ƒ·1225348
155
231
623
411
246
210
317
219
135
310
124
595
143
141
244
118
202
289
317
603
179
221
144
143
220
684
465
217
217
129
220
527
165
387
1278
120
1216
366
157
162
162
257
1240
718
324
1280
362
666
435
207
207
242
242
383
479
1427
149
149
690
117
117
123
123
123
172
135
248
214
199
285
430
751
274
202
658
182
116
419
130
326
143
280
496
176
318
251
158
305
146
114
113
114
109
911
109
331
113
1346
501
284
845
871
276
107
150
253
502
246
967
249
188
388
99
550
118
249
115
143
1028
346
254
99
260
234
167
118
191
288
465
242
348
264
1569
1865
125
288
208
126
951
114
116
115
251
1148
233
1148
296
181
382
439
338
167
138
323
205
278

K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-02T18:37:48.113575Z
K 7
svn:log
V 133
Remove an incorrect apostrophe.

PR:		160267
Submitted by:	Warren Block <wblock@wonkity.com>
Approved by:	re (kib)
MFC after:	1 week

END
K 10
svn:author
V 3
mdf
K 8
svn:date
V 27
2011-09-02T18:44:57.977178Z
K 7
svn:log
V 530
ufs: Don't assume struct dirent == struct direct

Remove offset alignment requirement, forthcoming changes in struct dirent
will make alignment requirements extremely hard to follow.

Work around clients supplying too small buffer. Don't allocate large
buffer for dirent conversion, it can be used for DoS attack.

Obtained from: ext2_readdir and DragonflyBSD commit
e088dc32f085e0201c97af25afcf361190292de3

GSoC r223158.
Code by: Gleb Kurtsou.

Also eliminate the BYTE_ORDER checking as the struct dirent is filled in
directly.

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-02T18:50:44.283192Z
K 7
svn:log
V 313
This patch adds automatic detection of USB mass storage devices
which does not support the no synchronize cache SCSI command.

The __FreeBSD_version version macro has been bumped and
external kernel modules needs to be recompiled after
this patch.

Approved by:    re (kib)
MFC after:      1 week
PR:		usb/160299

END
K 10
svn:author
V 3
mdf
K 8
svn:date
V 27
2011-09-02T18:52:17.071168Z
K 7
svn:log
V 153
ufs: Stop mangling dirent in ufs_extattr_iterate_directory.
It's meaningless after switching to new dirent format.

GSoC r223159.
Code by: Gleb Kurtsou.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-02T18:52:28.063678Z
K 7
svn:log
V 112
Wording, grammar and markup cleanup.

PR:		159948
Submitted by:	Ben Kaduk <kaduk@mit.edu>
Approved by:	re (kib)

END
K 10
svn:author
V 2
jh
K 8
svn:date
V 27
2011-09-02T19:29:37.875326Z
K 7
svn:log
V 225
Restore behavior of the autoboot_delay="-1" boot menu setting to the
pre-r222417 state. The behavior was essentially reversed in r222417
which can cause confusion.

PR:		159775
Submitted by:	Devin Teske
Approved by:	re (kib)

END
K 10
svn:author
V 4
stas
K 8
svn:date
V 27
2011-09-02T20:35:22.099125Z
K 7
svn:log
V 125
- Fix NULL pointer dereference when a packet of uneven size is being
  transmitted.

Approved by:	re (kib)
MFC after:	3 days

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-02T23:51:55.602711Z
K 7
svn:log
V 41
- Fix two more casts

Submitted by:	ache

END
K 10
svn:author
V 8
rmacklem
K 8
svn:date
V 27
2011-09-03T00:28:53.698730Z
K 7
svn:log
V 212
Fix the NFS servers so that they can do a Lookup of "..",
which requires that ni_strictrelative be set to 0, post-r224810.

Tested by:	swills (earlier version), geo dot liaskos at gmail.com
Approved by:	re (kib)

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-03T00:47:28.300508Z
K 7
svn:log
V 30
- Add support for escaped dot

END
K 10
svn:author
V 8
rmacklem
K 8
svn:date
V 27
2011-09-03T01:45:37.255249Z
K 7
svn:log
V 497
MFC: r225049
Fix the NFSv4 server so that it returns NFSERR_SYMLINK when
an attempt to do an Open operation on any type of file other
than VREG is done. A recent discussion on the IETF working group's
mailing list (nfsv4@ietf.org) decided that NFSERR_SYMLINK
should be returned for all non-regular files and not just symlinks,
so that the Linux client would work correctly.
This change does not affect the FreeBSD NFSv4 client and is not
believed to have a negative effect on other NFSv4 clients.

END
K 10
svn:author
V 3
des
K 8
svn:date
V 27
2011-09-03T03:12:33.671800Z
K 7
svn:log
V 51
Forgot this nit in r221107.

Approved by:	re (kib)

END
K 10
svn:author
V 3
des
K 8
svn:date
V 27
2011-09-03T03:13:35.255326Z
K 7
svn:log
V 49
insta-mfh r225359: cosmetic nit in usage string.

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-03T07:13:45.643843Z
K 7
svn:log
V 149
Upgrade to BIND version 9.8.1. Release notes at:

https://deepthought.isc.org/article/AA-00446/81/
or
/usr/src/contrib/bind9/

Approved by:	re (kib)

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-03T07:15:14.459920Z
K 7
svn:log
V 24
BIND 9.8.1 upgrade over

END
K 10
svn:author
V 8
jchandra
K 8
svn:date
V 27
2011-09-03T08:03:25.407659Z
K 7
svn:log
V 104
HAL updates

Update hardware access register defs and functions to be more in sync
with other platforms

END
K 10
svn:author
V 5
trasz
K 8
svn:date
V 27
2011-09-03T08:08:24.012084Z
K 7
svn:log
V 194
Fix panic that happens when fork(2) fails due to a limit other than
the rctl one - for example, it happens when someone reaches maximum
number of processes in the system.

Approved by:	re (kib)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-03T08:31:59.453742Z
K 7
svn:log
V 224
MFC r225040:
Prevent the hiwatermark for the unix domain socket from becoming
effectively negative. Often seen as upstream fastcgi connection timeouts
in nginx when using sendfile over unix domain sockets for communication.

END
K 10
svn:author
V 3
dim
K 8
svn:date
V 27
2011-09-03T11:41:00.482841Z
K 7
svn:log
V 510
When libexec/rtld-elf/rtld.c is compiled with clang, the r_debug_state()
function (a hook necessary for gdb support), is inlined, but since the
function contains no code, no calls to it are generated.  When gdb is
debugging a dynamically linked program, this causes backtraces to be
corrupted.

Fix it by marking the function __noinline, and inserting an empty asm
statement, that pretends to clobber memory.  This forces the compiler to
emit calls to r_debug_state() throughout rtld.c.

