]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
12 years agoAdd three additional clang intrinsics headers, which I missed in the
dim [Mon, 20 Aug 2012 20:56:40 +0000 (20:56 +0000)]
Add three additional clang intrinsics headers, which I missed in the
previous import.

12 years agoAvoid segfault in the 'smpphylist' subcommand.
pluknet [Mon, 20 Aug 2012 20:40:14 +0000 (20:40 +0000)]
Avoid segfault in the 'smpphylist' subcommand.
Initialize devlist.dev_queue tail queue early enough before its any
potential traversal in freebusdevlist() when in smpphylist error path.

Reported by: Pavel Polyakov <bsd kobyla org> (on irc)
Reviewed by: ken
MFC after: 5 days

12 years agoFix a build issue when ATH_DEBUG isn't defined - just initialise and use
adrian [Mon, 20 Aug 2012 18:57:41 +0000 (18:57 +0000)]
Fix a build issue when ATH_DEBUG isn't defined - just initialise and use
qnum.

12 years agoAs of r232844 we no longer need the maxpoll 9 workaround.
delphij [Mon, 20 Aug 2012 18:45:16 +0000 (18:45 +0000)]
As of r232844 we no longer need the maxpoll 9 workaround.

MFC after: 3 days

12 years agoUpgrade our copy of llvm/clang to trunk r162107. With thanks to
dim [Mon, 20 Aug 2012 18:33:03 +0000 (18:33 +0000)]
Upgrade our copy of llvm/clang to trunk r162107.  With thanks to
Benjamin Kramer and Joerg Sonnenberger for their input and fixes.

12 years agoRefine the changes made in r208212 to avoid bogus failures from
jhb [Mon, 20 Aug 2012 16:00:33 +0000 (16:00 +0000)]
Refine the changes made in r208212 to avoid bogus failures from
if_delmulti() when clearing the configuration for a subinterface when
the parent interface is being detached.  The current code was still
triggering an assertion in if_delmulti() due to the parent interface being
partially detached.  Fix this by not calling if_delmulti() at all if the
parent interface is being detached.  Warn if if_delmulti() fails when the
parent is not being detached (but similar to 208212, still proceed with
tearing down the vlan state).

Tested by: ae@
MFC after: 1 month

12 years agoWrap debugging in #ifdef ATH_DEBUG
adrian [Mon, 20 Aug 2012 15:30:26 +0000 (15:30 +0000)]
Wrap debugging in #ifdef ATH_DEBUG

12 years agoDefine and exclude DRAM regions used by hardware/bootloder on XLP
jchandra [Mon, 20 Aug 2012 11:51:49 +0000 (11:51 +0000)]
Define and exclude DRAM regions used by hardware/bootloder on XLP

Fix xlp_mem_init() - remove the ad-hoc code for excluding memory regions
and use an array of regions.

12 years agoFlesh out some initial EDMA TX FIFO fill, complete and refill routines.
adrian [Mon, 20 Aug 2012 06:11:04 +0000 (06:11 +0000)]
Flesh out some initial EDMA TX FIFO fill, complete and refill routines.

Note: This is totally sub-optimal and a work in progress.

* Support filling an empty FIFO TXQ with frames from the ath_buf queue
  in the ath_txq list.  However, since there's (currently) no clean, easy
  way to separate the frames that are in the FIFO versus just waiting,
  the code waits for the FIFO to be totally empty before it attempts to
  queue more.  This is highly sub-optimal but is enough to get the ball
  rolling.

* A _lot_ of the code assumes that the TX status is filled out in the
  struct ath_buf bf_status field.  So for now, memcpy() the completion over.

* None of the TX drain / reset routines will attempt to complete completed
  frames before draining, so it can't be used for 802.11n TX aggregation.
  (This won't work anyway, as the aggregation TX descriptor API hasn't
  yet been converted; and that'll happen in some future commits.)

* Fix an issue where the FIFO counter wasn't being incremented, leading
  to the queue logic just plain not working.

* HAL_EIO means "descriptor wasn't valid", versus "not finished, don't
  continue." So don't stop processing descriptors when HAL_EIO is hit.

* Don't service frame completion from the beacon queue.  It isn't currently
  fully setup like a real queue and the first attempt at accessing the
  queue lock will panic the kernel.

Tested:

* AR9380, STA mode

This commit is brought to you by said AR9380 in STA mode.

