]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoMove the check for the error case (variable ends without a closing
harti [Wed, 2 Mar 2005 15:52:04 +0000 (15:52 +0000)]
Move the check for the error case (variable ends without a closing
paranthesis or brace) into the loop and don't leak the buffer in this
case. Remove the check for Var_Parse returning NULL - it can't.

Patch: 7.92

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoClean up the botching of the previous repo-copy. Reference the included
scottl [Wed, 2 Mar 2005 15:13:37 +0000 (15:13 +0000)]
Clean up the botching of the previous repo-copy.  Reference the included
headers from the correct location.

Submitted by: Tai-hwa Liang

19 years agoUse the shiny, new top-level 'distribution' target when populating the
jhb [Wed, 2 Mar 2005 14:57:45 +0000 (14:57 +0000)]
Use the shiny, new top-level 'distribution' target when populating the
chroot area.  This fixes make release of HEAD on systems that have a
/usr/src with bsd.endian.mk and a matching /usr/obj but haven't installed
the world in /usr/obj and thus have no bsd.endian.mk /usr/share/mk.

19 years agoFix indentation on a block of code.
harti [Wed, 2 Mar 2005 14:43:40 +0000 (14:43 +0000)]
Fix indentation on a block of code.

Patch: 7.91

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoPush the length computation down into VarParseShort().
harti [Wed, 2 Mar 2005 14:30:36 +0000 (14:30 +0000)]
Push the length computation down into VarParseShort().
Return always malloc()-ed strings from VarParseShort() to get
rid of warnings when returning string constants from a non-const char *
function.

Patch: 7.90

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoTweak the lapic timer code to get the performance closer to the pre-lapic
jhb [Wed, 2 Mar 2005 14:17:43 +0000 (14:17 +0000)]
Tweak the lapic timer code to get the performance closer to the pre-lapic
timer case:
- Remove the virtual fooclock interrupt counters as they have served their
  purpose.
- Adjust the dividers for the different clock such that profhz is now a
  multiple of stathz as in the non-lapic case, and the timer now runs at
  hz * 2 rather than hz * 3.  With the new divisors, the default clock
  rates are:

  kern.clockrate: { hz = 1000, tick = 1000, profhz = 666, stathz = 133 }

19 years agoIn kern_sigtimedwait, remove waitset bits for td_sigmask before
davidxu [Wed, 2 Mar 2005 13:43:51 +0000 (13:43 +0000)]
In kern_sigtimedwait, remove waitset bits for td_sigmask before
sleeping, so in do_tdsignal, we no longer need to test td_waitset.
now td_waitset is only used to give a thread higher priority when
delivering signal to multithreads process.
This also fixes a bug:
when a thread in sigwait states was suspended and later resumed
by SIGCONT, it can no longer receive signals belong to waitset.

19 years agoRepo-copy tools/regression/usr.bin/make to tools/build/make_check
harti [Wed, 2 Mar 2005 12:33:23 +0000 (12:33 +0000)]
Repo-copy tools/regression/usr.bin/make to tools/build/make_check
and adjust the path in the Makefile for the upgrade_checks target.
These checks are really feature upgrade checks that should be fast
and just find out whether we need to build a new make before
proceeding with other targets like buildworld. This makes the
place free for a real regression test suite in the old place.

19 years agoPowerPC support for kldxref by handling the PPC-specific relocations.
grehan [Wed, 2 Mar 2005 12:27:42 +0000 (12:27 +0000)]
PowerPC support for kldxref by handling the PPC-specific relocations.

19 years agoConvert a function call to Lst_ForEach with a macro call to LST_FOREACH.
harti [Wed, 2 Mar 2005 12:00:21 +0000 (12:00 +0000)]
Convert a function call to Lst_ForEach with a macro call to LST_FOREACH.

19 years agoWrap BSD r* commands in NO_RCMDS.
trhodes [Wed, 2 Mar 2005 11:53:22 +0000 (11:53 +0000)]
Wrap BSD r* commands in NO_RCMDS.
Change NO_RCMDNS to NO_RCMDS and do the switch in bsd.compat.mk.

Discussed with: ru, nectar

19 years ago- Allocate the interrupt resource as RF_SHAREABLE allowing uart(4) to work
marius [Wed, 2 Mar 2005 11:30:14 +0000 (11:30 +0000)]
- Allocate the interrupt resource as RF_SHAREABLE allowing uart(4) to work
  with shared IRQs in case the bus code, MD interrupt code, etc. permits.
  Together with sys/sparc64/sparc64/intr_machdep.c rev. 1.21 this fixes
  an endless loop in uart_intr() when using the second NS16550 on the ISA
  bus of sparc64 machines.
- Destroy the hardware mutex on detach and in case attaching fails.

