]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
13 years agocxgbe(4) - NIC driver for Chelsio T4 (Terminator 4) based 10Gb/1Gb adapters.
np [Fri, 18 Feb 2011 08:00:26 +0000 (08:00 +0000)]
cxgbe(4) - NIC driver for Chelsio T4 (Terminator 4) based 10Gb/1Gb adapters.

MFC after: 3 weeks

13 years ago- Make an enum for all the umidi USB transfers.
hselasky [Fri, 18 Feb 2011 07:37:30 +0000 (07:37 +0000)]
- Make an enum for all the umidi USB transfers.
- Use the USB stack's builtin clear-stall feature.
- Wrap some long lines.
- Use memcpy() instead of bcopy().
- Use memset() instead of bzero().
- Tested applications:
/usr/ports/audio/fluidsynth

MFC after: 7 days
Approved by: thompsa (mentor)

13 years agoRemove use case of DC_TX_ONE. It was used to limit queue just 1 TX
yongari [Fri, 18 Feb 2011 03:37:53 +0000 (03:37 +0000)]
Remove use case of DC_TX_ONE. It was used to limit queue just 1 TX
frame in DM910x controllers. In r67595(more than 10 years ago) it
was replaced to use "Store and Forward" mode and made controller
generate TX completion interrupt for every frame.

13 years agoApply special PHY initialization code for 21143 controllers before
yongari [Fri, 18 Feb 2011 02:58:09 +0000 (02:58 +0000)]
Apply special PHY initialization code for 21143 controllers before
any other media configuration. Otherwise some 21143 controller
cannot establish a link. While I'm here remove the PHY
initialization code in dc_setcfg(). Since dc_setcfg() is called
whenever link state is changed, having the PHY initialization code
in dc_setcfg() resulted in continuous link flips.
After driver resets SIA, use default SIA transmit/receive
configuration instead of disabling autosense/autonegotiation.
Otherwise, controller fails to establish a link as well as losing
auto-negotiation capability. For manual media configuration, always
configure 21143 controller with specified media to ensure media
change. This change makes ANA-6922 establish link with/without
auto-negotiation.

While I'm here be more strict on link UP/DOWN detection logic.

Many thanks to marius who fixed several bugs in initial patch and
even tested the patch on a couple of dc(4) controllers.

PR: kern/79262
Reviewed by: marius
Tested by: marius

13 years agoWhen driver have to use base softc due to lack of SROM on second
yongari [Fri, 18 Feb 2011 02:14:53 +0000 (02:14 +0000)]
When driver have to use base softc due to lack of SROM on second
port, copy SROM information from base softc as well and run SROM
parser again.  This change is necessary for some dual port
controllers to make dc(4) correctly detect PHY media based on first
port configuration table.
While I'm here add a check for validity of the base softc before
duplicating SROM contents from base softc. If driver failed to
attach to the first port it can access invalid area.

PR: kern/79262
Reviewed by: marius

13 years agoAlways check memory allocation failure. If driver encounter memory
yongari [Fri, 18 Feb 2011 01:56:25 +0000 (01:56 +0000)]
Always check memory allocation failure. If driver encounter memory
allocation error, do not attach driver.

Reviewed by: marius

13 years agoFix the R_ARM_ABS32 relocation implementation. The memory address
marcel [Thu, 17 Feb 2011 19:31:21 +0000 (19:31 +0000)]
Fix the R_ARM_ABS32 relocation implementation. The memory address
contains the addend that we need to include.

Obtained from: Juniper Networks.
Fixed by: Santhanakrishnan Balraj <sbalraj@juniper.net>

13 years agoJust be double-sure short-gi isn't being enabled in 20mhz mode.
adrian [Thu, 17 Feb 2011 17:35:09 +0000 (17:35 +0000)]
Just be double-sure short-gi isn't being enabled in 20mhz mode.

13 years agoDisable short-GI in 20mhz mode - the hardware doesn't support this.
adrian [Thu, 17 Feb 2011 17:31:36 +0000 (17:31 +0000)]
Disable short-GI in 20mhz mode - the hardware doesn't support this.

13 years agoSave a copy of errno before invoking syslog() if accept() or select() fail.
jhb [Thu, 17 Feb 2011 17:03:56 +0000 (17:03 +0000)]
Save a copy of errno before invoking syslog() if accept() or select() fail.
syslog() can trash the errno value causing nfsd to exit for non-fatal
errors like ECONNABORTED from accept().

