]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoMore prep work for the errata file.
bmah [Wed, 15 Jan 2003 15:15:09 +0000 (15:15 +0000)]
More prep work for the errata file.

Simulate the effects of a version number bump for just the errata
document.  This change will be backed out in a few days, during the
post release version number bump for the release documentation.

21 years agoPrepare to make the errata file on HEAD serve as the on-line (mutable)
bmah [Wed, 15 Jan 2003 15:10:43 +0000 (15:10 +0000)]
Prepare to make the errata file on HEAD serve as the on-line (mutable)
errata document for 5.0-RELEASE, since that is the development branch
on which 5.0-RELEASE is based.

This change consists primarily of whitespace diff reductions to
RELENG_4 plus deleting some text that isn't really applicable.

21 years agoDon't forget to disconnect object from class.
davidxu [Wed, 15 Jan 2003 14:58:07 +0000 (14:58 +0000)]
Don't forget to disconnect object from class.

21 years agoMerged from sys/isa/syscons_isa.c revision 1.21.
nyan [Wed, 15 Jan 2003 13:12:12 +0000 (13:12 +0000)]
Merged from sys/isa/syscons_isa.c revision 1.21.

21 years agoResolve conflicts.
murray [Wed, 15 Jan 2003 10:47:32 +0000 (10:47 +0000)]
Resolve conflicts.

21 years agoResolve conflicts.
murray [Wed, 15 Jan 2003 10:44:05 +0000 (10:44 +0000)]
Resolve conflicts.

(import approved by: re)

21 years agoThis commit was generated by cvs2svn to compensate for changes in r109298,
murray [Wed, 15 Jan 2003 10:31:26 +0000 (10:31 +0000)]
This commit was generated by cvs2svn to compensate for changes in r109298,
which included commits to RCS files with non-trunk default branches.

21 years agoImport ISC DHCP 3.0.1 RC10 client.
murray [Wed, 15 Jan 2003 10:31:26 +0000 (10:31 +0000)]
Import ISC DHCP 3.0.1 RC10 client.

21 years agoAdd a missing word.
tjr [Wed, 15 Jan 2003 09:48:15 +0000 (09:48 +0000)]
Add a missing word.

21 years agoAdd missing . to mandoc macro
keramida [Wed, 15 Jan 2003 08:24:45 +0000 (08:24 +0000)]
Add missing . to mandoc macro

PR: docs/47092
Submitted by: SUZUKI Koichi <koich@cac.co.jp>

21 years ago - Add mozilla to the package list since it would be nice to have a browser
scottl [Wed, 15 Jan 2003 08:11:35 +0000 (08:11 +0000)]
 - Add mozilla to the package list since it would be nice to have a browser
   that a) was from this century, b) is not Opera, and c) doesn't require
   KDE.
 - Don't include the compat22 libraries unless the PKG_ARCH is i386.  Same
   goes for compat3x and compat4x for i386 and alpha.
 - Define PKG_ARCH and make it overridable in the environment.

21 years agoSlightly better grammar.
chris [Wed, 15 Jan 2003 06:11:13 +0000 (06:11 +0000)]
Slightly better grammar.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoDV_PAL shouldn't be here.
simokawa [Wed, 15 Jan 2003 05:28:50 +0000 (05:28 +0000)]
DV_PAL shouldn't be here.

21 years agoAdd DV_PAL for PAL users.
simokawa [Wed, 15 Jan 2003 05:26:23 +0000 (05:26 +0000)]
Add DV_PAL for PAL users.

21 years agoMinimally document hw.syscons.sc_no_suspend_vtswitch.
mdodd [Wed, 15 Jan 2003 05:26:10 +0000 (05:26 +0000)]
Minimally document hw.syscons.sc_no_suspend_vtswitch.

Requested by:  Nate Lawson <nate@root.org>

21 years agoFix IT DMA underun while bus reset.
simokawa [Wed, 15 Jan 2003 04:21:16 +0000 (04:21 +0000)]
Fix IT DMA underun while bus reset.
- Reduce number of DELAY's

21 years agoMake the SC_NO_SUSPEND_VTYSWITCH kernel option available as a loader
mdodd [Wed, 15 Jan 2003 03:45:27 +0000 (03:45 +0000)]
Make the SC_NO_SUSPEND_VTYSWITCH kernel option available as a loader
tunable and sysctl (hw.syscons.sc_no_suspend_vtswitch).

