]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
11 years agoDon't send an ABORT chunk with verification 0.
tuexen [Fri, 17 May 2013 21:45:52 +0000 (21:45 +0000)]
Don't send an ABORT chunk with verification 0.

MFC after: 1 week

11 years agoFor ARM, MIPS, and PowerPC, default to 32-byte alignment, but allow it
imp [Fri, 17 May 2013 20:53:15 +0000 (20:53 +0000)]
For ARM, MIPS, and PowerPC, default to 32-byte alignment, but allow it
to be as small as 8.

11 years agoUse readlocking now that assertions on vm_page_lookup() are relaxed.
attilio [Fri, 17 May 2013 20:03:55 +0000 (20:03 +0000)]
Use readlocking now that assertions on vm_page_lookup() are relaxed.

Sponsored by: EMC / Isilon storage division
Reviewed by: alc
Tested by: flo, pho

11 years agoWe really need to get the bsd.own.mk from this tree so the
sjg [Fri, 17 May 2013 19:37:16 +0000 (19:37 +0000)]
We really need to get the bsd.own.mk from this tree so the
correct options are set.
Also defined NO_PWD_OVERRIDE to match behavior of fmake.

11 years agoo Retrive the part number (CP2103 etc) from the hardware on attach.
gavin [Fri, 17 May 2013 19:13:31 +0000 (19:13 +0000)]
o  Retrive the part number (CP2103 etc) from the hardware on attach.
o  The CP2101 and CP2102 do not support GPIO pin use at all, enforce this.
o  Support reading the GPIO status on the second port of the CP2105.  More
   work is needed before the CP2105 GPIO pins can be used as outputs.

Hardware donated by: Silicon Labs
MFC after: 3 weeks

11 years agoRelax the object locking in vm_fault_prefault(). A read lock suffices.
alc [Fri, 17 May 2013 19:02:36 +0000 (19:02 +0000)]
Relax the object locking in vm_fault_prefault().  A read lock suffices.

Reviewed by: attilio
Sponsored by: EMC / Isilon Storage Division

11 years agoRelax the object locking assertion in pmap_enter_locked().
alc [Fri, 17 May 2013 18:59:00 +0000 (18:59 +0000)]
Relax the object locking assertion in pmap_enter_locked().

Reviewed by: attilio
Sponsored by: EMC / Isilon Storage Division

11 years ago- Update the release notes to DocBook 4.5 XML to build with the
gabor [Fri, 17 May 2013 18:53:29 +0000 (18:53 +0000)]
- Update the release notes to DocBook 4.5 XML to build with the
  updated documentation tree

11 years agoRelax the object locking assertion in vm_page_lookup(). Now that a radix
alc [Fri, 17 May 2013 18:49:43 +0000 (18:49 +0000)]
Relax the object locking assertion in vm_page_lookup().  Now that a radix
tree is used to maintain the object's collection of resident pages,
vm_page_lookup() no longer needs an exclusive lock.

Reviewed by:    attilio
Sponsored by:   EMC / Isilon Storage Division

11 years agoRevert r250692. We'll use 1 kernel config file for the BeagleBone models.
rpaulo [Fri, 17 May 2013 15:20:45 +0000 (15:20 +0000)]
Revert r250692. We'll use 1 kernel config file for the BeagleBone models.

11 years agoDecode new HT 3.00 and 3.10 capabilities.
kib [Fri, 17 May 2013 14:05:31 +0000 (14:05 +0000)]
Decode new HT 3.00 and 3.10 capabilities.

Submitted by: Dmitry Luhtionov <dmitryluhtionov@gmail.com>
MFC after: 1 week

11 years agoAdd new capability types encodings from HyperTransport I/O Link
kib [Fri, 17 May 2013 14:04:31 +0000 (14:04 +0000)]
Add new capability types encodings from HyperTransport I/O Link
Specification revisions 3.00 and 3.10.

Submitted by: Dmitry Luhtionov <dmitryluhtionov@gmail.com>
MFC after: 1 week

11 years agoUpgrade to OpenSSH 6.2p2. Mostly a no-op since I had already patched
des [Fri, 17 May 2013 09:12:33 +0000 (09:12 +0000)]
Upgrade to OpenSSH 6.2p2.  Mostly a no-op since I had already patched
the issues that affected us.

11 years agoAdd a -N option that prints the jail name rather than its number.
des [Fri, 17 May 2013 08:48:16 +0000 (08:48 +0000)]
Add a -N option that prints the jail name rather than its number.

MFC after: 3 weeks

11 years agoAdd some more debugging printf()s to complain if the ath_buf tx queue
adrian [Fri, 17 May 2013 05:16:30 +0000 (05:16 +0000)]
Add some more debugging printf()s to complain if the ath_buf tx queue
doesn't match the actual hardware queue this frame is queued to.

