]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoMFS 1.64.2.22: Re-enable non pre-emptive ARP requests.
orion [Tue, 4 Feb 2003 05:28:08 +0000 (05:28 +0000)]
MFS 1.64.2.22: Re-enable non pre-emptive ARP requests.

Submitted by: "Diomidis Spinellis" <dds@aueb.gr>
PR:           kern/46116

21 years agoRFC 959 doesn't list reply code 550 as a valid responce to STOR/STOU,
yar [Tue, 4 Feb 2003 03:33:25 +0000 (03:33 +0000)]
RFC 959 doesn't list reply code 550 as a valid responce to STOR/STOU,
so return reply code 553 to indicate a error from open(2) for consistency,
as long as the code is used in the rest of the STOR/STOU handler.

21 years agoCorrect examples for stateful inspection
brueffer [Tue, 4 Feb 2003 01:33:25 +0000 (01:33 +0000)]
Correct examples for stateful inspection

PR: 47817
Submitted by: Simon L.Nielsen <simon@nitro.dk>
Reviewed by: ceri, luigi

21 years agoImprove the layout of the description of the various parameter expansion
fanf [Mon, 3 Feb 2003 23:03:05 +0000 (23:03 +0000)]
Improve the layout of the description of the various parameter expansion
modifiers. The paragraph that explains the difference between ${foo:-bar}
and ${foo-bar} etc. was not very visible.

21 years agoeven the default case in a switch requires some action (add break;)
billf [Mon, 3 Feb 2003 22:47:03 +0000 (22:47 +0000)]
even the default case in a switch requires some action (add break;)

EVP_sha1() returns a const EVP_MD *, so reflect that in the variable
we store its return value in.

21 years agoSplit the global timezone structure into two integer fields to
phk [Mon, 3 Feb 2003 19:49:35 +0000 (19:49 +0000)]
Split the global timezone structure into two integer fields to
prevent the compiler from optimizing assignments into byte-copy
operations which might make access to the individual fields non-atomic.

Use the individual fields throughout, and don't bother locking them with
Giant: it is no longer needed.

Inspired by:    tjr

21 years agoNew release notes: CanBe power management, volume labels, MAKEDEV gone,
bmah [Mon, 3 Feb 2003 19:29:12 +0000 (19:29 +0000)]
New release notes:  CanBe power management, volume labels, MAKEDEV gone,
sunlabel.

Modified release notes:  Use elfdump(1) entity.

21 years agoRemove unintended growth of bt848_card_signature.
orion [Mon, 3 Feb 2003 18:59:12 +0000 (18:59 +0000)]
Remove unintended growth of bt848_card_signature.

21 years agoSplit statclock into statclock and profclock, and made the method for driving
jake [Mon, 3 Feb 2003 17:53:15 +0000 (17:53 +0000)]
Split statclock into statclock and profclock, and made the method for driving
statclock based on profhz when profiling is enabled MD, since most platforms
don't use this anyway.  This removes the need for statclock_process, whose
only purpose was to subdivide profhz, and gets the profiling clock running
outside of sched_lock on platforms that implement suswintr.
Also changed the interface for starting and stopping the profiling clock to
do just that, instead of changing the rate of statclock, since they can now
be separate.

Reviewed by: jhb, tmm
Tested on: i386, sparc64

21 years agoAdd IPv6 support for Linuxlator.
ume [Mon, 3 Feb 2003 17:43:20 +0000 (17:43 +0000)]
Add IPv6 support for Linuxlator.

Reviewed by: dwmalone
MFC after: 10 days

21 years agoBreak out the bind and connect syscalls to intend to make calling
ume [Mon, 3 Feb 2003 17:36:52 +0000 (17:36 +0000)]
Break out the bind and connect syscalls to intend to make calling
these syscalls internally easy.
This is preparation for force coming IPv6 support for Linuxlator.

