reported hangs.

Approved by:  re (scottl)

21 years agoAdd information about the log file.
Tom Rhodes [Thu, 8 May 2003 00:28:36 +0000 (00:28 +0000)]
Add information about the log file.

PR: 35575
Submitted by: "Simon L. Nielsen" <simon@nitro.dk> (original version)

21 years agoFix some basic errors:
Tom Rhodes [Thu, 8 May 2003 00:14:48 +0000 (00:14 +0000)]
Fix some basic errors:

- Add a description of b0 / b1 fields.
- Do not use 'entry' to refer to both 'entry' and 'field'.
- Do not confuse people with heading 'Name' and entry 'Name'.

PR: 48104
Submitted by: Gary W. Swearingen <swear@attbi.com> (original version)
Approved by: re (blanket)

21 years agoLeave space for the 128 byte red-zone on the stack.
Peter Wemm [Thu, 8 May 2003 00:13:24 +0000 (00:13 +0000)]
Leave space for the 128 byte red-zone on the stack.

21 years ago#include <machine/metadata.h> was missing; add it
Peter Wemm [Thu, 8 May 2003 00:12:37 +0000 (00:12 +0000)]
#include <machine/metadata.h> was missing; add it

21 years agoFix a preemption race. I was reenabling interrupts in the fast system
Peter Wemm [Thu, 8 May 2003 00:05:00 +0000 (00:05 +0000)]
Fix a preemption race.  I was reenabling interrupts in the fast system
call handler before it was safe.  It was possible for to lose context
and for something else to clobber the PCPU scratch variable.  This
moves the interrupt enable *way* too late, but its better safe than
sorry for the moment.

21 years agoFix typo, even though this is unused.
Peter Wemm [Thu, 8 May 2003 00:02:47 +0000 (00:02 +0000)]
Fix typo, even though this is unused.

21 years agoLike ia64, amd64 has got a 16 byte sized and aligned 'long double'.
Peter Wemm [Thu, 8 May 2003 00:02:03 +0000 (00:02 +0000)]
Like ia64, amd64 has got a 16 byte sized and aligned 'long double'.

Obtained from: re (blanket amd64)

21 years agoTurn off alloca.S - it cannot possibly work like this since on AMD64, gcc
Peter Wemm [Wed, 7 May 2003 23:49:24 +0000 (23:49 +0000)]
Turn off alloca.S - it cannot possibly work like this since on AMD64, gcc
doesn't use stack frames.  It uses offsets relative to %rsp, not %rbp.  So
we cannot just change %rsp like this.

Approved by: re (blanket amd64)

21 years agoActually use the correct values for AMD64. It is a 64 bit platform,
Peter Wemm [Wed, 7 May 2003 23:48:05 +0000 (23:48 +0000)]
Actually use the correct values for AMD64.  It is a 64 bit platform,
configure gdtoa as such.

21 years ago- Change the full Asic revision defines to CHIPID to better since the
Paul Saab [Wed, 7 May 2003 21:51:13 +0000 (21:51 +0000)]
- Change the full Asic revision defines to CHIPID to better since the
  ASIC revision is really the major number of the CHIPID.  Also store
  the chipid, asic rev and chip revision in the softc for later use.

- The write twice to send producer index workaround only applies to
  the 5700_BX chips, so only do it there.
  Requested by: jdp

- Do not initalize the LED's to 0x00.  The default configuration
  the chip comes up in should yeild proper operation of the LED's.
  Confirmed by: John Cagle <john.cagle@hp.com>

Approved by: re (blanket)

21 years agoBring the dumpon(8) man page closer to reality:
Hiten Pandya [Wed, 7 May 2003 19:37:51 +0000 (19:37 +0000)]
Bring the dumpon(8) man page closer to reality:

    - dumpon utility has not used kern.dumpdev sysctl
      since rev. 1.14 (sbin/dumpon/dumpon.c) when phk@
      updated it to use the DIOCSKERNELDUMP ioctl [1]
    - remove obsolete reference to sysctl(3)

While I am there, fix two style nits:

    - use .Nm instead of `dumpon'
    - change NOTES to IMPLEMENTATION NOTES, to bring
      it in line with recommended section headings in
      mdoc(7)

Original patch by: Martin Faxer <gmh003532brfmasthugget.se> [1]

PR: docs/39293
Approved and Reviewed by: des (mentor), re (scottl, bmah)

21 years agoDump is hard-wired to believe that it can read disks on
Kirk McKusick [Wed, 7 May 2003 18:27:09 +0000 (18:27 +0000)]
Dump is hard-wired to believe that it can read disks on
1024-byte boundaries. For many years this was a reasonable
assumption. However, in recent years we have begun seeing
devices with 2048-byte sectors. These devices return errors
when dump tries to read starting in the middle of a sector
or when it tries to read only the first half of a sector.
Rather than change the native block size used by dump (and
thus create an incompatible dump format), this fix checks
for transfer requests that start and/or end on a non-sector
boundary. When such a read is detected, the new code reads
the entire sector and copies out just the part that dump
needs.

