]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoDiff to vendor reduction.
David E. O'Brien [Mon, 8 Dec 2003 01:52:09 +0000 (01:52 +0000)]
Diff to vendor reduction.

20 years agoThe last argument to mtx_init() should be MTX_DEF, not 0. This is not a
Don Lewis [Mon, 8 Dec 2003 01:18:04 +0000 (01:18 +0000)]
The last argument to mtx_init() should be MTX_DEF, not 0.  This is not a
functional change since MTX_DEF happens to be defined as 0.

20 years agoThe last argument to mtx_init() should be MTX_DEF, not 0. This is not a
Don Lewis [Mon, 8 Dec 2003 01:08:03 +0000 (01:08 +0000)]
The last argument to mtx_init() should be MTX_DEF, not 0.  This is not a
functional change since MTX_DEF happens to be defined as 0.

20 years agoAddendum to revision 1.174: In the case where vm_pager_allocate() is called
Alan Cox [Mon, 8 Dec 2003 00:47:33 +0000 (00:47 +0000)]
Addendum to revision 1.174: In the case where vm_pager_allocate() is called
to create a vnode-backed object, the vnode lock must be held by the caller.

Reported by: truckman
Discussed with: kan

20 years agoCopy the old dgb man page to digi.4, plus very minor hacking. Add a
Warner Losh [Mon, 8 Dec 2003 00:11:22 +0000 (00:11 +0000)]
Copy the old dgb man page to digi.4, plus very minor hacking.  Add a
note that says that this man page is sub-optimal.  Bruce Mah should be
happier about this, but someone that groks the cards supported by the
digi driver is encouraged to make this man page suck less.

20 years agoLock and unlock sched_lock when walking through thread list, current we
David Xu [Sun, 7 Dec 2003 23:47:15 +0000 (23:47 +0000)]
Lock and unlock sched_lock when walking through thread list, current we
insert kse upcall thread into thread list at mi_switch time, process lock
is not enough.

20 years agoCorrect usage of mtx_init() API. This is not a functional change since
Don Lewis [Sun, 7 Dec 2003 23:21:22 +0000 (23:21 +0000)]
Correct usage of mtx_init() API.  This is not a functional change since
the code happened to work because MTX_DEF and NULL are both defined as 0.

Reviewed by:    phk

20 years agoCorrect usage of mtx_init() API. This is not a functional change since
Don Lewis [Sun, 7 Dec 2003 23:20:53 +0000 (23:20 +0000)]
Correct usage of mtx_init() API.  This is not a functional change since
the code happened to work because MTX_DEF and NULL are both defined as 0.

Reviewed by: phk

20 years agoTry a little harder to catch when fixate finishes.
Søren Schmidt [Sun, 7 Dec 2003 23:15:22 +0000 (23:15 +0000)]
Try a little harder to catch when fixate finishes.
On the <QSI CD-RW/DVD-ROM SBW-242> that failed causing burncd to
wait forever...

20 years agoUrk, add arch attributes to items from my previous commit.
Bruce A. Mah [Sun, 7 Dec 2003 23:07:39 +0000 (23:07 +0000)]
Urk, add arch attributes to items from my previous commit.

20 years agoStyle change
Tom Rhodes [Sun, 7 Dec 2003 23:02:16 +0000 (23:02 +0000)]
Style change

20 years agoNew release notes: Record the demise of the lmc, dgb, hea, meteor,
Bruce A. Mah [Sun, 7 Dec 2003 22:58:35 +0000 (22:58 +0000)]
New release notes:  Record the demise of the lmc, dgb, hea, meteor,
and loran drivers.

20 years agoDon't remove the virtual-to-physical mapping when an sf_buf is freed.
Alan Cox [Sun, 7 Dec 2003 22:49:25 +0000 (22:49 +0000)]
Don't remove the virtual-to-physical mapping when an sf_buf is freed.
Instead, allow the mapping to persist, but add the sf_buf to a free list.
If a later sendfile(2) or zero-copy send resends the same physical page,
perhaps with the same or different contents, then the mapping overhead is
avoided and the sf_buf is simply removed from the free list.