MFC after: 1 week

13 years agoAdd "Extract Only" functionality to pc-sysinstall. This allows disk setup to
jpaetzel [Thu, 17 Feb 2011 16:33:41 +0000 (16:33 +0000)]
Add "Extract Only" functionality to pc-sysinstall.  This allows disk setup to
be done manually, pc-sysinstall is pointed to a mount-point for installation.

PR: bin/154685
Submitted by: kmoore
Approved by: kib (mentor)

13 years agoRemove pmap fields that are either unused or not fully implemented.
alc [Thu, 17 Feb 2011 15:36:29 +0000 (15:36 +0000)]
Remove pmap fields that are either unused or not fully implemented.

Discussed with: kib

13 years agoInstall iodev.h.
kib [Thu, 17 Feb 2011 15:10:13 +0000 (15:10 +0000)]
Install iodev.h.

Reviewed by: attilio
MFC after: 1 week

13 years agoFix the beginning day of the Great Lent, it starts on Monday for 7 weeks before Paskha.
osa [Thu, 17 Feb 2011 12:21:14 +0000 (12:21 +0000)]
Fix the beginning day of the Great Lent, it starts on Monday for 7 weeks before Paskha.
Fix the Exaltation of the Holy Cross day.
Add more Twelve holidays (dvunadesyatye prazdniki).

13 years agoImprove the check for ports which have gone missing, and just ignore
phk [Thu, 17 Feb 2011 08:54:22 +0000 (08:54 +0000)]
Improve the check for ports which have gone missing, and just ignore
them.  We want a run to perform as much work as possible before it
gives up.

13 years ago- Fix build of manual page and inclusion of mos driver into kernel config file.
hselasky [Thu, 17 Feb 2011 07:39:53 +0000 (07:39 +0000)]
- Fix build of manual page and inclusion of mos driver into kernel config file.
- Fix style compliancy by wrapping some long lines in if_mos.c

Approved by: thompsa (mentor)

13 years agoAdd in ANI parameters for the AR9280. These aren't enabled by default
adrian [Thu, 17 Feb 2011 05:56:03 +0000 (05:56 +0000)]
Add in ANI parameters for the AR9280. These aren't enabled by default
as they're likely not entirely correct, but they give people something
to toy with to compare behaviour/performance.

Disable the anti-noise part, as this apparently interferes with
RIFS. I haven't verified this.

13 years agoAdd a new parameter to selectively enable/disable the ANI operations.
adrian [Thu, 17 Feb 2011 05:52:53 +0000 (05:52 +0000)]
Add a new parameter to selectively enable/disable the ANI operations.

This was inspired by ath9k, which disables ANI anti-noise immunity
parameter tweaking (but leaves the rest of the ANI operations alone.)

13 years agoCall the right function.
adrian [Thu, 17 Feb 2011 05:30:38 +0000 (05:30 +0000)]
Call the right function.

13 years agoProperly propagate whether the channel is HT40 or not when calculating
adrian [Thu, 17 Feb 2011 05:16:59 +0000 (05:16 +0000)]
Properly propagate whether the channel is HT40 or not when calculating
packet duration for the ath_rate_sample module.

This doesn't affect the packet TX at all; only how much time the
sample rate module attributes to a completed TX.

13 years agoAdd initial support for RTL8401E PCIe Fast Ethernet.
yongari [Wed, 16 Feb 2011 21:59:42 +0000 (21:59 +0000)]
Add initial support for RTL8401E PCIe Fast Ethernet.

PR: 154789

13 years agoMfp4 CH=177274,177280,177284-177285,177297,177324-177325
bz [Wed, 16 Feb 2011 21:29:13 +0000 (21:29 +0000)]
Mfp4 CH=177274,177280,177284-177285,177297,177324-177325

  VNET socket push back:
  try to minimize the number of places where we have to switch vnets
  and narrow down the time we stay switched.  Add assertions to the
  socket code to catch possibly unset vnets as seen in r204147.

  While this reduces the number of vnet recursion in some places like
  NFS, POSIX local sockets and some netgraph, .. recursions are
  impossible to fix.

  The current expectations are documented at the beginning of
  uipc_socket.c along with the other information there.