21 years agosigh... revert the previous change; this isn't getpmac
chris [Wed, 15 Jan 2003 03:13:51 +0000 (03:13 +0000)]
sigh... revert the previous change; this isn't getpmac

Pointy hat: chris

21 years agoCross-reference the appropriate mac library functions.
chris [Wed, 15 Jan 2003 03:13:50 +0000 (03:13 +0000)]
Cross-reference the appropriate mac library functions.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoCross-reference mac(4).
chris [Wed, 15 Jan 2003 03:09:01 +0000 (03:09 +0000)]
Cross-reference mac(4).

Sponsored by: DARPA, Network Associates Laboratories

21 years agoDocument the named semaphore functions. This could still use a bit
tjr [Wed, 15 Jan 2003 03:07:40 +0000 (03:07 +0000)]
Document the named semaphore functions. This could still use a bit
of polishing.

21 years agoActually add mac_prepare.3.
chris [Wed, 15 Jan 2003 03:05:21 +0000 (03:05 +0000)]
Actually add mac_prepare.3.

Sponsored by: DARPA, Network Associates Laboratories

21 years agoCross-reference mac(4)
chris [Wed, 15 Jan 2003 03:03:05 +0000 (03:03 +0000)]
Cross-reference mac(4)

Sponsored by: DARPA, Network Associates Laboratories

21 years agos/SEE_ALSO/SEE ALSO/
chris [Wed, 15 Jan 2003 03:02:30 +0000 (03:02 +0000)]
s/SEE_ALSO/SEE ALSO/
Cross-reference mac(4) and mac(9)

Sponsored by: DARPA, Network Associates Laboratories

21 years agoUpdate cross-references to include mac(4).
chris [Wed, 15 Jan 2003 02:59:36 +0000 (02:59 +0000)]
Update cross-references to include mac(4).

Sponsored by: DARPA, Network Associates Laboratories

21 years agoWelcome mac(4), the central source of information for all things MAC.
chris [Wed, 15 Jan 2003 02:57:39 +0000 (02:57 +0000)]
Welcome mac(4), the central source of information for all things MAC.

Sponsored by:   DARPA, Network Associates Laboratories

21 years ago- GC a few more hand-rolled 'abs' macros.
mdodd [Wed, 15 Jan 2003 02:15:57 +0000 (02:15 +0000)]
- GC a few more hand-rolled 'abs' macros.
- GC a few hand-rolled min()/max() macros while I'm here.

21 years ago- Add inline functions for {ll,l,}abs() to libkern.
mdodd [Wed, 15 Jan 2003 02:02:33 +0000 (02:02 +0000)]
- Add inline functions for {ll,l,}abs() to libkern.
- Remove hand rolled abs() functions.

21 years agoActivate getpmac.8
chris [Wed, 15 Jan 2003 01:43:09 +0000 (01:43 +0000)]
Activate getpmac.8

21 years agoDocument getpmac.
chris [Wed, 15 Jan 2003 01:40:35 +0000 (01:40 +0000)]
Document getpmac.

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

21 years agos/^Sh/.Sh/ (macros begin with .)
chris [Wed, 15 Jan 2003 01:20:58 +0000 (01:20 +0000)]
s/^Sh/.Sh/ (macros begin with .)

21 years agoDocument setpmac.
chris [Wed, 15 Jan 2003 01:17:12 +0000 (01:17 +0000)]
Document setpmac.

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

21 years agoo Document mac_prepare() and associated functions
chris [Wed, 15 Jan 2003 00:45:31 +0000 (00:45 +0000)]
o Document mac_prepare() and associated functions
o Link mac_get_pid.3 to mac_get.3
o Update SEE ALSO to refer to mac_prepare, and added missing references
o Remove clause #3 on my work
o Update mac_get.3 for the updated MAC API

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

21 years agoMFR500: Fix up ACPI-disabling hint text.
bmah [Tue, 14 Jan 2003 23:37:14 +0000 (23:37 +0000)]
MFR500:  Fix up ACPI-disabling hint text.

21 years agoo Remove clause #3
chris [Tue, 14 Jan 2003 23:20:40 +0000 (23:20 +0000)]
o Remove clause #3
o Document mac_set_link().