In other words, the i386 sf_buf implementation now behaves as a cache of
virtual-to-physical translations using an LRU replacement policy on
inactive sf_bufs.  This is similar in concept to a part of
http://www.cs.princeton.edu/~yruan/debox/ patch, but much simpler in
implementation.  Note: none of this is required on alpha, amd64, or ia64.
They now use their direct virtual-to-physical mapping to avoid any
emphemeral mapping overheads in their sf_buf implementations.

20 years agoThe lmc, meteor, and loran drivers have been removed from the source
Bruce A. Mah [Sun, 7 Dec 2003 22:48:03 +0000 (22:48 +0000)]
The lmc, meteor, and loran drivers have been removed from the source
tree, so remove them from the hardware notes.

The dgb driver has been replaced by the digi driver.

The hea driver has been replaced by the en driver.

20 years agoBump various entities for 5.2-CURRENT, trim release notes and errata.
Bruce A. Mah [Sun, 7 Dec 2003 22:12:05 +0000 (22:12 +0000)]
Bump various entities for 5.2-CURRENT, trim release notes and errata.

20 years agoPass MTX_DEF as the last argument to mtx_init() instead of 0. This
Don Lewis [Sun, 7 Dec 2003 21:53:41 +0000 (21:53 +0000)]
Pass MTX_DEF as the last argument to mtx_init() instead of 0.  This
is not a functional change.  The code happened to work properly only
because MTX_DEF is defined as 0.

20 years agobandaid LOR in rt_setgate; a proper fix requires code refactoring
Sam Leffler [Sun, 7 Dec 2003 21:44:14 +0000 (21:44 +0000)]
bandaid LOR in rt_setgate; a proper fix requires code refactoring

20 years agoadd install rule
Sam Leffler [Sun, 7 Dec 2003 21:42:40 +0000 (21:42 +0000)]
add install rule

20 years agoo report 0 instead of -1 for xmit rate when in transition
Sam Leffler [Sun, 7 Dec 2003 21:40:52 +0000 (21:40 +0000)]
o report 0 instead of -1 for xmit rate when in transition
o report rssi in running display
o futz with column widths in running display

20 years agofix comment
Sam Leffler [Sun, 7 Dec 2003 21:38:28 +0000 (21:38 +0000)]
fix comment

20 years agoThe uuidgen(1) program is WARNS=6 clean, so flag it as such.
Maxime Henrion [Sun, 7 Dec 2003 21:34:56 +0000 (21:34 +0000)]
The uuidgen(1) program is WARNS=6 clean, so flag it as such.

Tested on: i386, sparc64

20 years agoChange the definition of NULL on ia64 (for LP64 compilations) from
Marcel Moolenaar [Sun, 7 Dec 2003 21:10:06 +0000 (21:10 +0000)]
Change the definition of NULL on ia64 (for LP64 compilations) from
an int constant to a long constant. This change improves consistency
in the following two ways:
1. The first 8 arguments are always passed in registers on ia64, which
   by virtue of the generated code implicitly widens ints to longs and
   allows the use of an 32-bit integral type for 64-bit arguments.
   Subsequent arguments are passed onto the memory stack, which does
   not exhibit the same behaviour and consequently do not allow this.
   In practice this means that variadic functions taking pointers
   and given NULL (without cast) work as long as the NULL is passed
   in one of the first 8 arguments. A SIGSEGV is more likely the
   result if such would be done for stack-based arguments. This is
   due to the fact that the upper 4 bytes remain undefined.
2. All 64-bit platforms that FreeBSD supports, with the obvious
   exception of ia64, allow 32-bit integral types (specifically NULL)
   when 64-bit pointers are expected in variadic functions by way of
   how the compiler generates code. As such, code that works correctly
   (whether rightfully so or not) on any platform other than ia64, may
   fail on ia64.

To more easily allow tweaking of the definition of NULL, this commit
removes the 12 definitions in the various headers and puts it in a
new header that can be included whenever NULL is to be made visible.

This commit fixes GNOME, emacs, xemacs and a whole bunch of ports
that I don't particularly care about at this time...

20 years agoAdd a manual page for the consolidated debugging commit.
Scott Long [Sun, 7 Dec 2003 20:48:00 +0000 (20:48 +0000)]
Add a manual page for the consolidated debugging commit.