  Sponsored by: The FreeBSD Foundation
  Sponsored by: CK Software GmbH
  Reviewed by:  jhb
  Tested by:    zec

Tested by: Mikolaj Golub (to.my.trociny gmail.com)
MFC after: 2 weeks

13 years agoRemove in-addr.arpa from the list of zones it is possible to slave locally
dougb [Wed, 16 Feb 2011 21:23:09 +0000 (21:23 +0000)]
Remove in-addr.arpa from the list of zones it is possible to slave locally

13 years agoOnly use -z noexecstack for these explicit arches: i386, amd64, powerpc
dim [Wed, 16 Feb 2011 21:04:47 +0000 (21:04 +0000)]
Only use -z noexecstack for these explicit arches: i386, amd64, powerpc
and powerpc64.  Other arches currently do not support this feature.

Reminded by: kib

13 years agoFix the last binary in the base system that still has an executable
dim [Wed, 16 Feb 2011 20:46:57 +0000 (20:46 +0000)]
Fix the last binary in the base system that still has an executable
stack, /usr/sbin/uathload.

Since this program links in a .o file containing a firmware blob, and
there is no clean way to add a .note.GNU-stack section to this .o file,
we simply use the -z noexecstack option to ld here.

13 years agoModify the spi flash driver to allow smaller read IO sizes, but enforce
adrian [Wed, 16 Feb 2011 20:07:44 +0000 (20:07 +0000)]
Modify the spi flash driver to allow smaller read IO sizes, but enforce
the larger, aligned write+erase sizes the driver currently implements.

This preserves write behaviour but makes the flash driver usable for things
like a read-only FFS or a geom_uzip/geom_compress .

Note that since GEOM will now return the sector size as being smaller,
writes of sector size/alignment will now fail with an EIO. Code which
writes to the flash device will have to be (for now) manually taught
about the flash write blocksize.

13 years agoRemove reading of symbols from a.out loaded files. Since we are tight
imp [Wed, 16 Feb 2011 18:05:10 +0000 (18:05 +0000)]
Remove reading of symbols from a.out loaded files.  Since we are tight
on space for clang and a.out support is only needed for /boot/loader,
they are excess bytes that serve no useful purpose other than to
support really old kernels (FreeBSD < 3.2 or so).  Prefer clang
support over support for these old kernels and remove this code.  We
gain about 100 bytes of space this way.

Reviewed by: rdivacky@

13 years agoTo avoid excessive code duplication create wrapper for fill regs
dchagin [Wed, 16 Feb 2011 17:50:21 +0000 (17:50 +0000)]
To avoid excessive code duplication create wrapper for fill regs
from stack frame. Change the trap() code to use newly created function
instead of explicit regs assignment.

13 years agoBump dummynet module version to meet dummynet schedulers' requirements,
pluknet [Wed, 16 Feb 2011 15:43:35 +0000 (15:43 +0000)]
Bump dummynet module version to meet dummynet schedulers' requirements,
and thus unbreak loading dummynet.ko via /boot/loader.conf.

Reported by: rihad <rihad att mail.ru> on freebsd-net
Approved by: kib (mentor)

13 years agoMFi386: revision 218713
nyan [Wed, 16 Feb 2011 15:20:37 +0000 (15:20 +0000)]
MFi386: revision 218713

  Apply a few small optimizations to boot2's code, to make it shrink a
  little further.

13 years agoFix build breakage in if_mos.c when USB_DEBUG option is set.
hselasky [Wed, 16 Feb 2011 09:26:56 +0000 (09:26 +0000)]
Fix build breakage in if_mos.c when USB_DEBUG option is set.

Approved by: thompsa (mentor)

13 years agoAdd support for new USB to ethernet controller:
hselasky [Wed, 16 Feb 2011 08:33:30 +0000 (08:33 +0000)]
Add support for new USB to ethernet controller:
Moschip MCS7730/MCS7830

Submitted by: Rick van der Zwet <info@rickvanderzwet.nl>
Approved by: thompsa (mentor)

13 years agoAdd the -j option to enable soft updates journaling when creating
mckusick [Wed, 16 Feb 2011 06:00:27 +0000 (06:00 +0000)]
Add the -j option to enable soft updates journaling when creating
a new file system.

Reviewed by: Kostik Belousov <kostikbel@gmail.com>

13 years agoWhen building libzpool on ia64 or sparc64, don't add the .note.GNU-stack
dim [Tue, 15 Feb 2011 22:28:15 +0000 (22:28 +0000)]
When building libzpool on ia64 or sparc64, don't add the .note.GNU-stack
section.