Sponsored by: DARPA, Network Associates Labs

21 years agoSuspend/resume support (mostly for MiniPCI Prism2.5 boards).
mdodd [Tue, 14 Jan 2003 23:19:32 +0000 (23:19 +0000)]
Suspend/resume support (mostly for MiniPCI Prism2.5 boards).

Reviewed by:  imp

21 years agoSome USB devices are not prepared to deal with a single byte string
joe [Tue, 14 Jan 2003 23:07:43 +0000 (23:07 +0000)]
Some USB devices are not prepared to deal with a single byte string
descriptor request, which usbd_get_string_desc() uses to get the
length of a descriptor.  One device for instance returns a full 8
byte long packet instead which confuses the rest of the stack and
leads to the USB port being reset.  The fix is to instead request
two bytes, but not to complain if we only get one.

Submitted by: kan
MFC after: 3 days

21 years agoAdd a very simple but functional GEOM mirror class.
phk [Tue, 14 Jan 2003 22:44:48 +0000 (22:44 +0000)]
Add a very simple but functional GEOM mirror class.

This is committed more as an instructive tool than as a production
facility, but this will change over time.

21 years agoAdd missing #include
dillon [Tue, 14 Jan 2003 21:31:31 +0000 (21:31 +0000)]
Add missing #include

Submitted by: "Sam Leffler" <sam@errno.com>

21 years agoNow that we have non-geom_disk based drivers, we need to cover for those,
phk [Tue, 14 Jan 2003 21:31:00 +0000 (21:31 +0000)]
Now that we have non-geom_disk based drivers, we need to cover for those,
in case they return EOPNOTSUPP on an ioctl.

Found by: jhb

21 years agoMinor hardware note updates: xl(4) 3C555, an(4) mini-PCI.
bmah [Tue, 14 Jan 2003 20:31:31 +0000 (20:31 +0000)]
Minor hardware note updates:  xl(4) 3C555, an(4) mini-PCI.

21 years agoNew release notes: pc98 SCSI MO support in boot loader (+MFC), make(1)
bmah [Tue, 14 Jan 2003 20:29:18 +0000 (20:29 +0000)]
New release notes: pc98 SCSI MO support in boot loader (+MFC), make(1)
-C (+MFC), rarpd(8) -t (+MFC).

MFCs noted:  ubsa driver.

21 years agoFor the cpu throttling message, s/enabled/available
njl [Tue, 14 Jan 2003 19:39:41 +0000 (19:39 +0000)]
For the cpu throttling message, s/enabled/available

Requested by: many

21 years agoIntroduce the ability to flag a sysctl for operation at secure level 2 or 3
dillon [Tue, 14 Jan 2003 19:35:33 +0000 (19:35 +0000)]
Introduce the ability to flag a sysctl for operation at secure level 2 or 3
in addition to secure level 1.  The mask supports up to a secure level of 8
but only add defines through CTLFLAG_SECURE3 for now.

As per the missif in the log entry for 1.11 of ip_fw2.c which added the
secure flag to the IPFW sysctl's in the first place, change the secure
level requirement from 1 to 3 now that we have support for it.

Reviewed by: imp
With Design Suggestions by: imp

21 years agoRemove bogus locking from dc(4). Instead, move interrupt allocation
njl [Tue, 14 Jan 2003 19:31:27 +0000 (19:31 +0000)]
Remove bogus locking from dc(4).  Instead, move interrupt allocation
and ether_ifattach() to end.  This fixes a "could sleep" case and
simplifies error exit cases as well.  Also be sure to set errno
and clean up resources in !mac error case.

Tested by: Ryan Beasley

21 years agoComment correction. NetBSD uses 'readblksize' as the multiplier for the
dillon [Tue, 14 Jan 2003 19:26:08 +0000 (19:26 +0000)]
Comment correction.  NetBSD uses 'readblksize' as the multiplier for the
-r option, which defaults to 32K (not 512 bytes).  Still just as bad
from a UI standpoint, though.

21 years agoNew release notes: dump(8) -C, newsyslog(8) W flag.
bmah [Tue, 14 Jan 2003 19:06:11 +0000 (19:06 +0000)]
New release notes:  dump(8) -C, newsyslog(8) W flag.

21 years agoMFen: add changes between 1.469->1.470 forgotten in my previous commit.
hrs [Tue, 14 Jan 2003 17:37:52 +0000 (17:37 +0000)]
MFen: add changes between 1.469->1.470 forgotten in my previous commit.