I'm trying to ensure that the holding buffers are actually being queued
to the same TX queue as the holding buffer that they end up on.
I'm pretty sure this is all correct so if this complains, it'll be due
to some kind of subtle broken-ness that needs fixing.

This is only done for legacy hardware, not EDMA hardware.

Tested:

* AR5416 STA mode, very lightly

11 years agoComplete revision 243104.
grog [Fri, 17 May 2013 03:52:20 +0000 (03:52 +0000)]
Complete revision 243104.

11 years agoDrop any connection to newsyslog. I haven't worked on it for quite
gad [Fri, 17 May 2013 03:14:55 +0000 (03:14 +0000)]
Drop any connection to newsyslog.  I haven't worked on it for quite
some time.  Note that I do want to keep the pre-commit review for
usr.sbin/lpr.  I am actively working on some updates for that.

11 years agoguard against MK_SHARED_TOOLCHAIN not being defined.
delphij [Thu, 16 May 2013 22:35:33 +0000 (22:35 +0000)]
guard against MK_SHARED_TOOLCHAIN not being defined.

11 years agoSimilar to r250719: guard against MK_BMAKE not being defined.
delphij [Thu, 16 May 2013 22:25:17 +0000 (22:25 +0000)]
Similar to r250719: guard against MK_BMAKE not being defined.

11 years agoFix cut-n-paste error.
julian [Thu, 16 May 2013 21:27:47 +0000 (21:27 +0000)]
Fix cut-n-paste error.

11 years agoReverse the sense of the test wrt bmake, and guard against
sjg [Thu, 16 May 2013 21:26:46 +0000 (21:26 +0000)]
Reverse the sense of the test wrt bmake, and guard against
MK_BMAKE not being defined.

11 years agoConstify string pointers.
delphij [Thu, 16 May 2013 21:04:56 +0000 (21:04 +0000)]
Constify string pointers.

Verified with: sha256(1)
MFC after: 2 weeks

11 years agoMinor mdoc improvements.
joel [Thu, 16 May 2013 20:55:20 +0000 (20:55 +0000)]
Minor mdoc improvements.

11 years agoWhen running the -m option to generate a newfs(8) command suitable for
mckusick [Thu, 16 May 2013 20:07:08 +0000 (20:07 +0000)]
When running the -m option to generate a newfs(8) command suitable for
recreating the filesystem, check for and output the -i, -k, and -l
options if appropriate.

Note the remaining deficiencies of the -m option in the dumpfs(8)
manual page. Specifically that newfs(8) options -E, -R, -S, and -T
options are not handled and that -p is not useful so is omitted.
Also document that newfs(8) options -n and -r are neither checked
for nor output but should be. The -r flag is needed if the filesystem
uses gjournal(8).

PR:           bin/163992
Reported by:  Dieter <freebsd@sopwith.solgatos.com>
Submitted by: Andy Kosela <akosela@andykosela.com>
MFC after:    1 week

11 years agoWhen reporting the user readable size, round up. Several SD cards not
imp [Thu, 16 May 2013 19:44:51 +0000 (19:44 +0000)]
When reporting the user readable size, round up. Several SD cards not
only use SI units, but also are a couple of percent short. If you need
to know the exact size, diskinfo will return exact results.

11 years agoClean up trailing whitespace.
mckusick [Thu, 16 May 2013 19:39:11 +0000 (19:39 +0000)]
Clean up trailing whitespace.

Submitted by: Andy Kosela
MFC after:    1 week

11 years agoImplement NO_ROOT support for distributekernel and packagekernel.
brooks [Thu, 16 May 2013 18:37:14 +0000 (18:37 +0000)]
Implement NO_ROOT support for distributekernel and packagekernel.

11 years agoA library function shall not set errno to 0.
jh [Thu, 16 May 2013 18:13:10 +0000 (18:13 +0000)]
A library function shall not set errno to 0.

Reviewed by: mdf

11 years agoTidy up the debugging - don't bother printing out TID pointers; now
adrian [Thu, 16 May 2013 17:53:12 +0000 (17:53 +0000)]
Tidy up the debugging - don't bother printing out TID pointers; now
that we are printing out the MAC address in these fields, just printing
out the TID is enough.

11 years agoLimit the number of software queued frames when doing non-aggregation.
adrian [Thu, 16 May 2013 17:46:32 +0000 (17:46 +0000)]
Limit the number of software queued frames when doing non-aggregation.

This should prevent the TX queue being filled with non-aggregate frames,
causing starvation and non-fair queue behaviour.

11 years agoDump out the holding buffer descriptor contents and addresses stopping DMA.
adrian [Thu, 16 May 2013 17:45:01 +0000 (17:45 +0000)]
Dump out the holding buffer descriptor contents and addresses stopping DMA.

