]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
13 years agoMake freebsd-update(8) smarter in how it handles $FreeBSD$ tags in
cperciva [Wed, 11 May 2011 15:23:27 +0000 (15:23 +0000)]
Make freebsd-update(8) smarter in how it handles $FreeBSD$ tags in
configuration files.

If the current file differs from the canonical version from the old release
only due to differences in the $FreeBSD$ tag (which can happen if the system
was installed from source code, depending on how the src tree was checked out)
then freebsd-update will treat the file as "unmodified" and silently update
it to the "clean" version in the new release.

If the only change being made to a configuration file is in the $FreeBSD$
tag (e.g., for any configuration files which have been modified locally, now
that we're using SVN and the $FreeBSD$ tag changes when a branch is created),
freebsd-update will no longer print the diff and prompt "Does this look
reasonable (y/n)?".

Nagged by: pgollucci
MFC after: 1 month

13 years agoMake the NF calibration logic (hopefully!) more resistive to noisy
adrian [Wed, 11 May 2011 13:40:13 +0000 (13:40 +0000)]
Make the NF calibration logic (hopefully!) more resistive to noisy
environments.

In setups where NF calibration can take a while, don't load the CCA
and kick off a new NF calibration if the previous one hasn't yet
completed. This shouldn't happen unless the environment is noisy but
those exist (hi phk!).

Here, if the previous NF hasn't completed when ar5416LoadNf() is run
(which reads the NF), it skips updating the history buffer, loading
the NF CCA array and kicking off the next NF cal. It's hoped it'll
occur in the next long calibration interval.

Obtained from: Atheros, ath9k, my local HAL

13 years agoAlways log if the NF CCA load fails; so users with debugging enabled
adrian [Wed, 11 May 2011 13:25:43 +0000 (13:25 +0000)]
Always log if the NF CCA load fails; so users with debugging enabled
can see they're likely in a very noisy environment.

13 years agoMake sure the chip is awake before writing to it to finally detach
adrian [Wed, 11 May 2011 13:24:17 +0000 (13:24 +0000)]
Make sure the chip is awake before writing to it to finally detach
it.

Obtained from: Atheros

13 years agoAdd a new flag - HAL_DEBUG_UNMASKABLE - which always logs a debug message
adrian [Wed, 11 May 2011 13:22:41 +0000 (13:22 +0000)]
Add a new flag - HAL_DEBUG_UNMASKABLE - which always logs a debug message
(when debug is enabled) no matter what.

13 years agoRemove unused variable
adrian [Wed, 11 May 2011 13:20:25 +0000 (13:20 +0000)]
Remove unused variable

13 years agoRemove the initial NF completion check.
adrian [Wed, 11 May 2011 11:02:20 +0000 (11:02 +0000)]
Remove the initial NF completion check.

This is taking quite a while for some people in some situations
(eg AR5418 in phk's Abusive Radio Environment).

Instead, the rest of the calibration related code should
ensure that a NF calibration has occured before reading NF
values and kicking off another NF calibration.

The channel should also likely be marked as "noisy" (CWINT)
if the NF calibration takes too long.

13 years agoRemove a now unneeded comment..
adrian [Wed, 11 May 2011 10:30:31 +0000 (10:30 +0000)]
Remove a now unneeded comment..

13 years agoRestore the RSSI threshold after writing the board values.
adrian [Wed, 11 May 2011 09:47:48 +0000 (09:47 +0000)]
Restore the RSSI threshold after writing the board values.

This would be overwritten by the board initvals written in ah->writeIni().

13 years agoUse new option names for NFS client and NFS server
gonzo [Wed, 11 May 2011 00:52:41 +0000 (00:52 +0000)]
Use new option names for NFS client and NFS server

See UPDATING entry for 20110427

13 years agoAdd an ATOMIC_CLEAR_LONG.
marius [Tue, 10 May 2011 21:18:45 +0000 (21:18 +0000)]
Add an ATOMIC_CLEAR_LONG.