21 years agoDon't call destroy_dev it a channel has children.
cognet [Tue, 14 Jan 2003 17:13:52 +0000 (17:13 +0000)]
Don't call destroy_dev it a channel has children.
vchan creation doesn't lead to /dev entry creation if the new vchan is the
first child of a channel,
This fix a panic that happens when loading a sound driver module, creating
vchans and unloading the driver.

Approved by: cg
MFC after: 3 days

21 years agoSave 4 more bytes by not initializing opts to 0. This moves it from
imp [Tue, 14 Jan 2003 16:33:37 +0000 (16:33 +0000)]
Save 4 more bytes by not initializing opts to 0.  This moves it from
the data section to the bss section givig us initialization for free.

Noticed by: bde

21 years ago o Typo/Grammar fixes
mtm [Tue, 14 Jan 2003 16:02:56 +0000 (16:02 +0000)]
o Typo/Grammar fixes
o Added mini-function to correctly handle singular/plural
  of words ending in 'ly'

Approved by: markm (mentor)
Not objected to by: -audit

21 years agoAdd rc_debug knob to rc.conf. The code for it has been in rc.subr for
mtm [Tue, 14 Jan 2003 15:50:17 +0000 (15:50 +0000)]
Add rc_debug knob to rc.conf. The code for it has been in rc.subr for
some time now.
Document all knobs introduced by rc.d

Approved by: markm (mentor)
Reviewd by: gordon (earlier revision)

21 years agoFinish merging in rev. 1.124 of rc.network, so that natd can be used
mtm [Tue, 14 Jan 2003 15:43:02 +0000 (15:43 +0000)]
Finish merging in rev. 1.124 of rc.network, so that natd can be used
withough the $natd_interface having to be explicitly specified on the
command line.

Approved by: markm (mentor)
Submitted by: Aaron D. Gifford <agifford@infowest.com>
PR: conf/47024

MFC: upon re approval

21 years agoMake the base pcib_route_interrupt method available to other pci-pci bridge
benno [Tue, 14 Jan 2003 11:37:56 +0000 (11:37 +0000)]
Make the base pcib_route_interrupt method available to other pci-pci bridge
sub-classes.

This allows the powerpc kernel to build again.

Forgotten by: benno
Spotted by: grehan

21 years agoFix for FreeBSD-4.
simokawa [Tue, 14 Jan 2003 08:35:45 +0000 (08:35 +0000)]
Fix for FreeBSD-4.
- Protect whole fw_tbuf_update() with splfw().

21 years agoBack out previous; sharing semaphores between processes only works
tjr [Tue, 14 Jan 2003 07:14:06 +0000 (07:14 +0000)]
Back out previous; sharing semaphores between processes only works
in certain special cases.

21 years agoMFbed: MFen: id attributes for sect1 entities of the release notes
ue [Tue, 14 Jan 2003 07:12:31 +0000 (07:12 +0000)]
MFbed: MFen: id attributes for sect1 entities of the release notes

21 years ago - Update vm_pageout_deficit using atomic operations. It's a simple
alc [Tue, 14 Jan 2003 06:57:03 +0000 (06:57 +0000)]
 - Update vm_pageout_deficit using atomic operations.  It's a simple
   counter outside the scope of existing locks.
 - Eliminate a redundant clearing of vm_pageout_deficit.

21 years agoSharing semaphores between processes works now, so remove the stale comments
tjr [Tue, 14 Jan 2003 04:12:33 +0000 (04:12 +0000)]
Sharing semaphores between processes works now, so remove the stale comments
about it always returning EPERM. Document that ENFILE occurs when the
limit on kernel semaphores is reached.

21 years agoCross reference lio_listio(2).
tjr [Tue, 14 Jan 2003 03:42:16 +0000 (03:42 +0000)]
Cross reference lio_listio(2).

21 years agoCross reference sem(4) so users know which kernel options are required
tjr [Tue, 14 Jan 2003 03:39:09 +0000 (03:39 +0000)]
Cross reference sem(4) so users know which kernel options are required
to use these semaphore functions.

21 years agoAdd the newly created semaphore to the named semaphore list in sem_open()
tjr [Tue, 14 Jan 2003 03:36:45 +0000 (03:36 +0000)]
Add the newly created semaphore to the named semaphore list in sem_open()
so that multiple opens of the same semaphore without an intervening
sem_close() return the same object, and so that sem_close() does not
segfault while trying to remove the item from the list.