20 years agoSimplify the contexts created by the kernel and remove the related
Marcel Moolenaar [Sun, 7 Dec 2003 20:47:33 +0000 (20:47 +0000)]
Simplify the contexts created by the kernel and remove the related
flags. We now create asynchronous contexts or syscall contexts only.
Syscall contexts differ from the minimal ABI dictated contexts by
having the scratch registers saved and restored because that's where
we keep the syscall arguments and syscall return values.
Since this change affects KSE, have it use kse_switchin(2) for the
"new" syscall context.

20 years agoMake the DIAGNOSTIC code which complains about long {call|time}out(9)
Poul-Henning Kamp [Sun, 7 Dec 2003 20:03:28 +0000 (20:03 +0000)]
Make the DIAGNOSTIC code which complains about long {call|time}out(9)
functions less noisy:  We printf if a new function took longer than
the previous record holder, or of the previous record holder took
more than twice as long as the current record.

20 years agoRegen due to kse_switchin(2).
Marcel Moolenaar [Sun, 7 Dec 2003 19:36:16 +0000 (19:36 +0000)]
Regen due to kse_switchin(2).

20 years agoAdd kse_switchin(2). This syscall can be used by KSE implementations
Marcel Moolenaar [Sun, 7 Dec 2003 19:34:29 +0000 (19:34 +0000)]
Add kse_switchin(2). This syscall can be used by KSE implementations
to have the kernel switch to a new thread, instead of doing it in
userland. It is in fact needed on ia64 where syscall restarts do not
return to userland first. It's completely handled inside the kernel.
As such, any context created by the kernel as part of an upcall and
caused by some syscall needs to be restored by the kernel.

20 years agoFinish the transition from libkvm to sysctl that I started a while ago.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:46:14 +0000 (17:46 +0000)]
Finish the transition from libkvm to sysctl that I started a while ago.
The use of libkvm for post-mortem analysis is still supported (though it
could use more testing).  We can now remove vmstat's setgid bit.

While I'm here, hack the interrupt listing code to not display interrupts
that haven't occurred unless the -a option was given on the command line,
and document this change.

20 years agoLog all results to a history file in the log directory.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:43:30 +0000 (17:43 +0000)]
Log all results to a history file in the log directory.

20 years agoConstify, and add an API function to find a named node in a directory.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:41:19 +0000 (17:41 +0000)]
Constify, and add an API function to find a named node in a directory.

20 years agoMinor whitespace and style issues.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:40:00 +0000 (17:40 +0000)]
Minor whitespace and style issues.

20 years agoUse mp_ncpus instead of the hw.ncpu sysctl.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:38:20 +0000 (17:38 +0000)]
Use mp_ncpus instead of the hw.ncpu sysctl.

20 years agoRemove useless SMP check code.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:37:44 +0000 (17:37 +0000)]
Remove useless SMP check code.

20 years agoForced commit: previous log entry should have said "fix some warnings and
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:35:41 +0000 (17:35 +0000)]
Forced commit: previous log entry should have said "fix some warnings and
style issues, no functional changes"

20 years agoUse WARNS level 2 instead of setting CFLAGS explicitly to -Wall.
Dag-Erling Smørgrav [Sun, 7 Dec 2003 17:34:52 +0000 (17:34 +0000)]
Use WARNS level 2 instead of setting CFLAGS explicitly to -Wall.

20 years agoPrint out the file system access statistics using uintmax_t types
Ian Dowse [Sun, 7 Dec 2003 17:00:14 +0000 (17:00 +0000)]
Print out the file system access statistics using uintmax_t types
instead of casting the unsigned 64-bit values to longs.

Suggested by: bde

20 years agoMFR: Bump manpages revision to 5.2.
Ruslan Ermilov [Sun, 7 Dec 2003 15:46:50 +0000 (15:46 +0000)]
MFR: Bump manpages revision to 5.2.

20 years agoAdd support for timeout: and attempts: resolver options.
Murray Stokely [Sun, 7 Dec 2003 12:32:24 +0000 (12:32 +0000)]
Add support for timeout: and attempts: resolver options.

Submitted by: Paul Vixie <paul@vix.com> / ISC
MFC After: 1 week

20 years agouse callout_*() rather than timeout().
Hajimu UMEMOTO [Sun, 7 Dec 2003 11:23:59 +0000 (11:23 +0000)]
use callout_*() rather than timeout().