Approved by:	re (kib)

END
K 10
svn:author
V 5
gavin
K 8
svn:date
V 27
2011-09-03T12:59:05.693040Z
K 7
svn:log
V 85
Add my doc mentor/mentee relationships.

Approved by:	re (kib), bcr, jkois (mentors)

END
K 10
svn:author
V 8
jchandra
K 8
svn:date
V 27
2011-09-03T15:34:18.538035Z
K 7
svn:log
V 123
HAL updates

mmio.h renamed to haldefs.h - contains more than mmio now
use cpufunc.h functions for status register changes

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-03T20:46:30.247343Z
K 7
svn:log
V 47
MFC: r225215

Fix memory leaks in error cases.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-03T21:28:26.607664Z
K 7
svn:log
V 46
MFC: r225215

Fix memory leak in error cases.

END
K 10
svn:author
V 5
trasz
K 8
svn:date
V 27
2011-09-04T05:04:34.680145Z
K 7
svn:log
V 125
Since r224036 the cputime and wallclock are supposed to be in seconds,
not microseconds.  Make it so.

Approved by:	re (kib)

END
K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2011-09-04T13:07:02.426529Z
K 7
svn:log
V 587
Interrupts are disabled/enabled when entering and exiting the KDB context.
While this is generally good, it brings along a serie of problems,
like clocks going off sync and in presence of SW_WATCHDOG, watchdogs
firing without a good reason (missed hardclock wdog ticks update).

Fix the latter by kicking the watchdog just before to re-enable the interrupts.
Also, while here, not rely on users to stop the watchdog manually when
entering DDB but do that when entering KDB context.

Sponsored by: Sandvine Incorporated
Reviewed by:	emaste, rstone
Approved by:	re (kib)
MFC after:	1 week

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-04T14:09:59.035505Z
K 7
svn:log
V 369
When RX'ing a high throughput UDP stream, NF calibration loads will almost
always fail. For now, just drop the number of times we try the NF load
to a low value. I'll have to investigate splitting the longcal into two
halves and pushing the NF threshold cap (ie loading in -50 dBm) code into
something which occurs in a second phase, rather than doing this tight loop.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-04T14:28:06.216198Z
K 7
svn:log
V 119
MFC: r225228

Remove broken link to supported wifi cards. There doesn't appear to
be an equivalent list anywhere else.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-04T14:30:22.372867Z
K 7
svn:log
V 119
MFC: r225228

Remove broken link to supported wifi cards. There doesn't appear to
be an equivalent list anywhere else.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-04T14:53:40.968344Z
K 7
svn:log
V 34
hide this behind NFCAL debugging.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-04T14:55:51.379278Z
K 7
svn:log
V 124
Commit some RX interrupt mitigation values which work better on this board.
I should really make this runtime configurable.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-04T14:58:57.671126Z
K 7
svn:log
V 431
A previous commit from me used the aggregate length for packet completion.

This made for some very bad performing 11n throughput because the rate
used for selection (ie, the first frame length) didn't match up with the
rate completion (ie, the aggregate length.)

For now, just use the frame length of the first frame. This results in
better throughput - which isn't 100% reliable, but it's better than what
I'm currently seeing.

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-04T16:26:43.772557Z
K 7
svn:log
V 71
- Add a simple test program to generate and print out regex heuristics

END
K 10
svn:author
V 7
thompsa
K 8
svn:date
V 27
2011-09-04T22:06:32.316312Z
K 7
svn:log
V 290
On the first loop for generating a bridge MAC address use the local
hostid, this gives a good chance of keeping the same address over
reboots. This is intended to help IPV6 and similar which generate
their addresses from the mac.

PR:		kern/160300
Submitted by:	mdodd
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T00:10:37.463436Z
K 7
svn:log
V 1181
Migrate the rate control selection out from the packet setup code and
instead call it just before the packet is TXed.

For non-aggregate packets, call it just before the frame is queued
to hardware.

For aggregate packets, call it on the first frame in the aggregate
list so the aggregate code itself has a rate control selection to
use when building the aggregate.

This allows for some useful behaviour:

* rate selection is now not done per-frame when doing aggregation.
  Aggregation only requires a rate lookup for the first frame.
  The rest are ignored.

* The rate control module now gets a more "balanced" idea of how many
  times findrate is called versus completion. Beforehand, findrate
  would be called for each frame, but completion would be called for
  the entire aggregate.

* software retransmission now allows for a new rate lookup - that way
  the retransmits aren't done using the same failing rates.

This improves the aggregate throughput stability but it doesn't fix
things entirely - primarily because ath_rate_sample is still making
some very poor rate control decisions (eg dipping down to single-stream
rates and trying MCS14/15 when it doesn't work.)

END
K 10
svn:author
V 7
linimon
K 8
svn:date
V 27
2011-09-05T03:43:43.984347Z
K 7
svn:log
V 24
Clean up some comments.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T06:11:17.610502Z
K 7
svn:log
V 1119
Make the default rate choice semi 11n aware.

There's a couple of bugs here which I'm trying to fix:

* There's a bit of code which marks all larger size bins as having failures
  if a smaller bin has a failure.
* If there's no large packet TX (say I'm doing an iperf TCP test in the opposite
  direction) then the 1600-byte bins all get marked with both TX failures and
  _successive_ TX failures.
* Thus when it's time to TX a large packet, all the rates are marked as
  invalid.

This commit fixes the first problem - that the default rate selection was simply
choosing the highest MCS rate (and thus MCS15 was being used on a two stream NIC)
and so things like DHCP may take a few retransmits.

It also fixes the problem where all those MCS rates are slowly marked as invalid
by failures in small packets; the only rate left over is the base rate
(6mbit for 11a).

This doesn't fix a related problem - where subsequent TXes doesn't ever probe the
higher rates. Because the successive failures exist, those rates aren't ever
selected. pick_sample_rate() should be trying those after a while - 10 seconds by
default.

END
K 10
svn:author
V 3
art
K 8
svn:date
V 27
2011-09-05T06:54:13.528614Z
K 7
svn:log
V 273
MFC r225234:

Make sure RPC calls over UDP return RPC_INTR status if the process has
been interrupted in a restartable syscall. Otherwise we could end up
in an (almost) endless loop in clnt_reconnect_call().

PR: kern/160198
Reviewed by: rmacklem
Approved by: avg (mentor)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T07:36:16.772377Z
K 7
svn:log
V 62
Fix the ht40/preamble flag, I had those around the wrong way.

END
K 10
svn:author
V 7
pluknet
K 8
svn:date
V 27
2011-09-05T07:57:56.294911Z
K 7
svn:log
V 66
MFC r225330:

 Cosmetic cleanup: remove unused #define LIBMEMSTAT

END
K 10
svn:author
V 7
pluknet
K 8
svn:date
V 27
2011-09-05T07:59:23.445303Z
K 7
svn:log
V 66
MFC r225330:

 Cosmetic cleanup: remove unused #define LIBMEMSTAT

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-05T08:41:57.419932Z
K 7
svn:log
V 164
MFC r225076:
Apply the limit to avoid the overflows in the radix tree subr_blist.c
after the conversion of the swap device size to the page size units,
not before.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T08:44:46.975961Z
K 7
svn:log
V 1143
Begin teaching sample about MCS stuff.