Reviewed by: Poul-Henning Kamp <phk@critter.freebsd.dk>
Approved by: re (John Baldwin <jhb@FreeBSD.org>)
Sponsored by:   DARPA & NAI Labs.

21 years agoRev 1.67 accidently added "axe".
David E. O'Brien [Wed, 7 May 2003 18:14:00 +0000 (18:14 +0000)]
Rev 1.67 accidently added "axe".
Seeing as how it was just added to GENERIC and may be the thing to get
the release builds working I'm leaving it in.

21 years agoBack out rev 1.63. Things in this file should be in the same order as in
David E. O'Brien [Wed, 7 May 2003 18:12:38 +0000 (18:12 +0000)]
Back out rev 1.63.  Things in this file should be in the same order as in
GENERIC.  This greatly simplifies the hunting for the next thing to kill.

Approved by: re(blanket as alpha re-builder)

21 years agoClean up locking for the MAC Framework:
Robert Watson [Wed, 7 May 2003 17:49:24 +0000 (17:49 +0000)]
Clean up locking for the MAC Framework:

(1) Accept that we're now going to use mutexes, so don't attempt
    to avoid treating them as mutexes.  This cleans up locking
    accessor function names some.

(2) Rename variables to _mtx, _cv, _count, simplifying the naming.

(3) Add a new form of the _busy() primitive that conditionally
    makes the list busy: if there are entries on the list, bump
    the busy count.  If there are no entries, don't bump the busy
    count.  Return a boolean indicating whether or not the busy
    count was bumped.

(4) Break mac_policy_list into two lists: one with the same name
    holding dynamic policies, and a new list, mac_static_policy_list,
    which holds policies loaded before mac_late and without the
    unload flag set.  The static list may be accessed without
    holding the busy count, since it can't change at run-time.

(5) In general, prefer making the list busy conditionally, meaning
    we pay only one mutex lock per entry point if all modules are
    on the static list, rather than two (since we don't have to
    lower the busy count when we're done with the framework).  For
    systems running just Biba or MLS, this will halve the mutex
    accesses in the network stack, and may offer a substantial
    performance benefits.

(6) Lay the groundwork for a dynamic-free kernel option which
    eliminates all locking associated with dynamically loaded or
    unloaded policies, for pre-configured systems requiring
    maximum performance but less run-time flexibility.

These changes have been running for a few weeks on MAC development
branch systems.

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

21 years agoProperly cleanup the stack before jumping to cerror() if rfork(2) fails.
John Baldwin [Wed, 7 May 2003 17:23:25 +0000 (17:23 +0000)]
Properly cleanup the stack before jumping to cerror() if rfork(2) fails.

Submitted by: Igor Sysoev <is@rambler-co.ru>
Approved by: re (scottl)

21 years agoStyle nits.
John Baldwin [Wed, 7 May 2003 17:21:38 +0000 (17:21 +0000)]
Style nits.

Approved by: re (bmah)

21 years agoDon't use logger by default just yet for unmatched devices. It is a
Warner Losh [Wed, 7 May 2003 15:48:20 +0000 (15:48 +0000)]
Don't use logger by default just yet for unmatched devices.  It is a
little too chatty on boot and doesn't give enough information.

Approved by: re@ (bmah)

21 years ago#include <sys/resource.h> to limit ports damage.
Poul-Henning Kamp [Wed, 7 May 2003 15:26:43 +0000 (15:26 +0000)]
#include <sys/resource.h> to limit ports damage.

Approved by: re/rwatson

21 years agoAdd my birthday.
Michael Landin [Wed, 7 May 2003 13:28:16 +0000 (13:28 +0000)]
Add my birthday.

Approved by: roberto (mentor), re(bmah)

21 years agoHide the "ENOMEM" notice messages behind bootverbose. They are still
Poul-Henning Kamp [Wed, 7 May 2003 05:37:31 +0000 (05:37 +0000)]
Hide the "ENOMEM" notice messages behind bootverbose.  They are still
a valuable debugging tool for certain kinds of problems.

