dragonfly.git
15 years agoAdd .gitignore file
Matthias Schmidt [Mon, 1 Dec 2008 12:43:53 +0000 (13:43 +0100)]
Add .gitignore file

OK-and-input-by: corecode@
15 years agoFix wrong path
Matthias Schmidt [Mon, 1 Dec 2008 10:59:39 +0000 (11:59 +0100)]
Fix wrong path

15 years agoUpdate committer(7) to reflect the change to git
Matthias Schmidt [Mon, 1 Dec 2008 08:59:54 +0000 (09:59 +0100)]
Update committer(7) to reflect the change to git

In-collab-with: swildner@

15 years agoCheckout master after clone
Matthias Schmidt [Mon, 1 Dec 2008 08:12:11 +0000 (09:12 +0100)]
Checkout master after clone

15 years agoReplace CVS targets with GIT targets
Matthias Schmidt [Mon, 1 Dec 2008 08:03:22 +0000 (09:03 +0100)]
Replace CVS targets with GIT targets

Note: The initial clone is only possible if no previous src/ directory is present in
/usr (this is the default after cf2bfc4158dbb2accb11861dd3a96d0d79237f63 for new
installations).  If you already have a src/ directory remove it before starting cloning.

15 years agoReference development(7).
Sascha Wildner [Mon, 1 Dec 2008 02:20:16 +0000 (03:20 +0100)]
Reference development(7).

15 years agoUpdate some documents for git usage.
Sascha Wildner [Mon, 1 Dec 2008 01:45:30 +0000 (02:45 +0100)]
Update some documents for git usage.

15 years agoUpdate development(7) for git usage.
Sascha Wildner [Mon, 1 Dec 2008 01:43:35 +0000 (02:43 +0100)]
Update development(7) for git usage.

I've vaporized most of the old content. If people feel that something
should be there, we can gradually add/adapt documentation back.

Kinda-based-on: aggelos' wiki page

15 years agoAxe various remains from the CVS era.
Sascha Wildner [Sun, 30 Nov 2008 23:21:58 +0000 (00:21 +0100)]
Axe various remains from the CVS era.

* DragonFly supfiles in share/examples/cvsup (those for other BSDs remain).

* 'make update' target and associated variables.

More will follow.

15 years agoRemove handling of CVS IDs.
Sascha Wildner [Sun, 30 Nov 2008 21:52:05 +0000 (22:52 +0100)]
Remove handling of CVS IDs.

15 years agogit(1) doesn't allow to clone into an existing directory so it doesn't
Sascha Wildner [Sun, 30 Nov 2008 20:49:58 +0000 (21:49 +0100)]
git(1) doesn't allow to clone into an existing directory so it doesn't
make sense to create /usr/src anymore.

15 years agoMerge branch 'master' of ssh://swildner@crater.dragonflybsd.org/repository/git/dragon...
Sascha Wildner [Sun, 30 Nov 2008 20:16:53 +0000 (21:16 +0100)]
Merge branch 'master' of ssh://swildner@crater.dragonflybsd.org/repository/git/dragonfly into posix

15 years agoLet make upgrade install new files in etc/ if they don't exist.
Peter Avalos [Sun, 30 Nov 2008 18:23:19 +0000 (13:23 -0500)]
Let make upgrade install new files in etc/ if they don't exist.

Reviewed-by: dillon, corecode

15 years agoUse ${INSTALL} so installworld works for non-root users.
Peter Avalos [Sun, 30 Nov 2008 16:56:24 +0000 (11:56 -0500)]
Use ${INSTALL} so installworld works for non-root users.

15 years ago<utime.h> shall define time_t.
Sascha Wildner [Sun, 30 Nov 2008 14:21:03 +0000 (15:21 +0100)]
<utime.h> shall define time_t.