Approved by: marcel

19 years ago- Allow multiple INTR_FAST handlers for the same source. The motivation
marius [Wed, 2 Mar 2005 11:27:13 +0000 (11:27 +0000)]
- Allow multiple INTR_FAST handlers for the same source. The motivation
  for this are the on-board SCCs and UARTs that use a shared IRQ. [1]
- Rework the interrupt counting code to account for shared interrupts. [1]
- In case ithread_add_handler() failed in inthand_add() just return with
  the error code instead of setting up a non-fast handler regardless or
  setting up a non-fast handler instead of a fast handler. I can't think
  of a situation where the former behaviour would do the right thing.

Reviewed by: marcel [1]
Based on: sys/i386/i386/intr_machdep.c [1]

19 years agoAssorted style fixes and minor changes:
marius [Wed, 2 Mar 2005 10:55:56 +0000 (10:55 +0000)]
Assorted style fixes and minor changes:
- Use FBSDID.
- Use uintXX_t instead of u_intXX_t.
- Be consistent with white-space.
- Mark some globals as static.
- Add a missing prototype.
- Remove a unused variable.
- etc.

19 years agoOnly install acpiio.h in /usr/include. That's all we want to export to users.
njl [Wed, 2 Mar 2005 10:45:09 +0000 (10:45 +0000)]
Only install acpiio.h in /usr/include.  That's all we want to export to users.

Submitted by: ru (any bugs by me)
MFC after: 1 day

19 years agoFix SCM ID's.
obrien [Wed, 2 Mar 2005 09:22:34 +0000 (09:22 +0000)]
Fix SCM ID's.

19 years agoFix typo. Unbreak build. Take pointy hat.
glebius [Wed, 2 Mar 2005 09:11:18 +0000 (09:11 +0000)]
Fix typo. Unbreak build. Take pointy hat.

19 years agoAdd an entry for myself.
jcamou [Wed, 2 Mar 2005 09:08:17 +0000 (09:08 +0000)]
Add an entry for myself.

Approved by: trhodes (mentor)

19 years agoUse the LST_FOREACH macro instead of the Lst_ForEach function. This
harti [Wed, 2 Mar 2005 08:30:49 +0000 (08:30 +0000)]
Use the LST_FOREACH macro instead of the Lst_ForEach function. This
saves function calls and reduces void casting.

19 years agoSync the list of headers visible with SHARED=symlinks with those
ru [Wed, 2 Mar 2005 07:40:18 +0000 (07:40 +0000)]
Sync the list of headers visible with SHARED=symlinks with those
visible with SHARED=copies.

Inspired by: njl

19 years agoUse correct byte order when parsing the size of the gzip "Extra data" field.
kientzle [Wed, 2 Mar 2005 05:34:05 +0000 (05:34 +0000)]
Use correct byte order when parsing the size of the gzip "Extra data" field.
In particular, this correctly allows bsdtar (and pkg_add) to skip
package signatures.

Thanks to: Theo Schlossnagle

19 years agoDocument NO_RCMDS, bump doc date.
trhodes [Wed, 2 Mar 2005 05:22:27 +0000 (05:22 +0000)]
Document NO_RCMDS, bump doc date.

19 years agoMove all of the hptmv files to /sys/dev/hptmv so that they won't be mistaken
scottl [Wed, 2 Mar 2005 05:14:28 +0000 (05:14 +0000)]
Move all of the hptmv files to /sys/dev/hptmv so that they won't be mistaken
for being on a CVS vendor branch.  The files were moved via a repo-copy.

19 years agoignores ICMPv6 code field in case of ICMPv6 Packet-Too-Big (as specified in RFC2463...
suz [Wed, 2 Mar 2005 05:14:15 +0000 (05:14 +0000)]
ignores ICMPv6 code field in case of ICMPv6 Packet-Too-Big (as specified in RFC2463 and draft-ietf-ipngwg-icmp-v3-06.txt)

Obtained from: KAME
MFC after: 1 day

19 years agoProtect acpivar.h with _KERNEL. No user parts inside currently.
njl [Wed, 2 Mar 2005 04:36:24 +0000 (04:36 +0000)]
Protect acpivar.h with _KERNEL.  No user parts inside currently.