11 years agoFix a typo in a comment.
dteske [Thu, 16 May 2013 16:51:52 +0000 (16:51 +0000)]
Fix a typo in a comment.

11 years agoAdd a handy function for truncating variables to a specific byte-length. It
dteske [Thu, 16 May 2013 16:46:02 +0000 (16:46 +0000)]
Add a handy function for truncating variables to a specific byte-length. It
should be noted that newlines are both preserved and included in said byte-
count. If you want to truncate single-line values without regard to line
termination, there's always f_substr() which already exists herein.

11 years agoFinally change the mbuf to have its own fib field instead of stealing
julian [Thu, 16 May 2013 16:20:17 +0000 (16:20 +0000)]
Finally change the mbuf to have its own fib field instead of stealing
4 flag bits. This was supposed to happen in 8.0, and again in 2012..

MFC after: never

11 years agoBuild bmake by default.
sjg [Thu, 16 May 2013 15:28:38 +0000 (15:28 +0000)]
Build bmake by default.
If someone explicitly builds usr.bin/make while MK_BMAKE==yes,
install it as 'fmake'.

Reviewed by: brooks

11 years agoAdd a sample script to create filesystem images from an installed tree
brooks [Thu, 16 May 2013 14:38:12 +0000 (14:38 +0000)]
Add a sample script to create filesystem images from an installed tree
created by installworld+distribution with the -DNO_ROOT option.

11 years agoAdd dependencies on the firmware, which allows the loading of the cxgb
kib [Thu, 16 May 2013 13:07:02 +0000 (13:07 +0000)]
Add dependencies on the firmware, which allows the loading of the cxgb
and cxgbe modules.

Reviewed and approved by: np
MFC after: 1 week

11 years agoFix L2 cache write-back invalidate for Sheeva core.
gber [Thu, 16 May 2013 09:43:04 +0000 (09:43 +0000)]
Fix L2 cache write-back invalidate for Sheeva core.

Submitted by: Michal Dubiel
Obtained from: Netasq, Semihalf

11 years agoAdd tcp header flags ECE and CWR defined in RFC 3168.
hiren [Thu, 16 May 2013 05:16:56 +0000 (05:16 +0000)]
Add tcp header flags ECE and CWR defined in RFC 3168.

PR: 140349
Submitted by: Alan Amesbury <amesbury@umn.edu> (earlier version)
Reviewed by: delphij
Approved by: sbruno (mentor)

11 years agoAdd a kernel config file for the BeableBone Black SoC.
rpaulo [Thu, 16 May 2013 03:51:00 +0000 (03:51 +0000)]
Add a kernel config file for the BeableBone Black SoC.

11 years agoReturn one-based key so that user can check if the key is ever allocated
davidxu [Thu, 16 May 2013 03:01:04 +0000 (03:01 +0000)]
Return one-based key so that user can check if the key is ever allocated
in the first place.

Initial patch submitted by: phk

11 years agoConvert a couple of helper scripts used to test the ip provider to work on
markj [Wed, 15 May 2013 22:56:24 +0000 (22:56 +0000)]
Convert a couple of helper scripts used to test the ip provider to work on
FreeBSD. In the IPv6 case, try each interface before returning an error;
each IPv6-enabled interface will have a link-local address even if the link
isn't up.

MFC after: 1 week

11 years agoRename O_DIRECTORY to O_TMP_DIRECTORY to avoid shadowing the fcntl.h
jh [Wed, 15 May 2013 18:41:49 +0000 (18:41 +0000)]
Rename O_DIRECTORY to O_TMP_DIRECTORY to avoid shadowing the fcntl.h
O_DIRECTORY flag.

PR: bin/173924
Obtained from: git://repo.or.cz/nvi.git

11 years agoConstify parameters, no functional or binary change.
delphij [Wed, 15 May 2013 18:38:28 +0000 (18:38 +0000)]
Constify parameters, no functional or binary change.

Verified with: sha256(1)
MFC after: 2 weeks

11 years agoImplement my first cut at "correct" node power-save and
adrian [Wed, 15 May 2013 18:33:05 +0000 (18:33 +0000)]
Implement my first cut at "correct" node power-save and
PS-POLL support.

This implements PS-POLL awareness i nthe

* Implement frame "leaking", which allows for a software queue
  to be scheduled even though it's asleep
* Track whether a frame has been leaked or not
* Leak out a single non-AMPDU frame when transmitting aggregates
* Queue BAR frames if the node is asleep
* Direct-dispatch the rest of control and management frames.
  This allows for things like re-association to occur (which involves
  sending probe req/resp as well as assoc request/response) when
  the node is asleep and then tries reassociating.
* Limit how many frames can set in the software node queue whilst
  the node is asleep.  net80211 is already buffering frames for us
  so this is mostly just paranoia.
* Add a PS-POLL method which leaks out a frame if there's something
  in the software queue, else it calls net80211's ps-poll routine.
  Since the ath PS-POLL routine marks the node as having a single frame
  to leak, either a software queued frame would leak, OR the next queued
  frame would leak. The next queued frame could be something from the
  net80211 power save queue, OR it could be a NULL frame from net80211.