15 years agoTesting demultiplexing 2/2
Simon Schubert [Sun, 30 Nov 2008 13:10:25 +0000 (14:10 +0100)]
Testing demultiplexing 2/2

15 years agoTesting push demultiplexing 1/2
Simon Schubert [Sun, 30 Nov 2008 13:09:55 +0000 (14:09 +0100)]
Testing push demultiplexing 1/2

15 years agoMore testing.
Simon Schubert [Sun, 30 Nov 2008 12:33:15 +0000 (13:33 +0100)]
More testing.

15 years agoTest mail hook
Simon Schubert [Sun, 30 Nov 2008 12:32:18 +0000 (13:32 +0100)]
Test mail hook

15 years agoAdd a gaussian filter for maximum beeping pleasure.
Simon Schubert [Sat, 29 Nov 2008 23:41:05 +0000 (00:41 +0100)]
Add a gaussian filter for maximum beeping pleasure.

15 years agoRecognize TTTech MC322
Sepherosa Ziehau [Sun, 30 Nov 2008 10:54:10 +0000 (18:54 +0800)]
Recognize TTTech MC322

Obtained-from: OpenBSD

15 years agoregen
Sepherosa Ziehau [Sun, 30 Nov 2008 10:51:11 +0000 (18:51 +0800)]
regen

15 years agoAdd TTTech
Sepherosa Ziehau [Sun, 30 Nov 2008 10:46:54 +0000 (18:46 +0800)]
Add TTTech

Obtained-from: OpenBSD

15 years agoFilter out 8139 chips by PCI revid. It looks like certain 8139 does have
Sepherosa Ziehau [Sun, 30 Nov 2008 10:33:13 +0000 (18:33 +0800)]
Filter out 8139 chips by PCI revid.  It looks like certain 8139 does have
same hardware revision (in TXCFG) as 8139C+.

Reported-by: Antonio Huete Jimenez <ahuete.devel@gmail.com>
Obtained-from: FreeBSD

15 years agoRemove jumbo frame support for 8169; it never works (except slow ping) and
Sepherosa Ziehau [Sun, 30 Nov 2008 10:08:27 +0000 (18:08 +0800)]
Remove jumbo frame support for 8169; it never works (except slow ping) and
pullute the TX path a lot.

15 years ago- Use a longer timeout value for firmware handshaking.
Sepherosa Ziehau [Sun, 30 Nov 2008 09:49:49 +0000 (17:49 +0800)]
- Use a longer timeout value for firmware handshaking.
- Nuke the RX CPU self-diag, which prevents working cards from working.(*)

This commit intends to fix the firmware handshaking failure seen on PCI cards.
PCI-E cards do not seem to be affected.

(*) Linux tg3 does not have this diag neither does OpenBSD's bge(4)

15 years agotest git
Matthias Schmidt [Sun, 30 Nov 2008 09:33:06 +0000 (10:33 +0100)]
test git

15 years agotesting
Nuno Antunes [Sun, 30 Nov 2008 08:06:23 +0000 (08:06 +0000)]
testing

15 years agotest lance
Sepherosa Ziehau [Sun, 30 Nov 2008 07:54:27 +0000 (15:54 +0800)]
test lance

15 years agoTest commit/push access for master.
Peter Avalos [Sun, 30 Nov 2008 07:10:19 +0000 (02:10 -0500)]
Test commit/push access for master.

15 years agotest
Sascha Wildner [Sat, 29 Nov 2008 20:49:35 +0000 (21:49 +0100)]
test

15 years agoFix disklabel(1) core dump.
Nicolas Thery [Fri, 28 Nov 2008 16:52:27 +0000 (17:52 +0100)]
Fix disklabel(1) core dump.

The core dump occured when label line countained no fstype.  Now
displays an error message instead.