The basic differences include:

* the "goodness" function is based purely on the average tx time
  calculation. For 11bg and 11a, the "best" tx time for a given
  rate is roughly half of the previous rate. So the metric of
  "is the best rate tx time better than double the current rate
  tx time" doesn't work for MCS.

* 2-stream MCS rates have slightly higher best TX times than
  one-stream rates.

This commit begins to flesh out some more average tx time based
calculations, but it allows MCS rate avg times to be slightly
higher than the "best", as long as the MCS rate is larger.
This is absolutely not correct - as higher MCS rate doesn't
at all mean "better throughput" but it at least means it'll
be considered.

This now finds and settles on an MCS rate that's giving a low
average tx time, rather than wildly oscillating all over the
place. The problem now is that the average TX time calculation
in update_stats() doesn't take into account the aggregate
frame behaviour, so any successful subframe will mean the average
TX time will stay quite low. I'll address this in a followup
commit.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T09:31:26.638911Z
K 7
svn:log
V 622
Add a work-around for 11n aggregation packet behaviour and the lack of
correct average tx time calculation.

The average tx time calculation doesn't take into account the
sub-frame success rate. This means that in some instances, the
success rate may be lower than useful (say, around 60-70%) but
because the A-MPDU frame was partially successful, the calculated
TX time is low.

I'm not sure what the correct behaviour should be, but for now,
just enforce that the packet TX success rate must be higher.

Note: this is NOT correct (ie, a slightly higher failure rate is
ok if the aggregate throughput/latency is better.)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T09:39:35.304611Z
K 7
svn:log
V 228
Disable the code which also sets the successive_retries field
for the larger size buckets.

This is causing issues with 11n TX where aggregate frames can
sometimes entirely fail. This started marking other rates as
unavailable.

END
K 10
svn:author
V 2
mm
K 8
svn:date
V 27
2011-09-05T09:49:28.037071Z
K 7
svn:log
V 1187
MFC r225167, r225329:

MFC r225167:
Import bugfix for reading and extracting of FreeBSD ISO images with tar.

Vendor revision 3645 (merge of 3642):
Change the mechanism handling a rr_moved directory,
which is Rockridge extension that can exceed the limitation of
a maximum directory depth of ISO 9660.
  - Stop reading all entries at a time.
  - Connect "CL" entry to "RE" entry dynamically, which "CL" and "RE"
    have information to rebuild a full directory tree.
  - Tweak some related tests since we use Headsort for re-ordering
    entries and it cannot make a steady order when the keies of
    the entries are the same.

http://code.google.com/p/libarchive/issues/detail?id=168

MFC r225329:
Import additional bugfix for reading and extracting makefs-created
ISO images with tar.

Vendor revision 3648 (merge of 3647):
Additional fix to issue 168 because the change of r3642 was not sufficient.
  - Make sure "CL" entry appear after its "RE" entry which the "CL" entry
    should be connected with.
  - Give consideration to the case that the top level "RE" entry has
    already been exposed outside before its tree.

Obtained from:	libarchive (release/2.8, svn rev 3645, 3648)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T10:09:53.414140Z
K 7
svn:log
V 266
Hacky work around for now - disable the last rate series entry
when forming aggregates. This is a very low MCS rate in all MCS
instances and the 4ms frame length limit ends up being quite low.
Thus larger aggregates don't get a chance to form.

This isn't optimal..

END
K 10
svn:author
V 8
jchandra
K 8
svn:date
V 27
2011-09-05T10:45:29.550127Z
K 7
svn:log
V 568
MIPS XLP platform code update.

* Update the hardware access register definitions and functions to bring
  them in line with other Netlogic software.
* Update the platform bus to use PCI even for on-chip devices. Add a dummy
  PCI driver to ignore on-chip devices which do not need driver.
* Provide memory and IRQ resource allocation code for on-chip devices
  which cannot get it from PCI config.
* add support for on-chip PCI and USB interfaces.
* update conf files, enable pci and retain old MAXCPU until we can support
  >32 cpus.

Approved by:	re(kib), jmallett

END
K 10
svn:author
V 2
ed
K 8
svn:date
V 27
2011-09-05T11:08:33.881444Z
K 7
svn:log
V 343
MFC r224659, r224674 and r224765:

  Remove NIS entries from /etc/nsswitch.conf for WITHOUT_NIS.

  We already modify various configuration files in /etc based on build
  configuration. This is not done for nsswitch.conf right now when setting
  WITHOUT_NIS. This breaks various utilities, including crond, that depend
  on working databases.

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T12:36:10.597170Z
K 7
svn:log
V 111
MFC: r198132

A small change to avoid calling gettimeofday() too often (hardwired to once
every 20us at most).

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T12:36:11.629747Z
K 7
svn:log
V 111
MFC: r198132

A small change to avoid calling gettimeofday() too often (hardwired to once
every 20us at most).

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T12:39:14.642734Z
K 7
svn:log
V 146
MFC: r198136

Support the specification of a range of destination ports e.g.

	netsend 127.0.0.1 6666-7777 [payloadsize] [packet_rate] [duration]

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T12:39:15.942659Z
K 7
svn:log
V 146
MFC: r198136

Support the specification of a range of destination ports e.g.

	netsend 127.0.0.1 6666-7777 [payloadsize] [packet_rate] [duration]

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-05T14:37:59.095244Z
K 7
svn:log
V 285
Some USB mass storage devices requires that the sense information
is retrieved after a failed SCSI command to continue normal
operation. Else this sense information is retrived at the next
SCSI command.

Approved by:	re (kib)
Reported by:	Alex Kozlov
MFC after:	1 week
PR:		usb/160299

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T15:11:01.363794Z
K 7
svn:log
V 383
Begin fleshing out an EWMA packet rate percentage figure and use
that when considering the best rate.

This is a more optimal figure than using the overall percentage,
which doesn't adapt quickly over time.

This is a lot more stable than before (between 100-130mbit on my netbook)
but as the rate sampling for MCS still samples (almost) all rates, performance
can jump quite a bit.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-05T15:21:18.574378Z
K 7
svn:log
V 1330
As a temporary workaround, treat MCS error rates 5% higher than
the best rate as "ok" when picking a best rate.

This is all a hack to avoid having to implement the "correct"
solution, whatever that is. For example, minstrel calculates
a predicted A-MPDU throughput value based on the
average A-MPDU sub-frame count and per-rate TX success probability.

This, along with lowering sample_stats to 2 from 10 (ie, spend 2%
of TX time trying to sample alternate rates) seems to keep my TCP
TX mostly stable at around 120-130mbit for my testing.