TODO:

* Don't transmit further BAR frames (eg via a timeout) if the node is
  currently asleep.  Otherwise we may end up exhausting management frames
  due to the lots of queued BAR frames.

  I may just undo this bit later on and direct-dispatch BAR frames
  even if the node is asleep.

* It would be nice to burst out a single A-MPDU frame if both ends
  support this.  I may end adding a FreeBSD IE soon to negotiate
  this power save behaviour.

* I should make STAs timeout of power save mode if they've been in power
  save for more than a handful of seconds.  This way cards that get
  "stuck" in power save mode don't stay there for the "inactivity" timeout
  in net80211.

* Move the queue depth check into the driver layer (ath_start / ath_transmit)
  rather than doing it in the TX path.

* There could be some naughty corner cases with ps-poll leaking.
  Specifically, if net80211 generates a NULL data frame whilst another
  transmitter sends a normal data frame out net80211 output / transmit,
  we need to ensure that the NULL data frame goes out first.
  This is one of those things that should occur inside the VAP/ic TX lock.
  Grr, more investigations to do..

Tested:

* STA: AR5416, AR9280
* AP: AR5416, AR9280, AR9160

11 years agosbin/camcontrol/camcontrol.c
asomers [Wed, 15 May 2013 17:26:50 +0000 (17:26 +0000)]
sbin/camcontrol/camcontrol.c
If an expander returns 0x00 (no device attached) in the ATTACHED DEVICE
field of the SMP DISCOVER response, ignore the value of ATTACHED SAS
ADDRESS, because it is invalid.  Some expanders zero out the address
when the attached device is removed, but others do not.  Section
9.4.3.10 of the SAS Protocol Layer 2 revision 04b does not require them
to do so.

Approved by: ken (mentor)
MFC after: 3 weeks

11 years agoAdd Qlogic 10Gigabit Ethernet & CNA Adapter Driver Version 3.10.10 for
davidcs [Wed, 15 May 2013 17:03:09 +0000 (17:03 +0000)]
Add Qlogic 10Gigabit Ethernet & CNA Adapter Driver Version 3.10.10 for
QLogic 8300 Series Adapters

Submitted by: David C Somayajulu (davidcs@freebsd.org) QLogic Corporation
Approved by: George Neville-Neil (gnn@freebsd.org)

11 years agoDocument WITHOUT_CROSS_COMPILER and WITHOUT_FORMAT_EXTENSIONS.
brooks [Wed, 15 May 2013 14:35:02 +0000 (14:35 +0000)]
Document WITHOUT_CROSS_COMPILER and WITHOUT_FORMAT_EXTENSIONS.

11 years agoAdd support for an external cross compiler. The cross compiler is
brooks [Wed, 15 May 2013 14:30:03 +0000 (14:30 +0000)]
Add support for an external cross compiler.  The cross compiler is
specified by passing the XCC, XCXX, and XCPP variables (corresponding to
CC, CXX, and CPP) to buildworld/buildkernel.  The compiler must be clang
or be configured to target the appropriate architecture.

To speed build times, if XCC is an absolute path or
WITHOUT_CROSS_COMPILER is defined then no cross compiler will be built
during the cross-tools stage.

Limited documentation of this feature can currently be found at:

https://wiki.freebsd.org/ExternalToolchain

This functionality should be considered experimental and is subject to
change without notice.

Sponsored by: DARPA, AFRL
Discussed with: imp, sjg

11 years agoAdd a new option WITHOUT_FORMAT_EXTENSIONS to disable flags related to
brooks [Wed, 15 May 2013 13:04:10 +0000 (13:04 +0000)]
Add a new option WITHOUT_FORMAT_EXTENSIONS to disable flags related to
checking our kernel printf extensions.  This is useful to allow
compilers without these extensions to build kernels.

Sponsored by: DARPA, AFRL

11 years agoFix typo in comment.
des [Wed, 15 May 2013 08:38:49 +0000 (08:38 +0000)]
Fix typo in comment.

Submitted by: Alex Weber <alexwebr@gmail.com>
MFC after: 1 week

11 years agoFix a typo: It should be "strtoll" and not "stroll".
hiren [Tue, 14 May 2013 20:25:07 +0000 (20:25 +0000)]
Fix a typo: It should be "strtoll" and not "stroll".