Submitted by: kib

13 years agotest: Note that this is used both as a normal program and a shell builtin.
jilles [Tue, 15 Feb 2011 22:17:47 +0000 (22:17 +0000)]
test: Note that this is used both as a normal program and a shell builtin.

MFC after: 1 week

13 years agoFix some leftover binaries and shared libraries in the system that still
dim [Tue, 15 Feb 2011 22:03:09 +0000 (22:03 +0000)]
Fix some leftover binaries and shared libraries in the system that still
have an executable stack, due to linking in hand-assembled .S or .s
files, that have no .GNU-stack sections:

RWX --- ---  /lib/libcrypto.so.6
RWX --- ---  /lib/libmd.so.5
RWX --- ---  /lib/libz.so.6
RWX --- ---  /lib/libzpool.so.2
RWX --- ---  /usr/lib/liblzma.so.5

These were found using scanelf, from the sysutils/pax-utils port.

Reviewed by: kib

13 years agoFor realtime signals fill the sigval value.
dchagin [Tue, 15 Feb 2011 21:46:36 +0000 (21:46 +0000)]
For realtime signals fill the sigval value.

13 years agoMake a linux_rt_sigtimedwait() system call is actually working.
dchagin [Tue, 15 Feb 2011 21:42:48 +0000 (21:42 +0000)]
Make a linux_rt_sigtimedwait() system call is actually working.

1) Translate the native signal number in the appropriate Linux signal.
2) Remove bogus code, which can lead to a panic as it calls
   kern_sigtimedwait with same ksiginfo.
3) Return the corresponding signal number.

13 years agoStyle(9) fix. Wrap long lines in linux_rt_sigtimedwait().
dchagin [Tue, 15 Feb 2011 21:24:50 +0000 (21:24 +0000)]
Style(9) fix. Wrap long lines in linux_rt_sigtimedwait().

13 years agoPut the macro declaration in the relevant include file for future use.
dchagin [Tue, 15 Feb 2011 21:22:09 +0000 (21:22 +0000)]
Put the macro declaration in the relevant include file for future use.

13 years agoIn sys/boot/common/ufsread.c, use uint8_t instead of u_int8_t.
dim [Tue, 15 Feb 2011 21:01:13 +0000 (21:01 +0000)]
In sys/boot/common/ufsread.c, use uint8_t instead of u_int8_t.

Submitted by: mdf

13 years agoApply a few small optimizations to boot2's code, to make it shrink a
dim [Tue, 15 Feb 2011 20:18:52 +0000 (20:18 +0000)]
Apply a few small optimizations to boot2's code, to make it shrink a
little further.  This gets us further on the way to be able to build it
successfully with clang.  Using in-tree gcc, this shrinks boot2.bin with
60 bytes, the in-tree clang shaves off 72 bytes, and ToT clang 84 bytes.

Submitted by: rdivacky
Reviewed by: imp

13 years agoDon't forget to bump man page date for r218711.
ed [Tue, 15 Feb 2011 20:07:35 +0000 (20:07 +0000)]
Don't forget to bump man page date for r218711.

13 years agoRemove dead reference to regexp(3). Use regex(3) instead.
ed [Tue, 15 Feb 2011 20:04:13 +0000 (20:04 +0000)]
Remove dead reference to regexp(3). Use regex(3) instead.

PR: docs/149950
Submitted by: arundel@

13 years agoFix a regression introduced in r215906. The change made in r215906
yongari [Tue, 15 Feb 2011 18:16:04 +0000 (18:16 +0000)]
Fix a regression introduced in r215906. The change made in r215906
caused link re-negotiation whenever application joins or leaves a
multicast group.  If driver is running, it would have established a
link so there is no need to start re-negotiation. The re-negotiation
broke established link which in turn stopped multicast application
working while re-negotiation is in progress.

PR: kern/154667
MFC after: 1 week

13 years agoDisable flipping antennas for AR9280.
adrian [Tue, 15 Feb 2011 13:29:52 +0000 (13:29 +0000)]
Disable flipping antennas for AR9280.

Flipping antennas when doing 11n would cause all kinds of strange issues.
Just don't do it for now and when it comes time to do it, don't do it here.

13 years agoProperly check for errors from waitpid().
jhb [Tue, 15 Feb 2011 12:42:18 +0000 (12:42 +0000)]
Properly check for errors from waitpid().