20 years agolink-local multicast address must be converted to KAME specific
Hajimu UMEMOTO [Sun, 7 Dec 2003 11:11:26 +0000 (11:11 +0000)]
link-local multicast address must be converted to KAME specific
embeded scopeid form.

Reported by: dwmalone
MFC after: 3 days

20 years agoKASSERT against multiple orphanings of providers.
Poul-Henning Kamp [Sun, 7 Dec 2003 10:04:43 +0000 (10:04 +0000)]
KASSERT against multiple orphanings of providers.

20 years agoA too good Feynman quote to pass up.
Poul-Henning Kamp [Sun, 7 Dec 2003 10:03:20 +0000 (10:03 +0000)]
A too good Feynman quote to pass up.

20 years agorqb_bits[] may be an int64_t (eg: on alpha, and recently on amd64).
Peter Wemm [Sun, 7 Dec 2003 09:57:51 +0000 (09:57 +0000)]
rqb_bits[] may be an int64_t (eg: on alpha, and recently on amd64).
Be sure to shift (long)1 << 33 and higher, not (int)1.  Otherwise bad
things happen(TM).  This is why beast.freebsd.org paniced with ULE.

Reviewed by:  jeff

20 years agoAdd PowerPC CFLAGS.
David E. O'Brien [Sun, 7 Dec 2003 09:56:30 +0000 (09:56 +0000)]
Add PowerPC CFLAGS.

Submitted by: gallatin

20 years agoAdd IPv6 support to pppctl by using getaddrinfo() and trying each address
Tim J. Robbins [Sun, 7 Dec 2003 08:39:29 +0000 (08:39 +0000)]
Add IPv6 support to pppctl by using getaddrinfo() and trying each address
it returns. This allows it to connect to the server side again, which
has been listening on IPv6 addresses exclusively for more than 2 years.

PR: 59369

20 years agoUpdate to reflect eni removal
Warner Losh [Sun, 7 Dec 2003 08:04:05 +0000 (08:04 +0000)]
Update to reflect eni removal

20 years agohea is gone, remove its module
Warner Losh [Sun, 7 Dec 2003 07:04:39 +0000 (07:04 +0000)]
hea is gone, remove its module

20 years agoRemove hea module.
Warner Losh [Sun, 7 Dec 2003 07:03:07 +0000 (07:03 +0000)]
Remove hea module.

20 years agoDon't call em_stop() from the watchdog since it requires the controller
Daniel Eischen [Sun, 7 Dec 2003 06:50:04 +0000 (06:50 +0000)]
Don't call em_stop() from the watchdog since it requires the controller
mutex to be locked.  It is redundant since em_init() is called and this
correctly locks the mutex and calls em_stop().

5.2 release candidate since this can cause a panic if the watchdog
expires.

Tested by: kuriyama

20 years agolmc.4 is now gone
Warner Losh [Sun, 7 Dec 2003 06:49:10 +0000 (06:49 +0000)]
lmc.4 is now gone

20 years agoOoops. These are still used by the bktr driver. David O'Brien has
Warner Losh [Sun, 7 Dec 2003 06:37:32 +0000 (06:37 +0000)]
Ooops.  These are still used by the bktr driver.  David O'Brien has
plans for dealing, but I'll let him deal.

Pointy hat to: imp@

20 years agoSplit multibyte(3) into separate manual pages for each function.
Tim J. Robbins [Sun, 7 Dec 2003 06:33:52 +0000 (06:33 +0000)]
Split multibyte(3) into separate manual pages for each function.
Instead of just deleting it, turn the original page into a general
overview of the multibyte character conversion functions, somewhat
similar to stdio(3).

20 years agoSplit the documentation for localeconv() off into a separate manual page.
Tim J. Robbins [Sun, 7 Dec 2003 06:00:00 +0000 (06:00 +0000)]
Split the documentation for localeconv() off into a separate manual page.

20 years agoMake the if_broadcastaddr const. All the drivers in the tree which
Warner Losh [Sun, 7 Dec 2003 05:49:21 +0000 (05:49 +0000)]
Make the if_broadcastaddr const.  All the drivers in the tree which
violated the constness were corrected before the freeze.  This was
suggested by mdodd@, I think, and sam@ and others have signed off on
this if I recall my conversations with them correctly.

