]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agofix bug in previous commit related to rule deletion
luigi [Wed, 31 Mar 2010 02:20:22 +0000 (02:20 +0000)]
fix bug in previous commit related to rule deletion
(stable/8 just fixed moments ago)

14 years agoBe like stdlib.h and bring in a wchar_t definition and use it to prototype
jmallett [Wed, 31 Mar 2010 02:12:24 +0000 (02:12 +0000)]
Be like stdlib.h and bring in a wchar_t definition and use it to prototype
wcstoimax and wcstoumax, rather than spelling it __wchar_t.  This is necessary
to use these functions in C++ where wchar_t is different to __wchar_t and is
a built-in type.

It may be better to use __wchar_t here and to simply define __wchar_t as being
wchar_t in C++ mode rather than to bring in wchar_t, but this is less invasive
and follows our existing practice, and restores wchar_t usage in this file to
what it was before r1.8.

14 years agoMultiple fixes related to queue set sizing and resources:
np [Wed, 31 Mar 2010 00:27:49 +0000 (00:27 +0000)]
Multiple fixes related to queue set sizing and resources:

- Only the tunnelq (TXQ_ETH) requires a buf_ring, an ifq, and the watchdog/timer
  callouts.  Do not allocate these for the other tx queues.

- Use 16k jumbo clusters only on offload capable cards by default.

- Do not allocate a full tx ring for the offload queue if the card is not
  offload capable.

- Slightly better freelist size calculation.

- Fix nmbjumbo4 typo, remove unneeded global variables.

MFC after: 3 days

14 years agoFix signed/unsigned mix-up that allowed txq->in_use to grow beyond txq->size.
np [Wed, 31 Mar 2010 00:26:35 +0000 (00:26 +0000)]
Fix signed/unsigned mix-up that allowed txq->in_use to grow beyond txq->size.

14 years agoFix tx drop statistics.
np [Wed, 31 Mar 2010 00:26:02 +0000 (00:26 +0000)]
Fix tx drop statistics.

MFC after: 3 days

14 years agoFix build with "nooptions INET"
np [Wed, 31 Mar 2010 00:24:44 +0000 (00:24 +0000)]
Fix build with "nooptions INET"

Requested by: bz
MFC after: 3 days

14 years agoDo not attempt to retrieve interrupt information before it is available.
np [Wed, 31 Mar 2010 00:22:58 +0000 (00:22 +0000)]
Do not attempt to retrieve interrupt information before it is available.

MFC after: 3 days

14 years agoImproved PHY EDC settings.
np [Wed, 31 Mar 2010 00:21:56 +0000 (00:21 +0000)]
Improved PHY EDC settings.

MFC after: 3 days

14 years agoRefresh the firmware version immediately after it is upgraded (or downgraded).
np [Wed, 31 Mar 2010 00:19:39 +0000 (00:19 +0000)]
Refresh the firmware version immediately after it is upgraded (or downgraded).

MFC after: 3 days

14 years agoMerge OpenBSD revisions 1.4 through 1.9, mostly style cleanups.
delphij [Tue, 30 Mar 2010 23:14:35 +0000 (23:14 +0000)]
Merge OpenBSD revisions 1.4 through 1.9, mostly style cleanups.

Obtained from: OpenBSD
MFC after: 3 month

14 years agoThis patch should fix handling of byte range locks locally
rmacklem [Tue, 30 Mar 2010 23:11:50 +0000 (23:11 +0000)]
This patch should fix handling of byte range locks locally
on the server for the experimental nfs server. When enabled
by setting vfs.newnfs.locallocks_enable to non-zero, the
experimental nfs server will now acquire byte range locks
on the file on behalf of NFSv4 clients, such that lock
conflicts between the NFSv4 clients and processes running
locally on the server, will be recognized and handled correctly.

MFC after: 2 weeks

14 years agoRemove unused files.
delphij [Tue, 30 Mar 2010 23:10:56 +0000 (23:10 +0000)]
Remove unused files.

14 years agoSync with OpenBSD:
delphij [Tue, 30 Mar 2010 21:54:25 +0000 (21:54 +0000)]
Sync with OpenBSD:
 - avoid coredump when there's only one token on a line;
 - Use calloc();
 - Remove a line inherited from example mdoc.