Submitted by: gcooper
MFC after: 1 week

13 years agoSince r218070 reenabled the call to vm_map_simplify_entry() from
kib [Tue, 15 Feb 2011 09:03:58 +0000 (09:03 +0000)]
Since r218070 reenabled the call to vm_map_simplify_entry() from
vm_map_insert(), the kmem_back() assumption about newly inserted
entry might be broken due to interference of two factors. In the low
memory condition, when vm_page_alloc() returns NULL, supplied map is
unlocked. If another thread performs kmem_malloc() meantime, and its
map entry is placed right next to our thread map entry in the map,
both entries wire count is still 0 and entries are coalesced due to
vm_map_simplify_entry().

Mark new entry with MAP_ENTRY_IN_TRANSITION to prevent coalesce.
Fix some style issues, tighten the assertions to account for
MAP_ENTRY_IN_TRANSITION state.

Reported and tested by: pho
Reviewed by: alc

13 years agoNote that a UFS filesystem can have up to 20 active snapshots.
keramida [Tue, 15 Feb 2011 07:25:54 +0000 (07:25 +0000)]
Note that a UFS filesystem can have up to 20 active snapshots.

PR: docs/151104
Submitted by: Aldis Berjoza <aldis@bsdroot.lv>
MFC after: 3 days

13 years agoDocument TRYBROKEN in ports(7).
keramida [Tue, 15 Feb 2011 07:16:27 +0000 (07:16 +0000)]
Document TRYBROKEN in ports(7).

PR: docs/153542
Submitted by: Eitan Adler <lists@eitanadler.com>
MFC after: 3 days

13 years ago- Add Dell Wireless 1390 BCM4311 Mini PCI to supported list.
miwi [Tue, 15 Feb 2011 05:57:53 +0000 (05:57 +0000)]
- Add Dell Wireless 1390 BCM4311 Mini PCI to supported list.

PR: 154500
Submitted by: Ruslan Mahmatkhanov <cvs-src at yandex.ru>
Approved by: rwatson

13 years ago- Use the correct DMA tag/map pair for synchronize the FC scratch area.
marius [Mon, 14 Feb 2011 21:50:51 +0000 (21:50 +0000)]
- Use the correct DMA tag/map pair for synchronize the FC scratch area.
- Allocate coherent DMA memory for the request/response queue area and
  and the FC scratch area.

These changes allow isp(4) to work properly on sparc64 with usage of the
IOMMU streaming buffers enabled.

Approved by: mjacob
MFC after: 2 weeks

13 years agobring this in line with what ath9k does.
adrian [Mon, 14 Feb 2011 21:35:11 +0000 (21:35 +0000)]
bring this in line with what ath9k does.

13 years agoSome statistics additions - prepare for error codes > 32 (since the AR5416
adrian [Mon, 14 Feb 2011 21:24:54 +0000 (21:24 +0000)]
Some statistics additions - prepare for error codes > 32 (since the AR5416
error mask is > 5 bits) and add some extra CRC/HT40/ShortGI counters to
help debug 802.11n issues.

13 years agoMfp4 CH=177256:
bz [Mon, 14 Feb 2011 20:49:37 +0000 (20:49 +0000)]
Mfp4 CH=177256:

  Catch a set vnet upon return to user space. This usually
  means return paths with CURVNET_RESTORE() missing.

  If VNET_DEBUG is turned on we can even tell the function
  that did the CURVNET_SET() which is really helpful; else
  we print "N/A".

  Sponsored by: The FreeBSD Foundation
  Sponsored by: CK Software GmbH
  Reviewed by:  jhb

MFC after: 11 days

13 years agoRework r218685. Copy just enough data for the resource type.
jkim [Mon, 14 Feb 2011 20:05:37 +0000 (20:05 +0000)]
Rework r218685.  Copy just enough data for the resource type.

Reviewed by: jhb, mdf

13 years agoStyle(9) fix. Do not initialize variables in the declarations.
dchagin [Mon, 14 Feb 2011 17:24:58 +0000 (17:24 +0000)]
Style(9) fix. Do not initialize variables in the declarations.

13 years agoPrevent reading from the ACPI_RESOURCE past its actual end. For
mdf [Mon, 14 Feb 2011 17:20:20 +0000 (17:20 +0000)]
Prevent reading from the ACPI_RESOURCE past its actual end.  For
paranoia limit to the size of the ACPI_RESOURCE as well.