21 years agoManual page for sem.ko/options P1003_1B_SEMAPHORES, similar to aio(4).
tjr [Tue, 14 Jan 2003 02:57:54 +0000 (02:57 +0000)]
Manual page for sem.ko/options P1003_1B_SEMAPHORES, similar to aio(4).

21 years agoIncluding <time.h> before <aio.h> has not been necessary for a while now.
tjr [Tue, 14 Jan 2003 02:37:06 +0000 (02:37 +0000)]
Including <time.h> before <aio.h> has not been necessary for a while now.

21 years agoMake vm_pageout_page_free() static.
alc [Tue, 14 Jan 2003 02:28:39 +0000 (02:28 +0000)]
Make vm_pageout_page_free() static.

21 years agoNote that the Dlink Air 660 is supported.
trhodes [Tue, 14 Jan 2003 01:14:08 +0000 (01:14 +0000)]
Note that the Dlink Air 660 is supported.

21 years agoAdd an entry for the Dlink Air 660 Wireless PC Card.
trhodes [Tue, 14 Jan 2003 00:50:30 +0000 (00:50 +0000)]
Add an entry for the Dlink Air 660 Wireless PC Card.

PR: 46977
Submitted by: gioria
Approved by: imp

21 years agoAdd my birthday
brueffer [Tue, 14 Jan 2003 00:18:06 +0000 (00:18 +0000)]
Add my birthday

Approved by: ceri (mentor)

21 years agoAdd a mistakenly removed part.
hrs [Mon, 13 Jan 2003 23:58:17 +0000 (23:58 +0000)]
Add a mistakenly removed part.

21 years agoPrevent infinite substitution of the empty string by forcing non-
marcel [Mon, 13 Jan 2003 23:53:46 +0000 (23:53 +0000)]
Prevent infinite substitution of the empty string by forcing non-
global substitution. In general it's a makefile bug to globally
substitute the empty string, but it's a bug in make(1) if a bug
in the makefile yields an infinite running time of make(1).

Not objected to by: arch@

21 years agoSince reseting the SCSI busses via the passthrough interface usually
scottl [Mon, 13 Jan 2003 23:51:14 +0000 (23:51 +0000)]
Since reseting the SCSI busses via the passthrough interface usually
confuses the controller, tell CAM not to do it.  Also report the
correct error condition to CAM when it tries to probe a target that
doesn't exists.
This should make the CAM interface less risky to use.

MFC After: 3 days

21 years agoMerge the following from the English version:
hrs [Mon, 13 Jan 2003 23:37:21 +0000 (23:37 +0000)]
Merge the following from the English version:

1.444 -> 1.475 relnotes/common/new.sgml

21 years agoIt is possible for an active aio to prevent shared memory from being
dillon [Mon, 13 Jan 2003 23:04:32 +0000 (23:04 +0000)]
It is possible for an active aio to prevent shared memory from being
dereferenced when a process exits due to the vmspace ref-count being
bumped.  Change shmexit() and shmexit_myhook() to take a vmspace instead
of a process and call it in vmspace_dofree().  This way if it is missed
in exit1()'s early-resource-free it will still be caught when the zombie is
reaped.

Also fix a potential race in shmexit_myhook() by NULLing out
vmspace->vm_shm prior to calling shm_delete_mapping() and free().

MFC after: 7 days

21 years agoApply bandaid to bring svr4_sys_waitsys() in line with exit1(). This
dillon [Mon, 13 Jan 2003 22:44:23 +0000 (22:44 +0000)]
Apply bandaid to bring svr4_sys_waitsys() in line with exit1().  This
routine really need to be gutted and merged with exit1().

Reviewed by: jhb

21 years agoEven if the permissions deny it, a process should be allowed to
phk [Mon, 13 Jan 2003 22:20:36 +0000 (22:20 +0000)]
Even if the permissions deny it, a process should be allowed to
access its controlling terminal.

In essense, history dictates that any process is allowed to open
/dev/tty for RW, irrespective of credential, because by definition
it is it's own controlling terminal.

Before DEVFS we relied on a hacky half-device thing (kern/tty_tty.c)
which did the magic deep down at device level, which at best was
disgusting from an architectural point of view.