12 years agoAdvance the descriptor pointer by sc->sc_tx_desclen bytes, rather than
adrian [Mon, 20 Aug 2012 06:02:09 +0000 (06:02 +0000)]
Advance the descriptor pointer by sc->sc_tx_desclen bytes, rather than
sizeof(struct ath_desc).  This isn't correct for EDMA TX descriptors.

This popped up during iperf tests. Ping tests never created frames that
had enough segments to overflow into a second descriptor.  However,
an iperf TCP test would do that after a few seconds; the second descriptor
would almost always certainly have garbage.

Tested:

* AR9380, STA mode
* AR9280, STA mode (802.11n TX, legacy TX)

12 years agoMake sure all of the buffers are printed, rather than (n-1).
adrian [Mon, 20 Aug 2012 05:47:07 +0000 (05:47 +0000)]
Make sure all of the buffers are printed, rather than (n-1).

12 years agoAdd the ds1631 temperature driver.
andreast [Sun, 19 Aug 2012 19:44:13 +0000 (19:44 +0000)]
Add the ds1631 temperature driver.

12 years agoAdd a new sound driver for PowerMacs, found here on my Quad G5.
andreast [Sun, 19 Aug 2012 19:40:33 +0000 (19:40 +0000)]
Add a new sound driver for PowerMacs, found here on my Quad G5.
It allows simple playback and volume control like the other Mac drivers,
not more.

12 years agoAdd a new temperature driver for certain PowerMacs. Found here on my Quad G5.
andreast [Sun, 19 Aug 2012 19:37:14 +0000 (19:37 +0000)]
Add a new temperature driver for certain PowerMacs. Found here on my Quad G5.

12 years agoUnify the sysctl description with the other PowerMac temperature drivers.
andreast [Sun, 19 Aug 2012 19:34:10 +0000 (19:34 +0000)]
Unify the sysctl description with the other PowerMac temperature drivers.

12 years agoAvoid using the degree symbol. Looks ugly on the console.
andreast [Sun, 19 Aug 2012 19:32:38 +0000 (19:32 +0000)]
Avoid using the degree symbol. Looks ugly on the console.

12 years agoDo the ADC init only at startup and not during every sensor read call.
andreast [Sun, 19 Aug 2012 19:31:36 +0000 (19:31 +0000)]
Do the ADC init only at startup and not during every sensor read call.
This reduces the number of interrupts.

12 years agoFix typo. Not a win in terms of functionality but in terms of completeness.
andreast [Sun, 19 Aug 2012 19:17:54 +0000 (19:17 +0000)]
Fix typo. Not a win in terms of functionality but in terms of completeness.

12 years agoThough I disagree, I conceed to jhb & Rui. Note
rrs [Sun, 19 Aug 2012 11:54:02 +0000 (11:54 +0000)]
Though I disagree, I conceed to jhb & Rui. Note
that we still have a problem with this whole structure of
locks and in_input.c [it does not lock which it should not, but
this *can* lead to crashes]. (I have seen it in our SQA
testbed.. besides the one with a refcnt issue that I will
have SQA work on next week ;-)

12 years agoUpdate zfs(8) manpage with illumos version of "zfs diff"
mm [Sun, 19 Aug 2012 10:34:40 +0000 (10:34 +0000)]
Update zfs(8) manpage with illumos version of "zfs diff"

Illumos issue:
  2399 zfs manual page does not document use of "zfs diff"

References:
  https://www.illumos.org/issues/2399

PR: docs/170764
Obtained from: ssh://anonhg@hg.illumos.org/illumos-gate
MFC after: 1 week

12 years agoVendor import of llvm trunk r162107:
dim [Sun, 19 Aug 2012 10:31:50 +0000 (10:31 +0000)]
Vendor import of llvm trunk r162107:
http://llvm.org/svn/llvm-project/llvm/trunk@162107

12 years agoBackport fix for vendor issue #3085
mm [Sun, 19 Aug 2012 09:59:41 +0000 (09:59 +0000)]
Backport fix for vendor issue #3085
3085 zfs diff panics, then panics in a loop on booting

References:
  https://www.illumos.org/issues/3085

PR: kern/170763
Obtained from: ssh://anonhg@hg.illumos.org/illumos-gate (r13772)
MFC after: 1 week

12 years agoIn ip6_ctloutput() guard inp_flags modifications with INP_WLOCK.
trociny [Sun, 19 Aug 2012 08:16:13 +0000 (08:16 +0000)]
In ip6_ctloutput() guard inp_flags modifications with INP_WLOCK.