Reviewd by: jhb (in spirit)
MFC after: 1 week

13 years agoFurther simplify tmpfs_reg_resize(). Also, update its comments, including
alc [Mon, 14 Feb 2011 15:36:38 +0000 (15:36 +0000)]
Further simplify tmpfs_reg_resize().  Also, update its comments, including
style fixes.

13 years agoAdd a check to make sure the provider name is "mdN" before printing the
brucec [Mon, 14 Feb 2011 09:58:47 +0000 (09:58 +0000)]
Add a check to make sure the provider name is "mdN" before printing the
unit number.

Suggested by:  jh
MFC after: 3 days

13 years ago* Fix page fault caused by referring freed node.
hselasky [Mon, 14 Feb 2011 08:14:06 +0000 (08:14 +0000)]
* Fix page fault caused by referring freed node.

While updating Tx stats, already freed node could be referred and cause
page fault. To avoid such panic, spool Tx stats in driver's softc. Then,
on every ratectl interval, grab node though ieee80211_iterate_nodes() and
update ratectl stats.

* Simplify some code in run_iter_func().

* Fix typo

* Use memset instead of bzero (hselasky @)

PR: kern/153938
Submitted by: PseudoCylon <moonlightakkiy@yahoo.ca>
Approved by: thompsa (mentor)

13 years agoCorrect a subtle bug in the 'gsched_rr' disk scheduler.
luigi [Mon, 14 Feb 2011 08:09:02 +0000 (08:09 +0000)]
Correct a subtle bug in the 'gsched_rr' disk scheduler.
The algorithm is supposed to work as follows:
in order to prevent starvation, when a new client starts being served we
record the start time and reset the counter of bytes served.
We then switch to a new client after a certain amount of time or bytes,
even if the current one still has pending requests.
To avoid charging a new client the time of the first seek,
we start counting time when the first request is served.

Unfortunately a bug in the previous version of the code failed
to set the start time in certain cases, resulting in some processes
exceeding their timeslice.

The fix (in this patch) is trivial, though it took a while to find
out and replicate the bug.
Thanks to Tommaso Caprai for investigating and fixing the problem.

Submitted by: Tommaso Caprai
MFC after: 1 week

13 years agoRetire mp_fixme(). It's no longer used.
alc [Mon, 14 Feb 2011 02:37:27 +0000 (02:37 +0000)]
Retire mp_fixme().  It's no longer used.

13 years agoAdd back soon-to-be-release FreeBSD 7.4 which got clobbered in the previous
uqs [Sun, 13 Feb 2011 22:17:49 +0000 (22:17 +0000)]
Add back soon-to-be-release FreeBSD 7.4 which got clobbered in the previous
merges.

13 years agoLock the vnode around clearing of VV_TEXT flag. Remove mp_fixme() note
kib [Sun, 13 Feb 2011 21:52:26 +0000 (21:52 +0000)]
Lock the vnode around clearing of VV_TEXT flag. Remove mp_fixme() note
mentioning that vnode lock is needed.

Reviewed by: alc
Tested by: pho
MFC after: 1 week

13 years agoSort include files in the alphabetical order.
dchagin [Sun, 13 Feb 2011 20:07:48 +0000 (20:07 +0000)]
Sort include files in the alphabetical order.

13 years agoOops, wasn't supposed to commit this.
cognet [Sun, 13 Feb 2011 20:04:29 +0000 (20:04 +0000)]
Oops, wasn't supposed to commit this.

13 years agoCall init_param1() much earlier, so that msgbufsize is non-zero when we want
cognet [Sun, 13 Feb 2011 20:02:46 +0000 (20:02 +0000)]
Call init_param1() much earlier, so that msgbufsize is non-zero when we want
to map and use the msgbuf.

13 years agoUse the preload_fetch_addr() and preload_fetch_size() convenience
marcel [Sun, 13 Feb 2011 19:46:55 +0000 (19:46 +0000)]
Use the preload_fetch_addr() and preload_fetch_size() convenience
functions to obtain the address and size of the preloaded pool
configuration file/repository.

Sponsored by: Juniper Networks.

13 years agoUse the preload_fetch_addr() and preload_fetch_size() convenience
marcel [Sun, 13 Feb 2011 19:34:48 +0000 (19:34 +0000)]
Use the preload_fetch_addr() and preload_fetch_size() convenience
functions to obtain the address and size of the preloaded key files.

Sponsored by: Juniper Networks.