Approved by: re/scottl

21 years agoCorrect a bug introduced with reduced TCP state handling; make
Robert Watson [Wed, 7 May 2003 05:26:27 +0000 (05:26 +0000)]
Correct a bug introduced with reduced TCP state handling; make
sure that the MAC label on TCP responses during TIMEWAIT is
properly set from either the socket (if available), or the mbuf
that it's responding to.

Unfortunately, this is made somewhat difficult by the TCP code,
as tcp_twstart() calls tcp_twrespond() after discarding the socket
but without a reference to the mbuf that causes the "response".
Passing both the socket and the mbuf works arounds this--eventually
it might be good to make sure the mbuf always gets passed in in
"response" scenarios but working through this provided to
complicate things too much.

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

21 years agoNew release notes: Kerberos 5 built by default, krb5 distributed
Bruce A. Mah [Wed, 7 May 2003 04:15:57 +0000 (04:15 +0000)]
New release notes:  Kerberos 5 built by default, krb5 distributed
folded into crypto.

Submitted by: markm
Approved by: re (implicitly)

21 years agodf(1) and ls(1) print units in 'four or fewer' not 'three or less'.
Tom Rhodes [Tue, 6 May 2003 21:54:46 +0000 (21:54 +0000)]
df(1) and ls(1) print units in 'four or fewer' not 'three or less'.

PR: 35523
Submitted by: Tomas Svensson <tsn@gbdev.net>

21 years agoVersion number bump: We're at 5.1-BETA now.
Bruce A. Mah [Tue, 6 May 2003 21:30:16 +0000 (21:30 +0000)]
Version number bump:  We're at 5.1-BETA now.

Note that the previous commit message for this file should have read:
"4.8-RELEASE is the latest release from the 4-STABLE branch." [1]

Pointed out by: Stijn Hoop <stijn@win.tue.nl> [1]
Approved by re (implicitly)

21 years agoTrim a call to mac_create_mbuf_from_mbuf() since m_tag meta-data
Robert Watson [Tue, 6 May 2003 20:34:04 +0000 (20:34 +0000)]
Trim a call to mac_create_mbuf_from_mbuf() since m_tag meta-data
copying for mbuf headers now works properly in m_dup_pkthdr(), so
we don't need to do an explicit copy.

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

21 years agoUpdate supported architectures, use &unix; where appropriate, more
Bruce A. Mah [Tue, 6 May 2003 20:01:07 +0000 (20:01 +0000)]
Update supported architectures, use &unix; where appropriate, more
majordomo->mailman fixes, other misc. updates.

Approved by: re (implicitly)

21 years ago4.8-RELEASE was the last release from the 4-STABLE branch.
Bruce A. Mah [Tue, 6 May 2003 19:58:10 +0000 (19:58 +0000)]
4.8-RELEASE was the last release from the 4-STABLE branch.

21 years agoFix the WARNING for wrong rawoffset, I tested incompatible units.
Poul-Henning Kamp [Tue, 6 May 2003 19:36:13 +0000 (19:36 +0000)]
Fix the WARNING for wrong rawoffset, I tested incompatible units.

Approved by: re/jhb

21 years agoAdd PCI ID's for the Intel ICH5 (82801EB) chipset.
John Baldwin [Tue, 6 May 2003 19:31:56 +0000 (19:31 +0000)]
Add PCI ID's for the Intel ICH5 (82801EB) chipset.

Approved by: re (murray)
Sponsored by: The Weather Channel

21 years agoAdd PCI ID's for the 4 USB hubs on the ICH5 controller.
John Baldwin [Tue, 6 May 2003 19:30:41 +0000 (19:30 +0000)]
Add PCI ID's for the 4 USB hubs on the ICH5 controller.

Approved by: re (murray)

21 years agoConsistentify our NO_KERBEROS documentation.
David E. O'Brien [Tue, 6 May 2003 19:26:55 +0000 (19:26 +0000)]
Consistentify our NO_KERBEROS documentation.

Approved by: re(bmah)

21 years agomalloc(9) types can be used to identify multiple allocations.
Hiten Pandya [Tue, 6 May 2003 19:10:17 +0000 (19:10 +0000)]
malloc(9) types can be used to identify multiple allocations.