Obtained from: OpenBSD
MFC after: 1 month

14 years agoFor where possible, make the dates of the Tasmanian holidays properly
edwin [Tue, 30 Mar 2010 21:11:16 +0000 (21:11 +0000)]
For where possible, make the dates of the Tasmanian holidays properly
variable. The leftover ones are the Thursday before the Fourth
Saturday of October (Oct/SatFourth-2) which is not possible yet.

14 years agonit: xpt_bus_deregister has to be called with the sim lock held.
mjacob [Tue, 30 Mar 2010 20:39:57 +0000 (20:39 +0000)]
nit: xpt_bus_deregister has to be called with the sim lock held.

MFC after: 1 month

14 years agofix my clobber of the copyright date :)
jfv [Tue, 30 Mar 2010 19:54:29 +0000 (19:54 +0000)]
fix my clobber of the copyright date :)

14 years agoUse CACHE_LINE_SIZE alignment for 'struct pcpu' rather than hardcoding 128.
jhb [Tue, 30 Mar 2010 19:41:38 +0000 (19:41 +0000)]
Use CACHE_LINE_SIZE alignment for 'struct pcpu' rather than hardcoding 128.

Reviewed by: jeff

14 years agoThanks to Michael Tuexen for adding SCTP support for 82599,
jfv [Tue, 30 Mar 2010 19:09:18 +0000 (19:09 +0000)]
Thanks to Michael Tuexen for adding SCTP support for 82599,
also for finding a one character bug that kept TSO from working.

Sometimes with direct attach cables a failure can occur in init,
the old method of calling detach was broken, there is no way to
return an error to the system from init, so I have changed it to
return failure thru the ioctl.

And, have fixed the ALTQ code changes of Max Laier, sorry Max :)

14 years agoUse panic() (which the environment is required to provide to libstand) to
jhb [Tue, 30 Mar 2010 19:07:41 +0000 (19:07 +0000)]
Use panic() (which the environment is required to provide to libstand) to
implement assert() instead of relying on a non-required exit().  The exit()
invocation also did not match the semantics of the exit() routine that
current boot environments happen to require.

PR: kern/144749
Discussed with: bde
MFC after: 1 week

14 years agoDefer freeing a kevent list until after dropping kqueue locks.
jhb [Tue, 30 Mar 2010 18:31:55 +0000 (18:31 +0000)]
Defer freeing a kevent list until after dropping kqueue locks.

LOR: 185
Submitted by: Matthew Fleming @ Isilon
MFC after: 1 week

14 years agoVarious and sundry style, whitespace, and comment fixes.
jhb [Tue, 30 Mar 2010 18:29:19 +0000 (18:29 +0000)]
Various and sundry style, whitespace, and comment fixes.

Submitted by: bde (mostly)

14 years agoFix lint build problem.
jfv [Tue, 30 Mar 2010 17:05:24 +0000 (17:05 +0000)]
Fix lint build problem.

14 years ago- Handle calloc() allocation failures.
ru [Tue, 30 Mar 2010 13:46:40 +0000 (13:46 +0000)]
- Handle calloc() allocation failures.
- Fixed a comment.
- 2 -> EXIT_FAILURE in some places.
- errx() -> err() where appropriate.

PR: 144644
Submitted by: Garrett Cooper

14 years agoo Make sockstat -6 output more readable for long ipv6
maxim [Tue, 30 Mar 2010 09:58:21 +0000 (09:58 +0000)]
o Make sockstat -6 output more readable for long ipv6
addresses (most of them apart from ::1): put a whitespace
between local and remote address:port pairs.

PR: bin/145194
Submitted by: Fedor Dikarev
MFC after: 2 weeks

14 years agoo Remove a duplicate reference to -f flag in the SYNOPSYS section.
maxim [Tue, 30 Mar 2010 09:22:33 +0000 (09:22 +0000)]
o Remove a duplicate reference to -f flag in the SYNOPSYS section.
o Improve wording.

PR: docs/144407
Submitted by: gcooper
MFC after: 1 week

14 years agoOn request of joel@:
edwin [Tue, 30 Mar 2010 06:42:01 +0000 (06:42 +0000)]
On request of joel@:

- Remove the 3rd clause of the UC Berkeley copyrighted files.
- For the files added copyrighted by me, move the "All rights
  reserved" to the next line.