13 years agoFix whitespace.
marius [Tue, 10 May 2011 18:41:46 +0000 (18:41 +0000)]
Fix whitespace.

13 years agoFix a bug in r221407; this driver doesn't add the media itself.
marius [Tue, 10 May 2011 18:38:01 +0000 (18:38 +0000)]
Fix a bug in r221407; this driver doesn't add the media itself.

PR: 156893

13 years agoAdd SC_PIXEL_MODE to GENERIC for amd64 and i386.
jkim [Tue, 10 May 2011 16:44:16 +0000 (16:44 +0000)]
Add SC_PIXEL_MODE to GENERIC for amd64 and i386.

Requested by: many

13 years agobitcount32: replace lengthy comment with a reference to SWAR
avg [Tue, 10 May 2011 15:08:13 +0000 (15:08 +0000)]
bitcount32: replace lengthy comment with a reference to SWAR

MFC after: 5 days

13 years agodtrace: remove unused code
avg [Tue, 10 May 2011 15:05:27 +0000 (15:05 +0000)]
dtrace: remove unused code

Which is also useless, IMO.

MFC after: 5 days

13 years agoOnly try to set up IPIs at boot on systems that actually have more than one
nwhitehorn [Tue, 10 May 2011 14:03:05 +0000 (14:03 +0000)]
Only try to set up IPIs at boot on systems that actually have more than one
CPU. This fixes a panic observed on Heathrow-based systems without
SMP-capable PICs when the kernel had both options SMP and INVARIANTS.

MFC after: 5 days

13 years agoRegen.
ru [Tue, 10 May 2011 13:25:42 +0000 (13:25 +0000)]
Regen.

13 years agoReassurance for impatient.
ru [Tue, 10 May 2011 13:01:11 +0000 (13:01 +0000)]
Reassurance for impatient.

13 years agoStyle.
ru [Tue, 10 May 2011 12:57:39 +0000 (12:57 +0000)]
Style.

13 years agoAdd an entry for the SIIG Quartet Serial 850 which uses an Oxford
jhb [Tue, 10 May 2011 12:40:35 +0000 (12:40 +0000)]
Add an entry for the SIIG Quartet Serial 850 which uses an Oxford
chip with a non-default clock.

PR: kern/147583
MFC after: 1 week

13 years ago- There now exist options that have different defaults depending on
ru [Tue, 10 May 2011 11:14:40 +0000 (11:14 +0000)]
- There now exist options that have different defaults depending on
  the architecture, reflect this in documentation.  For such
  options, both WITH_FOO and WITHOUT_FOO description files should
  be provided.

  Prodded by: des

- Setting a build option may enforce other build options, try harder
  to detect this case.

- Setting a build option may change other option's default value,
  try harder to detect this case.

13 years agoThe Objective C support was removed in r220755.
ru [Tue, 10 May 2011 11:06:14 +0000 (11:06 +0000)]
The Objective C support was removed in r220755.

Submitted by: des

13 years agoRemoved unsupported options MK_BINUTILS and MK_GCC that were
ru [Tue, 10 May 2011 11:00:40 +0000 (11:00 +0000)]
Removed unsupported options MK_BINUTILS and MK_GCC that were
silently added in r218936.

13 years agoFlush the scriptfile whenever we see a non-graphical character to get
phk [Tue, 10 May 2011 10:58:57 +0000 (10:58 +0000)]
Flush the scriptfile whenever we see a non-graphical character to get
more real-time logging, without forcing a write(2) on every single
character.

13 years agoAnother fallout from r220359: MK_GNU_CPIO was mistakenly re-added
ru [Tue, 10 May 2011 10:48:43 +0000 (10:48 +0000)]
Another fallout from r220359: MK_GNU_CPIO was mistakenly re-added
after being removed in r205702.  Remove it again.

13 years agoAR9285 (Kite) fixes.
adrian [Tue, 10 May 2011 04:32:27 +0000 (04:32 +0000)]
AR9285 (Kite) fixes.

* Correct some of the silicon revision checks to match what
  the Atheros HAL does. (See [1] below.)