MFC after: 2 weeks

12 years ago- Allow to pass extra parameters for each jails.
kuriyama [Sun, 19 Aug 2012 08:15:32 +0000 (08:15 +0000)]
- Allow to pass extra parameters for each jails.
- To achieve above, convert jail(8) invocation to use new style
  command line "-c" flag.

Reviewed at: freebsd-jail@

12 years agoExtend the TX descriptor debug printing to be properly aware of
adrian [Sun, 19 Aug 2012 02:22:16 +0000 (02:22 +0000)]
Extend the TX descriptor debug printing to be properly aware of
EDMA code.

* create a new TX EDMA descriptor struct to represent TX EDMA descriptors
  when doing debugging;
* implement an EDMA printing function which:
  + hardcodes the TX map size to 4 for now;
  + correctly prints out the number of segments - there's one descriptor
    for up to 4 buffers (segments), not one for each segment;
  + print out 4 DS buffer and len pointers;
  + print out the correct number of DWORDs in the TX descriptor.

TODO:

* Remove all of the hard-coded stuff. Ew.

12 years agoWhen assembling the descriptor list, make sure that the "first" descriptor
adrian [Sun, 19 Aug 2012 02:16:22 +0000 (02:16 +0000)]
When assembling the descriptor list, make sure that the "first" descriptor
is marked correctly.

The existing logic assumed that the first descriptor is i == 0, which
doesn't hold for EDMA TX.  In this instance, the first time filltxdesc()
is called can be up to i == 3.

So for a two-buffer descriptor:

* firstSeg is set to 0;
* lastSeg is set to 1;
* the ath_hal_filltxdesc() code will treat it as the last segment in
  a descriptor chain and blank some of the descriptor fields, causing
  the TX to stop.

When firstSeg is set to 1 (regardless of lastSeg), it overrides the
lastSeg setting.  Thus, ath_hal_filltxdesc() won't blank out these
fields.

Tested: AR9380, STA mode.  With this, association is successful.

12 years agoUse pmap_kextract(x) rather than pmap_extract(kernel_pmap, x). The
marcel [Sat, 18 Aug 2012 23:28:34 +0000 (23:28 +0000)]
Use pmap_kextract(x) rather than pmap_extract(kernel_pmap, x). The
former knows about all the special mappings, like PBVM. The kernel
text and data are in the PBVM.

12 years agoBuild zfs on ia64 as well. We don't build a loader with zfs support
marcel [Sat, 18 Aug 2012 23:23:14 +0000 (23:23 +0000)]
Build zfs on ia64 as well. We don't build a loader with zfs support
just yet, but building it makes sure we catch (build) regressions.

12 years agoDon't build the ski loader anymore. The code is not deleted just yet,
marcel [Sat, 18 Aug 2012 23:17:42 +0000 (23:17 +0000)]
Don't build the ski loader anymore. The code is not deleted just yet,
as it explains the distinction between the common and the platform-
specific code.

12 years agoRemove support for SKI: HP's Itanium simulator. It's pretty much not
marcel [Sat, 18 Aug 2012 22:59:06 +0000 (22:59 +0000)]
Remove support for SKI: HP's Itanium simulator. It's pretty much not
used, serves very little value given that FreeBSD runs on real H/W
for a long time.
Note that SKI is open-source (see http://ski.sourceforge.net), so
if there's interest and value again, then this code can be revived.

Discussed with: jhb

12 years agoAdd drm and i915 ioctl translations for 32 bit process on 64 bit host.
kib [Sat, 18 Aug 2012 18:26:25 +0000 (18:26 +0000)]
Add drm and i915 ioctl translations for 32 bit process on 64 bit host.

Submitted by: meowthink@gmail.com
MFC after: 2 weeks

12 years agoDeliver SIGSYS to the guilty thread, not to the process.
kib [Sat, 18 Aug 2012 18:17:10 +0000 (18:17 +0000)]
Deliver SIGSYS to the guilty thread, not to the process.

MFC after: 1 week

12 years agoCleanup use of 'host' when running 'rpcinfo -p'.
kevlo [Sat, 18 Aug 2012 16:14:50 +0000 (16:14 +0000)]
Cleanup use of 'host' when running 'rpcinfo -p'.

Obtained from: NetBSD

12 years agoFix typo
kevlo [Sat, 18 Aug 2012 16:13:16 +0000 (16:13 +0000)]
Fix typo

12 years agoAdd mvts(4) driver for internal thermal sensor found on 88F6282 and 88F6283.
hrs [Sat, 18 Aug 2012 12:37:07 +0000 (12:37 +0000)]
Add mvts(4) driver for internal thermal sensor found on 88F6282 and 88F6283.
The temperature value will be exported via sysctl like this:

dev.mvts.0.temperature: 52.1C

12 years agoSort IDs.
hrs [Sat, 18 Aug 2012 12:20:51 +0000 (12:20 +0000)]
Sort IDs.