14 years agoUpdate to igb and em:
jfv [Mon, 29 Mar 2010 23:36:34 +0000 (23:36 +0000)]
Update to igb and em:

em revision 7.0.0:
- Using driver devclass, seperate legacy (pre-pcie) code
  into a seperate source file. This will at least help
  protect against regression issues. It compiles along
  with em, and is transparent to end use, devices in each
  appear to be 'emX'. When using em in a modular form this
  also allows the legacy stuff to be defined out.
- Add tx and rx rings as in igb, in the 82574 this becomes
  actual multiqueue for the first time (2 queues) while in
  other PCIE adapters its just make code cleaner.
- Add RX mbuf handling logic that matches igb, this will
  eliminate packet drops due to temporary mbuf shortage.

igb revision 1.9.3:
- Following the ixgbe code, use a new approach in what
  was called 'get_buf', the routine now has been made
  independent of rxeof, it now does the update to the
  engine TDT register, this design allows temporary
  mbuf resources to become non-critical, not requiring
  a packet to be discarded, instead it just returns and
  does not increment the tail pointer.
- With the above change it was also unnecessary to keep
  'spare' maps around, since we do not have the discard
  issue.
- Performance tweaks and improvements to the code also.

MFC in a week

14 years agoRefine r204265. We want the standard VGA palette for packed pixel mode.
jkim [Mon, 29 Mar 2010 22:41:30 +0000 (22:41 +0000)]
Refine r204265.  We want the standard VGA palette for packed pixel mode.

14 years agoBuild modules for this config to make sure they stay buildable...
imp [Mon, 29 Mar 2010 22:29:41 +0000 (22:29 +0000)]
Build modules for this config to make sure they stay buildable...

14 years agoEnable module builds now that the build completes for them. This
imp [Mon, 29 Mar 2010 22:03:55 +0000 (22:03 +0000)]
Enable module builds now that the build completes for them.  This
should get them into the universe rotation.

14 years agoUpdate the Australian calendar WRT DST.
edwin [Mon, 29 Mar 2010 21:17:03 +0000 (21:17 +0000)]
Update the Australian calendar WRT DST.
Layout changes.

14 years agog_vfs_open: correctly set devvp.v_bufobj.bo_bsize to DEV_BSIZE
avg [Mon, 29 Mar 2010 20:34:25 +0000 (20:34 +0000)]
g_vfs_open: correctly set devvp.v_bufobj.bo_bsize to DEV_BSIZE

Because of how breadn -> bufstrategy -> g_vfs_strategy are currently
implemented, bread on devvp always expects DEV_BSIZE block size.
Thus, devvp bo_bsize must always be DEV_BSIZE irrespective of media
properties or filesystem implementation details.

Reviewed by: mckusick
MFC after: 2 weeks

14 years agoStart copyright notices with /*-
joel [Mon, 29 Mar 2010 20:27:17 +0000 (20:27 +0000)]
Start copyright notices with /*-

14 years agoCheck the pointer to JIT binary filter before its de-allocation.
jkim [Mon, 29 Mar 2010 20:24:03 +0000 (20:24 +0000)]
Check the pointer to JIT binary filter before its de-allocation.

Submitted by: Alexander Sack (asack at niksun dot com)
MFC after: 3 days

14 years agoPrint memory model of the video mode except for planar memory model.
jkim [Mon, 29 Mar 2010 20:12:44 +0000 (20:12 +0000)]
Print memory model of the video mode except for planar memory model.
'P', 'D', 'C', 'H', 'V' mean packed pixel, direct color, CGA, Hercules,
VGA X memory models respectively where they have fixed number of planes.

14 years agocalendar.freebsd: add my entry
avg [Mon, 29 Mar 2010 20:04:01 +0000 (20:04 +0000)]
calendar.freebsd: add my entry

MFC after: 4 days

14 years agoAdd myself.
trasz [Mon, 29 Mar 2010 19:52:36 +0000 (19:52 +0000)]
Add myself.

MFC after: 2 weeks

14 years agoArm defines intr_disable() as a macro, which causes problems for cxgb.
imp [Mon, 29 Mar 2010 19:36:37 +0000 (19:36 +0000)]
Arm defines intr_disable() as a macro, which causes problems for cxgb.
Rather than remove it for all architectures (which was a botch in
r205845), remove it only for arm and mips until that can be sorted out.