* Move the PA cal and init cal method assignment to -after-
  the mac version/revision IDs are stored. The AR9285 init
  cal was never being called.

* Enable ANI.

Note Kite 1.0 and 1.1 were prototypes that shouldn't be seen
in the wild. Linux ath9k simply removed the prototype code from
their codebase. I'm going to leave it in there for now but
make it conditionally compilable in the future.

Obtained from: Atheros

13 years agoAdd VID for Simtec Electronics.
bms [Tue, 10 May 2011 02:38:44 +0000 (02:38 +0000)]
Add VID for Simtec Electronics.
Add PID for Simtec Electronics EntropyKey, a hardware random number generator.

13 years agoTypo. For USB devices, 'serial' should be 'sernum'.
bms [Tue, 10 May 2011 02:34:11 +0000 (02:34 +0000)]
Typo. For USB devices, 'serial' should be 'sernum'.
See sys/dev/usb/usb_device.c for what devctl_notify() gets.

13 years agoRegenerated with latest configure script.
delphij [Tue, 10 May 2011 01:05:41 +0000 (01:05 +0000)]
Regenerated with latest configure script.

MFC after: 1 month

13 years agoMFV: Update to less v443.
delphij [Mon, 9 May 2011 21:51:59 +0000 (21:51 +0000)]
MFV: Update to less v443.

MFC after: 1 month

13 years agoAnother long standing vm bug found at Isilon:
mlaier [Mon, 9 May 2011 20:27:49 +0000 (20:27 +0000)]
Another long standing vm bug found at Isilon:
 Fix a race between vm_object_collapse and vm_fault.

Reviewed by: alc@
MFC after: 3 days

13 years agoRecognize BCM5719C PHY.
yongari [Mon, 9 May 2011 20:20:43 +0000 (20:20 +0000)]
Recognize BCM5719C PHY.

Submitted by: Geans Pin at Broadcom

13 years agoSince r117657, bge(4) does not enable buffer manager for BCM5705 or
yongari [Mon, 9 May 2011 20:10:46 +0000 (20:10 +0000)]
Since r117657, bge(4) does not enable buffer manager for BCM5705 or
newer controllers.  However, all data sheet I have access has no
indication that buffer manager should not be touched on these
controllers.  It seems the buffer manager always runs on BCM5705 or
newer controllers. Some controller(e.g. BCM5719) needs other buffer
manager configuration so driver should enable buffer manager for
all controllers.  Both Linux and OpenBSD/NetBSD use the same
approach.
This change polls enable bit of block to know whether specified
block was really stopped as well as enabling buffer manager for all
controllers in driver initialization.

Obtained from: NetBSD

13 years ago- Simplify multicast address programming.
davidch [Mon, 9 May 2011 18:46:53 +0000 (18:46 +0000)]
- Simplify multicast address programming.
- Fix an incorrect "uint32_t *" cast in bxe_set_rx_mode().

Submitted by:   yongari@
Approved by:    davidch@
MFC after:      Two weeks

13 years agoMove VT switching hack for suspend/resume from bus drivers to syscons.c
jkim [Mon, 9 May 2011 18:46:49 +0000 (18:46 +0000)]
Move VT switching hack for suspend/resume from bus drivers to syscons.c
using event handlers.  A different version was

Submitted by: Taku YAMAMOTO (taku at tackymt dot homeip dot net)

13 years agoImplement boot-time TSC synchronization test for SMP. This test is executed
jkim [Mon, 9 May 2011 17:34:00 +0000 (17:34 +0000)]
Implement boot-time TSC synchronization test for SMP.  This test is executed
when the user has indicated that the system has synchronized TSCs or it has
P-state invariant TSCs.  For the former case, we may clear the tunable if it
fails the test to prevent accidental foot-shooting.  For the latter case, we
may set it if it passes the test to notify the user that it may be usable.

13 years agoDisable diversity combining support until I can get a firm answer
adrian [Mon, 9 May 2011 17:30:25 +0000 (17:30 +0000)]
Disable diversity combining support until I can get a firm answer
from Atheros as to what/when this is supposed to be enabled.