12 years agoFix build when DEBUG is defined.
hrs [Sat, 18 Aug 2012 12:08:04 +0000 (12:08 +0000)]
Fix build when DEBUG is defined.

12 years ago- MV_DEV_88F6282 has 256KB 4-way L2 cache.
hrs [Sat, 18 Aug 2012 11:40:55 +0000 (11:40 +0000)]
- MV_DEV_88F6282 has 256KB 4-way L2 cache.
- Sort IDs in win_cpu_can_remap() and remove MV_DEV_MV78100 because it is
  included in MV_DEV_DISCOVERY.
- Add MV_DEV_MV78XXX to xor_max_eng().

12 years agoFix a bug that could fail to initialize GPIO pins specified in "gpios" because
hrs [Sat, 18 Aug 2012 11:33:21 +0000 (11:33 +0000)]
Fix a bug that could fail to initialize GPIO pins specified in "gpios" because
"gpio-controller" property at the controller node was always ignored.

12 years ago- Add OF_hasprop() and ofw_bus_has_prop(). These can be used to check
hrs [Sat, 18 Aug 2012 11:25:07 +0000 (11:25 +0000)]
- Add OF_hasprop() and ofw_bus_has_prop().  These can be used to check
  existence of the property.

- Fix ofw_bus_is_compatible{,_strict}() to prevent substring match in the
  compatible string.

Reviewed by: raj

12 years agoFix two instances of the file names generated by revnetgroup(8).
zeising [Sat, 18 Aug 2012 09:26:51 +0000 (09:26 +0000)]
Fix two instances of the file names generated by revnetgroup(8).

Approved by: joel (mentor)

12 years agoRemove machine from the LN2410SBC config, it is set by std.s3c2410
andrew [Sat, 18 Aug 2012 05:52:17 +0000 (05:52 +0000)]
Remove machine from the LN2410SBC config, it is set by std.s3c2410

12 years agoSet machine correctly on ARM. This allows universe to use the correct world
andrew [Sat, 18 Aug 2012 05:48:19 +0000 (05:48 +0000)]
Set machine correctly on ARM. This allows universe to use the correct world
when building each kernel.

Reviewed by: imp

12 years agoEliminate another vestige of page coloring.
alc [Fri, 17 Aug 2012 20:15:01 +0000 (20:15 +0000)]
Eliminate another vestige of page coloring.

12 years agoCatch up with the subversion conversion and apply these build-time patches.
obrien [Fri, 17 Aug 2012 18:20:38 +0000 (18:20 +0000)]
Catch up with the subversion conversion and apply these build-time patches.

12 years agoRemove unused member of struct indir (in_exists) from UFS and EXT2 code.
mjg [Fri, 17 Aug 2012 17:45:27 +0000 (17:45 +0000)]
Remove unused member of struct indir (in_exists) from UFS and EXT2 code.

Reviewed by: mckusick
Approved by: trasz (mentor)
MFC after: 1 week

12 years agoAdd new USB device ID.
hselasky [Fri, 17 Aug 2012 16:27:11 +0000 (16:27 +0000)]
Add new USB device ID.

PR: usb/170688
MFC after: 1 week

12 years agoUnexpand a couple of TAILQ_FOREACH()s.
jhb [Fri, 17 Aug 2012 16:01:24 +0000 (16:01 +0000)]
Unexpand a couple of TAILQ_FOREACH()s.

12 years agoFix dhclient to properly exit and teardown the configured lease when
jhb [Fri, 17 Aug 2012 15:53:43 +0000 (15:53 +0000)]
Fix dhclient to properly exit and teardown the configured lease when
link is lost.  devd will start a new dhclient instance when link is
restored.

PR: bin/166656
Submitted by: Peter Jeremy (mostly)
Reviewed by: brooks (earlier version from Peter)
MFC after: 1 month

12 years agoAllow static DMA allocations that allow for enough segments to do page-sized
jhb [Fri, 17 Aug 2012 14:14:25 +0000 (14:14 +0000)]
Allow static DMA allocations that allow for enough segments to do page-sized
segments for the entire allocation to use kmem_alloc_attr() to allocate
KVM rather than using kmem_alloc_contig().  This avoids requiring
a single physically contiguous chunk in this case.