Pointy Hat: imp

Submitted by: Navdeep Parhar
MFC after: 7 days

14 years agoAdd a handler for the local APIC error interrupt. For now it just prints
jhb [Mon, 29 Mar 2010 19:13:34 +0000 (19:13 +0000)]
Add a handler for the local APIC error interrupt.  For now it just prints
out the current value of the local APIC error register when the interrupt
fires.

MFC after: 1 week

14 years agoCosmetic tweak to use a type suffix instead of a cast to force a constant
jhb [Mon, 29 Mar 2010 18:47:04 +0000 (18:47 +0000)]
Cosmetic tweak to use a type suffix instead of a cast to force a constant
to be a long.

14 years agoChange how multipath labels are created and managed. This makes it easier
mjacob [Mon, 29 Mar 2010 18:04:06 +0000 (18:04 +0000)]
Change how multipath labels are created and managed. This makes it easier
to support various storage boxes which really aren't active-active.

We only write the label on the *first* provider. For all other providers
we just "add" the disk. This also allows for an "add" verb.

A usage implication is that you should specificy the currently active
storage path as the first provider.

Note that this does not add RDAC-like functionality, but better allows for
autovolumefailover configurations (additional checkins elsewhere will support
this).

Sponsored by: Panasas
MFC after: 1 month

14 years agoFix references.
trasz [Mon, 29 Mar 2010 17:39:38 +0000 (17:39 +0000)]
Fix references.

14 years agoMake all the modules build on arm (modulo ipfilter, which I'm looking
imp [Mon, 29 Mar 2010 17:31:55 +0000 (17:31 +0000)]
Make all the modules build on arm (modulo ipfilter, which I'm looking
into):

o Don't build vpo and syscons on mips or arm either
o Add a section for mips and arm breakages, and document why.
  This is easier than opting in on all the other architectures:
  # no BUS_SPACE_UNSPECIFIED: bce, bwi, bwn, mfi, mpt, siba_bwn
  # No barrier instruction support (specific to this driver): sym
  # no uart_cpu_$MACHINE_ARCH: uart

(mips has, by inspection, the same issues as arm, so ditto for it)

MFC after: 7 days

14 years agoCast the bus_size_t to a intmax_t rather than assuming type-punning to
imp [Mon, 29 Mar 2010 17:26:24 +0000 (17:26 +0000)]
Cast the bus_size_t to a intmax_t rather than assuming type-punning to
a size_t.  Switch from %z to %j.

14 years agoCast bus_space_t to a intmax_t and use %j to print it instead of
imp [Mon, 29 Mar 2010 17:25:06 +0000 (17:25 +0000)]
Cast bus_space_t to a intmax_t and use %j to print it instead of
assuming it is size_t sized and using %z.

14 years agoCast a bus_size_t to a (size_t) to use %zd specifier. Not quite the
imp [Mon, 29 Mar 2010 17:24:23 +0000 (17:24 +0000)]
Cast a bus_size_t to a (size_t) to use %zd specifier.  Not quite the
right thing to do, but it is in compat code I don't want to sort out
at the moment.

14 years agoAdd AR9285.
rpaulo [Mon, 29 Mar 2010 17:09:04 +0000 (17:09 +0000)]
Add AR9285.

Sponsored by: iXsystems, inc

14 years agoMark the vtoc.h structure as packed so that it is the right size and
imp [Mon, 29 Mar 2010 15:59:44 +0000 (15:59 +0000)]
Mark the vtoc.h structure as packed so that it is the right size and
layout on arm.

MFC after: 7 days

14 years agosize_t is not equivalent to int on 64-bit platforms.
des [Mon, 29 Mar 2010 13:20:52 +0000 (13:20 +0000)]
size_t is not equivalent to int on 64-bit platforms.

Pointy hat to: edwin@

14 years agoremove a leftover debugging message
luigi [Mon, 29 Mar 2010 12:27:49 +0000 (12:27 +0000)]
remove a leftover debugging message

14 years agoFix handling of set manipulations.
luigi [Mon, 29 Mar 2010 12:19:23 +0000 (12:19 +0000)]
Fix handling of set manipulations.
This patch has two fixes for potential kernel panics (one wrong
index, one access to the wrong lock) and two fixes to wrong logic
in a conditional. The potential panics are also on stable/8,
so I am going to MFC the fix quickly.

14 years agoUse new syntax for the dutch calendar: Use local month names instead
edwin [Mon, 29 Mar 2010 07:09:27 +0000 (07:09 +0000)]
Use new syntax for the dutch calendar: Use local month names instead
of english ones.