19 years agoWhen resubmitting a timed out request, reset donecount.
mdodd [Wed, 2 Mar 2005 04:01:37 +0000 (04:01 +0000)]
When resubmitting a timed out request, reset donecount.

Submitted by:   Nate Lawson <nate AT root.org>

19 years agoThis will not compile without:
ambrisko [Wed, 2 Mar 2005 04:00:55 +0000 (04:00 +0000)]
This will not compile without:
        http://www.ambrisko.com/doug/listio_kqueue/listio_kqueue.patch

Note: it is a good idea to run this against a physical drive to
exercise the physio fast path (ie. lio_kqueue /dev/<something safe>)
This will ensure op's counting per LIO request is correct.  It is
currently broken the above patch fixes it.

Sponsored by:   IronPort

19 years agoHandle PIO timeouts in ata_end_transaction() by immediately returning.
mdodd [Wed, 2 Mar 2005 03:59:28 +0000 (03:59 +0000)]
Handle PIO timeouts in ata_end_transaction() by immediately returning.
Failure to do this will result in following ata_pio_read() calls walking
off the end of the read buffer.

This resolves the "memory modified after free" panics common with Thinkpads
and CD/DVD drives.

Submitted by:  Nate Lawson <nate AT root.org>

19 years agoAdd NO_RCMDS to the list.
trhodes [Wed, 2 Mar 2005 03:49:02 +0000 (03:49 +0000)]
Add NO_RCMDS to the list.

19 years agoIn ata_generic_reset() while waiting for both master & slave to become
mdodd [Wed, 2 Mar 2005 03:34:51 +0000 (03:34 +0000)]
In ata_generic_reset() while waiting for both master & slave to become
idle the 'mask' variable could be set to 0, resulting in the timeout loop
running for the full 31 seconds.

Handling this case eliminates long hangs on resume on some systems.

Submitted by: Nate Lawson <nate AT root.org>

19 years agoAdd an AIO & kqueue regression test. It is a good idea to run this
ambrisko [Wed, 2 Mar 2005 03:32:01 +0000 (03:32 +0000)]
Add an AIO & kqueue regression test.  It is a good idea to run this
against a disk as the argument.  If you don't it will use a temp file.
The raw disk will use the kernel physio fast path method until the
max number of pending op's is reached then it will queue them.  File
system op's are always queued.  This is more important with LIO since
operation can get split across and accounting of op's is broken with LIO.

Note that this was broken when locking was added to kqueue (ie. 5.3)
My fix needs to be better integrated with FreeBSD.

Next is an LIO test and implementation.

Sponsored by: IronPort

19 years agoAllow chkprintcap(8) to be run before lpd is started. Disabled by
brooks [Wed, 2 Mar 2005 02:46:47 +0000 (02:46 +0000)]
Allow chkprintcap(8) to be run before lpd is started.  Disabled by
default for now.  Default flags create missing directories.

Remove comment about doing this in etc/rc.d/var.

Unlike in the PR, I chose to do this in the lpd script where we reliably
have /usr available.

PR: conf/71488
Submitted by: RZ-FreeBSD0904 at fh-karlsruhe dot de

19 years agoUse a signal-safe type for two variables that are used to synchronise
iedowse [Wed, 2 Mar 2005 02:30:08 +0000 (02:30 +0000)]
Use a signal-safe type for two variables that are used to synchronise
with a signal handler. This fixes a race condition introduced by
compiler reordering that caused dump to sometimes get stuck,
especially while dumping large filesystems.

19 years agoIf we don't have /usr/sbin/mtree, try to mount /usr. We're only likely
brooks [Wed, 2 Mar 2005 00:58:05 +0000 (00:58 +0000)]
If we don't have /usr/sbin/mtree, try to mount /usr.  We're only likely
to hit this case when /usr is remote and thus hasn't been mounted (since
you're supposed to have /var before mounting remote file systems).
Normal machines that don't have a /var for some reason will have /usr
already available because it's local.

19 years ago- Fix braino introduced in rev. 1.17, unlike the X1032A (HME-ISP1040-combo)
marius [Wed, 2 Mar 2005 00:49:37 +0000 (00:49 +0000)]
- Fix braino introduced in rev. 1.17, unlike the X1032A (HME-ISP1040-combo)
  and the X1034A (quad HME; QFE) cards the X1033A (single HME) don't have a
  PCI-PCI-bridge so we can't rely on the PCI slot number being useable as
  index for the network address to read from the VPD on the latter. Use
  the end tag to determine whether it is a QFE VPD with 4 NAs and only use
  the slot number as index in this case.
- Remove a useless check.