PR: docs/38618
Approved by: des (mentor)

21 years agoHook devfs.conf up to the build.
Doug Barton [Tue, 6 May 2003 07:25:17 +0000 (07:25 +0000)]
Hook devfs.conf up to the build.

Approved by: re (scottl)

21 years agoStart mailing -CURRENT tinderbox failures to the lists.
Dag-Erling Smørgrav [Tue, 6 May 2003 04:04:02 +0000 (04:04 +0000)]
Start mailing -CURRENT tinderbox failures to the lists.

21 years agoWe are now in 5.1-BETA
Scott Long [Tue, 6 May 2003 03:55:24 +0000 (03:55 +0000)]
We are now in 5.1-BETA

21 years agoFix a printf() format error which broke the ia64 GENERIC build.
Dag-Erling Smørgrav [Tue, 6 May 2003 03:55:12 +0000 (03:55 +0000)]
Fix a printf() format error which broke the ia64 GENERIC build.

21 years agoPrevent a 'make SHARED=symlinks install' followed by 'make install' from
Gregory Neil Shapiro [Tue, 6 May 2003 03:38:24 +0000 (03:38 +0000)]
Prevent a 'make SHARED=symlinks install' followed by 'make install' from
failing by removing the symlink before trying to install the actual files.

Noticed by: bde
MFC after: 5 days

21 years agoLock the vm_object when performing vm_pager_deallocate().
Alan Cox [Tue, 6 May 2003 02:45:28 +0000 (02:45 +0000)]
Lock the vm_object when performing vm_pager_deallocate().

21 years agoFix a null dereference leading to a core dump when
Mike Makonnen [Tue, 6 May 2003 02:33:49 +0000 (02:33 +0000)]
Fix a null dereference leading to a core dump when
the number of threads exceeds the number of open slots
in ldt_entries[].

Approved by: markm (mentor)(implicit)
Reviewed by: jeff

21 years agoo Correct a debug message that refered to the wrong function
Mike Makonnen [Tue, 6 May 2003 02:30:52 +0000 (02:30 +0000)]
o Correct a debug message that refered to the wrong function
o Remove an unncecesary if clause

Approved by: markm (mentor)(implicit)
Reviewd by: jeff

21 years agos/kadmind5/kadmin/
Mike Makonnen [Tue, 6 May 2003 02:28:03 +0000 (02:28 +0000)]
s/kadmind5/kadmin/

Approved by: markm (mentor)

21 years agoDon't call timeout() in sis_tick(), this is done earlier by mii_tick(), and it
Olivier Houchard [Tue, 6 May 2003 02:00:01 +0000 (02:00 +0000)]
Don't call timeout() in sis_tick(), this is done earlier by mii_tick(), and it
leads to a panic at unload time, as we own 2 instances of callout and
untimeout() only one.
Will I'm there, remove a call to callout_handler_init(), one is enough.

Reviewed by: wpaul

21 years agoAllow users to configure stuff in /dev once again by re-doing the devfs
Doug Barton [Tue, 6 May 2003 01:10:33 +0000 (01:10 +0000)]
Allow users to configure stuff in /dev once again by re-doing the devfs
script to read /etc/devfs.conf, and take appropriate actions based on
what it finds there.

The (commented out) examples in the new devfs.conf file are taken in
part from the old rc.devfs script.

21 years agoBack out the amd64 additions. Hammer doesn't do APM, and we certainly aren't
David E. O'Brien [Tue, 6 May 2003 00:09:51 +0000 (00:09 +0000)]
Back out the amd64 additions.  Hammer doesn't do APM, and we certainly aren't
ready for any ibcs2_compat bits yet.

21 years agoMake pthread_join() async-cancel-safe. David was going to commit
Daniel Eischen [Tue, 6 May 2003 00:02:54 +0000 (00:02 +0000)]
Make pthread_join() async-cancel-safe.  David was going to commit
this, but I think he's asleep and want to be sure it gets in before
the freeze.

Submitted by: davidxu

21 years agoWe won't do APM on Hammer -- ACPI is the game there.
David E. O'Brien [Mon, 5 May 2003 23:55:43 +0000 (23:55 +0000)]
We won't do APM on Hammer -- ACPI is the game there.

21 years agoAdd an example of how to create an fstab entry.
Doug Barton [Mon, 5 May 2003 23:41:17 +0000 (23:41 +0000)]
Add an example of how to create an fstab entry.