Using the default RX fast diversity settings seems to help quite
a bit.

Whilst I'm here, change the prototype to return HAL_BOOL rather than int.

13 years agoFix a regression I introduced - only swap analog chains if the RX chainmask
adrian [Mon, 9 May 2011 17:10:48 +0000 (17:10 +0000)]
Fix a regression I introduced - only swap analog chains if the RX chainmask
is 0x5.

13 years agoDisable TX STBC - it isn't used for now, but it isn't supported on Kite.
adrian [Mon, 9 May 2011 16:49:40 +0000 (16:49 +0000)]
Disable TX STBC - it isn't used for now, but it isn't supported on Kite.

13 years agoAdd proper build infrastructure for teken.
ed [Mon, 9 May 2011 16:27:39 +0000 (16:27 +0000)]
Add proper build infrastructure for teken.

I'm not sure whether we should install teken as a library on any stock
FreeBSD installation, but I can imagine people want to tinker around
with it now and then. Create a /sys/teken/libteken, which holds a
Makefile to install a shared library version of the terminal emulator,
complete with a manpage.

Also add Makefiles for the demo/stress applications, to build it against
the shared library.

13 years agoWorkaround for broken no-name USB audio devices sold by dealextreme
hselasky [Mon, 9 May 2011 15:57:04 +0000 (15:57 +0000)]
Workaround for broken no-name USB audio devices sold by dealextreme
called "3D sound" and the alike.

MFC after: 14 days

13 years agoImport some initial Kite fixed diversity code from Atheros.
adrian [Mon, 9 May 2011 15:19:49 +0000 (15:19 +0000)]
Import some initial Kite fixed diversity code from Atheros.

For now, the diversity settings are controlled by 'txantenna',
-not- rxantenna. This is because the earlier chipsets had
controllable TX diversity; the RX antenna setting twiddles
the default antenna register. I'll try sort that stuff out at
some point.

Call the antenna switch function from the board setup function
so scans, channel changes, mode changes, etc don't set the
diversity back to a default state too far from what's intended.

Things to todo:

* Squirrel away the last antenna diversity/combining parameters
  and restore them during board setup if HAL_ANT_VARIABLE is
  defined. That way scans, etc don't reset the diversity settings.

* Add some more public facing statistics, rather than what's
  simply logged under HAL_DEBUG_DIVERSITY.

For now, the fixed antenna settings behave better than variable
settings for me. I have some further fiddling to do..

Obtained from: Atheros

13 years agoRemove an un-needed PA cal call here.
adrian [Mon, 9 May 2011 14:04:49 +0000 (14:04 +0000)]
Remove an un-needed PA cal call here.

13 years agoRelease SP's refcount in key_get_spdbyid().
vanhu [Mon, 9 May 2011 13:16:21 +0000 (13:16 +0000)]
Release SP's refcount in key_get_spdbyid().

PR: 156676
Submitted by: Tobias Brunner (tobias@strongswan.org)
MFC after: 1 week

13 years agoThis is a full 32 bit field for kite.
adrian [Mon, 9 May 2011 10:39:15 +0000 (10:39 +0000)]
This is a full 32 bit field for kite.

13 years agoRefactor TCP ISN increment logic. Instead of firing callout at 100Hz to
mav [Mon, 9 May 2011 07:37:47 +0000 (07:37 +0000)]
Refactor TCP ISN increment logic. Instead of firing callout at 100Hz to
keep constant ISN growth rate, do the same directly inside tcp_new_isn(),
taking into account how much time (ticks) passed since the last call.

On my test systems this decreases idle interrupt rate from 140Hz to 70Hz.

13 years agosh: Add \u/\U support (in $'...') for UTF-8.
jilles [Sun, 8 May 2011 17:40:10 +0000 (17:40 +0000)]
sh: Add \u/\U support (in $'...') for UTF-8.

Because we have no iconv in base, support for other charsets is not
possible.