TODO:

* The EWMA doesn't raise quickly enough and seems a bit wrong, eg:

[ 0 MCS: 250]      329:329      (100%) (EWMA  63.6%) T       22 F    0 avg   407 last 19144

.. how'd the EWMA come out to be 63% given a 100% success rate overall?
It made an incorrect judgement early on and hasn't yet had enough time
to correct it.

The sampling of other MCS rates is also not optimal - it shouldn't try
sampling all of them during the sample interval. Instead, come up with
a more intelligent way of picking the MCS rates to sample.

Finally, comparing MCS rates by number is incorrect. This really should
compare the per-rate throughput value, as higher MCS rate != higher
throughput.

But, as said above, this is "good enough" for local testing.
It shouldn't be merged into -HEAD like this.

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T17:45:22.367589Z
K 7
svn:log
V 54
MFC: r225334

Fix alignment assumptions.

PR:		160289

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-05T17:45:24.088874Z
K 7
svn:log
V 54
MFC: r225334

Fix alignment assumptions.

PR:		160289

END
K 10
svn:author
V 6
qingli
K 8
svn:date
V 27
2011-09-05T17:54:19.263245Z
K 7
svn:log
V 594
The maximum read size of incoming packets is done in 1024-byte increments.
The current code was rounding down the maximum frame size instead of
routing up, resulting in a read size of 1024 bytes, in the non-jumbo
frame case, and splitting the packets across multiple mbufs.

Consequently the above problem exposed another issue, which is when
packets were splitted across multiple mbufs, and all of the mbufs in the
chain have the M_PKTHDR flag set.

Submitted by:	original patch by Ray Ruvinskiy at BlueCoat dot com
Reviewed by:	jfv, kmacy, rwatson
Approved by:	re (rwatson)
MFC after:	5 days

END
K 10
svn:author
V 6
marcel
K 8
svn:date
V 27
2011-09-06T01:05:17.177528Z
K 7
svn:log
V 22
Sync with head@225405

END
K 10
svn:author
V 6
marcel
K 8
svn:date
V 27
2011-09-06T01:33:20.124391Z
K 7
svn:log
V 22
Sync with head@225405

END
K 10
svn:author
V 6
marcel
K 8
svn:date
V 27
2011-09-06T01:40:09.945361Z
K 7
svn:log
V 28
Fixup sync with head@25405.

END
K 10
svn:author
V 6
marcel
K 8
svn:date
V 27
2011-09-06T01:40:45.388576Z
K 7
svn:log
V 28
Fixup sync with head@225405

END
K 10
svn:author
V 6
marcel
K 8
svn:date
V 27
2011-09-06T01:42:49.419462Z
K 7
svn:log
V 28
Fixup sync with head@225405

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T02:31:57.115013Z
K 7
svn:log
V 77
Add some debugging to keep track when TX buffers are either busy or missing.

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-06T03:17:11.018918Z
K 7
svn:log
V 41
The -a option needs the same check as -r

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T05:06:49.275179Z
K 7
svn:log
V 152
Two fixes:

* Clarify the software queue handling a little better;
* Don't add a frame to the BAW if it's not in the BAW - specifically,
  NULL frames.

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-06T08:15:55.385054Z
K 7
svn:log
V 116
Fix precedence warning when compiling kernel with clang.

Approved by:	re (kib)
Submitted by:	dim
MFC after:	1 week

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-06T10:17:02.478289Z
K 7
svn:log
V 106
Fix for the obsolete libraries list after the 9.0 bump.

Partially noted by:	pluknet
Approved by:	re (bz)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-06T10:19:01.198445Z
K 7
svn:log
V 192
dd -t switch for mdmfs to enable TRIM on the configured filesystem.
While there, fix minor style issues.

Submitted by:	Alex Kozlov <spam rm-rf kiev ua>
MFC after:	1 week
Approved by:	re (bz)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-06T10:21:33.715513Z
K 7
svn:log
V 337
Do not use the function pointers for the internal operation of rtld_printf()
functions. The _rtld_error() function might be called early during the rtld
bootstrap, in which case function pointers are not yet functional on ia64
due to required relocations not yet performed.

Reported, reviewed and tested by:	marcel
Approved by:	re (bz)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-06T10:30:11.402898Z
K 7
svn:log
V 658
Split the vm_page flags PG_WRITEABLE and PG_REFERENCED into atomic
flags field. Updates to the atomic flags are performed using the atomic
ops on the containing word, do not require any vm lock to be held, and
are non-blocking. The vm_page_aflag_set(9) and vm_page_aflag_clear(9)
functions are provided to modify afalgs.

Document the changes to flags field to only require the page lock.

Introduce vm_page_reference(9) function to provide a stable KPI and
KBI for filesystems like tmpfs and zfs which need to mark a page as
referenced.

Reviewed by:    alc, attilio
Tested by:      marius, flo (sparc64); andreast (powerpc, powerpc64)
Approved by:	re (bz)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-06T10:40:21.554797Z
K 7
svn:log
V 181
Document vm_page_aflag_set(9), vm_page_aflag_clear(9) and vm_page_reference(9).
Retire vm_page_flag_set() and vm_page_flag_clear() functions.

Reviewed by:	alc
Approved by:	re (bz)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T10:49:05.855969Z
K 7
svn:log
V 106
Fix the addac serial load register write for AR5416.

Obtained from:	Linux, Atheros
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T10:54:56.345993Z
K 7
svn:log
V 562
Fix 5ghz calibration logic when using AR9280 w/ fast clock.

When the fast clock (44mhz) is enabled for 5ghz HT20, the
dual ADCs aren't enabled. Trying to do the ADC calibrations
here would result in calibration never completing; this
resulted in IQ calibration never running and thus performance
issues in 11a/11n HT20 mode.

Leave it enabled for non-fastclock (40mhz) 11a mode and
HT40 modes.

This has been fixed in discussion with Felix Fietkau (nbd)
and discussions with the Atheros baseband team.

Linux ath9k now has a similar fix.

Approved by:	re (kib)

END
K 10
svn:author
V 3
jhb
K 8
svn:date
V 27
2011-09-06T14:49:38.048267Z
K 7
svn:log
V 90
Take Hellmuth's commit bit into safe keeping per his request to core@.

Approved by:	core

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T15:12:51.522730Z
K 7
svn:log
V 21
Fix some typos, eep.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-06T16:37:02.670334Z
K 7
svn:log
V 323
Convert some packet fields to 64 bit values.

Since I can now tx/rx ~ 18,000 pps, the 32 bit counters overflow
the percentage calculations in a relatively easily obtained period
of time (hours.) So given the likely uptime of some 802.11 links
will be measured in days if not weeks, assume it'll overflow in
the real world.