20 years agoReinstate 1.40 -- swap avgfilesize and avgfpdir column order.
Don Lewis [Sun, 7 Dec 2003 05:27:27 +0000 (05:27 +0000)]
Reinstate 1.40 -- swap avgfilesize and avgfpdir column order.

MFC after: 3 days

20 years agounifdef old interface support out to prevent false positives.
Warner Losh [Sun, 7 Dec 2003 05:17:13 +0000 (05:17 +0000)]
unifdef old interface support out to prevent false positives.

Suggested by: jeffr, obrien, and others

20 years agoSet fs_ronly to the correct value in ffs_reload() when reloading the file
Don Lewis [Sun, 7 Dec 2003 05:16:52 +0000 (05:16 +0000)]
Set fs_ronly to the correct value in ffs_reload() when reloading the file
system super block after fsck has repaired the file system.  The value of
fs_ronly was getting overwritten, which caused ffs_update() to attempt to
update inode timestamps even though the file system was still mounted
read-only.

This fixes the "giving up on N buffers" error that is triggered by running
fsck on the root file system and then rebooting without mounting the file
system read-write.

20 years agoFix prototype for getchar_unlocked().
Tim J. Robbins [Sun, 7 Dec 2003 05:12:51 +0000 (05:12 +0000)]
Fix prototype for getchar_unlocked().

20 years agoRe-arrange and consolidate some random debugging stuff
Scott Long [Sun, 7 Dec 2003 05:04:49 +0000 (05:04 +0000)]
Re-arrange and consolidate some random debugging stuff

20 years agoIncrease VM_KMEM_SIZE_MAX from 200MB to 400MB.
Alan Cox [Sun, 7 Dec 2003 04:51:04 +0000 (04:51 +0000)]
Increase VM_KMEM_SIZE_MAX from 200MB to 400MB.

Discussed with: peter

20 years agoRemote meteor driver. It hasn't compiled in over 3 years. If someone
Warner Losh [Sun, 7 Dec 2003 04:41:11 +0000 (04:41 +0000)]
Remote meteor driver.  It hasn't compiled in over 3 years.  If someone
makes it compile again, and can test it, we can restore the driver to
the tree.

20 years agoNow that we have the en(4) driver, we no longer need the hea driver.
Warner Losh [Sun, 7 Dec 2003 04:27:08 +0000 (04:27 +0000)]
Now that we have the en(4) driver, we no longer need the hea driver.

Approved by: harti@

20 years agoThe dgb driver is redundant with the digi driver in the tree. It uses
Warner Losh [Sun, 7 Dec 2003 04:18:52 +0000 (04:18 +0000)]
The dgb driver is redundant with the digi driver in the tree.  It uses
lots of old interfaces, and digi now supports all cards that dgb
supported.  The author of the driver says that this is no longer
necessary.

Approved by: babkin@

20 years ago - Giant is no longer required by vm_thread_new().
Alan Cox [Sun, 7 Dec 2003 04:16:49 +0000 (04:16 +0000)]
 - Giant is no longer required by vm_thread_new().

20 years agoThis commit was generated by cvs2svn to compensate for changes in r123202,
Sam Leffler [Sun, 7 Dec 2003 04:07:35 +0000 (04:07 +0000)]
This commit was generated by cvs2svn to compensate for changes in r123202,
which included commits to RCS files with non-trunk default branches.

20 years agoVersion 0.9.6.3:
Sam Leffler [Sun, 7 Dec 2003 04:07:35 +0000 (04:07 +0000)]
Version 0.9.6.3:

o relax analog radio hardware revision checking for 5212 cards
  so newer revs are allowed to attach

20 years agoContinue to remove drivers that don't compile and haven't compiled in
Warner Losh [Sun, 7 Dec 2003 04:05:19 +0000 (04:05 +0000)]
Continue to remove drivers that don't compile and haven't compiled in
a long time: lmc The LAN Media Corp PCI WAN driver based on tulip.
This driver hasn't compiled for 3 years since the PCI compat shims
were removed, and Lan Media appears to have gone out of business.
These cards appear to be rare (a recent search of ebay had no hits).

Should someone wish to revive this driver, submitting patches to make
it compile plus a testing report will bring it back.