21 years agoFix age_old_log checking so that it will notice log files which were
Garance A Drosehn [Mon, 5 May 2003 23:38:03 +0000 (23:38 +0000)]
Fix age_old_log checking so that it will notice log files which were
rotated and then compressed with bzip2 instead of gzip.  Otherwise,
any file which had a time-interval specified for 'when' and also
specified the 'J' flag would be rotated every time newsyslog was run.
(this is a quick-fix, trying to beat the code-freeze for 5.1-release)

PR: bin/51519
MFC after: 1 week

21 years agoCentralize _PATH_* definitions.
David E. O'Brien [Mon, 5 May 2003 22:49:23 +0000 (22:49 +0000)]
Centralize _PATH_* definitions.

Submitted by: Tim Kientzle <kientzle@acm.org> (embellished by me)

21 years agoSlightly improve the checking of the 'size' field, so people will get
Garance A Drosehn [Mon, 5 May 2003 22:27:37 +0000 (22:27 +0000)]
Slightly improve the checking of the 'size' field, so people will get
a warning message if they switch the values for 'size' and 'when'.

MFC after: 1 week

21 years ago* Add the nut port from the IANA list (for the nut port)
Doug Barton [Mon, 5 May 2003 22:10:14 +0000 (22:10 +0000)]
* Add the nut port from the IANA list (for the nut port)
* Add the rndc port, by common usage (TCP only since rndc doesn't
use UDP, and the port is not officially assigned)

21 years agoAdd traditional BSD names (sio[0-3]) to the finger-friendly com[1-4]
Doug Barton [Mon, 5 May 2003 21:37:43 +0000 (21:37 +0000)]
Add traditional BSD names (sio[0-3]) to the finger-friendly com[1-4]
entries so that you can do things like 'tip sio1' without having
to remember the DOS equivalent.

21 years agoCompile bsdlabel on all platforms.
Poul-Henning Kamp [Mon, 5 May 2003 21:28:08 +0000 (21:28 +0000)]
Compile bsdlabel on all platforms.

Install a link to the disklabel(8) name on i386 and alpha platforms.

Leave old disklabel(8) sources intact but disconnected from the build
for now.

21 years agoTweak the clearing of TDF_DEADLKTREAT so that we only bother grabbing the
John Baldwin [Mon, 5 May 2003 21:27:29 +0000 (21:27 +0000)]
Tweak the clearing of TDF_DEADLKTREAT so that we only bother grabbing the
lock and clearing the flag if it was clear when uiomove() was called.

21 years agoMostly sort the includes.
John Baldwin [Mon, 5 May 2003 21:26:25 +0000 (21:26 +0000)]
Mostly sort the includes.

21 years agoPut descriptive comments on the GEOM_* options
Poul-Henning Kamp [Mon, 5 May 2003 21:21:31 +0000 (21:21 +0000)]
Put descriptive comments on the GEOM_* options

21 years agoLock the proc lock around calls to tdsignal() in the sigwait() family of
John Baldwin [Mon, 5 May 2003 21:18:10 +0000 (21:18 +0000)]
Lock the proc lock around calls to tdsignal() in the sigwait() family of
syscalls.

21 years agoMake issignal() private to kern_sig.c since it is only called from cursig()
John Baldwin [Mon, 5 May 2003 21:16:28 +0000 (21:16 +0000)]
Make issignal() private to kern_sig.c since it is only called from cursig()
and cursig() is now a function rather than a macro.

21 years agoOpenPAM is WANRS6-clean.
Dag-Erling Smørgrav [Mon, 5 May 2003 21:15:35 +0000 (21:15 +0000)]
OpenPAM is WANRS6-clean.

21 years agoWhitespace cleanup (1.15 had spaces instead of tabs)
Dag-Erling Smørgrav [Mon, 5 May 2003 21:14:36 +0000 (21:14 +0000)]
Whitespace cleanup (1.15 had spaces instead of tabs)

21 years agoDon't use -pedantic unless we also set -std of some kind.
Dag-Erling Smørgrav [Mon, 5 May 2003 21:13:32 +0000 (21:13 +0000)]
Don't use -pedantic unless we also set -std of some kind.

21 years agoRemove TD_ON_RUNQ() from a check to make sure Giant is not held when
John Baldwin [Mon, 5 May 2003 21:12:36 +0000 (21:12 +0000)]
Remove TD_ON_RUNQ() from a check to make sure Giant is not held when
calling mi_switch().  The kernel would panic on an earlier KASSERT() in
mi_switch() if TD_ON_RUNQ() was true.