Submitted by: Peter Jeremy (original version)
MFC after: 1 month

12 years agoOk jhb, lets move the ifa_free() down to the bottom to
rrs [Fri, 17 Aug 2012 05:51:46 +0000 (05:51 +0000)]
Ok jhb, lets move the ifa_free() down to the bottom to
assure that *all* tables and such are removed before
we start to free. This won't protect the Hash in ip_input.c
but in theory should protect any other uses that *do* use locks.

MFC after: 1 week (or more)

12 years agoFix two problems with pmap_clear_modify().
alc [Fri, 17 Aug 2012 05:02:29 +0000 (05:02 +0000)]
Fix two problems with pmap_clear_modify().

First, pmap_clear_modify() is write protecting all mappings to the specified
page, not just clearing the modified bit.  Specifically, it sets PTE_RO on
the PTE, which is wrong.  Moreover, it is calling vm_page_dirty(), which is
not the expected behavior for pmap_clear_modify().  Generally speaking, the
machine-independent VM layer masks these mistakes.  For example, setting
PTE_RO will result in additional soft faults, but not a catastrophe.

Second, pmap_clear_modify() may not clear the modified bits because it only
iterates over the PV list when the page has the PV_TABLE_MOD flag set and
elsewhere the pmap clears the PV_TABLE_MOD flag anytime a modified mapping
is write protected or destroyed.  However, the page may still have other
mappings with the modified bit set.

Eliminate a stale comment.

12 years agoThe GPIO drivers were initialising their mutexes with type of
rpaulo [Fri, 17 Aug 2012 04:44:57 +0000 (04:44 +0000)]
The GPIO drivers were initialising their mutexes with type of
MTX_NETWORK_LOCK. This is wrong since these mutexes have nothing to do
with networking.

12 years agoregen.
davidxu [Fri, 17 Aug 2012 02:47:16 +0000 (02:47 +0000)]
regen.

12 years agoCorrect a regression introduced during the import of file(1) 5.11.
mckay [Fri, 17 Aug 2012 02:27:17 +0000 (02:27 +0000)]
Correct a regression introduced during the import of file(1) 5.11.

Magic tests containing "search" or "regex" directives were incorrectly
compiled by "mkmagic" and were effectively ignored.  This caused troff
files (for example) to be detected as simply "ASCII text" instead of
as "troff or preprocessor input, ASCII text".

PR: bin/170415
Approved by: consensus on developers@
MFC after: 3 days

12 years agoImplement syscall clock_getcpuclockid2, so we can get a clock id
davidxu [Fri, 17 Aug 2012 02:26:31 +0000 (02:26 +0000)]
Implement syscall clock_getcpuclockid2, so we can get a clock id
for process, thread or others we want to support.
Use the syscall to implement POSIX API clock_getcpuclock and
pthread_getcpuclockid.

PR: 168417

12 years agoThe TCP PAWS fix for kernels with fast tick rates (r231767) changed the TCP
lstewart [Fri, 17 Aug 2012 01:49:51 +0000 (01:49 +0000)]
The TCP PAWS fix for kernels with fast tick rates (r231767) changed the TCP
timestamp related stack variables to reference ms directly instead of ticks.
The h_ertt(4) Khelp module relies on TCP timestamp information in order to
calculate its enhanced RTT estimates, but was not updated as part of r231767.

Consequently, h_ertt has not been calculating correct RTT estimates since
r231767 was comitted, which in turn broke all delay-based congestion control
algorithms because they rely on the h_ertt RTT estimates.

Fix the breakage by switching h_ertt to use tcp_ts_getticks() in place of all
previous uses of the ticks variable. This ensures all timestamp related
variables in h_ertt use the same units as the TCP stack and therefore results in
meaningful comparisons and RTT estimate calculations.

Reported & tested by: Naeem Khademi (naeemk at ifi uio no)
Discussed with: bz
MFC after: 3 days

12 years agoMake 'junk' volatile so that compilers won't be tempted to optimize
kevlo [Fri, 17 Aug 2012 01:05:56 +0000 (01:05 +0000)]
Make 'junk' volatile so that compilers won't be tempted to optimize

Reviewed by: ache
MFC after: 3 days

12 years agoSupport for TCP DDP (Direct Data Placement) in the T4 TOE module.
np [Fri, 17 Aug 2012 00:49:29 +0000 (00:49 +0000)]
Support for TCP DDP (Direct Data Placement) in the T4 TOE module.