Submitted-by: Antonio Huete Jimenez <ahuete.devel@gmail.com>
15 years agotest xanadu
Sepherosa Ziehau [Sat, 29 Nov 2008 11:01:32 +0000 (19:01 +0800)]
test xanadu

15 years agotest enigma
Sepherosa Ziehau [Sat, 29 Nov 2008 11:02:08 +0000 (19:02 +0800)]
test enigma

15 years agotest local mirror
Sepherosa Ziehau [Sat, 29 Nov 2008 10:52:35 +0000 (18:52 +0800)]
test local mirror

15 years agotest push
Sepherosa Ziehau [Sat, 29 Nov 2008 10:16:33 +0000 (18:16 +0800)]
test push

15 years agoMerge branch 'master' of git://chlamydia.fs.ei.tum.de/dragonfly
Nicolas Thery [Fri, 28 Nov 2008 12:16:55 +0000 (13:16 +0100)]
Merge branch 'master' of git://chlamydia.fs.ei.tum.de/dragonfly

15 years agotest
Charlie [Sat, 29 Nov 2008 02:51:22 +0000 (10:51 +0800)]
test

15 years agoTest switch to git repo.
Nicolas Thery [Fri, 28 Nov 2008 12:12:21 +0000 (13:12 +0100)]
Test switch to git repo.

15 years agoBump .Dd to when the page was updated.
Sascha Wildner [Wed, 26 Nov 2008 17:17:16 +0000 (17:17 +0000)]
Bump .Dd to when the page was updated.

15 years agoRemove wrong parentheses.
Sascha Wildner [Wed, 26 Nov 2008 15:27:24 +0000 (15:27 +0000)]
Remove wrong parentheses.

15 years agoResurrect priority message support; there is always +/-5~10K measurement error.
Sepherosa Ziehau [Wed, 26 Nov 2008 15:05:42 +0000 (15:05 +0000)]
Resurrect priority message support; there is always +/-5~10K measurement error.

15 years ago- In sys_setsockopt():
Sepherosa Ziehau [Wed, 26 Nov 2008 13:10:56 +0000 (13:10 +0000)]
- In sys_setsockopt():
  o  Don't test sopt.sopt_val against NULL; this is not setup yet.  This fixes
     random setsockopt(2) failure with EINVAL.
  o  Clear sopt.sopt_val earlier
  o  Don't do copyout in setsockopt; the user arg is const
- In sys_getsockopt():
  o  Clear sopt.sopt_val earlier
  o  Centralize sopt.sopt_valsize check

Reviewed-by: aggelos@
15 years ago- Proper handle flags passed to bridge_rtflush() and bridge_rtdelete(),
Sepherosa Ziehau [Wed, 26 Nov 2008 12:49:43 +0000 (12:49 +0000)]
- Proper handle flags passed to bridge_rtflush() and bridge_rtdelete(),
  instead of handle it as a boolean.
- Add IFBF_FLUSHSYNC flag so bridge_rtflush() and bridge_rtdelete() could
  perform async operation (used by STP code).

15 years ago- JMC260 with full mask revision 2 and ECO revision 0 does not support
Sepherosa Ziehau [Wed, 26 Nov 2008 11:55:18 +0000 (11:55 +0000)]
- JMC260 with full mask revision 2 and ECO revision 0 does not support
  64bits DMA operatoin.
- Newer parts (full mask revision >= 2) of JMC250/JMC260 require driver
  to configure TX offload engine and TX MAC clock source according to the
  current link speed.
- New datasheet no longer uses PCI revision ID to identify chip generation,
  but chipmode's full mask revision and ECO revision are used; update
  driver accordingly.

Thank Ethan <ethanhsiao@jmicron.com> for pointing out necessary updates
and sending me an EVB.

15 years agoRemove /usr/include/libusbhid.h.
Hasso Tepper [Mon, 24 Nov 2008 17:40:02 +0000 (17:40 +0000)]
Remove /usr/include/libusbhid.h.

15 years agoSync libusbhid with other BSDs (breaks API compatibility). Sync usbhidctl
Hasso Tepper [Mon, 24 Nov 2008 17:15:17 +0000 (17:15 +0000)]
Sync libusbhid with other BSDs (breaks API compatibility). Sync usbhidctl
with FreeBSD.