Submitted by: dwmalone
MFC after: 10 days

21 years agoAdd config glue to add an optional GEOM_VOL to add optional volume support.
gordon [Mon, 3 Feb 2003 17:35:54 +0000 (17:35 +0000)]
Add config glue to add an optional GEOM_VOL to add optional volume support.

Reviewed by: jake (mentor)

21 years agoCorrect a comment. GEOM modules do not create /dev entries. They create
gordon [Mon, 3 Feb 2003 17:21:43 +0000 (17:21 +0000)]
Correct a comment. GEOM modules do not create /dev entries. They create
providers.

Pointed out by: phk

21 years agoAdd the GEOM module that makes volume labels useful. A kernel compiled with
gordon [Mon, 3 Feb 2003 17:14:29 +0000 (17:14 +0000)]
Add the GEOM module that makes volume labels useful. A kernel compiled with
this will cause volume labels to be exposed in /dev/vol/<volname>. Currently,
there is no conflict resolution if more than one FS has the same volume name.

Reviewed by: phk

21 years agoFixup capitalization in some of the Startup menu entries.
jhb [Mon, 3 Feb 2003 16:14:33 +0000 (16:14 +0000)]
Fixup capitalization in some of the Startup menu entries.

Submitted by: Hiten Pandya <hiten@unixdaemons.com>

21 years agoNo need to lock Giant around call to nanosleep1() in nanosleep().
tjr [Mon, 3 Feb 2003 15:31:57 +0000 (15:31 +0000)]
No need to lock Giant around call to nanosleep1() in nanosleep().

21 years agoAvoid zero padding when feeding read channels. chn_rdfeed has no way
orion [Mon, 3 Feb 2003 15:02:28 +0000 (15:02 +0000)]
Avoid zero padding when feeding read channels.  chn_rdfeed has no way
of knowing data size transformations of feeder chain and in some cases
this means too much data is pulled through chain, eg converting input
stream from 16bits to 8bits on 16bit only h/w.

PR: kern/37831
Submitted by: Harti Brandt <brandt@fokus.fraunhofer.de>

21 years agoAvoid holding Giant across copyout() in gettimeofday() and getitimer().
tjr [Mon, 3 Feb 2003 14:47:22 +0000 (14:47 +0000)]
Avoid holding Giant across copyout() in gettimeofday() and getitimer().

21 years agoAdd CanBe power management controller support.
nyan [Mon, 3 Feb 2003 14:46:26 +0000 (14:46 +0000)]
Add CanBe power management controller support.

Submitted by: KIYOHARA Takashi <kiyohara@kk.iij4u.or.jp>

21 years agoDon't enable pam_krb5 by default - most people don't have it since most
des [Mon, 3 Feb 2003 14:45:02 +0000 (14:45 +0000)]
Don't enable pam_krb5 by default - most people don't have it since most
people don't build with MAKE_KERBEROS5 defined.  Provide commented-out
usage examples instead, like we do everywhere else.

Pointy hat to: des

21 years agoLinux-PAM's pam_start(3) fails with a bogus error message if passed the
des [Mon, 3 Feb 2003 14:10:28 +0000 (14:10 +0000)]
Linux-PAM's pam_start(3) fails with a bogus error message if passed the
pam_conv argument is NULL.  OpenPAM doesn't care, but to make things
easier for people porting this code to other systems (or -STABLE), use
a dummy struct pam_conv instead of NULL.

Pointed out by: Damien Miller <djm@mindrot.org>

21 years agoBump patch date to 2003-02-01 (the day after I fixed PAM authentication
des [Mon, 3 Feb 2003 11:11:36 +0000 (11:11 +0000)]
Bump patch date to 2003-02-01 (the day after I fixed PAM authentication
for ssh1)

21 years agoPark & Miller PRNG can be safely initialized with any value but 0 and stuck
ache [Mon, 3 Feb 2003 10:29:47 +0000 (10:29 +0000)]
Park & Miller PRNG can be safely initialized with any value but 0 and stuck
at 0 as designed. Its BSD adaptation tries to fight it by mapping 0 to
2147483647 after calculation, but this method not works since 2147483647
seed returns to 0 again on the next interation. Instead of after calculation
mapping, map 0 to another value _before_ calculation, so it never stucks.