PR: 178642
Reported by: Michael Galassi (michaelgalassi@gmail.com)
Approved by: sbruno (mentor)
MFC after: 1 week

11 years agoPort the new PV entry allocator from amd64/i386/mips to armv6/v7.
gber [Tue, 14 May 2013 09:47:58 +0000 (09:47 +0000)]
Port the new PV entry allocator from amd64/i386/mips to armv6/v7.

PV entries are now roughly half the size.
Instead of using a shared UMA zone for 28 byte pv entries
(two 8-byte tailq nodes, a 4 byte pointer, a 4 byte address and 4 byte
flags), we allocate a page at a time per process.
This provides 252 pv entries per process (actually, per pmap address space)
and eliminates one of the 8-byte tailq entries since we now can track
per-process pv entries implicitly.
The pointer to the pmap can be eliminated by doing address arithmetic to
find the metadata on the page headers to find a single pointer shared by
all 252 entries. There is an 8-int bitmap for the freelist of those 252
entries.
When in serious low memory condition, allocation of another pv_chunk is
possible by freeing some pages in pmap_pv_reclaim().

Added pv_entry/pv_chunk related statistics to pmap.
pv_entry/pv_chunk statistics can be accessed via sysctl vm.pmap.

Ported PTE freelist of KVA allocation and maintenance from i386.
Using an idea from Stephan Uphoff, use the empty pte's that correspond
to the unused kva in the pv memory block to thread a freelist through.
This allows us to free pages that used to be used for pv entry chunks
since we can now track holes in the kva memory block.

As both ARM pmap.c and pmap-v6.c use the same header and pv_entry, pmap and
md_page structures are different, it was needed to separate code designed
for ARMv6/7 from the one for other ARMs.

Submitted by: Zbigniew Bodek <zbb@semihalf.com>
Reviewed by: alc
Sponsored by: The FreeBSD Foundation, Semihalf

11 years agoCentralize standard getopts arguments, both for convenience and to correct
dteske [Tue, 14 May 2013 03:21:13 +0000 (03:21 +0000)]
Centralize standard getopts arguments, both for convenience and to correct
a bug in which certain combinations of arguments produced unexpected results
such as `-dX' (now properly produces debugging and X11), `-XS' (now properly
produces X11 in secure mode), `-df-' (enables debugging when reading a
script from standard-input, etc. Multi-word variations such as `-d -X',
`-X -S', `-d -f-', `-d -f -', etc. also work as expected. Also tested were
variations in argument order, which are now working as expected.

11 years agoAdd less to rescue build. On amd64, this increases rescue size by about
delphij [Mon, 13 May 2013 22:16:33 +0000 (22:16 +0000)]
Add less to rescue build.  On amd64, this increases rescue size by about
130KB or 2.4%.

MFC after: 1 month

11 years agoImprove readability of static assertions for OFFSET_* macros.
ed [Mon, 13 May 2013 21:47:17 +0000 (21:47 +0000)]
Improve readability of static assertions for OFFSET_* macros.

Instead of doing all sorts of weird casting of constants to
pointer-pointers, simply use the standard C offsetof() macro to obtain
the offset of the respective fields in the structures.

11 years agoRework the way C11 keywords are defined.
ed [Mon, 13 May 2013 21:46:07 +0000 (21:46 +0000)]
Rework the way C11 keywords are defined.

Instead of only checking the __STDC_VERSION__, we can also use Clang's
__has_extension() to check for features specifically. This allows us to,
say, use Clang's native _Static_assert() instead of the typedef hack,
making the compiler error messages a lot more readable.

Reviewed by: theraven

11 years agoAdd ALQ beacon debugging.
adrian [Mon, 13 May 2013 21:18:00 +0000 (21:18 +0000)]
Add ALQ beacon debugging.

11 years agoSupport sending ATH_ALQ messages with no payload.
adrian [Mon, 13 May 2013 21:17:27 +0000 (21:17 +0000)]
Support sending ATH_ALQ messages with no payload.

11 years agoetc/rc.d/syslogd
asomers [Mon, 13 May 2013 20:28:24 +0000 (20:28 +0000)]
etc/rc.d/syslogd
Add netif as a requirement of syslogd to get lo0 up.  Currently, this
doesn't affect the rc order, because mountcritremote already depends on
netif.

Reviewed by: eadler
Approved by: kenm (mentor)
MFC after: 2 weeks

11 years agoUse an ugly hack to get around bootstrapping problems when building
dim [Mon, 13 May 2013 20:14:58 +0000 (20:14 +0000)]
Use an ugly hack to get around bootstrapping problems when building
clang on head between r239347 and r245428.

The former revision introduced CLOCK_PROCESS_CPUTIME_ID as a clock id
for the clock_gettime() function and friends, but it was only added in
<sys/time.h>, not in <time.h>.  Any program including <time.h> would
therefore not be able to use CLOCK_PROCESS_CPUTIME_ID, even though the
value of _POSIX_CPUTIME indicates its existence.  The latter revision
synchronized the defines again.