14 years agoBump WARNS to 7
edwin [Mon, 29 Mar 2010 07:08:47 +0000 (07:08 +0000)]
Bump WARNS to 7

14 years agoLong awaited update to the calendar system:
edwin [Mon, 29 Mar 2010 06:49:20 +0000 (06:49 +0000)]
Long awaited update to the calendar system:

- Repeating events which span multiple years (because of -A, -B or
  just the three days before the end of the year).

- Support for lunar events (full moon, new moon) and solar events
  (equinox and solstice, chinese new year). Because of this, the
  options -U (UTC offset) and -l (longitude) are available to
  compensate if reality doesn't match the calculated values.

MFC after: 1 month

14 years agoWait for pmc name in the log before displaying data.
fabient [Sun, 28 Mar 2010 22:46:56 +0000 (22:46 +0000)]
Wait for pmc name in the log before displaying data.
This will solve an abort in case of low throughput PMCs.

MFC after: 3days

14 years agoDo not sync cache for the PL2506
thompsa [Sun, 28 Mar 2010 20:09:04 +0000 (20:09 +0000)]
Do not sync cache for the PL2506

PR: usb/144915
Submitted by: Monty Hall

14 years agoDo not swap Apple keys when detecting Apple-FN keyboards.
thompsa [Sun, 28 Mar 2010 20:07:08 +0000 (20:07 +0000)]
Do not swap Apple keys when detecting Apple-FN keyboards.

Reported by: Steven Noonan
Submitted by: Hans Petter Selasky

14 years agoMake sure the bsd_urb_list gets initialised and that new URB's are queued at
thompsa [Sun, 28 Mar 2010 20:04:03 +0000 (20:04 +0000)]
Make sure the bsd_urb_list gets initialised and that new URB's are queued at
the end of the list.

Submitted by: Hans Petter Selasky

14 years agoAdd PCI IDs for two more nForce controllers.
thompsa [Sun, 28 Mar 2010 20:02:50 +0000 (20:02 +0000)]
Add PCI IDs for two more nForce controllers.

Submitted by: Dmitry Luhtionov @ gmail.com

14 years agoAdd a couple of usb product IDs.
thompsa [Sun, 28 Mar 2010 20:02:13 +0000 (20:02 +0000)]
Add a couple of usb product IDs.

Submitted by: Dmitry Luhtionov @ gmail.com

14 years agoMake acl_to_text_np(3) not crash on long group or user names in NFSv4 ACLs.
trasz [Sun, 28 Mar 2010 17:29:15 +0000 (17:29 +0000)]
Make acl_to_text_np(3) not crash on long group or user names in NFSv4 ACLs.

PR: amd64/145091
MFC after: 2 weeks

14 years agoSet hw.ofwfb.relax_mmap=1 by default. While these checks may be a good
nwhitehorn [Sun, 28 Mar 2010 17:17:32 +0000 (17:17 +0000)]
Set hw.ofwfb.relax_mmap=1 by default. While these checks may be a good
idea in principle, X does not work without them on basically any hardware,
and this is probably the most frequent problem people run into on PowerPC.

Prodded by: rnoland
MFC after: 1 week

14 years agoDon't forget to bump the date in the man page.
ed [Sun, 28 Mar 2010 13:40:13 +0000 (13:40 +0000)]
Don't forget to bump the date in the man page.

14 years agoChange all our own code to use st_*tim instead of st_*timespec.
ed [Sun, 28 Mar 2010 13:16:08 +0000 (13:16 +0000)]
Change all our own code to use st_*tim instead of st_*timespec.

Also remove some local patches to diff(1) which are now unneeded.

14 years agoRename st_*timespec fields to st_*tim for POSIX 2008 compliance.
ed [Sun, 28 Mar 2010 13:13:22 +0000 (13:13 +0000)]
Rename st_*timespec fields to st_*tim for POSIX 2008 compliance.

A nice thing about POSIX 2008 is that it finally standardizes a way to
obtain file access/modification/change times in sub-second precision,
namely using struct timespec, which we already have for a very long
time. Unfortunately POSIX uses different names.

This commit adds compatibility macros, so existing code should still
build properly. Also change all source code in the kernel to work
without any of the compatibility macros. This makes it all a less
ambiguous.