21 years agoPark & Miller PRNG can be safely initialized with any value but 0 and stuck
ache [Mon, 3 Feb 2003 10:22:12 +0000 (10:22 +0000)]
Park & Miller PRNG can be safely initialized with any value but 0 and stuck
at 0 as designed. Its BSD adaptation tries to fight it by mapping 0 to
2147483647 after calculation, but this method not works since 2147483647
seed returns to 0 again on the next interation. Instead of after calculation
mapping, map 0 to another value _before_ calculation, so it never stucks.

21 years agoDocument additional behaviour of this function (see
phantom [Mon, 3 Feb 2003 10:10:40 +0000 (10:10 +0000)]
Document additional behaviour of this function (see
rev 1.3 of uthread_attr_get_np.c)

21 years agoImprove pthread_attr_get_np() by enabling it to return thread's real stack
phantom [Mon, 3 Feb 2003 10:08:45 +0000 (10:08 +0000)]
Improve pthread_attr_get_np() by enabling it to return thread's real stack
address instead of specified by pthread_attr_t passed to pthread_create().

Suggested by: deischen

21 years agoIn pam_sm_acct_mgmt(), retrieve the cached credentials before trying to
des [Mon, 3 Feb 2003 09:45:41 +0000 (09:45 +0000)]
In pam_sm_acct_mgmt(), retrieve the cached credentials before trying to
initialize the context.  This way, a failure to initialize the context is
not fatal unless we actually have work to do - because if we don't, we
return PAM_SUCCESS without even trying to initialize the context.

21 years agoWhitespace cleanup
des [Mon, 3 Feb 2003 09:43:28 +0000 (09:43 +0000)]
Whitespace cleanup

21 years agoRemove unnecessary M_NOWAIT.
simokawa [Mon, 3 Feb 2003 09:41:42 +0000 (09:41 +0000)]
Remove unnecessary M_NOWAIT.

21 years agoUse vaccess() instead of rolling our own access checks. This fixes a bug
tjr [Mon, 3 Feb 2003 09:04:34 +0000 (09:04 +0000)]
Use vaccess() instead of rolling our own access checks. This fixes a bug
where requests to open a file in append mode were always denied, and
will also be useful when capabilities and auditing are implemented.

21 years agoAdd missing 'static'.
simokawa [Mon, 3 Feb 2003 08:10:50 +0000 (08:10 +0000)]
Add missing 'static'.

21 years ago- Take malloc type as an argument in fw_xfer_alloc().
simokawa [Mon, 3 Feb 2003 07:33:31 +0000 (07:33 +0000)]
- Take malloc type as an argument in fw_xfer_alloc().
- Fix overwrite problem of freed buffers. It was rare but could happen
when fwohci_arcv() is called before fwohci_txd() is called for
the transcation.
- Drain AT queues and pend AR queues on SID receive rather than BUS reset to
make sure DMA actually stops.
- Do agent reset in sbp_timeout().

21 years agoMake the variable types, the sysctl macros and the sysctl handler for
harti [Mon, 3 Feb 2003 06:50:59 +0000 (06:50 +0000)]
Make the variable types, the sysctl macros and the sysctl handler for
kern.ipc.{maxsockbuf,sockbuf_waste_factor} to agree that those variables
are of type unsigned long.

PR: sparc64/47389
Approved by: jake (mentor)

21 years ago - Make some context switches conditional on SCHED_STRICT_RESCHED. This may
jeff [Mon, 3 Feb 2003 05:30:07 +0000 (05:30 +0000)]
 - Make some context switches conditional on SCHED_STRICT_RESCHED.  This may
   have some negative effect on interactivity but it yields great perf. gains.
   This also brings the conditions under which ULE context switches inline
   with SCHED_4BSD.
 - Define some new kseq_* functions for manipulating the run queue.
 - Add a new kseq member ksq_rslices and ksq_bload.  rslices is the sum of
   the slices of runnable kses.  This will be used for push load balance
   decisions.  bload is the number of threads blocked waiting on IO.