Note that \u/\U are processed using the locale that was active when the
shell started. This is necessary to avoid behaviour that depends on the
parse/execute split (for example when placing braces around an entire
script). Therefore, UTF-8 encoding is implemented manually.

13 years agosh: Optimize variable code by storing the length of the name.
jilles [Sun, 8 May 2011 16:15:50 +0000 (16:15 +0000)]
sh: Optimize variable code by storing the length of the name.

Obtained from: NetBSD

13 years agoFix the 5ghz fast clock logic.
adrian [Sun, 8 May 2011 15:55:52 +0000 (15:55 +0000)]
Fix the 5ghz fast clock logic.

The macro which I incorrectly copied into ah_internal.h assumed
that it'd be called with an AR_SREV_MERLIN_20() check to ensure
it was only enabled for Merlin (AR9280) silicon revision 2.0 or
later.

Trouble is, the 5GHz fast clock EEPROM flag is only valid for
EEPROM revision 16 or greater; it's assumed to be enabled
by default for Merlin rev >= 2.0. This meant it'd be incorrectly
set for AR5416 and AR9160 in 5GHz mode.

This would have affected non-default clock timings such as SIFS,
ACK and slot time. The incorrect slot time was very likely wrong
for 5ghz mode.

13 years ago* Add AR_SREV_KITE macro for later use
adrian [Sun, 8 May 2011 15:25:22 +0000 (15:25 +0000)]
* Add AR_SREV_KITE macro for later use
* Modify AR_SREV_MERLIN_20() to match the Atheros/Linux ath9k behaviour -
  its supposed to match Merlin 2.0 and later Merlin chips.
  AR_SREV_MERLIN_20_OR_LATER() matches AR9280 2.0 and later chips
  (AR9285, AR9287, etc.)

13 years agoRevert r221655:
bcr [Sun, 8 May 2011 14:57:01 +0000 (14:57 +0000)]
Revert r221655:

Various people voiced their concerns about these changes.
Until this is resolved, we should use the old version.

13 years agosh(1): Update BUGS section for UTF-8 support.
jilles [Sun, 8 May 2011 14:03:44 +0000 (14:03 +0000)]
sh(1): Update BUGS section for UTF-8 support.

13 years agoWe now have multiple filesystems (UFS, ZFS, ...), so for tools that only
gavin [Sun, 8 May 2011 12:34:31 +0000 (12:34 +0000)]
We now have multiple filesystems (UFS, ZFS, ...), so for tools that only
operate on one type of filesystem, mention this.
While here, capitalise the use of "UFS" in growfs.8 to match other uses of
the term in other man pages.

MFC after: 1 week

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 12:28:13 +0000 (12:28 +0000)]
Limit number of sectors that can be addressed.

MFC after: 1 week

13 years agoEnable 11n (sans HT40) support.
bschmidt [Sun, 8 May 2011 12:23:01 +0000 (12:23 +0000)]
Enable 11n (sans HT40) support.

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 12:20:30 +0000 (12:20 +0000)]
Limit number of sectors that can be addressed.

MFC after: 1 week

13 years agoJails have a problem in that if the jail directory is world-readable,
bcr [Sun, 8 May 2011 12:16:39 +0000 (12:16 +0000)]
Jails have a problem in that if the jail directory is world-readable,
an attacker with root access to the jail can create a setuid binary for
their own use in the host environment (if they also have this access),
thus breaking root in the host.

This exploit is impossible if the jail's files are not world-readable.
Add instructions to the man page on how to create a jail with the
correct permissions set.

PR: docs/156853
Submitted by: Chris Rees (utisoft at gmail dot com)
Reviewed by: cperciva (security parts)
MFC after: 9 days

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 12:16:39 +0000 (12:16 +0000)]
Limit number of sectors that can be addressed.
Reject table if blkcount from metadata is greater than provider.

13 years agoNotify firmware about various HT parameters once associated.
bschmidt [Sun, 8 May 2011 12:11:20 +0000 (12:11 +0000)]
Notify firmware about various HT parameters once associated.