I am also renaming st_birthtime to st_birthtim, even though it was a
local extension anyway. It seems Cygwin also has a st_birthtim.

14 years agoAdd a comment explaining the previous commit.
rpaulo [Sun, 28 Mar 2010 12:55:31 +0000 (12:55 +0000)]
Add a comment explaining the previous commit.

Submitted by: sam
> Description of fields to fill in above:                     76 columns --|
> PR:            If a GNATS PR is affected by the change.
> Submitted by:  If someone else sent in the change.
> Reviewed by:   If someone else reviewed your modification.
> Approved by:   If you needed approval for this commit.
> Obtained from: If the change is from a third party.
> MFC after:     N [day[s]|week[s]|month[s]].  Request a reminder email.
> Security:      Vulnerability reference (one per line) or description.
> Empty fields above will be automatically removed.

M    ieee80211_hostap.c

14 years agoWhen tearing down IPsec as part of a (virtual) network stack,
bz [Sun, 28 Mar 2010 06:51:50 +0000 (06:51 +0000)]
When tearing down IPsec as part of a (virtual) network stack,
do not try to free the same list twice but free both the
acquiring list and the security policy acquiring list.

Reviewed by: anchie
MFC after: 3 days

14 years agoUse standard types in preference to BSD types so that these header files
gibbs [Sun, 28 Mar 2010 00:10:09 +0000 (00:10 +0000)]
Use standard types in preference to BSD types so that these header files
can be used in applications compiled with only POSIX types visible.

MFC after: 2 weeks

14 years agoCorrectly handle preemption of pmap_update_pde_invalidate().
alc [Sat, 27 Mar 2010 23:53:47 +0000 (23:53 +0000)]
Correctly handle preemption of pmap_update_pde_invalidate().

X-MFC after: r205573

14 years agoSimplify pmap_growkernel(), making the i386 version more like the amd64
alc [Sat, 27 Mar 2010 18:24:27 +0000 (18:24 +0000)]
Simplify pmap_growkernel(), making the i386 version more like the amd64
version.

MFC after: 3 weeks

14 years agoDon't hold connection lock when doing reconnects as it makes I/Os wait for
pjd [Sat, 27 Mar 2010 16:35:07 +0000 (16:35 +0000)]
Don't hold connection lock when doing reconnects as it makes I/Os wait for
connection timeouts.

Reported by: Kevin Day <toasty@dragondata.com>

14 years agoFix lock leakage.
mav [Sat, 27 Mar 2010 15:39:19 +0000 (15:39 +0000)]
Fix lock leakage.

PR: kern/145081

14 years ago(S)LIST_HEAD_INITIALIZER takes a (S)LIST_HEAD as an argument.
antoine [Sat, 27 Mar 2010 13:43:18 +0000 (13:43 +0000)]
(S)LIST_HEAD_INITIALIZER takes a (S)LIST_HEAD as an argument.
Fix some wrong usages.
Note: this does not affect generated binaries as this argument is not used.

Approved by: harti@

14 years agoMerge improvements from kernel HID parser to the userland usbhid(3)
kaiw [Sat, 27 Mar 2010 08:00:16 +0000 (08:00 +0000)]
Merge improvements from kernel HID parser to the userland usbhid(3)
parser.  This merge does not change any API and should not break any
native or thirdparty applications.

Changes include:

* Merge multiple report ID support and other improvements from kernel
  HID parser.
* Ignore rid argument in hid_start_parser, parse all the report items since
  we now support multiple report ID.
* Skip report ID byte in hid_get_data() and set report ID byte in
  hid_set_data(), if report ID is non-zero.
* Reimplement hid_get_report_id: instead get report id from uhid device
  (which is always 0), try parsing the report descriptor and return the
  first report ID encountered.

Reviewed by: hps
Silent on: -usb mailing list

14 years agoBring up-to-date:
marcel [Sat, 27 Mar 2010 06:53:11 +0000 (06:53 +0000)]
Bring up-to-date:
o   Switch to ITANIUM2 has the cpu. This has absolutely no effect
    on the code, but makes for a better example.
o   Drop COMPAT_FREEBSD6. We're tier 2, so you're supposed to run
    8-stable or newer.
o   Add PREEMPTION. It works now.
o   Remove HWPMC_HOOKS. We don't have support for hwpmc yet.

o   Add a bunch of new devices: atapist, hptiop, amr, ips, twa, igb,
    ixgbe, ae, age, alc, ale, bce, bfe, et, jme, msk, nge, sk, ste,
    stge, tx, vge, axe, rue, udav, fwip, and all USB serial.
o   Remove "legacy" devices: le, vx, dc, pcn, rl, sis.