My first shot at this was to use the cloning mechanism to simply
give people the right tty when they ask for /dev/tty, that's why
you get this, slightly counter intuitive result:

        syv# ls -l /dev/tty `tty`
        crw--w----  1 u1  tty    5,   0 Jan 13 22:14 /dev/tty
        crw--w----  1 u1  tty    5,   0 Jan 13 22:14 /dev/ttyp0

Trouble is, when user u1 su(1)'s to user u2, he cannot open
/dev/ttyp0 anymore because he doesn't have permission to do so.

The above fix allows him to do that.

The interesting side effect is that one was previously only able
to access the controlling tty by indirection:
        date > /dev/tty
but not by name:
        date > `tty`

This is now possible, and that feels a lot more like DTRT.

PR:             46635
MFC candidate:  could be.

21 years agoIf you don't create a /usr filesystem, / will need 200MB.
kuriyama [Mon, 13 Jan 2003 21:57:07 +0000 (21:57 +0000)]
If you don't create a /usr filesystem, / will need 200MB.

21 years agoWe can get past here on a normal vnode as well, so use VOP_STRATEGY if so.
phk [Mon, 13 Jan 2003 21:32:16 +0000 (21:32 +0000)]
We can get past here on a normal vnode as well, so use VOP_STRATEGY if so.

21 years agoFix interactive booting:
imp [Mon, 13 Jan 2003 21:28:24 +0000 (21:28 +0000)]
Fix interactive booting:
o Revision 1.38 introduced the -n flag.  It conflicted with the
  RB_BOOTINFO flag, so was in effect always on.  Change the -n flag to
  be bit 0x1c instead of 0x1f.  This also had the consequence that a mal-formed
  /boot.config would render the system unbootable because the user was
  unable to enter anything at all on the command line.
o Remove the initialization of opt to be RB_BOOTINFO since we filter that bit
  out and do not otherwise use it.

Reviewed by: jhb
MFC after: 3 days

21 years agoOK Ok, so I didn't check the NO_GEOM case for the final version...
phk [Mon, 13 Jan 2003 20:19:04 +0000 (20:19 +0000)]
OK Ok, so I didn't check the NO_GEOM case for the final version...

Stumbled on by: bde

21 years agoList the Davicom 9009 as supported.
trhodes [Mon, 13 Jan 2003 20:02:40 +0000 (20:02 +0000)]
List the Davicom 9009 as supported.

Discussed with: bmah

21 years agoreally, this time for sure. Fix formatting in usage().
dillon [Mon, 13 Jan 2003 19:58:05 +0000 (19:58 +0000)]
really, this time for sure.  Fix formatting in usage().

21 years agoGrr. I keep forgetting things. Include -C in dump's usage() .
dillon [Mon, 13 Jan 2003 19:56:03 +0000 (19:56 +0000)]
Grr. I keep forgetting things.  Include -C in dump's usage() .

21 years agoAdd support for obsolete option form for -C
dillon [Mon, 13 Jan 2003 19:50:46 +0000 (19:50 +0000)]
Add support for obsolete option form for -C

21 years agoAdd a caching option to dump. Use -C. Note that NetBSD has a caching option
dillon [Mon, 13 Jan 2003 19:42:41 +0000 (19:42 +0000)]
Add a caching option to dump.  Use -C.  Note that NetBSD has a caching option
called -r but it takes 512 byte blocks instead of megabytes, and I felt a
megabytes specification would be far more useful so I did not use the same
option character.

This will *greatly* improve dump performance at the cost of possibly
missing filesystem changes that occur between passes, and does a fairly
good job making up for the loss of buffered block devices.  Caching is disabled
by default to retain historical behavior.

In tests, dump performance improved by about 40% when dumping / or /usr.

Beware that dump forks and the cache may wind up being larger then you
specify, but a more complex shared memory implementation would not produce
results that are all that much better so I kept it simple for now.

MFC after: 3 days

21 years agoMention the oddities and requirements for mount operations executed by
joerg [Mon, 13 Jan 2003 19:42:21 +0000 (19:42 +0000)]
Mention the oddities and requirements for mount operations executed by
non-root users.