Obtained-from: FreeBSD

15 years agoUnbreak installworld.
Hasso Tepper [Mon, 24 Nov 2008 13:35:57 +0000 (13:35 +0000)]
Unbreak installworld.

15 years agoAdd/sync some feature strings.
Sascha Wildner [Mon, 24 Nov 2008 13:14:21 +0000 (13:14 +0000)]
Add/sync some feature strings.

Taken-from: FreeBSD

15 years agoAdd hardware type value define for IP over firewire. Not used yet.
Hasso Tepper [Mon, 24 Nov 2008 12:19:14 +0000 (12:19 +0000)]
Add hardware type value define for IP over firewire. Not used yet.

15 years agoInstall acpiio.h.
Hasso Tepper [Mon, 24 Nov 2008 10:27:25 +0000 (10:27 +0000)]
Install acpiio.h.

15 years agoUse .Fn for functions and macros with parameters.
Sascha Wildner [Sun, 23 Nov 2008 21:55:52 +0000 (21:55 +0000)]
Use .Fn for functions and macros with parameters.

15 years agoSave the value read from TX queue CSR, so we don't test against a stale one.
Sepherosa Ziehau [Sun, 23 Nov 2008 04:28:27 +0000 (04:28 +0000)]
Save the value read from TX queue CSR, so we don't test against a stale one.

15 years agoAdd comment about how to survive blocking operation during percpu member
Sepherosa Ziehau [Sun, 23 Nov 2008 02:58:26 +0000 (02:58 +0000)]
Add comment about how to survive blocking operation during percpu member
interface list iteration.

15 years agoBring in a devclass_get_count(9) manual page.
Sascha Wildner [Sat, 22 Nov 2008 20:08:35 +0000 (20:08 +0000)]
Bring in a devclass_get_count(9) manual page.

Taken-from: FreeBSD

15 years agoCorrect my previous commit which only preserved snapshots/ but not
Sascha Wildner [Sat, 22 Nov 2008 18:36:18 +0000 (18:36 +0000)]
Correct my previous commit which only preserved snapshots/ but not
its contents. While here, expand the comments a bit.

15 years agoRemove MSGF_PRIORITY support. The flag testing and message queue selection
Sepherosa Ziehau [Sat, 22 Nov 2008 11:03:35 +0000 (11:03 +0000)]
Remove MSGF_PRIORITY support.  The flag testing and message queue selection
on the hot code path introduce noticeable performance regression during ip
forwarding (from 667Kpps to 655Kpps w/ 64bytes packet and fastforwarding
enabled on Phenom 9550).

15 years agoParallelize bridge_input step 2/2:
Sepherosa Ziehau [Sat, 22 Nov 2008 09:54:28 +0000 (09:54 +0000)]
Parallelize bridge_input step 2/2:
- Remove bridge's serializer from bridge_input()/bridge_output() path
- Push down bridge's serializer holding into bridge_span()
- Use bridge_handoff() whenever possible
- Add a simplified version of bridge_broadcast() to be called by bridge_start()
- Remove functions which are no longer used

15 years agoParallelize bridge_input step 1.85/2
Sepherosa Ziehau [Sat, 22 Nov 2008 05:57:31 +0000 (05:57 +0000)]
Parallelize bridge_input step 1.85/2
Add a boolean field in bridge_softc to indicate whether we have span ports
to tap packets to.  Once bridge's serializer is removed from bridge_input,
we could push the serializer holding down into bridge_span.

15 years agobstp_input() always returns NULL, so nuke its return value and let caller
Sepherosa Ziehau [Sat, 22 Nov 2008 04:30:28 +0000 (04:30 +0000)]
bstp_input() always returns NULL, so nuke its return value and let caller
set m to NULL.

15 years agoFactor out ether_reinput_oncpu(). Mainly used by pseudo interfaces to change
Sepherosa Ziehau [Sat, 22 Nov 2008 04:00:53 +0000 (04:00 +0000)]
Factor out ether_reinput_oncpu().  Mainly used by pseudo interfaces to change
mbuf's input interface.