END
K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2011-09-06T17:21:39.617945Z
K 7
svn:log
V 34
MFC (for sucking in aflags in vm)

END
K 10
svn:author
V 5
trasz
K 8
svn:date
V 27
2011-09-06T17:22:40.900817Z
K 7
svn:log
V 231
Work around a kernel panic triggered by forkbomb with an rctl rule
such as j:name:maxproc:sigkill=100.  Proper fix - deferring psignal
to a taskqueue - is somewhat complicated and thus will happen
after 9.0.

Approved by:	re (kib)

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-07T00:16:36.612611Z
K 7
svn:log
V 47
Use %s as format string when calling syslog().

END
K 10
svn:author
V 2
bz
K 8
svn:date
V 27
2011-09-07T00:43:28.886837Z
K 7
svn:log
V 188
Shorten a variable access some using a cached value rather than
casting the input argument again, making the code more readable.

Reviewed by:	jhb
MFC after:	2 weeks
Approved by:	re (kib)

END
K 10
svn:author
V 2
bz
K 8
svn:date
V 27
2011-09-07T00:45:15.376203Z
K 7
svn:log
V 404
Fix two problems:
1) if no default gateway is given the route command will fail
   and the script will abort with an error leading to an endless
   loop unless configuration is cancelled or changed. [1]
2) In the IPv6 case add -inet6 to not alter the legacy IP default
   route.

Reported by:	Oleg Ginzburg (olevole olevole.ru) [1]
Tested by:	Oleg Ginzburg (olevole olevole.ru) [1]
Approved by:	re (kib)

END
K 10
svn:author
V 2
bz
K 8
svn:date
V 27
2011-09-07T00:48:58.477105Z
K 7
svn:log
V 85
Correct a typo.

Submitted by:	gcooper
Reviewed by:	nwhitehorn
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-07T03:00:58.778085Z
K 7
svn:log
V 222
Add a definition for ASYNC_CAUSE_CLR. It's not used yet, but the
reference driver does clear the async interrupts after each service.
I'll tinker with this in a future commit.

Obtained from:	Atheros
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-07T07:24:46.605779Z
K 7
svn:log
V 155
Correct the retry bit handling during software TX retransmission.

* only update the bit if retries == 0;
* if the retry bit is set, resync the descriptor

END
K 10
svn:author
V 5
trasz
K 8
svn:date
V 27
2011-09-07T07:52:45.528643Z
K 7
svn:log
V 64
Fix whitespace.

Submitted by:	amdmi3
Approved by:	re (rwatson)

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T12:53:18.592207Z
K 7
svn:log
V 210
- Make the heuristic code loosly coupled to the fast matcher by providing
  properly escaped patterns instead of using an internal flag
- Add some struct fields for escaped dots, forgotten in a previous commit

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T13:01:26.669115Z
K 7
svn:log
V 52
- Merge some fixes from TRE
- Silence some warnings

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T13:26:20.122966Z
K 7
svn:log
V 20
- Fix some warnings

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T14:04:51.481559Z
K 7
svn:log
V 19
- Correct BC shift

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T14:20:36.064283Z
K 7
svn:log
V 20
- Fix some warnings

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-07T14:37:51.670309Z
K 7
svn:log
V 15
- Fix warnings

END
K 10
svn:author
V 7
yongari
K 8
svn:date
V 27
2011-09-07T16:57:43.873780Z
K 7
svn:log
V 814
vge(4) hardwares poll media status and generates an interrupt
whenever the link state is changed.  Using software based polling
for media status tracking is known to cause MII access failure
under certain conditions once link is established so vge(4) used to
rely on link status change interrupt.
However DEVICE_POLLING completely disables generation of all kind
of interrupts on vge(4) such that this resulted in not detecting
link state change event.  This means vge(4) does not correctly
detect established/lost link with DEVICE_POLLING.  Losing the
interrupt made vge(4) not to send any packets to peer since vge(4)
does not try to send any packets when there is no established link.

Work around the issue by generating link state change interrupt
with DEVICE_POLLING.

PR:		kern/160442
Approved by:	re (kib)

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-07T18:00:05.762805Z
K 7
svn:log
V 13
IFC @225440.

END
K 10
svn:author
V 8
rmacklem
K 8
svn:date
V 27
2011-09-08T00:45:28.426122Z
K 7
svn:log
V 233
MFC: r225109
Fix nfsstat(1) so that it prints out correct stats for the
new NFS server when the "-w" option is used. The problem was
spotted by hrs@ during testing where srvrpcnt[] must be indexed
by NFSV4OP_XXX and not NFSPROC_XXX.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-08T00:55:48.664281Z
K 7
svn:log
V 18
Fix casting issue

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-08T01:23:05.617830Z
K 7
svn:log
V 1249
Update the TSF and next-TBTT methods to work for the AR5416 and later NICs.
This is another commit in a series of TDMA support fixes for the 11n NICs.

* Move ath_hal_getnexttbtt() into the HAL; write methods for it.
  This returns a timer value in TSF, rather than TU.

* Move ath_hal_getcca() and ath_hal_setcca() into the HAL too, where they
  likely now belong.

* Create a new HAL capability: HAL_CAP_LONG_RXDESC_TSF.
  The pre-11n NICs write 15 bit TSF snapshots into the RX descriptor;
  the AR5416 and later write 32 bit TSF snapshots into the RX descriptor.
* Use the new capability to choose between 15 and 31 bit TSF adjustment
  functions in ath_extend_tsf().

* Write ar5416GetTsf64() and ar5416SetTsf64() methods.
  ar5416GetTsf64() tries to compensate for TSF changes at the 32 bit boundary.

According to yin, this fixes the TDMA beaconing on 11n chipsets and TDMA
stations can now associate/talk, but there are still issues with traffic
stability which need to be investigated.

The ath_hal_extendtsf() function is also used in RX packet timestamping;
this may improve adhoc mode on the 11n chipsets. It also will affect the
timestamps seen in radiotap frames.

Submitted by:	Kang Yin Su <cantona@cantona.net>
Approved by:	re (kib)

END
K 10
svn:author
V 2
ae
K 8
svn:date
V 27
2011-09-08T04:14:16.728630Z
K 7
svn:log
V 409
Don't use the whole free space when resizing  partition to a larger size
on a disk with non zero stripesize (e.g. disks with 4k sector size)[1].
Also do not use automatic alignment when size is exactly specified, but
an alignment is not. Use automatic alignment only for case when user
omits both "-s" and "-a" options.

Reported by:	Mikael Fridh <frimik at gmail> [1]
Approved by:	re (kib)
MFC after:	1 week

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-08T09:33:49.454126Z
K 7
svn:log
V 187
Fix SSL memory handlig for (EC)DH cipher suites, in particular for
multi-threaded use of ECDH.

Security:	CVE-2011-3210
Reviewed by:	stas
Obtained from:	OpenSSL CVS
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-08T12:19:58.052713Z
K 7
svn:log
V 749
Add txq stop debugging; always memzero the 11n descriptors before chaining.