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 12:11:16 +0000 (12:11 +0000)]
Limit number of sectors that can be addressed.

MFC after: 1 week

13 years agoAdd support for TX packet aggregation.
bschmidt [Sun, 8 May 2011 12:06:12 +0000 (12:06 +0000)]
Add support for TX packet aggregation.

13 years agoAdd support for RX packet aggregation.
bschmidt [Sun, 8 May 2011 11:58:23 +0000 (11:58 +0000)]
Add support for RX packet aggregation.

13 years agoAdd support for transmitting frames at MCS rates.
bschmidt [Sun, 8 May 2011 11:54:38 +0000 (11:54 +0000)]
Add support for transmitting frames at MCS rates.

13 years agoPrepare for transmitting frames at MCS rates:
bschmidt [Sun, 8 May 2011 11:49:50 +0000 (11:49 +0000)]
Prepare for transmitting frames at MCS rates:
- instead of calling iwn_plcp_signal() for every frame, map the expected
  value directly within wn->ridx
- concat plcp, rflags and xrflags, there is no clean byte boundary within
  the flags, for example the antenna setting uses bit 6, 7 and 8
- there is still need for a custom rate to plcp mapping, as those expected
  by the hardware are not conform to the std

13 years agoReplace UINT_MAX to UINT32_MAX.
ae [Sun, 8 May 2011 11:42:51 +0000 (11:42 +0000)]
Replace UINT_MAX to UINT32_MAX.

Pointed out by: kib
MFC after: 1 week

13 years agosh: Add UTF-8 support to pattern matching.
jilles [Sun, 8 May 2011 11:32:20 +0000 (11:32 +0000)]
sh: Add UTF-8 support to pattern matching.

?, [...] patterns match codepoints instead of bytes. They do not match
invalid sequences. [...] patterns must not contain invalid sequences
otherwise they will not match anything. This is so that ${var#?} removes the
first codepoint, not the first byte, without putting UTF-8 knowledge into
the ${var#pattern} code. However, * continues to match any string and an
invalid sequence matches an identical invalid sequence. (This differs from
fnmatch(3).)

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 11:20:27 +0000 (11:20 +0000)]
Limit number of sectors that can be addressed.

MFC after: 1 week

13 years agoLimit number of sectors that can be addressed.
ae [Sun, 8 May 2011 11:16:17 +0000 (11:16 +0000)]
Limit number of sectors that can be addressed.

MFC after: 1 week

13 years agoAllow to specify remote as 'none' again which was broken by r219351, where
pjd [Sun, 8 May 2011 11:10:56 +0000 (11:10 +0000)]
Allow to specify remote as 'none' again which was broken by r219351, where
'none' was defined as a value for checksum.

Reported by: trasz
MFC after: 1 week

13 years agoRead chainmask information before announcing it.
bschmidt [Sun, 8 May 2011 11:05:03 +0000 (11:05 +0000)]
Read chainmask information before announcing it.

13 years agoAdd HT capabilities to probe requests.
bschmidt [Sun, 8 May 2011 11:03:16 +0000 (11:03 +0000)]
Add HT capabilities to probe requests.

13 years agoDisable background scan support for 4965 adapters.
bschmidt [Sun, 8 May 2011 11:01:53 +0000 (11:01 +0000)]
Disable background scan support for 4965 adapters.

On legacy channels every once in a while the firmware throws a SYSASSERT
on line 208. On HT channels though this does always happen and I'm not
aware of any workaround currently.

13 years agoRX aggregation is slightly different then the legacy path, we will only
bschmidt [Sun, 8 May 2011 10:57:44 +0000 (10:57 +0000)]
RX aggregation is slightly different then the legacy path, we will only
receive one RX_PHY for each aggregate and not one RX_PHY per frame.

13 years agoAllocate all TX rings, those will be use for TX packet aggregation.
bschmidt [Sun, 8 May 2011 10:54:50 +0000 (10:54 +0000)]
Allocate all TX rings, those will be use for TX packet aggregation.