21 years agoRemove another unused Makefile.
trhodes [Mon, 3 Feb 2003 04:15:53 +0000 (04:15 +0000)]
Remove another unused Makefile.

21 years agoRemove mono encodings from vchan format and mixer description. Fixes
orion [Mon, 3 Feb 2003 03:20:27 +0000 (03:20 +0000)]
Remove mono encodings from vchan format and mixer description.  Fixes
mono formats at 44.1kHz playing at double speed when vchans are
enabled.

21 years ago - Stop abusing oncpu for our cpu binding. Define a scheduler local element
jeff [Mon, 3 Feb 2003 02:26:28 +0000 (02:26 +0000)]
 - Stop abusing oncpu for our cpu binding.  Define a scheduler local element
   in the kse datastructure called ke_cpu.  This is the cpu which we are
   currently bound to.  Some flags may be added later to support hard binding.

21 years agoAdd a man page.
obrien [Mon, 3 Feb 2003 01:59:27 +0000 (01:59 +0000)]
Add a man page.

21 years agoPrint ac97 name/id on normal boot.
orion [Mon, 3 Feb 2003 01:38:43 +0000 (01:38 +0000)]
Print ac97 name/id on normal boot.

Fix typo reported in pr misc/47794.

21 years agoMake WARNS=5 clean even on 64-bit platforms.
obrien [Mon, 3 Feb 2003 01:30:47 +0000 (01:30 +0000)]
Make WARNS=5 clean even on 64-bit platforms.

21 years agoMake WARNS=5 on i386, (WARNS=1 on 64-bit platforms).
obrien [Mon, 3 Feb 2003 01:22:58 +0000 (01:22 +0000)]
Make WARNS=5 on i386, (WARNS=1 on 64-bit platforms).

Submitted by: dwmalone (tweaked by me)

21 years agoDon't use the 'c' partition for mounting root. A disklabel is very
marcel [Mon, 3 Feb 2003 01:10:01 +0000 (01:10 +0000)]
Don't use the 'c' partition for mounting root. A disklabel is very
likely not present under the simulator. If multiple partitions are
present on the virtual disk, then the 'a' partition would be the
most logical choice. Nowadays partitions are GPT based, which would
make the assumption of a disklabel even more questionable. Given
all the possible scenarios, assuming a raw "device" seems best.

21 years ago - Make allpmaps static.
alc [Mon, 3 Feb 2003 00:05:11 +0000 (00:05 +0000)]
 - Make allpmaps static.
 - Use atomic subtract to update the global wired pages count.  (See
   also vm/vm_page.c revision 1.233.)
 - Assert that the page queue lock is held in pmap_remove_entry().

21 years agoSome small enhancments
ambrisko [Sun, 2 Feb 2003 22:49:31 +0000 (22:49 +0000)]
Some small enhancments
      - Implement TX power control with hints to code in Linux driver
by Douglas S. J. De Couto <decouto@lcs.mit.edu>
      - Update ifmedia support to use ieee80211_rate2media and
ieee80211_media2rate.  Note doesn't work with stock ifconfig
since there seems to be an issue with the setmedia code.
"ancontrol -t" works now, before it did nothing.
      - Fix spelling error in header.

21 years agoTry to tighten up the types a little bit to help debugging with GDB.
obrien [Sun, 2 Feb 2003 22:16:05 +0000 (22:16 +0000)]
Try to tighten up the types a little bit to help debugging with GDB.

21 years agoAdd the TCP flags to the log message whenever log_in_vain is 1, not
cjc [Sun, 2 Feb 2003 22:06:56 +0000 (22:06 +0000)]
Add the TCP flags to the log message whenever log_in_vain is 1, not
just when set to 2.