20 years agoHEAD __FreeBSD_version is at 502100
Scott Long [Sun, 7 Dec 2003 03:56:54 +0000 (03:56 +0000)]
HEAD __FreeBSD_version is at 502100

20 years agoHEAD is at 5.2-CURRENT now
Scott Long [Sun, 7 Dec 2003 03:56:12 +0000 (03:56 +0000)]
HEAD is at 5.2-CURRENT now

20 years agoFirst part of the removal of drivers for hardware that isn't relevant
Warner Losh [Sun, 7 Dec 2003 03:51:19 +0000 (03:51 +0000)]
First part of the removal of drivers for hardware that isn't relevant
or whose drivers haven't even compiled for years.

The loran hardware was very unique, and only a few copies of it ever
existed.  It used the old COMPAT_ISA_DRIVER and when the author was
contacted, he indicated that he had no intention of ever updating this
driver and it was no longer relevant to the FreeBSD world and can be
removed without impact to anybody.

Approved by: phk

20 years agoUpdate with the 4.9 pam modules, and add pam_opieaccess.so to the mix.
David E. O'Brien [Sun, 7 Dec 2003 03:02:27 +0000 (03:02 +0000)]
Update with the 4.9 pam modules, and add pam_opieaccess.so to the mix.

Approved by: re(scottl)

20 years agoUpdate with the 4.9 pam modules, and add pam_opieaccess.so to the mix.
David E. O'Brien [Sun, 7 Dec 2003 02:38:01 +0000 (02:38 +0000)]
Update with the 4.9 pam modules, and add pam_opieaccess.so to the mix.

Approved by: re(scottl)

20 years agoUpdate with the 4.9 libraries, and add libdevstat.so.2 libmp.so.3
David E. O'Brien [Sun, 7 Dec 2003 02:11:19 +0000 (02:11 +0000)]
Update with the 4.9 libraries, and add libdevstat.so.2 libmp.so.3
libskey.so.2 libusbhid.so.0 libutil.so.3 libvgl.so.2 to the mix.

Approved by: re(scottl)

20 years agoWhee!
Bruce A. Mah [Sun, 7 Dec 2003 01:11:33 +0000 (01:11 +0000)]
Whee!

Modified release note:  kdump(1) -E.

Approved by: re (implicitly)

20 years agoGrrr...get the semantics of these entities right.
Bruce A. Mah [Sun, 7 Dec 2003 01:08:24 +0000 (01:08 +0000)]
Grrr...get the semantics of these entities right.

Approved by: re (implicitly)

20 years agoAdd a -E (elapsed time) flag to kdump. This is like -T, except it is
Peter Wemm [Sun, 7 Dec 2003 01:06:32 +0000 (01:06 +0000)]
Add a -E (elapsed time) flag to kdump.  This is like -T, except it is
relative to start of the dump.

Approved by:  re (scottl)

20 years agoUse sab(4) manpage entity.
Bruce A. Mah [Sun, 7 Dec 2003 00:54:22 +0000 (00:54 +0000)]
Use sab(4) manpage entity.

Approved by: re (implicitly)

20 years agoHook the sab(4) manpage up to the build.
Bruce A. Mah [Sun, 7 Dec 2003 00:46:37 +0000 (00:46 +0000)]
Hook the sab(4) manpage up to the build.

Reviewed by: bms
Approved by: re (implicitly)

20 years agoMore correctly handle cross-references to USB keyboards and mice.
Bruce A. Mah [Sun, 7 Dec 2003 00:37:35 +0000 (00:37 +0000)]
More correctly handle cross-references to USB keyboards and mice.

Approved by: re (implicitly)

20 years agoMore hardware notes device trimming: Remove devices supported by
Bruce A. Mah [Sun, 7 Dec 2003 00:06:50 +0000 (00:06 +0000)]
More hardware notes device trimming:  Remove devices supported by
ohci(4), uhci(4), ubsa(4), uftdi(4), uvscom(4), fwohci(4), and sbp(4)
in favor of the device lists in their respective manual pages.