Basically, this is automatic rx zero copy when feasible.  TCP payload is
DMA'd directly into the userspace buffer described by the uio submitted
in soreceive by an application.

- Works with sockets that are being handled by the TCP offload engine
  of a T4 chip (you need t4_tom.ko module loaded after cxgbe, and an
  "ifconfig +toe" on the cxgbe interface).
- Does not require any modification to the application.
- Not enabled by default.  Use hw.t4nex.<X>.toe.ddp="1" to enable it.

12 years agoInitialize various DDP parameters in the main cxgbe(4) driver:
np [Thu, 16 Aug 2012 22:33:56 +0000 (22:33 +0000)]
Initialize various DDP parameters in the main cxgbe(4) driver:

- Setup multiple DDP page sizes.  When the driver attempts DDP it will
  try to combine physically contiguous pages into regions of these sizes.

- Set the indicate size such that the payload carried in the indicate can
  be copied in the header mbuf (and the 16K rx buffer can be recycled).

- Set DDP threshold to the max payload that the chip will coalesce and
  deliver to the driver (this is ~16K by default, which is also why the
  offload rx queue is backed by 16K buffers).  If the chip is able to
  coalesce up to the max it's allowed to, it's a good sign that the peer
  is transmitting in bulk without any TCP PSH.

MFC after: 2 weeks

12 years agoMerge ACPICA 20120816.
jkim [Thu, 16 Aug 2012 20:54:52 +0000 (20:54 +0000)]
Merge ACPICA 20120816.

12 years agoMake room for DDP page pods in the default configuration profile. While
np [Thu, 16 Aug 2012 20:30:14 +0000 (20:30 +0000)]
Make room for DDP page pods in the default configuration profile.  While
here, bump up the L2 table's size to 4K entries.

MFC after: 2 weeks

12 years agoAdd a routine (t4_set_tcb_field) to update arbitrary parts of a hardware
np [Thu, 16 Aug 2012 20:15:29 +0000 (20:15 +0000)]
Add a routine (t4_set_tcb_field) to update arbitrary parts of a hardware
TCB.  Filters are programmed by modifying the TCB too (via a different
routine) and the reply to any TCB update is delivered via a
CPL_SET_TCB_RPL.  Figure out whether the reply is for a filter-write or
something else and route it appropriately.

MFC after: 2 weeks

12 years ago- Typo fix
gonzo [Thu, 16 Aug 2012 19:22:34 +0000 (19:22 +0000)]
- Typo fix
- style(9) fix

Spotted by: kib@, Andrey Zonov

12 years agoAllow for a different handler for each type of firmware message.
np [Thu, 16 Aug 2012 18:31:50 +0000 (18:31 +0000)]
Allow for a different handler for each type of firmware message.

MFC after: 2 weeks

12 years agoRemove D_NEEDGIANT from dead_devsw. biofinish() (and thus dead_strategy)
jhb [Thu, 16 Aug 2012 18:04:33 +0000 (18:04 +0000)]
Remove D_NEEDGIANT from dead_devsw.  biofinish() (and thus dead_strategy)
does not need Giant.

MFC after: 1 month

12 years agoIts never a good idea to double free the same
rrs [Thu, 16 Aug 2012 17:55:16 +0000 (17:55 +0000)]
Its never a good idea to double free the same
address.

MFC after: 1 week (after the other commits ahead of this gets MFC'd)

12 years agoImport ACPICA 20120816.
jkim [Thu, 16 Aug 2012 17:26:22 +0000 (17:26 +0000)]
Import ACPICA 20120816.

12 years agoAdd locking for sscdisk(4) and mark it MPSAFE. Since this driver just
jhb [Thu, 16 Aug 2012 17:17:08 +0000 (17:17 +0000)]
Add locking for sscdisk(4) and mark it MPSAFE.  Since this driver just
makes calls out to the emulator, the locking is fairly simple.  A global
mutex protects the list of ssc disks, and each ssc disk has a mutex
to protect it's bioq.

Approved by: marcel

12 years agoOn lun disable, complete all INOTs and ATIOs with CAM_REQ_ABORTED.
mjacob [Thu, 16 Aug 2012 15:32:16 +0000 (15:32 +0000)]
On lun disable, complete all INOTs and ATIOs with CAM_REQ_ABORTED.

Reviewed by: ken (silently), chuck
MFC after: 3 weeks

12 years agoAs a safety measure, disable lowering pid_max too much.
kib [Thu, 16 Aug 2012 13:04:21 +0000 (13:04 +0000)]
As a safety measure, disable lowering pid_max too much.