15 years ago- IFF_MONITOR processing will happen in ether_input_oncpu()
Sepherosa Ziehau [Fri, 21 Nov 2008 13:37:02 +0000 (13:37 +0000)]
- IFF_MONITOR processing will happen in ether_input_oncpu()
- Don't count parent interface's if_noproto if vlan interface is not up;
  just silently discard the incoming packet like a real interface

15 years agoWhite space changes
Sepherosa Ziehau [Fri, 21 Nov 2008 12:43:42 +0000 (12:43 +0000)]
White space changes

15 years ago- Update bridge ifnet's if_imcasts
Sepherosa Ziehau [Fri, 21 Nov 2008 12:13:02 +0000 (12:13 +0000)]
- Update bridge ifnet's if_imcasts
- Update comment
- Minor white space changes

15 years agoParallelize bridge_input step 1.8/2:
Sepherosa Ziehau [Fri, 21 Nov 2008 11:11:03 +0000 (11:11 +0000)]
Parallelize bridge_input step 1.8/2:
Parallelize bridge member interfaces list --
o  Split bridge_iflist into percpu part and shared part (bridge_ifinfo).
   The shared part contains STP related information.
o  Put create bridge_iflist on each CPU and put it onto percpu member
   interface list.
o  All of the STP operation is still serialized by bridge's serializer,
   except testing member interface's STP state.
o  Span interfaces no longer have unused STP information.

15 years agoSince kmalloc's limit testing is not synchronized across CPUs, we raise
Sepherosa Ziehau [Thu, 20 Nov 2008 14:21:01 +0000 (14:21 +0000)]
Since kmalloc's limit testing is not synchronized across CPUs, we raise
the limit by another 1/8 to take the loosememuse into account.

15 years agoMake apps using '#define _POSIX_C_SOURCE' compile.
Hasso Tepper [Thu, 20 Nov 2008 11:27:24 +0000 (11:27 +0000)]
Make apps using '#define _POSIX_C_SOURCE' compile.

15 years agoAdjust list width.
Sascha Wildner [Wed, 19 Nov 2008 19:30:02 +0000 (19:30 +0000)]
Adjust list width.

15 years agoRaise WARNS to 6 and silence resulting warnings.
Sascha Wildner [Wed, 19 Nov 2008 18:41:30 +0000 (18:41 +0000)]
Raise WARNS to 6 and silence resulting warnings.

15 years agoRaise WARNS to 6 and silence resulting warnings.
Sascha Wildner [Wed, 19 Nov 2008 17:46:55 +0000 (17:46 +0000)]
Raise WARNS to 6 and silence resulting warnings.

15 years agoDon't use "timeout" mode for RX HC. The "collect" mode gives much simpler
Sepherosa Ziehau [Wed, 19 Nov 2008 13:57:49 +0000 (13:57 +0000)]
Don't use "timeout" mode for RX HC.  The "collect" mode gives much simpler
RX HC formula:
  if (packets > rx_bds || time since 1st packet > rx_ticks)
    generate interrupt

Adjust RX BDs to 128, and RX ticks to 125.  This gives 12000~14000/s interrupt
rate across all ranges of packet size; this also gives good RX performance:
- 1GB line rate for bulk data (TCP/UDP)
- Peak packet per second rate of the hardware

Thank Brad <brad@comstyle.com> very much!

15 years agoHopefully more bulletproof workaround to fix problems with SATA ATAPI
Hasso Tepper [Wed, 19 Nov 2008 08:12:16 +0000 (08:12 +0000)]
Hopefully more bulletproof workaround to fix problems with SATA ATAPI
drives - don't allow DMA for requests with length not multiple of 16 bytes.
Matches the behaviour of the Linux ATA stack.