Work around this problem by defining the id on the command line for the
particular .cpp file that needs it.  If the id ever changes value, this
hack will need to be updated.

11 years agoRemove a bogus re-assignment.
jkim [Mon, 13 May 2013 20:03:59 +0000 (20:03 +0000)]
Remove a bogus re-assignment.

MFC after: 3 days

11 years agoDeal correctly with 40G ports that don't have any transceiver plugged
np [Mon, 13 May 2013 20:00:03 +0000 (20:00 +0000)]
Deal correctly with 40G ports that don't have any transceiver plugged
in.  Do not claim that they have unknown tranceivers.

MFC after: 3 days

11 years agoFix typo in net.inet.tcp.minmss sysctl description.
jimharris [Mon, 13 May 2013 19:55:27 +0000 (19:55 +0000)]
Fix typo in net.inet.tcp.minmss sysctl description.

MFC after: 3 days

11 years agoAs python3 does not have raw_input(), convert it to input() when we
hiren [Mon, 13 May 2013 19:53:19 +0000 (19:53 +0000)]
As python3 does not have raw_input(), convert it to input() when we
are using python3.

PR: 177214
Reviewed by: gnn
Approved by: sbruno (mentor)

11 years agoImprove the debugging output - use the MAC address rather than various
adrian [Mon, 13 May 2013 19:52:35 +0000 (19:52 +0000)]
Improve the debugging output - use the MAC address rather than various
pointer values everywhere.

11 years agoSince the node state is 100% back under the TX lock, just kill the use
adrian [Mon, 13 May 2013 19:03:12 +0000 (19:03 +0000)]
Since the node state is 100% back under the TX lock, just kill the use
of atomics.

I'll re-think this nonsense later.

11 years agoOops, commit the other half of r250606.
adrian [Mon, 13 May 2013 19:02:22 +0000 (19:02 +0000)]
Oops, commit the other half of r250606.

11 years agoThis lock only protects the rate control state for now, mention this.
adrian [Mon, 13 May 2013 18:57:18 +0000 (18:57 +0000)]
This lock only protects the rate control state for now, mention this.

11 years agoBegin tidying up the reassociation and node sleep/wakeup paths.
adrian [Mon, 13 May 2013 18:56:04 +0000 (18:56 +0000)]
Begin tidying up the reassociation and node sleep/wakeup paths.

* Move the node sleep/wake state under the TX lock rather than the
  node lock.  Let's leave the node lock protecting rate control only
  for now.

* When reassociating, various state needs to be cleared.  For example,
  the aggregate session needs to be torn down, including any pending
  aggregation negotiation and BAR TX waiting.

* .. and we need to do a "cleanup" pass since frames in the hardware
  TX queue need to be transmitted.

Modify ath_tx_tid_cleanup() to be called with the TX lock held and push
frames into a completion list.  This allows for the cleanup to be
done atomically for all TIDs in a node rather than grabbing and
releasing the TX lock each time.

11 years agoSet st_nlink in the stat structure within the inode to 1 as well.
marcel [Mon, 13 May 2013 18:34:33 +0000 (18:34 +0000)]
Set st_nlink in the stat structure within the inode to 1 as well.
The cd9660 file system uses that field for the link count and it
was 0. This impacts pwd_mkdb(8) as it checks for st_nlink not being
0 as part of closing a race.

11 years agomdoc sweep
joel [Mon, 13 May 2013 18:13:50 +0000 (18:13 +0000)]
mdoc sweep

11 years agoo Add accessor functions to add and remove pages from a specific
attilio [Mon, 13 May 2013 15:40:51 +0000 (15:40 +0000)]
o Add accessor functions to add and remove pages from a specific
  freelist.
o Split the pool of free pages queues really by domain and not rely on
  definition of VM_RAW_NFREELIST.
o For MAXMEMDOM > 1, wrap the RR allocation logic into a specific
  function that is called when calculating the allocation domain.
  The RR counter is kept, currently, per-thread.
  In the future it is expected that such function evolves in a real
  policy decision referee, based on specific informations retrieved by
  per-thread and per-vm_object attributes.
o Add the concept of "probed domains" under the form of vm_ndomains.
  It is responsibility for every architecture willing to support multiple
  memory domains to correctly probe vm_ndomains along with mem_affinity
  segments attributes.  Those two values are supposed to remain always
  consistent.
  Please also note that vm_ndomains and td_dom_rr_idx are both int
  because segments already store domains as int.  Ideally u_int would
  have much more sense. Probabilly this should be cleaned up in the
  future.
o Apply RR domain selection also to vm_phys_zero_pages_idle().