PR: kern/43348
MFC after: 5 days

21 years agoSet si_drv1 to our softc for all the six dev_t's we create for a serial port.
phk [Sun, 2 Feb 2003 21:25:22 +0000 (21:25 +0000)]
Set si_drv1 to our softc for all the six dev_t's we create for a serial port.

21 years agoAdd OPIE and PAM libs to the mix.
obrien [Sun, 2 Feb 2003 21:11:15 +0000 (21:11 +0000)]
Add OPIE and PAM libs to the mix.

21 years agoOPIE and PAM bits to agument LukeMftpd.
obrien [Sun, 2 Feb 2003 21:06:10 +0000 (21:06 +0000)]
OPIE and PAM bits to agument LukeMftpd.

Submitted by: mikeh (reworked by me)

21 years agoMerge our rev 1.2 PAM support into 'nbsd_20030122'.
obrien [Sun, 2 Feb 2003 21:03:28 +0000 (21:03 +0000)]
Merge our rev 1.2 PAM support into 'nbsd_20030122'.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r110242,
obrien [Sun, 2 Feb 2003 20:58:44 +0000 (20:58 +0000)]
This commit was generated by cvs2svn to compensate for changes in r110242,
which included commits to RCS files with non-trunk default branches.

21 years agoImport of LukeM's ftpd taken from the NetBSD CVS repo on 22-Jan-2003.
obrien [Sun, 2 Feb 2003 20:58:44 +0000 (20:58 +0000)]
Import of LukeM's ftpd taken from the NetBSD CVS repo on 22-Jan-2003.
This includes the DoS fix of rev 1.133 libexec/ftpd/ftpd.c.

21 years agoTweak the definition of the EV_SET macro so that it evaluates each
nectar [Sun, 2 Feb 2003 19:39:51 +0000 (19:39 +0000)]
Tweak the definition of the EV_SET macro so that it evaluates each
of its arguments exactly once.  (Previously it evaluated the first
argument six times.)

MFC after: 1 week

21 years agoOpenPAMify.
des [Sun, 2 Feb 2003 18:43:58 +0000 (18:43 +0000)]
OpenPAMify.

21 years agoEnable pam_krb5 for sshd. I've had this in my tree for ages.
des [Sun, 2 Feb 2003 18:41:26 +0000 (18:41 +0000)]
Enable pam_krb5 for sshd.  I've had this in my tree for ages.

21 years agoA minor stylistic change to make it more clear to lint-like tools.
phk [Sun, 2 Feb 2003 18:15:28 +0000 (18:15 +0000)]
A minor stylistic change to make it more clear to lint-like tools.

21 years agoAdd BCTV3/PCI entry.
orion [Sun, 2 Feb 2003 17:46:00 +0000 (17:46 +0000)]
Add BCTV3/PCI entry.

Rename functions bctv_* to bctv2_* and macros BCTV_* to BCTV2_* for
coexistance of BCTV2 and BCTV3.

Rename CARD_IO_GV to CARD_TO_BCTV2.

Add auto recognition of BCTV3/PCI.

PR:             i386/43879
Submitted by:   MOROHOSHI Akihiko <moro@remus.dti.ne.jp>
Approved by:    roger
MFC after:      6 days

21 years agoFor some combinations of variable sizes and RAND_MAX value rand_r()
ache [Sun, 2 Feb 2003 14:27:51 +0000 (14:27 +0000)]
For some combinations of variable sizes and RAND_MAX value rand_r()
may store less amount bits for seed, than available. Fix it.

21 years agochase more of the MIN/MAX mess. *sigh*
alfred [Sun, 2 Feb 2003 13:52:25 +0000 (13:52 +0000)]
chase more of the MIN/MAX mess. *sigh*

21 years agoCatch more uses of MIN().
alfred [Sun, 2 Feb 2003 13:30:00 +0000 (13:30 +0000)]
Catch more uses of MIN().