Approved by: re (implicitly)

20 years agoReconfigure the runq macros to use the 64 bit ffs/bsf routines instead
Peter Wemm [Sat, 6 Dec 2003 23:24:18 +0000 (23:24 +0000)]
Reconfigure the runq macros to use the 64 bit ffs/bsf routines instead
of doing a loop and taking two 32 bit passes at the runqueue bits.  All
the 64 bit platforms should probably do this since there are 64 run queues.

Approved by:  re (scottl)

20 years agoAdd 64 bit bsf*/ffs* routines. Have the ffs() inline use gcc's builtin
Peter Wemm [Sat, 6 Dec 2003 23:22:43 +0000 (23:22 +0000)]
Add 64 bit bsf*/ffs* routines.  Have the ffs() inline use gcc's builtin
because it uses the better cmove instructions to avoid branches.

20 years agoVarious whitespace and cosmetic sync-up's with i386.
Peter Wemm [Sat, 6 Dec 2003 23:19:47 +0000 (23:19 +0000)]
Various whitespace and cosmetic sync-up's with i386.

Approved by:  re (scottl)

20 years agoamd64_protection_init and the protection_codes[] array was overkill.
Peter Wemm [Sat, 6 Dec 2003 23:18:42 +0000 (23:18 +0000)]
amd64_protection_init and the protection_codes[] array was overkill.
Inline it instead.

Approved by:  re (scottl)

20 years agoKill the ASM versions of the mtx_lock_spin and friends. They were never
Peter Wemm [Sat, 6 Dec 2003 23:17:18 +0000 (23:17 +0000)]
Kill the ASM versions of the mtx_lock_spin and friends.  They were never
used on amd64, and were actually totally broken.  They had the wrong
calling conventions.  I believe the i386 versions are going away too.

Approved by:  re (scottl)

20 years agoMFi386: put the apic disable hook in a better place.
Peter Wemm [Sat, 6 Dec 2003 23:14:44 +0000 (23:14 +0000)]
MFi386: put the apic disable hook in a better place.

Approved by:  re (scottl)

20 years agoNew release notes: Logical CPUs for HTT now enabled using BIOS
Bruce A. Mah [Sat, 6 Dec 2003 23:14:35 +0000 (23:14 +0000)]
New release notes:  Logical CPUs for HTT now enabled using BIOS
settings, options IPFILTER requires options PFIL_HOOKS, last bits of
a.out compiler toolchain are gone, acpiconf(8) -i, asf(8), .snap
directory for snapshots, kdump(1) -p, last(1) -n, libwrap/tcpdchk(8)
support extended tcp_wrappers syntax, repquota(8) -n, savecore(8) -C,
uname(1) -i, hardware notes device trimmage, Turkish translation for
docs.

Modified release notes:  Spell "Cronyx-Sigma" correctly in cx(4)
item, spell "FireWire" consistently, fix grammer in puc(4) item, move
uart(4) note from NIC to generic kernel section, spell "NetGraph"
consistently, fix typo in xe(4) note, dynamic linking note more
precise,

Deleted release notes:  -pthread removed (this was reverted).

Other:  Correct a reference to recent security advisories.

Approved by: re (implicitly)

20 years agoRevert some amd64 changes that cached curthread and converge back to the
Peter Wemm [Sat, 6 Dec 2003 23:13:22 +0000 (23:13 +0000)]
Revert some amd64 changes that cached curthread and converge back to the
i386 version.  The curthread special case in pcpu.h solves my complaint
about the verbose macro expansion in this case.  Note that the i386
version still has some OBE comments, I didn't re-add them back again.

Approved by:  re (scottl)

20 years agoAdd Japan's real entry into the Second World War.
Greg Lehey [Sat, 6 Dec 2003 22:49:30 +0000 (22:49 +0000)]
Add Japan's real entry into the Second World War.

http://www.abcmalaysia.com/tour_malaysia/kltn_japivsn.htm

20 years agoRename mac_create_cred() MAC Framework entry point to mac_copy_cred(),
Robert Watson [Sat, 6 Dec 2003 21:48:03 +0000 (21:48 +0000)]
Rename mac_create_cred() MAC Framework entry point to mac_copy_cred(),
and the mpo_create_cred() MAC policy entry point to
mpo_copy_cred_label().  This is more consistent with similar entry
points for creation and label copying, as mac_create_cred() was
called from crdup() as opposed to during process creation.  For
a number of policies, this removes the requirement for special
handling when copying credential labels, and improves consistency.