Requested by: Peter Jeremy <peter@rulingia.com>
MFC after: 1 week

12 years agoFix grammar.
kib [Thu, 16 Aug 2012 13:01:56 +0000 (13:01 +0000)]
Fix grammar.

Submitted by: jh
MFC after: 1 week

12 years ago- When running out of swzone, instead of spewing an error message every
des [Thu, 16 Aug 2012 08:29:49 +0000 (08:29 +0000)]
- When running out of swzone, instead of spewing an error message every
  tick until the situation is resolved (if ever), just print a single
  message when running out and another when space becomes available.

- When adding more swap, warn if the total amount exceeds half the
  theoretical maximum we can handle.

12 years agoFix "speaker" volume control, broken at r230451.
mav [Thu, 16 Aug 2012 07:43:15 +0000 (07:43 +0000)]
Fix "speaker" volume control, broken at r230451.

Reported and tested by: Slawa Olhovchenkov <slw@zxy.spb.ru>
MFC after: 1 month

12 years agoAdd comment why the code has been disabled.
ae [Thu, 16 Aug 2012 06:45:58 +0000 (06:45 +0000)]
Add comment why the code has been disabled.

Requested by: rpaulo

12 years agoPreliminary Embest ATEB9200 support.
imp [Thu, 16 Aug 2012 05:03:59 +0000 (05:03 +0000)]
Preliminary Embest ATEB9200 support.

12 years agoHmmm, somehow this file was completely deleted, rather than just
imp [Thu, 16 Aug 2012 04:53:30 +0000 (04:53 +0000)]
Hmmm, somehow this file was completely deleted, rather than just
having the bogus lines being removed.

12 years agoRemove unused hints.
imp [Thu, 16 Aug 2012 04:49:23 +0000 (04:49 +0000)]
Remove unused hints.

12 years agoEliminate an unused parameter from init_pte_prot().
alc [Thu, 16 Aug 2012 04:41:15 +0000 (04:41 +0000)]
Eliminate an unused parameter from init_pte_prot().

Eliminate stray whitespace within init_pte_prot().

Eliminate a gratuitous variable initialization from pmap_enter().

12 years agoLimit popcorn limit to something sane (either 2ns or 2 ticks if that's
imp [Thu, 16 Aug 2012 02:35:44 +0000 (02:35 +0000)]
Limit popcorn limit to something sane (either 2ns or 2 ticks if that's
longer).

PR: 156481
Submitted by: Ian Lepore

12 years agoFix an incorrect comparison.
adrian [Thu, 16 Aug 2012 00:53:23 +0000 (00:53 +0000)]
Fix an incorrect comparison.

PR: kern/170098

12 years agoMerge somewhat modified r230399 from projects/armv6:
gonzo [Thu, 16 Aug 2012 00:51:50 +0000 (00:51 +0000)]
Merge somewhat modified r230399 from projects/armv6:

Add timeout to wait for network controllers to appear when netbooting.

USB ethernet adapter initialization usually is delayed and
they're not available immidiately after autoconfiguration. So we need
to wait a bit before giving up

Reviewed by: stas@

12 years agoReplace all uses of the vm page queues lock by a r/w lock that is private
alc [Wed, 15 Aug 2012 22:51:01 +0000 (22:51 +0000)]
Replace all uses of the vm page queues lock by a r/w lock that is private
to this pmap.

Tidy up the #include's.

Remove the (now) unused #define PMAP_SHPGPERPROC.  (This should have
been removed in r239236.)

Tested by: jchandra

12 years agoEliminate some unused declarations.
alc [Wed, 15 Aug 2012 22:25:57 +0000 (22:25 +0000)]
Eliminate some unused declarations.

12 years agoCorrect a KASSERT message.
alc [Wed, 15 Aug 2012 22:12:01 +0000 (22:12 +0000)]
Correct a KASSERT message.

Submitted by: bde

12 years agoVendor import of clang trunk r161861:
dim [Wed, 15 Aug 2012 20:02:54 +0000 (20:02 +0000)]
Vendor import of clang trunk r161861:
http://llvm.org/svn/llvm-project/cfe/trunk@161861

12 years agoDon't call the node iteration function inside the node table / node
adrian [Wed, 15 Aug 2012 20:01:28 +0000 (20:01 +0000)]
Don't call the node iteration function inside the node table / node
iterate lock.

This causes LORs and deadlocks as some code paths will have the com lock
held when calling ieee80211_iterate_nodes().