15 years agoWe don't have /dev/audio support any more, but make it symlink to /dev/dsp
Hasso Tepper [Wed, 19 Nov 2008 07:46:05 +0000 (07:46 +0000)]
We don't have /dev/audio support any more, but make it symlink to /dev/dsp
to preserve compatibility with older programs (all pkgsrc apps by default
up to 2008Q3).

15 years agoIf a user has turned on snapshots in /tmp and has clear_tmp_enable set,
Sascha Wildner [Tue, 18 Nov 2008 12:56:50 +0000 (12:56 +0000)]
If a user has turned on snapshots in /tmp and has clear_tmp_enable set,
do not remove the /tmp/snapshots/ dir (similar to lost+found/).

15 years agoMake ath(4) compilable into the kernel and add it to LINT.
Sascha Wildner [Mon, 17 Nov 2008 11:55:20 +0000 (11:55 +0000)]
Make ath(4) compilable into the kernel and add it to LINT.

Adapted-from: FreeBSD

15 years agoBring in some fixes from FreeBSD. Amongst other fixes, like panics in debug
Hasso Tepper [Mon, 17 Nov 2008 07:45:48 +0000 (07:45 +0000)]
Bring in some fixes from FreeBSD. Amongst other fixes, like panics in debug
mode, it should also workaround the problems (interrupt storms etc) with
SATA ATAPI devices many of us have seen lately.

Obtained-from: FreeBSD

15 years agoFix some typos in user visible messages.
Sascha Wildner [Sun, 16 Nov 2008 18:44:00 +0000 (18:44 +0000)]
Fix some typos in user visible messages.

15 years agoRestore rev1.6 in different form
Sepherosa Ziehau [Sat, 15 Nov 2008 12:34:38 +0000 (12:34 +0000)]
Restore rev1.6 in different form

15 years agoReturn the value returned by lwkt_domsg()
Sepherosa Ziehau [Sat, 15 Nov 2008 11:58:16 +0000 (11:58 +0000)]
Return the value returned by lwkt_domsg()

15 years agoParallelize bridge_input step 1/2:
Sepherosa Ziehau [Sat, 15 Nov 2008 11:46:37 +0000 (11:46 +0000)]
Parallelize bridge_input step 1/2:
Parallelize bridge route information.  See the commet at the head of
net/if_bridge.c for the detailed information.

15 years ago- Defer deep copyout which happens in bridge_ioctl_{gifs,rts}(), so that
Sepherosa Ziehau [Sat, 15 Nov 2008 04:50:23 +0000 (04:50 +0000)]
- Defer deep copyout which happens in bridge_ioctl_{gifs,rts}(), so that
  all bridge "get" ioctls could be dispatched to BRIDGE_CFGPORT.
  This could ease upcoming bridge_input parallelizing.
- Change the handling of ifbac_len==0 in bridge_ioctl_rts() a little bit
  to match bridge_ioctl_gifs()

15 years agoDeactivate -Werror for now.
Sascha Wildner [Fri, 14 Nov 2008 15:04:42 +0000 (15:04 +0000)]
Deactivate -Werror for now.

15 years agoDefault timeout is 1200
Sepherosa Ziehau [Fri, 14 Nov 2008 12:52:04 +0000 (12:52 +0000)]
Default timeout is 1200

15 years ago- Defer bridge callouts to BRIDGE_CFGPORT using dropable priority message.
Sepherosa Ziehau [Fri, 14 Nov 2008 12:48:06 +0000 (12:48 +0000)]
- Defer bridge callouts to BRIDGE_CFGPORT using dropable priority message.
- Remove unnecessary callout_stop(); bridge_stop() has already done those.

15 years agoSome mdoc cleanup: Improve markup, adjust list width.
Sascha Wildner [Thu, 13 Nov 2008 21:46:03 +0000 (21:46 +0000)]
Some mdoc cleanup: Improve markup, adjust list width.

15 years agoUpdate and switch to UTF-8.
Sascha Wildner [Thu, 13 Nov 2008 21:09:48 +0000 (21:09 +0000)]
Update and switch to UTF-8.