Prodded by: joerg
Additional testing by: joerg
MFC after: 1 day

19 years agoRemove a number of entries from the distribution target that are
brooks [Wed, 2 Mar 2005 00:44:39 +0000 (00:44 +0000)]
Remove a number of entries from the distribution target that are
automaticly created at boot.  There's no need to maintain a list of
files and permissions in multiple places.  This also means binary
updates won't stomp on log files.

For the record, utmp is created in etc/rc.d/cleanvar, wtmp and lastlog
in etc/rc.d/var, and the reset via etc/rc.d/newsyslog.

19 years ago - Update etc/rc.d/newsyslog to FreeBSD standards and install it.
brooks [Wed, 2 Mar 2005 00:40:55 +0000 (00:40 +0000)]
 - Update etc/rc.d/newsyslog to FreeBSD standards and install it.
 - Enable it by default, running newsyslog with -CN which creates files
   that have the C flag specified in /etc/newsyslog.conf.
 - Remove the "newsyslog -CC" call from etc/rc.d/var and the check for
   newsyslog.
 - Add the C flag to entries in /etc/newsyslog.conf that are currently
   installed as part of the base system.

There are two effects from this change:
 - Users who delete default syslog files to stop logging to them
   will need to set newsyslog_enable=NO in rc.conf or remove the C
   flag from those file in /etc/newsyslog.conf or they will come back
   on the next boot.
 - Diskless systems now create the same set of files that ordinary
   systems have by default instead of every file in newsyslog.conf.

19 years agoFix incorrect comment.
davidxu [Tue, 1 Mar 2005 23:42:00 +0000 (23:42 +0000)]
Fix incorrect comment.

19 years agoReduce the maximum supported save state size to 4k, as the 8k limit
iedowse [Tue, 1 Mar 2005 23:17:42 +0000 (23:17 +0000)]
Reduce the maximum supported save state size to 4k, as the 8k limit
was not implemented correctly and needs further work.

19 years ago- Remove the dependency of /usr/bin/touch by using "cp /dev/null <target>"
brooks [Tue, 1 Mar 2005 22:08:15 +0000 (22:08 +0000)]
- Remove the dependency of /usr/bin/touch by using "cp /dev/null <target>"
  to create /var/log/lastlog.
- Also create /var/log/wtmp if missing.
- Attempt to create these files unless populate_var is NO rather then
  only when /var is empty or populate_var=YES.

19 years agoAttempt to doff the pointy hat: implement 'hw.realmem' on remaining
wes [Tue, 1 Mar 2005 21:55:27 +0000 (21:55 +0000)]
Attempt to doff the pointy hat: implement 'hw.realmem' on remaining
architectures.  Pointed out by O'Brien, ScottL via email.

Reviewed by: obrien (various)

19 years agoIn cpu_throw(), correctly calculate td->td_md.md_tp.
cognet [Tue, 1 Mar 2005 20:51:37 +0000 (20:51 +0000)]
In cpu_throw(), correctly calculate td->td_md.md_tp.
In cpu_switch(), set the DACR even if we're switching to a kernel thread.

19 years ago.It the -v option.
obrien [Tue, 1 Mar 2005 20:41:38 +0000 (20:41 +0000)]
.It the -v option.

19 years agoFix typo in a comment.
stefanf [Tue, 1 Mar 2005 20:32:05 +0000 (20:32 +0000)]
Fix typo in a comment.

19 years agoIntroduce realmem.
cognet [Tue, 1 Mar 2005 20:12:52 +0000 (20:12 +0000)]
Introduce realmem.

19 years agoPull up to ETHER_HDR_LEN before passing an mbuf to ether_input().
ru [Tue, 1 Mar 2005 19:39:57 +0000 (19:39 +0000)]
Pull up to ETHER_HDR_LEN before passing an mbuf to ether_input().
The bug was found by running ng_dummy(4) node configured with a
delay, in front of the ng_eiface(4) node.

19 years agoResort newfs(8) -f entry, no content changes.
bmah [Tue, 1 Mar 2005 18:45:34 +0000 (18:45 +0000)]
Resort newfs(8) -f entry, no content changes.