Sponsored by: EMC / Isilon storage division
Partly obtained from: jeff
Reviewed by: alc
Tested by: jeff

11 years agoAdd a remark to the effect that a manually started relearn will always
markj [Mon, 13 May 2013 15:00:36 +0000 (15:00 +0000)]
Add a remark to the effect that a manually started relearn will always
result in the battery being completely drained, even in transparent learning
mode.

MFC in: 3 days
Sponsored by: Sandvine Inc.

11 years agoThe HPN patch added a new BUG bit for SSH_BUG_LARGEWINDOW
bdrewery [Mon, 13 May 2013 11:32:20 +0000 (11:32 +0000)]
The HPN patch added a new BUG bit for SSH_BUG_LARGEWINDOW
and the update to 6.1 added SSH_BUG_DYNAMIC_RPORT with the
same value.

Fix the HPN SSH_BUG_LARGEWINDOW bit so it is unique.

Approved by: des
MFC after: 2 weeks

11 years agoBandaid for compiling with gcc, which happens to be the default compiler
peter [Mon, 13 May 2013 07:09:31 +0000 (07:09 +0000)]
Bandaid for compiling with gcc, which happens to be the default compiler
for a number of platforms still.

11 years agoPull in r181286 from upstream llvm trunk:
dim [Mon, 13 May 2013 07:02:15 +0000 (07:02 +0000)]
Pull in r181286 from upstream llvm trunk:

  LoopVectorize: getConsecutiveVector must respect signed arithmetic

  We were passing an i32 to ConstantInt::get where an i64 was needed and we must
  also pass the sign if we pass negatives numbers. The start index passed to
  getConsecutiveVector must also be signed.

  Should fix PR15882.

This should fix Firefox crashes some people have been reporting, when it
is compiled with -O3.

11 years agoMFV: less v458.
delphij [Mon, 13 May 2013 06:52:46 +0000 (06:52 +0000)]
MFV: less v458.

MFC after: 2 weeks

11 years agomdoc sweep.
joel [Sun, 12 May 2013 22:22:12 +0000 (22:22 +0000)]
mdoc sweep.

11 years agoAdd support for the eofflag to nfs_readdir() in the new NFS
rmacklem [Sun, 12 May 2013 21:48:08 +0000 (21:48 +0000)]
Add support for the eofflag to nfs_readdir() in the new NFS
client so that it works under a unionfs mount.

Submitted by: Jared Yanovich (slovichon@gmail.com)
Reviewed by: kib
MFC after: 2 weeks

11 years agoRevert r250565 which causes issues for older CPUs
eadler [Sun, 12 May 2013 21:24:18 +0000 (21:24 +0000)]
Revert r250565 which causes issues for older CPUs

PR: conf/178504
Requested by: many

11 years ago - pctrie really only requires two byte alignment so that there is a single
jeff [Sun, 12 May 2013 20:44:28 +0000 (20:44 +0000)]
 - pctrie really only requires two byte alignment so that there is a single
   bit available for a flag in the pointer.  However, it felt more correct
   to enforce natural alignment of the key pointer.  Unfortunately on
   32bit architectures 64bit integers are not always naturally aligned.
   Change the assert to enforce only 32bit alignment of the 64bit key for
   now to fix the build.  A more correct fix would be to properly sort
   the struct buf fields which definitely suffer from bloat due to padding.

11 years agoRefactor vm_page_alloc()'s interactions with vm_reserv_alloc_page() and
alc [Sun, 12 May 2013 16:50:18 +0000 (16:50 +0000)]
Refactor vm_page_alloc()'s interactions with vm_reserv_alloc_page() and
vm_page_insert() so that (1) vm_radix_lookup_le() is never called while the
free page queues lock is held and (2) vm_radix_lookup_le() is called at most
once.  This change reduces the average time that the free page queues lock
is held by vm_page_alloc() as well as vm_page_alloc()'s average overall
running time.

Sponsored by: EMC / Isilon Storage Division

11 years agoFix several typos
eadler [Sun, 12 May 2013 16:43:26 +0000 (16:43 +0000)]
Fix several typos

PR: kern/176054
Submitted by: Christoph Mallon <christoph.mallon@gmx.de>
MFC after: 3 days

11 years agohead -n doesn't take negative arguments on FreeBSD, so instead use sed '$d'
markj [Sun, 12 May 2013 16:29:09 +0000 (16:29 +0000)]
head -n doesn't take negative arguments on FreeBSD, so instead use sed '$d'
to remove the last line of a file.

11 years agoBring back part of r249367 by adding DTrace's temporal option, which allows
markj [Sun, 12 May 2013 16:26:33 +0000 (16:26 +0000)]
Bring back part of r249367 by adding DTrace's temporal option, which allows
users to guarantee that the output of DTrace scripts will be time-ordered.
This option is enabled by adding the line

  #pragma D option temporal

to the beginning of a script, or by adding '-x temporal' to the arguments of
dtrace(1).