* To help me trace down some ath txq device timeout issues, print out some
  TXQ state before the TXQ is halted.

* Always memzero the descriptor in ar5416ChainTxDesc(). It was being called
  for the non-first descriptor in a (sub) frame.
  Problem is, I'm calling this function first, and THEN setting up the first
  descriptor, because of the way that ar5416ChainTxDesc() overwrites
  some fields that ar5416SetupFirstTxDesc() sets up.

  It's quite possible that ar5416SetupFirstTxDesc() should be called first;
  and ar5416ChainTxDesc() should be modified to merge in these changes
  rather than overwrite them. If I decide this, I'll revert this part of the
  patch.

END
K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2011-09-08T12:56:26.379882Z
K 7
svn:log
V 774
Improve the informations reported in case of busy buffers during the shutdown:
- Axe out the SHOW_BUSYBUFS option and uses a tunable for selectively
enable/disable it, which is defaulted for not printing anything (0
value) but can be changed for printing (1 value) and be verbose (2
value)
- Improves the informations outputed: right now, there is no track of
the actual struct buf object or vnode which are referenced by the
shutdown process, but it is printed the related struct bufobj object
which is not really helpful
- Add more verbosity about the state of the struct buf lock and the
vnode informations, with the latter to be activated separately by the
sysctl

Sponsored by:	Sandvine Incorporated
Reviewed by:	emaste, kib
Approved by:	re (ksmith)
MFC after:	10 days

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-08T16:22:58.454181Z
K 7
svn:log
V 179
MFC r225446:

Fix SSL memory handlig for (EC)DH cipher suites, in particular for
multi-threaded use of ECDH.

Security:	CVE-2011-3210
Reviewed by:	stas
Obtained from:	OpenSSL CVS

END
K 10
svn:author
V 3
mdf
K 8
svn:date
V 27
2011-09-08T17:05:51.054547Z
K 7
svn:log
V 15
IFC @ r225449.

END
K 10
svn:author
V 3
mdf
K 8
svn:date
V 27
2011-09-08T17:24:44.737837Z
K 7
svn:log
V 58
Fix the path to shlib-compat since I renamed it.  Whoops!

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-08T23:46:07.110624Z
K 7
svn:log
V 156
Sync pf.os with OpenBSD:

 - Update OpenBSD fingerprints through OpenBSD 4.9
 - Fix typos.

Obtained from:	OpenBSD
MFC after:	2 weeks
Approved by:	re (kib)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-08T23:59:16.759913Z
K 7
svn:log
V 406
* Printing the TX flags field ("F") needed fixing - it was printing an 04
  in front, rather than using 04 as part of the field description.

* Add support to dump a chain of aggregate buffers, rather than a single
  buffer. This is suboptimal as the calls to ieee80211_dump_pkt() only
  dump the first mbuf chain, rather than all aggregate packets.
  I'll have to fix this up before I commit it to -HEAD.

END
K 10
svn:author
V 7
delphij
K 8
svn:date
V 27
2011-09-09T01:39:19.596784Z
K 7
svn:log
V 149
MFC r225199:

Fix format strings for KTR_STATE in 4BSD ad ULE schedulers.

Submitted by:	Ivan Klymenko <fidaj@ukr.net>
PR:		kern/159904, kern/159905

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T04:40:20.281926Z
K 7
svn:log
V 871
I've been seeing situations where the TX queue has validly completed
packets in it. This shouldn't occur - the hardware should've sent an
interrupt to signal that a TX completion had occured.

So on a hunch, I decided to shuffle the interrupt mitigation bits
up a bit, because I figured that the hardware may be sending a
TX mitigation interrupt but not asserting TXOK at the time.

This seems to be doing the "right" thing - ie, at the time the
mitigation timer expiry has occured, the AR_ISR_S0/AR_ISR_S1/AR_ISR_S2
registers seem to have the correct TX queue interrupt cause bits set;
as they would stay set (once set) until AR_ISR_RAC is read.

I'm going to have to chase this up with the MAC guys at Atheros to see
exactly what the correct behaviour is.

(Now there are still device timeout messages, but with an empty TX
 queue- this signifies a different problem.)

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T04:41:14.999044Z
K 7
svn:log
V 153
Disable the current debugging and add something of my own; just to point out
if something stupid is occuring.

This shouldn't be merged back into -HEAD.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T04:42:11.977321Z
K 7
svn:log
V 93
.. add this little gem of a reminder. Yes, why is sc_lastrx being updated
in the TX handler?

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-09T07:44:14.483241Z
K 7
svn:log
V 290
MFC r225350 and r225400:

This patch adds automatic detection of USB mass storage devices
which does not support the no synchronize cache SCSI command.

The __FreeBSD_version version macro has been bumped and
external kernel modules needs to be recompiled after
this patch.

PR:	usb/160299

END
K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2011-09-09T09:39:26.294995Z
K 7
svn:log
V 4
MFC

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T11:24:04.754071Z
K 7
svn:log
V 454
Begin adding some further debugging to chase down the TX hangs that I've
seen under TCP TX load.

It looks like the underlying problem is that some packets that are -before-
the BAW are making it onto the software TX queue. These frames are outside
the BAW (but before it, not after it) and thus the TX scheduling stops.

The source of those grossly out of order (and likely already successfully
TXed, but that's currently unknown) is currently unknown.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T11:59:59.261246Z
K 7
svn:log
V 23
Fix a previous commit.

END
K 10
svn:author
V 6
tuexen
K 8
svn:date
V 27
2011-09-09T13:52:37.751940Z
K 7
svn:log
V 153
Improve implementation of the Nagle algorithm for SCTP:
Don't delay the final fragment of a fragmented user message.

Approved by: re
MFC after: 4 weeks

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-09T15:03:52.972935Z
K 7
svn:log
V 20
Correct formatting.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-09T19:39:54.036530Z
K 7
svn:log
V 46
MFC: r225348

Remove an incorrect apostrophe.

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-10T07:16:53.051311Z
K 7
svn:log
V 933
In find_and_delete_distfiles() (run after a successful install) change
the default pattern to be foo-[0-9]* instead of foo* [1]

This matches 19,305 out of 25,106 unique distfiles, and dramatically
reduces false positives for files like foo-1.23 vs. foo-bar-2.34. Of the
ones that this pattern misses, 4,713 don't match the traditional distfile
format anyway, so it's a huge net win; and should make running without -D
much less annoying.

When the new pattern doesn't match any existing distfiles, fall back to
the old one. Do this by calling with the pattern, not the file name. This
simplifies processing in the function a bit.

Also in find_and_delete_distfiles(), cache matches to $DI_FILES so that
we don't have to re-test that same file again.

Change the method used to print the prompt in get_answer_g() to echo -e
so that I can be more flexible in formatting prompt messages.