21 years agoremove MIN now that it's a standard kernel define
alfred [Sun, 2 Feb 2003 13:21:05 +0000 (13:21 +0000)]
remove MIN now that it's a standard kernel define

21 years agoConsolidate MIN/MAX macros into one place (param.h).
alfred [Sun, 2 Feb 2003 13:17:30 +0000 (13:17 +0000)]
Consolidate MIN/MAX macros into one place (param.h).

Submitted by: Hiten Pandya <hiten@unixdaemons.com>

21 years agoCorrect error message.
nyan [Sun, 2 Feb 2003 13:06:18 +0000 (13:06 +0000)]
Correct error message.

21 years agoAdd a bio_disk pointer for use between geom_disk and the device drivers.
phk [Sun, 2 Feb 2003 11:59:54 +0000 (11:59 +0000)]
Add a bio_disk pointer for use between geom_disk and the device drivers.

21 years agoWe don't need sscopen() and sscclose().
phk [Sun, 2 Feb 2003 10:22:34 +0000 (10:22 +0000)]
We don't need sscopen() and sscclose().

Register sscstrategy directly, instead of using a cdevsw{} for the purpose.

Tested by: marcel

21 years agoExport IA32 from opt_ia32.h to assembly so that we can eliminate
marcel [Sun, 2 Feb 2003 09:07:15 +0000 (09:07 +0000)]
Export IA32 from opt_ia32.h to assembly so that we can eliminate
saving and restoring ia32 specific registers when switching
context and ia32 support has not been compiled-in. The primary
reason for this change is that one of the ia32 registers (ar.fcr)
is wrongly marked as invalid by the simulator. Now that we avoid
using the register when possible, usability is improved. The
secundary reason is that it saves us 7 loads and stores.

Note that the PCB will continue to have room for these registers,
irrespective of the IA32 option. There are no benefits that make
it worthwhile.

21 years agoUse hz if stathz is zero. Adopted from sched_4bsd.
scottl [Sun, 2 Feb 2003 08:24:32 +0000 (08:24 +0000)]
Use hz if stathz is zero.  Adopted from sched_4bsd.

21 years ago - It's more accurate to say that vm_paging_needed() returns TRUE
alc [Sun, 2 Feb 2003 07:16:40 +0000 (07:16 +0000)]
 - It's more accurate to say that vm_paging_needed() returns TRUE
   than a positive number.
 - In pagedaemon_wakeup(), set vm_pages_needed to 1 rather than
   incrementing it to accomplish the same.

21 years agoUnlock the mutex in the error case in wi_init. Otherwise we can return
imp [Sun, 2 Feb 2003 06:35:46 +0000 (06:35 +0000)]
Unlock the mutex in the error case in wi_init.  Otherwise we can return
from an ioctl with the lock held.

Submitted by: iedowse (by way of sam)

21 years agoAdd device zs to GENERIC on powerpc.
benno [Sun, 2 Feb 2003 06:11:23 +0000 (06:11 +0000)]
Add device zs to GENERIC on powerpc.

21 years agoRegenerate after fixing duplicate device entries.
marcel [Sun, 2 Feb 2003 01:58:28 +0000 (01:58 +0000)]
Regenerate after fixing duplicate device entries.

21 years agoUnbreak kernel builds caused by what appears to be a merge conflict.
marcel [Sun, 2 Feb 2003 01:57:31 +0000 (01:57 +0000)]
Unbreak kernel builds caused by what appears to be a merge conflict.
This change has been tested.

21 years ago - Convert vm_pageout()'s tsleep()s to msleep()s with the page queue lock.
alc [Sun, 2 Feb 2003 01:11:21 +0000 (01:11 +0000)]
 - Convert vm_pageout()'s tsleep()s to msleep()s with the page queue lock.