13 years agoUse the enhanced TX power information availabe on newer EEPROMs.
bschmidt [Sun, 8 May 2011 10:35:16 +0000 (10:35 +0000)]
Use the enhanced TX power information availabe on newer EEPROMs.

13 years agoHook HT channel setup.
bschmidt [Sun, 8 May 2011 10:31:22 +0000 (10:31 +0000)]
Hook HT channel setup.

13 years agoThe 6000 series adapters have a slightly different offset for band 6,
bschmidt [Sun, 8 May 2011 10:21:42 +0000 (10:21 +0000)]
The 6000 series adapters have a slightly different offset for band 6,
2GHz HT40 channels.

13 years agoRe-add 2 device IDs which got lost.
bschmidt [Sun, 8 May 2011 10:19:29 +0000 (10:19 +0000)]
Re-add 2 device IDs which got lost.

Pointed out by: benjsc

13 years agoDocument the following sysctls:
pjd [Sun, 8 May 2011 09:46:09 +0000 (09:46 +0000)]
Document the following sysctls:

kern.geom.eli.version
kern.geom.eli.key_cache_limit
kern.geom.eli.key_cache_hits
kern.geom.eli.key_cache_misses

MFC after: 1 week

13 years agoFix isitme(), which is used to check if node-specific configuration
trociny [Sun, 8 May 2011 09:31:17 +0000 (09:31 +0000)]
Fix isitme(), which is used to check if node-specific configuration
belongs to our node, and was returning false positive if the first
part of a node name matches short hostname.

Approved by: pjd (mentor)

13 years agoExport GELI class version via sysctl kern.geom.eli.version.
pjd [Sun, 8 May 2011 09:29:21 +0000 (09:29 +0000)]
Export GELI class version via sysctl kern.geom.eli.version.

MFC after: 1 week

13 years agoVersion 6 is compatible with version 5 when it comes to control commands.
pjd [Sun, 8 May 2011 09:25:54 +0000 (09:25 +0000)]
Version 6 is compatible with version 5 when it comes to control commands.

MFC after: 1 week

13 years agoDetect and handle metadata of version 6.
pjd [Sun, 8 May 2011 09:25:16 +0000 (09:25 +0000)]
Detect and handle metadata of version 6.

MFC after: 1 week

13 years agoWhen support for multiple encryption keys was committed, GELI integrity mode
pjd [Sun, 8 May 2011 09:17:56 +0000 (09:17 +0000)]
When support for multiple encryption keys was committed, GELI integrity mode
was not updated to pass CRD_F_KEY_EXPLICIT flag to opencrypto. This resulted in
always using first key.

We need to support providers created with this bug, so set special
G_ELI_FLAG_FIRST_KEY flag for GELI provider in integrity mode with version
smaller than 6 and pass the CRD_F_KEY_EXPLICIT flag to opencrypto only if
G_ELI_FLAG_FIRST_KEY doesn't exist.

Reported by: Anton Yuzhaninov <citrin@citrin.ru>
MFC after: 1 week

13 years agoFix a locking issue showing up on Mac OS X when subscribing to
tuexen [Sun, 8 May 2011 09:11:59 +0000 (09:11 +0000)]
Fix a locking issue showing up on Mac OS X when subscribing to
authentication events. DTLS/SCTP renegotiations trigger the bug.

MFC after: 2 weeks.

13 years agoRemove prototype for a function that no longer exist.
pjd [Sun, 8 May 2011 09:11:04 +0000 (09:11 +0000)]
Remove prototype for a function that no longer exist.

MFC after: 1 week

13 years agoDrop proper key.
pjd [Sun, 8 May 2011 09:09:49 +0000 (09:09 +0000)]
Drop proper key.

MFC after: 1 week

13 years agoAdd magic field to the g_eli_key structure to detect if we are really
pjd [Sun, 8 May 2011 09:08:50 +0000 (09:08 +0000)]
Add magic field to the g_eli_key structure to detect if we are really
operating on proper structures.