Suggested by:	Lapo Luchini <lapo@lapo.it> [1]

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-10T08:00:26.902762Z
K 7
svn:log
V 251
Add a more-or-less painless test to deal with the issue of the current
working directory no longer existing. This problem usually manifests as
a "The value of DISTDIR cannot be empty" error because 'make -f' does
not handle this condition gracefully.

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-10T09:40:00.118557Z
K 7
svn:log
V 156
Add ftp4.se.freebsd.org to the IPv4 and IPv6 mirror lists.

PR:		157210, 157211
Submitted by:	Niclas Zeising <zeising@lysator.liu.se>
Approved by:	re (kib)

END
K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2011-09-10T12:36:11.411213Z
K 7
svn:log
V 4
MFC

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-10T15:55:36.770718Z
K 7
svn:log
V 162
Refactor auto-quirk solution so that we break as few external
drivers as possible.

PR:		usb/160299
Approved by:	re (kib)
Suggested by:	rwatson
MFC after:	0 days

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-10T16:52:15.749421Z
K 7
svn:log
V 136
MFC r225469:

Refactor auto-quirk solution so that we break as few external
drivers as possible.

PR:		usb/160299
Suggested by:	rwatson

END
K 10
svn:author
V 8
hselasky
K 8
svn:date
V 27
2011-09-10T16:56:38.227524Z
K 7
svn:log
V 70
MFC r225414:
Fix precedence warning when compiling kernel with clang.

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-11T07:26:36.821465Z
K 7
svn:log
V 26
Merge from HEAD @ 225469.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T09:43:13.950536Z
K 7
svn:log
V 96
Fix the order of parameters passed to the HT frame duration calculation.

Approved by:	re (kib)

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-11T16:05:09.539331Z
K 7
svn:log
V 195
Inline the syscallenter() and syscallret(). This reduces the time measured
by the syscall entry speed microbenchmarks by ~10% on amd64.

Submitted by:	jhb
Approved by:	re (bz)
MFC after:	2 weeks

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-11T16:08:10.659727Z
K 7
svn:log
V 372
Perform amd64-specific microoptimizations for native syscall entry
sequence. The effect is ~1% on the microbenchmark.

In particular, do not restore registers which are preserved by the
C calling sequence. Align the jump target. Avoid unneeded memory
accesses by calculating some data in syscall entry trampoline.

Reviewed by:	jhb
Approved by:	re (bz)
MFC after:	2 weeks

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T16:18:10.276948Z
K 7
svn:log
V 146
Fix some descriptor references I missed when I added bf_lastds to
ath_buf.

This resulted in the incorrect status being printed during debugging.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T16:20:59.321386Z
K 7
svn:log
V 252
Non-TDMA data TXQ's also require HAL_TXQ_TXEOLINT_ENABLE.

The interrupt mitigation method for non-aggregate frames
(and, until an upcoming commit, aggregate frames too) requires
the TXEOL interrupt to be fired when a TX queue reaches the
end of list.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T16:28:07.511406Z
K 7
svn:log
V 168
Just add a little warning if NOACK tagged frames end up in an aggregate
queue, and disable NOACK for now.

I'll worry about fixing this up at some point in the future.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T16:42:03.837792Z
K 7
svn:log
V 1472
Fix up the InterReq flag in the TX descriptor; and always set it when
doing aggregate frame TX.

It turns out that the AR5416 (and AR5212!) seems to want the InterReq
flag to be consistently set in all descriptors in the given frame.
This is what the datasheet says.

Both linux ath9k and the reference driver (at least when under linux)
seem to get away without this HAL change because the TX (sub) frame
is only ever one skb in length. Thus the intermediary frames never
needed this bit set.

FreeBSD's implementation here involves TX'ing mbuf chains, so each
descriptor in the subframe now needs this flag set consistently.

Secondly, always set the InterReq flag (HAL_TXDESC_INTREQ) on all
subframes in an aggregate - the interrupt mitigation hardware
(for AR5416 and later, but not AR9130) will handle limiting TX
interrupts for us. This is in line with what ath9k does.

Although I haven't tested this change separate from the stuff I'm
about to commit, this change was the final change needed to fix the
odd TX hangs I was seeing, where a TXQ would complete but no TX
interrupt occured. This wouldn't matter in the past as a subsequent
TX (any TX!) would have been enough to kick the queue along.
Without it, I'd get occasional device timeouts even though the TXQ
consisted of a frame (aggregate or otherwise) whose status descriptor
indicated it was completed.

Whilst I'm here, do the same for the AR5212 - the AR5213 datasheet
indicates much the same is needed.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T16:52:07.133092Z
K 7
svn:log
V 1768
Revamp how interrupts and handled and scheduled, bringing this driver
in line with what the reference driver and linux ath9k does.

Currently, ath_intr() would submit certain things to occur via
taskqueue_enqueue(). In the case of handling TX interrupts, it's
quite possible that a TX completion interrupt (EOL, OK, mitigated,
etc) could occur whilst the TX completion task was running.

This meant that a small race could occur where the TX completion
call to txqactive() (which would call into the HAL to see whether
the current TXQ had some active interrupt event to service, and
then clear that flag) could race with ath_intr() and the call
to ah_getPendingInterrupts(), which for ar5212/ar5416 will set
the TXQ active bits based on the contents of the relevant interrupt
registers.

This also means that ath_intr() won't occur during one of the TX, RX
or reset-during-fatal situations. The only current exception to this
are fatal events or SWBA, which still is enabled.

It didn't completely close up the initial TXQ hang issue I saw earlier,
but it along with the previous missing TXEOL change from earlier did
eliminate some of the TXQ hangs.

I'm going to be porting more code over from the reference driver and
ath9k, so having the interrupt handling model here match those will
make it easier.

Finally, the ath_start() calls have been moved from various places in
the receive/tx-complete path and into the deferred interrupt handler.

Finally finally - I've disabled the optimised queue proc handling
(q0 and q0123) versions of ath_tx_proc(), this needs to be reintroduced
before this is merged into HEAD.

Finally finally finally - I need to reschedule a read event after
SWBA (which still occurs in interrupt context) to kick along fast-frames
handling.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-11T17:07:33.801448Z
K 7
svn:log
V 30
Add another thing to research

END
K 10
svn:author
V 8
brueffer
K 8
svn:date
V 27
2011-09-11T17:39:51.653779Z
K 7
svn:log
V 190
Fix a zyd(4) comment typo that was copy+pasted into most kernel config files.

PR:		160276
Submitted by:	MATSUMIYA Ryo <matsumiya@mma.club.uec.ac.jp>
Approved by:	re (kib)
MFC after:	1 week

END
K 10
svn:author
V 3
kib
K 8
svn:date
V 27
2011-09-11T18:00:46.196339Z
K 7
svn:log
V 115
The jump target shall be after the padding, not into it.