Make sure to the module list is a superset of what goes into GENERIC.

14 years agoImplement interrupt to CPU binding. Assign interrupts to CPUs in a
marcel [Sat, 27 Mar 2010 05:40:50 +0000 (05:40 +0000)]
Implement interrupt to CPU binding. Assign interrupts to CPUs in a
round-robin fashion, starting with the highest priority interrupt
on the highest-numbered CPU and cycling downwards.

14 years ago<machine/nexusvar.h> removed on ia64.
marcel [Sat, 27 Mar 2010 03:17:39 +0000 (03:17 +0000)]
<machine/nexusvar.h> removed on ia64.

14 years agoRemove nx_pcibus from the nexus resource. Nexus is not involved
marcel [Sat, 27 Mar 2010 03:15:34 +0000 (03:15 +0000)]
Remove nx_pcibus from the nexus resource. Nexus is not involved
with PCI busses. Remove nexus_read_ivar() and nexus_write_ivar()
to give default behaviour. Remove <machine/nexusvar.h> as well,
because there's nothing in it that's being used.

14 years agoUpdate the driver to Intel version 2.1.6
jfv [Sat, 27 Mar 2010 00:21:40 +0000 (00:21 +0000)]
Update the driver to Intel version 2.1.6
- add some new hardware support for 82599
- Big change to interrupt architecture, it now
  uses a queue which contains an RX/TX pair as
  the recipient of the interrupt. This will reduce
  overall system interrupts/msix usage.
- Improved RX mbuf handling: the old get_buf routine
  is no longer synchronized with rxeof, this allows
  the elimination of packet discards due to mbuf
  allocation failure.
- Much simplified and improved AIM code, it now
  happens in the queue interrupt context and takes
  into account both the traffic on the RX AND TX
  side.
- variety of small tweaks, like ring size, that have
  been seen as performance improvements.
- Thanks to those that provided feedback or suggested
  changes, I hope I've caught all of them.

14 years agoRename disable_intr() to ia64_disable_intr() and rename enable_intr()
marcel [Fri, 26 Mar 2010 21:22:02 +0000 (21:22 +0000)]
Rename disable_intr() to ia64_disable_intr() and rename enable_intr()
to ia64_enable_intr(). This reduces confusion with intr_disable() and
intr_restore().

Have configure_final() call ia64_finalize_intr() instead of enable_intr()
in preparation of adding support for binding interrupts to all CPUs.

14 years agoD'oh- isp_handle_index' logic was reversed (not used in FreeBSD).
mjacob [Fri, 26 Mar 2010 20:22:18 +0000 (20:22 +0000)]
D'oh- isp_handle_index' logic was reversed (not used in FreeBSD).

MFC after: 1 week

14 years agoAllow building a cross-kgdb for ia64.
marcel [Fri, 26 Mar 2010 19:41:40 +0000 (19:41 +0000)]
Allow building a cross-kgdb for ia64.

14 years agoHandle cross-builds for gdbserver.
marcel [Fri, 26 Mar 2010 19:40:53 +0000 (19:40 +0000)]
Handle cross-builds for gdbserver.

14 years agoPass the correct pointer to fled_cb().
rpaulo [Fri, 26 Mar 2010 18:49:43 +0000 (18:49 +0000)]
Pass the correct pointer to fled_cb().

14 years agoRemove GNU cpio after fix of CVE-2010-0624.
delphij [Fri, 26 Mar 2010 17:02:32 +0000 (17:02 +0000)]
Remove GNU cpio after fix of CVE-2010-0624.

Note that this is actually a no-op for most users, as this GNU
cpio was broken on -HEAD and 8-STABLE since last March until
the recent fix.

FreeBSD 8.0+ uses BSD cpio by default and the code is being
actively maintained.

Blessed by: kientzle
With hat: secteam
MFC after: 3 days

14 years agoClean up some printing stuff so that we can have a bit finer control
mjacob [Fri, 26 Mar 2010 15:13:31 +0000 (15:13 +0000)]
Clean up some printing stuff so that we can have a bit finer control
on debug output. Add a new platform function requirement to allow
for printing based upon the ITL nexus instead of the isp unit plus
channel, target and lun. This allows some printouts and error messages
from the core code to appear in the same format as the platform's
subsystem (in FreeBSD's case, CAM path).