19 years agoNew release notes: LC_CTYPE files machine-independent, stack sizes
bmah [Tue, 1 Mar 2005 18:34:12 +0000 (18:34 +0000)]
New release notes:  LC_CTYPE files machine-independent, stack sizes
increased for libpthread and libc_r, new math functions (+MFC),
rtld(1) LD_LIBMAP (+MFC), telnet(1) +port number.

MFCs noted:  gshsec(8), fsck(8) -n (also fixed a mistake...mentioned
the wrong program before), mkuzip(8), Heimdal 0.6.3, OpenSSL 0.9.7e

Also mentioned that upgrades to -CURRENT are only possible from 5.3 or
newer.

19 years agoRemove linux_emul_find() and the CHECKALT*() macros as they are no longer
jhb [Tue, 1 Mar 2005 17:57:45 +0000 (17:57 +0000)]
Remove linux_emul_find() and the CHECKALT*() macros as they are no longer
used.

19 years agoVarious grammar, punctuation, and style fixups.
bmah [Tue, 1 Mar 2005 17:53:25 +0000 (17:53 +0000)]
Various grammar, punctuation, and style fixups.

19 years agoMake sure the length variable is initialized to 0 before passing
harti [Tue, 1 Mar 2005 17:52:18 +0000 (17:52 +0000)]
Make sure the length variable is initialized to 0 before passing
it to Var_Parse().

Patch: 7.85

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoUse kern_kevent instead of the stackgap for 32bit syscall wrapping.
ps [Tue, 1 Mar 2005 17:45:55 +0000 (17:45 +0000)]
Use kern_kevent instead of the stackgap for 32bit syscall wrapping.

Submitted by: jhb
Tested on: amd64

19 years agoregen
ps [Tue, 1 Mar 2005 17:44:34 +0000 (17:44 +0000)]
regen

19 years agoChange the prototype of kevent to remove the const from the changelist.
ps [Tue, 1 Mar 2005 17:43:08 +0000 (17:43 +0000)]
Change the prototype of kevent to remove the const from the changelist.

Reviewed by: jhb

19 years agoIn windrv_load(), I was allocating the driver object using
wpaul [Tue, 1 Mar 2005 17:21:25 +0000 (17:21 +0000)]
In windrv_load(), I was allocating the driver object using
malloc(sizeof(device_object), ...) by mistake. Correct this, and
rename "dobj" to "drv" to make it a bit clearer what this variable
is supposed to be.

Spotted by: Mikore Li at Sun dot comnospamplzkthx

19 years agoWe can specify device size in bytes. Document this in usage.
pjd [Tue, 1 Mar 2005 14:56:49 +0000 (14:56 +0000)]
We can specify device size in bytes. Document this in usage.

19 years agoAdd polish version of libc NLS catalog.
pjd [Tue, 1 Mar 2005 14:38:30 +0000 (14:38 +0000)]
Add polish version of libc NLS catalog.

19 years agoOoops. I will compile test before committing. The stackgap version
ps [Tue, 1 Mar 2005 13:50:57 +0000 (13:50 +0000)]
Ooops.  I will compile test before committing.  The stackgap version
of kevent32 will be going away shortly, so this is temporary until
I commit the non-stackgap version.

19 years agoAdd more locking when reading/writing to carp softc. When carp softc is
glebius [Tue, 1 Mar 2005 13:14:33 +0000 (13:14 +0000)]
Add more locking when reading/writing to carp softc. When carp softc is
attached to a parent interface we use its mutex to lock the softc. This
means that in several places like carp_ioctl() we lock softc conditionaly.
This should be redesigned.

To avoid LORs when MII announces us a link state change, we schedule
a quick callout and call carp_carpdev_state_locked() from it.

Initialize callouts using NET_CALLOUT_MPSAFE.

Sponsored by: Rambler
Reviewed by: mlaier

19 years ago- Add carp_mtx. Use it to protect list of all carp interfaces.
glebius [Tue, 1 Mar 2005 12:36:07 +0000 (12:36 +0000)]
- Add carp_mtx. Use it to protect list of all carp interfaces.
- In carp_send_ad_all() walk through list of all carp interfaces
  instead of walking through list of all interfaces.

Sponsored by: Rambler
Reviewed by: mlaier

19 years agoSplit out ParseModifiers from VarParseLong.
harti [Tue, 1 Mar 2005 12:26:32 +0000 (12:26 +0000)]
Split out ParseModifiers from VarParseLong.