This change fixes a bug in the original port of the temporal option. This
bug was causing some assertions to fail, so they had been disabled; in this
revision the assertions are working properly and are enabled.

The DTrace version number has been bumped from 1.9.0 to 1.9.1 to reflect
the language change that's being introduced.

This change corresponds to part of illumos-gate commit e5803b76927480:
  3021 option for time-ordered output from dtrace(1M)

Reviewed by: pfg
Obtained from: illumos
MFC after: 1 month

11 years agoFix typo in UPDATING
eadler [Sun, 12 May 2013 16:07:23 +0000 (16:07 +0000)]
Fix typo in UPDATING

Submitted by: Rainer Hurling <rhurlin@gwdg.de>

11 years agofdescfs: Supply a real value for d_type in readdir.
jilles [Sun, 12 May 2013 15:44:49 +0000 (15:44 +0000)]
fdescfs: Supply a real value for d_type in readdir.

All the fdescfs nodes (except . and ..) appear as character devices to
stat(), so DT_CHR is correct.

11 years agoAdd support for "d" floating-point suffix, as defined by draft N1312
pfg [Sun, 12 May 2013 15:29:35 +0000 (15:29 +0000)]
Add support for "d" floating-point suffix, as defined by draft N1312
of TR 24732. Emit pedantic warning if the feature is being used.

Should solve GCC bug 39027.

Obtained from: OpenBSD
MFC after: 2 weeks

11 years agoMake newsyslog compress logs with xz instead of bzip2 to save space.
eadler [Sun, 12 May 2013 15:23:59 +0000 (15:23 +0000)]
Make newsyslog compress logs with xz instead of bzip2 to save space.

PR: conf/178504
Submitted by: ak
Reviewed by: smh

11 years agoRevert r250394, which disabled snmp_hast due to libl dependency issue --
trociny [Sun, 12 May 2013 13:42:49 +0000 (13:42 +0000)]
Revert r250394, which disabled snmp_hast due to libl dependency issue --
the dependency has been removed in r250503.

11 years agoSuppress error printing for "PREVENT ALLOW MEDIUM REMOVAL" on da open.
mav [Sun, 12 May 2013 09:33:33 +0000 (09:33 +0000)]
Suppress error printing for "PREVENT ALLOW MEDIUM REMOVAL" on da open.
Change at r250208 exposed more errors here, hidden before.  The same flag
is used in cd driver.

11 years agoTypo.
pluknet [Sun, 12 May 2013 05:54:58 +0000 (05:54 +0000)]
Typo.

11 years agoAdd the letter 'x' to the font iso05-8x16.fnt
eadler [Sun, 12 May 2013 04:34:55 +0000 (04:34 +0000)]
Add the letter 'x' to the font iso05-8x16.fnt

PR: 169797
Submitted by: Takefu <takefu@airport.fm>
MFC after: 3 days

11 years ago - Add a new general purpose path-compressed radix trie which can be used
jeff [Sun, 12 May 2013 04:05:01 +0000 (04:05 +0000)]
 - Add a new general purpose path-compressed radix trie which can be used
   with any structure containing a uint64_t index.  The tree code
   auto-generates type safe wrappers.
 - Eliminate the buf splay and replace it with pctrie.  This is not only
   significantly faster with large files but also allows for the possibility
   of shared locking.

Reviewed by:    alc, attilio
Sponsored by:   EMC / Isilon Storage Division

11 years agoUpdate the gcc43 changelog.
pfg [Sun, 12 May 2013 03:36:28 +0000 (03:36 +0000)]
Update the gcc43 changelog.

The issue solved in r250392 actually originated upstream and was
fixed in upstream gcc43 branch in 2006/10/24 (under GPLv2).

Register it in the appropriate ChangeLog for reference.

MFC after: 3 days

11 years agoSome filesystems (NFS in particular) do not fill out the d_type field when
markj [Sun, 12 May 2013 01:58:04 +0000 (01:58 +0000)]
Some filesystems (NFS in particular) do not fill out the d_type field when
returning directory entries through readdir(3). In this case we need to
obtain the file type ourselves; otherwise newsyslog -t will not be able to
find archived log files and will fail to both delete old log files and to
do interval-based rotations properly.

Reported by: jilles
Reviewed by: jilles
MFC after: 2 weeks

11 years agoTidy up some CVS workarounds.
peter [Sun, 12 May 2013 01:53:47 +0000 (01:53 +0000)]
Tidy up some CVS workarounds.

11 years agoTurns out that there really isn't an RTC chip on this board, at all.
imp [Sun, 12 May 2013 01:29:18 +0000 (01:29 +0000)]
Turns out that there really isn't an RTC chip on this board, at all.

11 years agoComment.
dteske [Sun, 12 May 2013 00:50:18 +0000 (00:50 +0000)]
Comment.