MFC after: 1 week

14 years agoRevert r205683 to resolve some code quality issues which do not affect the
netchild [Fri, 26 Mar 2010 14:36:16 +0000 (14:36 +0000)]
Revert r205683 to resolve some code quality issues which do not affect the
build or use of linprocfs, before committing the reworked patch again.

Requested by: des

14 years agoHandling SIGPIPE will cause deadlock/crash.
fabient [Fri, 26 Mar 2010 14:35:48 +0000 (14:35 +0000)]
Handling SIGPIPE will cause deadlock/crash.
Return an error immediatly in case of hard shutdown.

MFC after: 3days

14 years agoDo not overflow the term in the case of multi-line display.
fabient [Fri, 26 Mar 2010 14:29:57 +0000 (14:29 +0000)]
Do not overflow the term in the case of multi-line display.

MFC after: 3days

14 years agoTighten my grip on pseudofs, procfs, linprocfs.
des [Fri, 26 Mar 2010 12:59:15 +0000 (12:59 +0000)]
Tighten my grip on pseudofs, procfs, linprocfs.

14 years agoFix some bogus values in linprocfs.
netchild [Fri, 26 Mar 2010 11:43:15 +0000 (11:43 +0000)]
Fix some bogus values in linprocfs.

Submitted by: Petr Salinger <Petr.Salinger@seznam.cz>
Verified on: GNU/kFreeBSD debian 8.0-1-686 (by submitter)
PR: 144584

14 years agoSupport only LOOKUP operation for "/" in relookup() because lookup()
jh [Fri, 26 Mar 2010 11:33:12 +0000 (11:33 +0000)]
Support only LOOKUP operation for "/" in relookup() because lookup()
can't succeed for CREATE, DELETE and RENAME.

Discussed with: bde

14 years ago- add some usb devices (scanner, printer, usb storage)
netchild [Fri, 26 Mar 2010 11:02:01 +0000 (11:02 +0000)]
- add some usb devices (scanner, printer, usb storage)
- add quirks for the usb storage

Reviewed by: hselasky

14 years agoUse last 16 bytes of serial number in metadata instead of first ones,
mav [Fri, 26 Mar 2010 10:18:19 +0000 (10:18 +0000)]
Use last 16 bytes of serial number in metadata instead of first ones,
same as Intel MatrixRAID does.

PR: kern/124064

14 years agoFix a typo in a comment.
netchild [Fri, 26 Mar 2010 08:47:07 +0000 (08:47 +0000)]
Fix a typo in a comment.

14 years agoFix some problems which may lead to a panic:
netchild [Fri, 26 Mar 2010 08:42:11 +0000 (08:42 +0000)]
Fix some problems which may lead to a panic:
 - right order of src and dst in memcpy
 - NULL out the clips after freeing to prevent an accident

Noticed by: hselasky

14 years agoReplace sb_store64()/sb_load64() with mips3_sd()/mips3_ld() respectively.
neel [Fri, 26 Mar 2010 07:15:27 +0000 (07:15 +0000)]
Replace sb_store64()/sb_load64() with mips3_sd()/mips3_ld() respectively.

Obtained from NetBSD.

Suggested by: jmallett@

14 years agoo Fix typo.
maxim [Fri, 26 Mar 2010 05:35:25 +0000 (05:35 +0000)]
o Fix typo.

PR: docs/145031
Submitted by: olgeni
MFC after: 1 week

14 years agoo Trim leading w/space.
maxim [Fri, 26 Mar 2010 05:34:13 +0000 (05:34 +0000)]
o Trim leading w/space.

PR: docs/145030
Submitted by: olgeni
MFC after: 1 week

14 years agoOnly use the interval timer for clock interrupts on the BSP and
marcel [Fri, 26 Mar 2010 02:29:15 +0000 (02:29 +0000)]
Only use the interval timer for clock interrupts on the BSP and
have the BSP use IPIs to trigger clock interrupts on the APs.
This allows us to run on hardware configurations for which the
ITC has non-uniform frequencies across CPUs.

While here, change the clock XIV to type IPI so as to protect
the interrupt delivery against CPU re-balancing once that's
implemented.