PR: docs/42651
Submitted by: Thomas Seck <tmseck@netcologne.de>
MFC after: 3 days

21 years agoMFR500: Updates for GCC version number, update 4.8-RELEASE scheduled
bmah [Mon, 13 Jan 2003 19:24:34 +0000 (19:24 +0000)]
MFR500:  Updates for GCC version number, update 4.8-RELEASE scheduled
release date, update Perl removal text.

Reminded by: keramida

21 years agoAdd the /boot prefix. I forgot this when updating the location.
trhodes [Mon, 13 Jan 2003 18:57:19 +0000 (18:57 +0000)]
Add the /boot prefix.  I forgot this when updating the location.

Neglected by: trhodes

21 years agoAdd Realtek ALC650 id.
orion [Mon, 13 Jan 2003 17:43:49 +0000 (17:43 +0000)]
Add Realtek ALC650 id.

Submitted by: "Mikko S. Hyvarinen" <morphy@morphy.iki.fi>
MFC after: 5 days

21 years agoAdd nForce2 device id.
orion [Mon, 13 Jan 2003 17:42:13 +0000 (17:42 +0000)]
Add nForce2 device id.

Submitted by: "Mikko S. Hyvarinen" <morphy@morphy.iki.fi>
MFC after: 5 days

21 years agoEnable the new h0h0magic code which on GEOM kernels make the md(4)
phk [Mon, 13 Jan 2003 17:31:46 +0000 (17:31 +0000)]
Enable the new h0h0magic code which on GEOM kernels make the md(4)
driver a _real_ GEOM driver.

21 years agoUpdate the kernel location in FILES.
trhodes [Mon, 13 Jan 2003 16:15:56 +0000 (16:15 +0000)]
Update the kernel location in FILES.

PR: 47015

21 years agoMinimal fix for DV part.
simokawa [Mon, 13 Jan 2003 16:08:09 +0000 (16:08 +0000)]
Minimal fix for DV part.
- Don't panic on contigmalloc failure.
- Calculate timestamp by feedforward rather than feedback which depends on
unreliable interrupt timing.
- Overwrite timestamp in CIP header correctly.
- Add debug code for timestamp synchronization.
- Add comments.

21 years agoMerge big endian and little endian case.
simokawa [Mon, 13 Jan 2003 15:08:48 +0000 (15:08 +0000)]
Merge big endian and little endian case.
Fill fdf bit fields and others.

21 years agostyle(9) fixes, mostly add parens around return arguments.
alfred [Mon, 13 Jan 2003 15:06:05 +0000 (15:06 +0000)]
style(9) fixes, mostly add parens around return arguments.

21 years agoAlways issue ioctls as BIO_GEATTR requests. The direction of data copies on
phk [Mon, 13 Jan 2003 11:34:35 +0000 (11:34 +0000)]
Always issue ioctls as BIO_GEATTR requests.  The direction of data copies on
ioctls are no reliable indication of the ioctls "set" or "get" nature or if
such simplistic categories can even be applied.

MFC candidate: boot0cfg issue.

21 years agoFix NewReno.
hsu [Mon, 13 Jan 2003 11:01:20 +0000 (11:01 +0000)]
Fix NewReno.

Reviewed by: Tom Henderson <thomas.r.henderson@boeing.com>

21 years agoRefer to 1003.1 instead of 1003.2 in the Standards section.
tjr [Mon, 13 Jan 2003 10:37:11 +0000 (10:37 +0000)]
Refer to 1003.1 instead of 1003.2 in the Standards section.

21 years agoAdd a mutex around the per unit bioqueue.
phk [Mon, 13 Jan 2003 08:50:23 +0000 (08:50 +0000)]
Add a mutex around the per unit bioqueue.

Only grab giant in the per unit kthread for SWAP and VNODE backed devices.

Initialize the bioq before the kthread gets a chance to study it.

Don't lock Giant in mddone_swap, we shouldn't need it.

21 years agoRemove four members of struct nameidata which have been commented
phk [Mon, 13 Jan 2003 08:49:36 +0000 (08:49 +0000)]
Remove four members of struct nameidata which have been commented
out since rev 1.1 (24 may 1994) of this file.

Add a nail to the K&R coffin by removing the #ifdef'ed K&R function
declaration for NDINIT.

21 years agoRemove g_silence(). It does not do anything anymore.
phk [Mon, 13 Jan 2003 08:46:32 +0000 (08:46 +0000)]
Remove g_silence().  It does not do anything anymore.