MFC after: 1 week

13 years agoCleanup usb_notify_addq_compat(). It should not
hselasky [Sun, 8 May 2011 08:22:11 +0000 (08:22 +0000)]
Cleanup usb_notify_addq_compat(). It should not
be needed any more.

MFC after: 7 days

13 years agoThese EEPROM bits actually defined whether HT/20 and HT/40 support
adrian [Sun, 8 May 2011 08:18:30 +0000 (08:18 +0000)]
These EEPROM bits actually defined whether HT/20 and HT/40 support
for the given channel is available.

It isn't used yet; ar5416GetWirelessModes() needs to be taught
about this rather than assuming HT20/HT40 is available.

13 years agoFiddle with the PLL initialisation order to match ath9k/Atheros HAL.
adrian [Sun, 8 May 2011 07:21:09 +0000 (07:21 +0000)]
Fiddle with the PLL initialisation order to match ath9k/Atheros HAL.

This seems to make the AR9160 behave better during heavy scanning,
where before it'd hang and require a hard reset to recover.

Obtained From: Linux ath9k, Atheros

13 years agoProperly indent the WAR code i pasted in from ath9k a few months
adrian [Sun, 8 May 2011 05:45:06 +0000 (05:45 +0000)]
Properly indent the WAR code i pasted in from ath9k a few months
ago.

13 years ago* Add in a comment about ar5416InitUserSettings() potentially
adrian [Sun, 8 May 2011 05:25:42 +0000 (05:25 +0000)]
* Add in a comment about ar5416InitUserSettings() potentially
  modifying AR_DIAG_SW.

  There's a hardware workaround which sets disabling some errors
  early at startup and clears said bits before the PCU begins
  receiving - it does this to avoid RX descriptor status errors.

  It's possible these bits aren't being completely properly twiddled
  in all instances; but in particular if the diag_reg HAL variable
  is set it won't be setting these bits correctly. I'll review this
  at some point.

 * Disable multicast search on mac address and key id - the driver
   doesn't use it at the moment and thus adhoc may be broken for
   merlin and later.

* Change this to be for Merlin 1.0 (which from what I understand
  wasn't ever publicly released) to be more correct.

13 years agoFiddle with the AR5416 1.0 chainmask setup.
adrian [Sun, 8 May 2011 03:24:17 +0000 (03:24 +0000)]
Fiddle with the AR5416 1.0 chainmask setup.

Apparently all three RX chains need to be enabled before initial calibration
is done, even if only two are configured.

Reorder the alt chain swap bit to match what the Atheros HAL is doing.

Obtained From: ath9k, Atheros

13 years agoChange the new NFS server so that it uses vfs.nfsd naming
rmacklem [Sun, 8 May 2011 01:01:27 +0000 (01:01 +0000)]
Change the new NFS server so that it uses vfs.nfsd naming
for its sysctls instead of vfs.newnfs. This separates the
names from the ones used by the client.

13 years agoFix the IS_5416 checks to actually work correctly.
adrian [Sat, 7 May 2011 18:42:41 +0000 (18:42 +0000)]
Fix the IS_5416 checks to actually work correctly.

I've verified that my AR5416 revision 2.2 (minor revision 0x0A) now
matches the correct checks.

13 years agoTo avoid duplicated warning, move WITNESS_WARN() added in r221597 to the
jh [Sat, 7 May 2011 17:59:07 +0000 (17:59 +0000)]
To avoid duplicated warning, move WITNESS_WARN() added in r221597 to the
branch which doesn't call malloc(9).

Suggested by: kib

13 years agoIn pmap_kextract(), return the physical address for PBVM virtual
marcel [Sat, 7 May 2011 17:23:13 +0000 (17:23 +0000)]
In pmap_kextract(), return the physical address for PBVM virtual
addresses as well (incl. the PBVM page table).

13 years agoAdd new USB ID.
hselasky [Sat, 7 May 2011 16:32:59 +0000 (16:32 +0000)]
Add new USB ID.

Submitted by: Dmitry Luhtionov
MFC after: 7 days