13 years agoUse the preload_fetch_addr() and preload_fetch_size() convenience
marcel [Sun, 13 Feb 2011 19:26:51 +0000 (19:26 +0000)]
Use the preload_fetch_addr() and preload_fetch_size() convenience
functions to obtain the address and size of the PCI vendor data.

Sponsored by: Juniper Networks.

13 years agoUse the preload_fetch_addr() and preload_fetch_size() convenience
marcel [Sun, 13 Feb 2011 19:25:48 +0000 (19:25 +0000)]
Use the preload_fetch_addr() and preload_fetch_size() convenience
functions to obtain the address and size of the bitmap splash image.

Sponsored by: Juniper Networks.

13 years agoUse the preload_fetch_addr() and preload_fetch_size() convenience
marcel [Sun, 13 Feb 2011 19:24:04 +0000 (19:24 +0000)]
Use the preload_fetch_addr() and preload_fetch_size() convenience
functions to obtain the address and size of the overriding DSDT.

Sponsored by: Juniper Networks.

13 years agoSort include files in the alphabetical order.
dchagin [Sun, 13 Feb 2011 19:07:48 +0000 (19:07 +0000)]
Sort include files in the alphabetical order.

13 years agoRemove comment about 'ftlk' LOR.
dchagin [Sun, 13 Feb 2011 18:46:34 +0000 (18:46 +0000)]
Remove comment about 'ftlk' LOR.

13 years agoStop printing the LOR, as this is expected behavior.
dchagin [Sun, 13 Feb 2011 18:41:40 +0000 (18:41 +0000)]
Stop printing the LOR, as this is expected behavior.

13 years agoOnly print the unit number when invoked with the -n flag.
brucec [Sun, 13 Feb 2011 18:30:17 +0000 (18:30 +0000)]
Only print the unit number when invoked with the -n flag.

PR: bin/144300
Submitted by: arundel
MFC after: 3 days

13 years agoMove potentially offensive fortune to fortunes-o.
brucec [Sun, 13 Feb 2011 18:18:56 +0000 (18:18 +0000)]
Move potentially offensive fortune to fortunes-o.

PR: bin/137702
MFC after: 3 days

13 years agoThe bitset field of freshly created futex should be initialized explicity.
dchagin [Sun, 13 Feb 2011 17:56:22 +0000 (17:56 +0000)]
The bitset field of freshly created futex should be initialized explicity.
Otherwise, REQUEUE operations fails.

13 years agoDocument some more sysconf(3) variables.
brucec [Sun, 13 Feb 2011 17:43:56 +0000 (17:43 +0000)]
Document some more sysconf(3) variables.

MFC after: 1 month

13 years agoThis should be TX stream, not RX stream.
adrian [Sun, 13 Feb 2011 15:14:13 +0000 (15:14 +0000)]
This should be TX stream, not RX stream.

13 years agoFix a bug reported by Jonathan Leighton in his web-sctp testing
rrs [Sun, 13 Feb 2011 14:48:11 +0000 (14:48 +0000)]
Fix a bug reported by Jonathan Leighton in his web-sctp testing
at the Univ-of-Del. Basically when a 1-to-1 socket did a
socket/bind/send(data)/close. If the timing was right
we would dereference a socket that is NULL.

MFC after: 1 month

13 years agoEliminate tn_reg.tn_aobj_pages. Instead, correctly maintain the vm
alc [Sun, 13 Feb 2011 14:46:39 +0000 (14:46 +0000)]
Eliminate tn_reg.tn_aobj_pages.  Instead, correctly maintain the vm
object's size field.  Previously, that field was always zero, even
when the object tn_reg.tn_aobj contained numerous pages.

Apply style fixes to tmpfs_reg_resize().

In collaboration with: kib

13 years agoFix several bugs related to stream scheduling.
tuexen [Sun, 13 Feb 2011 13:53:28 +0000 (13:53 +0000)]
Fix several bugs related to stream scheduling.

Obtained from: Robin Seggelmann
MFC after: 3 months.

13 years ago* add in new EEPROM fields from later revisions
adrian [Sun, 13 Feb 2011 13:11:00 +0000 (13:11 +0000)]
* add in new EEPROM fields from later revisions
* add in printing futureBase

13 years ago- Added following port committers:
miwi [Sun, 13 Feb 2011 06:19:38 +0000 (06:19 +0000)]
- Added following port committers:

Marcus von Appen (mva@) (mentor: beech, miwi)
Dennis Herrmann (dhn@) (mentor: tabthorpe, miwi)
Brendan Fabeny (bf@)  (mentor: makc, miwi)
Kris Moore (kmoore@)  (mentor: brooks, miwi)
Sylvio Cesar Teixeira <sylvio@> (mentor: itetcu, miwi)
Lars Engels (lme) (mentor: miwi)

13 years agoOops, revert an accidental local change that got added in
deischen [Sun, 13 Feb 2011 04:44:06 +0000 (04:44 +0000)]
Oops, revert an accidental local change that got added in
my last commit (r218627).  No damage was done in the last
commit, just some duplicated code was added (which is now
removed).