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

20 years agoTrim some unnecessary text from bt(4) and mly(4) listings, fix typo in
Bruce A. Mah [Sat, 6 Dec 2003 20:43:48 +0000 (20:43 +0000)]
Trim some unnecessary text from bt(4) and mly(4) listings, fix typo in
sk(4) listing, trim em(4) devices, trim dgb(4) devices, trim
stl(4)/stli(4) devices, trim uaudio(4) devices.

Approved by: re (implicitly)

20 years agoHTT logical processors are now enabled (not disabled) by default, so
Bruce A. Mah [Sat, 6 Dec 2003 20:39:58 +0000 (20:39 +0000)]
HTT logical processors are now enabled (not disabled) by default, so
reflect this in the hardware notes.

Approved by: re (implicitly)

20 years agoDo not panic when flushing dummynet firewall rules
Maksim Yevmenkin [Sat, 6 Dec 2003 09:01:25 +0000 (09:01 +0000)]
Do not panic when flushing dummynet firewall rules

Reviewed by: andre
Approved by: re (scottl)

20 years agoFix a deadlock between vm_fault() and vm_mmap(): The expected lock ordering
Alan Cox [Sat, 6 Dec 2003 05:45:32 +0000 (05:45 +0000)]
Fix a deadlock between vm_fault() and vm_mmap(): The expected lock ordering
between vm_map and vnode locks is that vm_map locks are acquired first.  In
revision 1.150 mmap(2) was changed to pass a locked vnode into vm_mmap().
This creates a lock-order reversal when vm_mmap() calls one of the vm_map
routines that acquires a vm_map lock.  The solution implemented herein is
to release the vnode lock in mmap() before calling vm_mmap() and reacquire
this lock if necessary in vm_mmap().

Approved by: re (scottl)
Reviewed by: jeff, kan, rwatson

20 years agoSlim down the package list some more. The KDE and Gnome metaports might
Scott Long [Sat, 6 Dec 2003 04:15:13 +0000 (04:15 +0000)]
Slim down the package list some more.  The KDE and Gnome metaports might
need to be tweaked a bit, but this gives the disc1 ISO a fighting chance.

20 years agoAdd support for ELECOM/Laneed CardBus FastEtherner Card(LD-CBL/TXA)
MIHIRA Sanpei Yoshiro [Sat, 6 Dec 2003 02:29:31 +0000 (02:29 +0000)]
Add support for ELECOM/Laneed CardBus FastEtherner Card(LD-CBL/TXA)

Submitted by: Masahiro Yamagishi <night@pluto.dti.ne.jp>
Approved by: re (scottl)

20 years agoNew release note: ip_fastforward.
Bruce A. Mah [Fri, 5 Dec 2003 21:34:31 +0000 (21:34 +0000)]
New release note:  ip_fastforward.

Approved by: re (implicitly)

20 years agoMove cx(4) entry from "Serial Devices" to "Miscellaneous Networks".
Bruce A. Mah [Fri, 5 Dec 2003 21:04:51 +0000 (21:04 +0000)]
Move cx(4) entry from "Serial Devices" to "Miscellaneous Networks".

Approved by: re (implicitly)
Reviewed by: imp

20 years agoRemove SMP option support from building the ACPI module as it is no longer
John Baldwin [Fri, 5 Dec 2003 15:49:01 +0000 (15:49 +0000)]
Remove SMP option support from building the ACPI module as it is no longer
needed.

Approved by: re (murray)
Requested by: njl

20 years agoDon't include the file system ID in the output of `mount -v' if it
Ian Dowse [Fri, 5 Dec 2003 09:36:56 +0000 (09:36 +0000)]
Don't include the file system ID in the output of `mount -v' if it
is all zeros. The kernel now consistently zeroes FSIDs for non-root
users, so there's no point in printing these.

Also fix a number of compiler warnings, including two real bugs:
- a bracket placement bug caused `mount -t ufs localhost:/foo /mnt'
  to override the `-t ufs' specification and use mount_nfs.
- an unitialised variable was used instead of _PATH_SYSPATH when
  warning that the mount_* program cound not be found.

Submitted by: Rudolf Cejka <cejkar@fit.vutbr.cz> (FSID part)
Approved by: re (scottl)