15 years agoSerialize bridge up/down by netisr0; this could ease the upcoming bridge_input
Sepherosa Ziehau [Thu, 13 Nov 2008 11:30:25 +0000 (11:30 +0000)]
Serialize bridge up/down by netisr0; this could ease the upcoming bridge_input
parallelizing work.

15 years agoMT_TAG has gone for a long timer, GC bridge_handoff_notag()
Sepherosa Ziehau [Thu, 13 Nov 2008 10:56:40 +0000 (10:56 +0000)]
MT_TAG has gone for a long timer, GC bridge_handoff_notag()

15 years agoAllow an error code to be returned in the head.error element of the ioctl.
Matthew Dillon [Thu, 13 Nov 2008 02:23:53 +0000 (02:23 +0000)]
Allow an error code to be returned in the head.error element of the ioctl.

15 years agoHAMMER: Correct minor oops in last commit. Do not allow downgrading.
Matthew Dillon [Thu, 13 Nov 2008 02:23:29 +0000 (02:23 +0000)]
HAMMER: Correct minor oops in last commit.  Do not allow downgrading.

15 years agoHAMMER:
Matthew Dillon [Thu, 13 Nov 2008 02:18:43 +0000 (02:18 +0000)]
HAMMER:

* Add version management ioctls to allow continued work on the filesystem
  without interfering with production systems.

* Refuse to mount an unsupported version.

15 years agoHAMMER utilities:
Matthew Dillon [Thu, 13 Nov 2008 02:04:27 +0000 (02:04 +0000)]
HAMMER utilities:

* Add the 'version' and 'version-upgrade' directives.

15 years agoSilence -Wold-style-definition.
Sascha Wildner [Wed, 12 Nov 2008 21:44:59 +0000 (21:44 +0000)]
Silence -Wold-style-definition.

15 years agoNo .Pp needed here.
Sascha Wildner [Wed, 12 Nov 2008 15:18:26 +0000 (15:18 +0000)]
No .Pp needed here.

15 years ago- Update comment: bridge_rtupdate() is no longer called within interrupt
Sepherosa Ziehau [Wed, 12 Nov 2008 13:46:04 +0000 (13:46 +0000)]
- Update comment: bridge_rtupdate() is no longer called within interrupt
  threads.
- Use M_WAITOK when allocating bridge route node.

15 years agoAllocate if_start_nmsg from M_LWKTMSG
Sepherosa Ziehau [Wed, 12 Nov 2008 11:03:28 +0000 (11:03 +0000)]
Allocate if_start_nmsg from M_LWKTMSG

15 years agoFix a bootstrapping issue, UF_NOHISTORY may not exist on older systems.
Matthew Dillon [Wed, 12 Nov 2008 00:16:46 +0000 (00:16 +0000)]
Fix a bootstrapping issue, UF_NOHISTORY may not exist on older systems.

Reported-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
15 years agoGrab some changes from FreeBSD:
Peter Avalos [Tue, 11 Nov 2008 18:06:12 +0000 (18:06 +0000)]
Grab some changes from FreeBSD:

-Use getopt.
-Use err/warn.

While I'm here, fix some whitespace, spelling, and unneeded (void) casts.

15 years agoRecord number of valid ifaddr containers in ifaddr, so we could use
Sepherosa Ziehau [Tue, 11 Nov 2008 13:48:01 +0000 (13:48 +0000)]
Record number of valid ifaddr containers in ifaddr, so we could use
atomic_fetchadd_int() to avoid N->0 race instead of using thread based
serialization.  Remove the netmsg embedded in ifaddr_container; it is
no longer needed/used.

15 years ago- Use priority message to carry out ipfw callout.
Sepherosa Ziehau [Tue, 11 Nov 2008 12:59:11 +0000 (12:59 +0000)]
- Use priority message to carry out ipfw callout.
- Mark the callout message with DROPABLE flag, so we don't need to
  do additional netmsg_service_sync() in module event function and
  the callout function itself could much straightforward.