21 years agoProperly capitalize the subsection title on setting MAC labels.
chris [Sat, 1 Feb 2003 23:11:57 +0000 (23:11 +0000)]
Properly capitalize the subsection title on setting MAC labels.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoRegen
sanpei [Sat, 1 Feb 2003 23:09:33 +0000 (23:09 +0000)]
Regen

21 years agoFix table formatting for printed man pages.
chris [Sat, 1 Feb 2003 23:06:30 +0000 (23:06 +0000)]
Fix table formatting for printed man pages.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoAdd Genesys Logic products
sanpei [Sat, 1 Feb 2003 22:50:50 +0000 (22:50 +0000)]
Add Genesys Logic products
- GL650 HUB
- GL641USB CompactFlash Card Reade
- GL641USB USB-IDE Bridge
Add MELCO product
- USB-IDE Bridge: DUB-PxxG

21 years agoRemove special casing for running in the simulator from the kernel
marcel [Sat, 1 Feb 2003 22:50:09 +0000 (22:50 +0000)]
Remove special casing for running in the simulator from the kernel
and instead add platform, firmware and EFI stubs to the loader.
The net effect of this change is that besides a special console and
disk driver, the kernel has no knowledge of the simulator. This has
the following advantages:
o  Simulator support is much harder to break,
o  It's easier to make use of more feature complete simulators.
   This would only need a change in the simulator specific loader,
o  Running SMP kernels within the simulator. Note that ski at this
   time does not simulate IPIs, so there's no way to start APs.

The platform, firmware and EFI stubs describe the following hardware:
o  4 CPU Itanium,
o  128 MB RAM within the 4GB address space,
o  64 MB RAM above the 4GB address space.

NOTE: The stubs in the skiloader describe a machine that should in
parts be defined by the simulator. Things like processor interrupt
block and AP wakeup vector cannot be choosen at random because they
require interpretation by the simulator. Currently the simulator is
ignorant of this.

This change introduces an unofficial SSC call SSC_SAL_SET_VECTORS
which is ignored by the simulator.

Tested with: ski (version 0.943 for linux)

21 years agoNODEVFS cleanup: Goodbye, and good riddance!
phk [Sat, 1 Feb 2003 22:22:50 +0000 (22:22 +0000)]
NODEVFS cleanup:  Goodbye, and good riddance!

21 years agoNODEVFS cleanup: Don't ever install MAKEDEV anymore.
phk [Sat, 1 Feb 2003 22:22:06 +0000 (22:22 +0000)]
NODEVFS cleanup: Don't ever install MAKEDEV anymore.

21 years agoThis took longer to test than expected:
phk [Sat, 1 Feb 2003 22:12:50 +0000 (22:12 +0000)]
This took longer to test than expected:

NODEVFS cleanup: don't bother with MAKEDEV.

Look for /boot/boot first, and arrange disklabel arguments accordingly.

21 years ago - Remove (some) unnecessary explicit initializations to zero.
alc [Sat, 1 Feb 2003 21:55:30 +0000 (21:55 +0000)]
 - Remove (some) unnecessary explicit initializations to zero.
 - Style changes to vm_pageout(): declarations and white-space.

21 years agoSSC calls use break immediate 0x80000. 0x80001 only works for
marcel [Sat, 1 Feb 2003 21:38:36 +0000 (21:38 +0000)]
SSC calls use break immediate 0x80000. 0x80001 only works for
break.i. Ski is rather broken in this respect.

21 years ago - Convert the tsleep()s in vm_wait() and vm_waitpfault() to msleep()s
alc [Sat, 1 Feb 2003 21:18:16 +0000 (21:18 +0000)]
 - Convert the tsleep()s in vm_wait() and vm_waitpfault() to msleep()s
   with the page queue lock.
 - Assert that the page queue lock is held in vm_page_free_wakeup().

21 years agoRemove this Makefile, these docs are built by the Makefiles in src/share/doc/smm.
trhodes [Sat, 1 Feb 2003 20:34:04 +0000 (20:34 +0000)]
Remove this Makefile, these docs are built by the Makefiles in src/share/doc/smm.

21 years agoPut replace spaces with tabs in keeping with the rest of the file.
joe [Sat, 1 Feb 2003 18:45:18 +0000 (18:45 +0000)]
Put replace spaces with tabs in keeping with the rest of the file.