Patch: 7.84
Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoVarParseLong: Create new else block for haveModifier. Move the call
harti [Tue, 1 Mar 2005 12:22:07 +0000 (12:22 +0000)]
VarParseLong: Create new else block for haveModifier. Move the call
to VarExpand down into the branches of the if as well as cleanup code.
Eliminate code that is now obviously dead.

Patch: 7.83

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoAvoid a couple of mutex operations in the process exit path for the
phk [Tue, 1 Mar 2005 12:20:49 +0000 (12:20 +0000)]
Avoid a couple of mutex operations in the process exit path for the
common case where procfs have never been mounted.

OK'ed by: des

19 years agoUse NET_CALLOUT_MPSAFE macro.
glebius [Tue, 1 Mar 2005 12:01:17 +0000 (12:01 +0000)]
Use NET_CALLOUT_MPSAFE macro.

19 years agoAdd macro NET_CALLOUT_MPSAFE, which should be used when initializing
glebius [Tue, 1 Mar 2005 11:54:46 +0000 (11:54 +0000)]
Add macro NET_CALLOUT_MPSAFE, which should be used when initializing
network related callouts.

Reviewed by: rwatson

19 years agoAdd myself to the calendar.
flz [Tue, 1 Mar 2005 11:48:36 +0000 (11:48 +0000)]
Add myself to the calendar.
Feel free to send presents :)

Approved by: pav (mentor)

19 years agoReplace NG_PARSE_APPEND() macro with ng_parse_append() function. Check
glebius [Tue, 1 Mar 2005 11:31:06 +0000 (11:31 +0000)]
Replace NG_PARSE_APPEND() macro with ng_parse_append() function. Check
its return value and free resources if function returns error. Plug
several memory leaks with this change.

Submitted by: archie
Found by: Coverity Prevent analysis tool

19 years agoRevert change to struct ifnet. Use ifnet pointer in softc. Embedding
glebius [Tue, 1 Mar 2005 10:59:14 +0000 (10:59 +0000)]
Revert change to struct ifnet. Use ifnet pointer in softc. Embedding
ifnet into smth will soon be removed.

Requested by: brooks

19 years agocorrect WARNS=6 fix to use cast to (void *).
ume [Tue, 1 Mar 2005 10:55:06 +0000 (10:55 +0000)]
correct WARNS=6 fix to use cast to (void *).
use of struct sockaddr_strage * is thought as not good manner. :)

19 years agoRemove debugging printf.
glebius [Tue, 1 Mar 2005 09:31:36 +0000 (09:31 +0000)]
Remove debugging printf.

Reviewed by: mlaier

19 years agoUse BUS_PROBE_DEFAULT in preference to 0 and BUS_PROBE_LOW_PRIORITY in
imp [Tue, 1 Mar 2005 08:58:06 +0000 (08:58 +0000)]
Use BUS_PROBE_DEFAULT in preference to 0 and BUS_PROBE_LOW_PRIORITY in
preference to some random negative number to allow other drivers a
bite at the apple.

19 years agoWhen mac_check_system_acct() fails, make sure to unlock as well as close
rwatson [Tue, 1 Mar 2005 08:56:13 +0000 (08:56 +0000)]
When mac_check_system_acct() fails, make sure to unlock as well as close
the vnode.

Pointed out by: jeff

19 years agoMerge from NetBSD.
sobomax [Tue, 1 Mar 2005 08:01:22 +0000 (08:01 +0000)]
Merge from NetBSD.

o usb_subr.c, add delta 1.119:

  Move usb_get_string() and make it public.

o usbdi.c, bring on par with 1.106, this includes:

  - Make an iterator abstraction for looping through all descriptors.

  - Whine about not being able to figure out default language if we are debugging.

  - Move usb_get_string() and make it public.

o usbdi.h, bring on par with 1.64, this includes:

  - Make an iterator abstraction for looping through all descriptors.

  - Move usb_get_string() and make it public.

o usbdi_util.c, bring on par with 1.42, this includes:

  - Add usbd_get_protocol().

  - Use NULL instead of 0.

  - Fix (mostly harmless) typo.

  - Move utility routine from uirda.c to usbdi_util.c.

o usbdi_util.h, bring on par with 1.31, this includes:

  - Add usbd_get_protocol().

  - Move utility routine from uirda.c to usbdi_util.c.