Here, the comlock isn't held during the node table and node iteration
locks; and the callback isn't called with any (extra) lock held.

PR: kern/170098
Submitted by: moonlightakkiy@yahoo.ca
MFC after: 4 weeks

12 years agoVendor import of llvm trunk r161861:
dim [Wed, 15 Aug 2012 19:34:23 +0000 (19:34 +0000)]
Vendor import of llvm trunk r161861:
http://llvm.org/svn/llvm-project/llvm/trunk@161861

12 years agoFix argument type for bus_space_map
gonzo [Wed, 15 Aug 2012 18:37:01 +0000 (18:37 +0000)]
Fix argument type for bus_space_map

12 years agoUnbreak build for the rest of AT91 platforms
gonzo [Wed, 15 Aug 2012 18:33:58 +0000 (18:33 +0000)]
Unbreak build for the rest of AT91 platforms

12 years agoSwitch lpc initarm() to use struct arm_boot_params and therefore fix
jceel [Wed, 15 Aug 2012 18:18:29 +0000 (18:18 +0000)]
Switch lpc initarm() to use struct arm_boot_params and therefore fix
EA3250 kernel build.

Approved by: gonzo

12 years agoSmall tweaks:
kib [Wed, 15 Aug 2012 18:00:52 +0000 (18:00 +0000)]
Small tweaks:
- fix macro argument
- wrap long line
- be more explicit about old pid_t type.

MFC after: 1 week

12 years agoMinor mdoc fixes and language tweaks.
joel [Wed, 15 Aug 2012 17:29:05 +0000 (17:29 +0000)]
Minor mdoc fixes and language tweaks.

12 years agoCustomer report of a panic on boot due to the old
jfv [Wed, 15 Aug 2012 17:12:40 +0000 (17:12 +0000)]
Customer report of a panic on boot due to the old
"m_getjcl:invalid cluster type" that occurred some
time back with the igb driver. This happens often when
booting over the net. I believe the NIC hardware is left
in a warm state when handed over to the driver, and a stray
RX interrupt happens earlier than the code is prepared for
it to happen. This change was verified to fix the problem,
its kind of a bandaid... but it is similar to what was done
in the igb code.

12 years agoStreamline use of cdevpriv and correct some corner cases.
hselasky [Wed, 15 Aug 2012 16:19:39 +0000 (16:19 +0000)]
Streamline use of cdevpriv and correct some corner cases.

1) It is not useful to call "devfs_clear_cdevpriv()" from
"d_close" callbacks, hence for example read, write, ioctl and
so on might be sleeping at the time of "d_close" being called
and then then freed private data can still be accessed.
Examples: dtrace, linux_compat, ksyms (all fixed by this patch)

2) In sys/dev/drm* there are some cases in which memory will
be freed twice, if open fails, first by code in the open
routine, secondly by the cdevpriv destructor. Move registration
of the cdevpriv to the end of the drm open routines.

3) devfs_clear_cdevpriv() is not called if the "d_open" callback
registered cdevpriv data and the "d_open" callback function
returned an error. Fix this.

Discussed with: phk
MFC after: 2 weeks

12 years agoAdd a short man page describing how to run a.out binaries on the
kib [Wed, 15 Aug 2012 16:01:45 +0000 (16:01 +0000)]
Add a short man page describing how to run a.out binaries on the
current kernels.

MFC after: 1 week

12 years agoAdd a sysctl kern.pid_max, which limits the maximum pid the system is
kib [Wed, 15 Aug 2012 15:56:21 +0000 (15:56 +0000)]
Add a sysctl kern.pid_max, which limits the maximum pid the system is
allowed to allocate, and corresponding tunable with the same
name. Note that existing processes with higher pids are left intact.

MFC after: 1 week

12 years agoFix build
kib [Wed, 15 Aug 2012 15:53:27 +0000 (15:53 +0000)]
Fix build

12 years agoRevert r239178 and implement two new functions, namely
hselasky [Wed, 15 Aug 2012 15:42:57 +0000 (15:42 +0000)]
Revert r239178 and implement two new functions, namely
"device_free_softc()" and "device_claim_softc()",
to allow USB serial drivers refcounting the softc.
These functions are used to grab the softc from
auto-free and to free the softc back to the correct
malloc type, respectivly.

Discussed with: jhb
MFC after: 2 weeks

12 years agoAdd new USB device quirk.
hselasky [Wed, 15 Aug 2012 15:35:20 +0000 (15:35 +0000)]
Add new USB device quirk.

Submitted by: Kra OTN
MFC after: 2 weeks