13 years agoAllow the SO_SETFIB socket option to select the default (0)
deischen [Sun, 13 Feb 2011 00:14:13 +0000 (00:14 +0000)]
Allow the SO_SETFIB socket option to select the default (0)
routing table.

Reviewed by: julian

13 years agosh: Detect dividing the smallest integer by -1.
jilles [Sat, 12 Feb 2011 23:44:05 +0000 (23:44 +0000)]
sh: Detect dividing the smallest integer by -1.

This overflows and on some architectures such as amd64 it generates SIGFPE.
Generate an error on all architectures.

13 years agoFix Incorrectly formatted ClientHello SSL/TLS handshake messages could
simon [Sat, 12 Feb 2011 21:30:46 +0000 (21:30 +0000)]
Fix Incorrectly formatted ClientHello SSL/TLS handshake messages could
cause OpenSSL to parse past the end of the message.

Note: Applications are only affected if they act as a server and call
SSL_CTX_set_tlsext_status_cb on the server's SSL_CTX. This includes
Apache httpd >= 2.3.3, if configured with "SSLUseStapling On".

Security: http://www.openssl.org/news/secadv_20110208.txt
Security: CVE-2011-0014
Obtained from: OpenSSL CVS

13 years agoAdd missing USB_HOST_ALIGN option.
hselasky [Sat, 12 Feb 2011 21:17:38 +0000 (21:17 +0000)]
Add missing USB_HOST_ALIGN option.

Approved by: thompsa (mentor)

13 years agoRename used_requeue and use it as bitwise field to store more flags.
dchagin [Sat, 12 Feb 2011 20:58:59 +0000 (20:58 +0000)]
Rename used_requeue and use it as bitwise field to store more flags.
Reimplement used_requeue logic with LINUX_XDEPR_REQUEUEOP flag.

13 years agoIf the pf.conf(5) example file is copied when setting up a firewall it's
brucec [Sat, 12 Feb 2011 20:42:53 +0000 (20:42 +0000)]
If the pf.conf(5) example file is copied when setting up a firewall it's
easy to forget about icmp. Update the file to show allowing icmp through
the firewall.

PR: docs/144986
MFC after: 1 month

13 years agoFix typos.
brucec [Sat, 12 Feb 2011 20:28:15 +0000 (20:28 +0000)]
Fix typos.

PR: docs/131625
Submitted by: Andrew Wright <andrew at qemg.org>
MFC after: 1 month

13 years agoSlightly rewrite linux_fork:
dchagin [Sat, 12 Feb 2011 20:16:25 +0000 (20:16 +0000)]
Slightly rewrite linux_fork:

1) Remove bogus error checking.
2) A new process exit from kernel through fork_trampoline(),
   so remove bogus check.

13 years agoRemove bogus include <machine/frame.h>
dchagin [Sat, 12 Feb 2011 19:14:57 +0000 (19:14 +0000)]
Remove bogus include <machine/frame.h>

13 years agoMove linux_clone(), linux_fork(), linux_vfork() to a MI path.
dchagin [Sat, 12 Feb 2011 18:17:12 +0000 (18:17 +0000)]
Move linux_clone(), linux_fork(), linux_vfork() to a MI path.

13 years agoIn preparation for moving linux_clone() to a MI path
dchagin [Sat, 12 Feb 2011 16:33:00 +0000 (16:33 +0000)]
In preparation for moving linux_clone() to a MI path
introduce linux_set_upcall_kse().

13 years agoIn preparation for moving linux_clone () to a MI path
dchagin [Sat, 12 Feb 2011 15:50:21 +0000 (15:50 +0000)]
In preparation for moving linux_clone () to a MI path
move the TLS code in a separate function.

Use function parameter instead of direct using register.

13 years agoRegen for r218610.
dchagin [Sat, 12 Feb 2011 15:36:25 +0000 (15:36 +0000)]
Regen for r218610.