MFC after: 3 days

19 years agoCatch up with "physical memory" sysctl change.
grehan [Tue, 1 Mar 2005 07:59:24 +0000 (07:59 +0000)]
Catch up with "physical memory" sysctl change.
(MFi386: rev 1.608)

19 years agoReverse a condition so that the else clause can be changed
harti [Tue, 1 Mar 2005 07:58:18 +0000 (07:58 +0000)]
Reverse a condition so that the else clause can be changed
to a fallthrough.

Patch: 7.82

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoUse BUS_PROBE_DEFAULT in preference to 0. Also for vx, return
imp [Tue, 1 Mar 2005 07:50:12 +0000 (07:50 +0000)]
Use BUS_PROBE_DEFAULT in preference to 0.  Also for vx, return
BUS_PROBE_LOW_PRIORITY in stead of ifdef for devices that xl and vx
both support so that xl will snarf them on up.

19 years ago - Fix anoter dyslexic moment; an atomic_set_int should've become ACTIVESET,
jeff [Tue, 1 Mar 2005 07:38:45 +0000 (07:38 +0000)]
 - Fix anoter dyslexic moment; an atomic_set_int should've become ACTIVESET,
   not ACTIVECLEAR.

Submitted by: iedowse

19 years agoSync with 1.9 from NetBSD, this includes:
sobomax [Tue, 1 Mar 2005 06:35:04 +0000 (06:35 +0000)]
Sync with 1.9 from NetBSD, this includes:

o Add Ethernet descriptor.

o Use attribute packed for on-the-wire data structures.

MFC after: 3 days

19 years agoCorrect the freebsd32_kevent prototype.
ps [Tue, 1 Mar 2005 06:32:53 +0000 (06:32 +0000)]
Correct the freebsd32_kevent prototype.

19 years agoAdd a manual page for the fm801 chipsets.
trhodes [Tue, 1 Mar 2005 05:58:16 +0000 (05:58 +0000)]
Add a manual page for the fm801 chipsets.

PR: 78143
Submitted by: Joel Dahl <joel@automatvapen.se> (original version)

19 years agoAdd sysdoc, a small set of scripts which will parse a kernel binary and
trhodes [Tue, 1 Mar 2005 05:48:37 +0000 (05:48 +0000)]
Add sysdoc, a small set of scripts which will parse a kernel binary and
modules to rip out the available sysctls.  It will then produce a manual
page which may be installed with 'make install'.  Currently typing 'make'
in the directory uses the default /boot/kernel files.  To use a specific
directory, run sysdoc -k [location].

19 years agoUse the kernel pmap's lock to guarantee that only one thread at a time is
alc [Tue, 1 Mar 2005 05:06:52 +0000 (05:06 +0000)]
Use the kernel pmap's lock to guarantee that only one thread at a time is
using either pmap_temp_map_1 or pmap_temp_map_2.

Tested by: kris@

19 years agoCatch up with the "physical memory" sysctl change.
obrien [Tue, 1 Mar 2005 04:18:32 +0000 (04:18 +0000)]
Catch up with the "physical memory" sysctl change.
(MFi386: rev 1.608)

19 years agoAdd rc.bsdextended. It's been tested enough.
trhodes [Tue, 1 Mar 2005 03:59:24 +0000 (03:59 +0000)]
Add rc.bsdextended.  It's been tested enough.

19 years agoSupport \H, \h, \w, \W, \$ string expansion in the prompt.
obrien [Tue, 1 Mar 2005 03:35:58 +0000 (03:35 +0000)]
Support \H, \h, \w, \W, \$ string expansion in the prompt.

Submitted by: mini

19 years agoMake an advise that a rebuild of fsck(8) is recommended for -CURRENT
delphij [Tue, 1 Mar 2005 02:33:34 +0000 (02:33 +0000)]
Make an advise that a rebuild of fsck(8) is recommended for -CURRENT
after 20050220 due to the superblock summary recomputation change.
Also make a note about how to go back to the old behavior.

MFC After: 1 day

19 years ago- Split tests into three rough categories.
das [Tue, 1 Mar 2005 01:43:20 +0000 (01:43 +0000)]
- Split tests into three rough categories.
- Use fesetround() instead of fpsetround(), and add tests for
  various rounding modes.
- Test that all NaNs generated are quiet.