Reported by:	alc
Approved by:	re (bz)
MFC after:	2 weeks

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-11T19:31:41.654430Z
K 7
svn:log
V 32
- Fix an ugly file handling bug

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-11T19:34:56.392166Z
K 7
svn:log
V 858
Add $ipv6_cpe_wanif to enable functionality required for IPv6 CPE.  When
setting an interface name to it, the following configurations will be enabled:

 1. "no_radr" is set to all IPv6 interfaces automatically.

 2. "-no_radr accept_rtadv" will be set only for $ipv6_cpe_wanif.

 3. The node will add RA-sending routers to the default router list even if
    net.inet6.ip6.forwarding=1.

This mode is added to conform to RFC 6204 (a router which connects the end-user
network to a service provider network).  To enable packet forwarding, you still
need to set ipv6_gateway_enable=YES.

Note that accepting router entries into the default router list when packet
forwarding capability and a routing daemon are enabled can result in messing
up the routing table.  To minimize such unexpected behaviors, "no_radr" is set
on all interfaces but $ipv6_cpe_wanif.

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-11T19:37:57.882325Z
K 7
svn:log
V 22
Remove an extra line.

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-11T19:55:49.043812Z
K 7
svn:log
V 22
- Increase WARNS to 3

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-11T20:11:00.364466Z
K 7
svn:log
V 21
- Recover to WARNS=6

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-11T20:17:44.163752Z
K 7
svn:log
V 158
- Add an warning when ifconfig_IF_ipv6 has no inet6 keyword in front of an IPv6
  address.

- Use eval for ${ifconfig_args} to fix an issue fixed in r223506.

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-11T20:25:57.904284Z
K 7
svn:log
V 1051
MFC: r221407 (partial)

- Let mii_phy_add_media() also announce the support of IFM_NONE.
- Add a PHY-specific reset routine is for lxtphy(4) which disables MII
  interrupts.
- Remove redundant checks in NIC drivers for mii->mii_instance not being
  zero before calling the reset routines because as soon as one PHY driver
  attaches mii->mii_instance is incremented and we hardly can end up in
  their media change callbacks etc if no PHY driver has attached as
  mii_attach() would have failed in that case and not attach a miibus(4)
  instance.
- Convert the mebers of the MII data structure to unsigned where
  appropriate. This is partly inspired by NetBSD/OpenBSD.
- Add MIIF_NOMANPAUSE and mii_phy_flowstatus() calls to drivers that
  weren't explicitly converted to support flow control before. It's
  unclear whether flow control actually works with these but typically
  it should and their net behavior should be more correct with these
  changes in place than without if the MAC driver sets MIIF_DOPAUSE.

Obtained from:	NetBSD (partially)

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-11T20:37:01.765014Z
K 7
svn:log
V 138
- According to re_format(7) `^' and `$' are supposed to match null-string.
  Fix to match this behavior.

Reported by:	aakuusta@gmail.com

END
K 10
svn:author
V 6
marius
K 8
svn:date
V 27
2011-09-11T20:38:33.875045Z
K 7
svn:log
V 1051
MFC: r221407 (partial)

- Let mii_phy_add_media() also announce the support of IFM_NONE.
- Add a PHY-specific reset routine is for lxtphy(4) which disables MII
  interrupts.
- Remove redundant checks in NIC drivers for mii->mii_instance not being
  zero before calling the reset routines because as soon as one PHY driver
  attaches mii->mii_instance is incremented and we hardly can end up in
  their media change callbacks etc if no PHY driver has attached as
  mii_attach() would have failed in that case and not attach a miibus(4)
  instance.
- Convert the mebers of the MII data structure to unsigned where
  appropriate. This is partly inspired by NetBSD/OpenBSD.
- Add MIIF_NOMANPAUSE and mii_phy_flowstatus() calls to drivers that
  weren't explicitly converted to support flow control before. It's
  unclear whether flow control actually works with these but typically
  it should and their net behavior should be more correct with these
  changes in place than without if the MAC driver sets MIIF_DOPAUSE.

Obtained from:	NetBSD (partially)

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-11T21:17:55.436353Z
K 7
svn:log
V 201
- Partly fix context after maximum count of matches. GNU grep does not
  include further matching lines in the context so the behavior is still
  not totally the same.

Reported by:	aakuusta@gmail.com

END
K 10
svn:author
V 6
jilles
K 8
svn:date
V 27
2011-09-11T21:38:33.149563Z
K 7
svn:log
V 86
MFC r223007,r223011,r223546,r224104: sh: New testcases that already work in
stable/8.

END
K 10
svn:author
V 7
thompsa
K 8
svn:date
V 27
2011-09-11T23:37:40.059104Z
K 7
svn:log
V 285
MFC r225380

 On the first loop for generating a bridge MAC address use the local
 hostid, this gives a good chance of keeping the same address over
 reboots. This is intended to help IPV6 and similar which generate
 their addresses from the mac.

PR:		kern/160300
Submitted by:	mdodd

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-12T00:50:33.527444Z
K 7
svn:log
V 344
Add the ability to retry package creation. This is mostly useful for
PM_SU_CMD after a build that took so long that your sudo ticket has
expired.

While I'm at it, apply the benefits of looking at old code with fresh
eyes to the other place where there are 3 options in the menu (and thus
neither of the get_answer*() variants are appropriate.

END
K 10
svn:author
V 5
gabor
K 8
svn:date
V 27
2011-09-12T01:16:57.568012Z
K 7
svn:log
V 243
- Reimplement reverse quick search algorithm. It can be used when
  REG_NOSUB is specified and it is more efficient when the distance
  of the last dot and the end of the pattern is less than the
  distance of the beginning and the first dot.

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-12T01:36:54.952563Z
K 7
svn:log
V 75
The rc.d/bridge script should be invoked before invoking daemon processes.

END
K 10
svn:author
V 3
hrs
K 8
svn:date
V 27
2011-09-12T01:38:24.634931Z
K 7
svn:log
V 46
Use resolvconf(8) to create /etc/resolv.conf.

END
K 10
svn:author
V 6
adrian
K 8
svn:date
V 27
2011-09-12T02:00:52.438735Z
K 7
svn:log
V 227
.. and add the last missing piece from last night's commit:

* Actually follow through on setting up those flags. unfortunately ctl0 was
  being 0'ed and I didn't pick up on it. This correctly sets CLRDMASK, VEOL and
  INTREQ.

END
K 10
svn:author
V 5
dougb
K 8
svn:date
V 27
2011-09-12T03:31:16.093641Z
K 7
svn:log
V 110
In the code that derives the proper value of PACKAGESITE remove 6.x and
add 9.x. Also add a 10-current entry.

END
K 10
svn:author
V 7
linimon
K 8
svn:date
V 27
2011-09-12T03:59:42.587554Z
K 7
svn:log
V 181
Remove clause that could never be executed and put it where it properly
belongs.  Now you can say 'build create amd64 9-exp-whatever' without
having to remember to append 'latest'.

END