21 years ago-kthread has been dead these many years. This PR has languished for a
imp [Sat, 1 Feb 2003 18:39:32 +0000 (18:39 +0000)]
-kthread has been dead these many years.  This PR has languished for a
long time.  Go ahead and commit the patch since a co-worker has been
working up a code patch in this area and wondered why it hasn't been
committed yet.

PR: 35800

21 years agoadd PST to i386 notes.
phk [Sat, 1 Feb 2003 17:23:20 +0000 (17:23 +0000)]
add PST to i386 notes.

21 years agoCatch up with new categories and note that ports-base is essential without
schweikh [Sat, 1 Feb 2003 17:16:59 +0000 (17:16 +0000)]
Catch up with new categories and note that ports-base is essential without
ports-all.

PR: 47774
Submitted by: Sergei Kolobov <sergei@kolobov.com>
MFC after: 3 days

21 years agoAdd new categories hungarian and portuguese.
schweikh [Sat, 1 Feb 2003 17:01:45 +0000 (17:01 +0000)]
Add new categories hungarian and portuguese.

PR: 47772
Submitted by: Sergei Kolobov <sergei@kolobov.com>
MFC after: 3 days

21 years agoDefine new malloc type M_FW and use it.
simokawa [Sat, 1 Feb 2003 15:04:33 +0000 (15:04 +0000)]
Define new malloc type M_FW and use it.

21 years agoFix some typos in 3 comments.
gj [Sat, 1 Feb 2003 14:43:24 +0000 (14:43 +0000)]
Fix some typos in 3 comments.

Reported by: marius@alchemy.franken.de

21 years ago- replace timeout with callout_*.
simokawa [Sat, 1 Feb 2003 14:42:49 +0000 (14:42 +0000)]
- replace timeout with callout_*.
- replace TAILQ with STAILQ for device list.
- some clean up.

21 years agoReversion of commit by Davidxu plus fixes since applied.
julian [Sat, 1 Feb 2003 12:17:09 +0000 (12:17 +0000)]
Reversion of commit by Davidxu plus fixes since applied.

I'm not convinced there is anything major wrong with the patch but
them's the rules..

I am using my "David's mentor" hat to revert this as he's
offline for a while.

21 years agoCheck status FIFO more closely to report error.
simokawa [Sat, 1 Feb 2003 12:06:57 +0000 (12:06 +0000)]
Check status FIFO more closely to report error.

21 years agoEliminate the sc_openmask, ccdopen() and ccdclose() functions, we
phk [Sat, 1 Feb 2003 11:17:22 +0000 (11:17 +0000)]
Eliminate the sc_openmask, ccdopen() and ccdclose() functions, we
can use the flag maintained by geom_disk.c

Having only a strategy method to intialize, don't waste space using
a cdevsw structure to do so.

21 years agoNO_GEOM cleanup: don't #include <sys/diskslice.h>
phk [Sat, 1 Feb 2003 10:26:23 +0000 (10:26 +0000)]
NO_GEOM cleanup: don't #include <sys/diskslice.h>

21 years agoUnder #ifdef DIAGNOSTIC, fill malloc(9) allocations which do not have
phk [Sat, 1 Feb 2003 10:07:49 +0000 (10:07 +0000)]
Under #ifdef DIAGNOSTIC, fill malloc(9) allocations which do not have
M_ZERO specified with 0x70.  (malloc_flags=J for the kernel :-)

21 years agoUnder DIAGNOSTIC, only report expensive timeouts if they are more expensive
phk [Sat, 1 Feb 2003 10:06:40 +0000 (10:06 +0000)]
Under DIAGNOSTIC, only report expensive timeouts if they are more expensive
than the last on we reported.

21 years agoAdd basic support for device wiring down to specific (CAM)
simokawa [Sat, 1 Feb 2003 08:55:33 +0000 (08:55 +0000)]
Add basic support for device wiring down to specific (CAM)
target id. It still needs to be hard-coded now but will
be changed to be configured by tunables or device hints.