Some of these tests won't pass until problems in vendor sources
(gdtoa and gcc) are fixed and new versions imported, but I
want to get these changes into the tree before I accidentally
blow them away again.  :-(

19 years ago- Split the printfloat test into 11 individually wrapped and packaged
das [Tue, 1 Mar 2005 01:43:05 +0000 (01:43 +0000)]
- Split the printfloat test into 11 individually wrapped and packaged
  tests.  (Buy 10, get one free!)  The separate categories were
  already there; they just weren't labeled.
- Use fesetround() instead of fpsetround(), since the former is
  standard and implemented on all supported architectures.  Add
  tests for each rounding mode.
- Add additional tests for subnormals.

Some of these tests won't pass until problems in vendor sources
(gdtoa and gcc) are fixed and new versions imported, but I
want to get these changes into the tree before I accidentally
blow them away.

19 years agoMFi386: Sync whitespace and an abbreviation
peter [Mon, 28 Feb 2005 23:39:58 +0000 (23:39 +0000)]
MFi386: Sync whitespace and an abbreviation

19 years agoMFi386: Update alc's copyright notice
peter [Mon, 28 Feb 2005 23:38:15 +0000 (23:38 +0000)]
MFi386:  Update alc's copyright notice

19 years agoMFi386: Bring over John's local apic timer code
peter [Mon, 28 Feb 2005 23:37:35 +0000 (23:37 +0000)]
MFi386:  Bring over John's local apic timer code

19 years agoBootstrap pwd_mkdb(8) and use its new feature during "make distribute".
ru [Mon, 28 Feb 2005 22:55:43 +0000 (22:55 +0000)]
Bootstrap pwd_mkdb(8) and use its new feature during "make distribute".

19 years agoAdd a sysctl that records the amount of physical memory in the machine.
wes [Mon, 28 Feb 2005 21:42:56 +0000 (21:42 +0000)]
Add a sysctl that records the amount of physical memory in the machine.

Submitted by: Nicko Dehaine <nicko@stbernard.com>
MFC after: 1 day

19 years agoSave and restore the VGA state across a suspend-resume cycle. This
iedowse [Mon, 28 Feb 2005 21:06:14 +0000 (21:06 +0000)]
Save and restore the VGA state across a suspend-resume cycle. This
is particularly useful when VESA is available (either `options VESA'
or load the vesa module), as BIOSes in some notebooks may correctly
save and restore LCD panel settings using VESA in cases where calling
the video BIOS POST is not effective. On some systems it may also
be necessary to set the hw.acpi.reset_video sysctl to 0.

19 years agoAdd endianness support.
ru [Mon, 28 Feb 2005 21:05:30 +0000 (21:05 +0000)]
Add endianness support.

While version 4 entries are architecture-independent, we
also store old (version 3) entries in native byte order.
Also, the hash itself is created in a native byte order.

With this change, pwd_mkdb(8) can be used to cross-build
*pwd.db files for another architecture.

Tested on: i386, amd64, alpha, sparc64

19 years agoAdd a missing bcopy() to make saving the VESA state actually work.
iedowse [Mon, 28 Feb 2005 20:40:44 +0000 (20:40 +0000)]
Add a missing bcopy() to make saving the VESA state actually work.
Also save the DAC state, increase the maximum save state size from
4k to 8k, and refuse to save the VESA state if the BIOS reports it
is larger than the maximum size we can handle.

It doesn't appear that anything currently uses this code, but it
turns out to be capable of restoring some notebook displays to a
working state after a suspend-resume cycle.

19 years agoDefine the _mtx_assert() function prototype as well as the MA_* constants
jhb [Mon, 28 Feb 2005 19:26:16 +0000 (19:26 +0000)]
Define the _mtx_assert() function prototype as well as the MA_* constants
if either INVARIANTS or INVARIANT_SUPPORT is defined so that kernel modules
that want to use mtx_assert() only need to define INVARIANTS.

MFC after: 1 week

19 years agoFix a grammar nit of mine.
obrien [Mon, 28 Feb 2005 18:32:25 +0000 (18:32 +0000)]
Fix a grammar nit of mine.

19 years ago-mandoc style tweak rev 1.13.
obrien [Mon, 28 Feb 2005 18:28:58 +0000 (18:28 +0000)]
-mandoc style tweak rev 1.13.
Submitted by: ru

19 years agoDon't extract the .gdbinit file from the distribution.
harti [Mon, 28 Feb 2005 17:29:10 +0000 (17:29 +0000)]
Don't extract the .gdbinit file from the distribution.