]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
16 years agoMFC rev 1.14 axe.4
remko [Sun, 11 Nov 2007 11:40:03 +0000 (11:40 +0000)]
MFC rev 1.14 axe.4

  Specify which Dlink chip is supported (D-Link DUB-E100 ver A) because
  others (D-Link DUB-E100 ver B1) are not.

  PR: docs/109664
  Submitted by: Martin Nilsson<martin at mullet dot se>
  Patch by: asmodai

16 years agoMFC: match.c: 1.21
krion [Sat, 10 Nov 2007 22:04:31 +0000 (22:04 +0000)]
MFC: match.c: 1.21

  date: 2007/11/02 20:18:47;  author: krion;  state: Exp;  lines: +57 -11
  Memoize the results of isinstalledpkg()

  PR:             bin/116452
  Submitted by:   wollmann

16 years agoMFC the following revisions:
csjp [Sat, 10 Nov 2007 17:08:41 +0000 (17:08 +0000)]
MFC the following revisions:

1.14      +1 -0      src/sys/security/audit/audit.h
1.15      +22 -0     src/sys/security/audit/audit_arg.c
1.19      +35 -2     src/sys/security/audit/audit_bsm.c
1.14      +2 -2      src/sys/security/audit/audit_bsm_token.c
1.21      +7 -2      src/sys/security/audit/audit_syscalls.c (partial)

- Add support for auditing arguments in setaudit_addr(2)
- Fix bug which which would result in in_addr_ex tokens not being
  parsed properly
- Calculate the size of an ip6 address correctly

16 years agoMFC the following revisions:
csjp [Sat, 10 Nov 2007 15:32:03 +0000 (15:32 +0000)]
MFC the following revisions:

1.25      +1 -1      src/sys/security/audit/audit.c
1.10      +1 -1      src/sys/security/audit/audit.h
1.11      +2 -2      src/sys/security/audit/audit_arg.c
1.16      +34 -10    src/sys/security/audit/audit_bsm.c
1.10      +12 -12    src/sys/security/audit/audit_bsm_token.c
1.13      +2 -0      src/sys/security/audit/audit_private.h
1.13      +52 -7     src/sys/security/audit/audit_syscalls.c
1.475     +1 -1      src/sys/sys/proc.h

This brings extended subject state support into RELENG_6 for audit
records.  This means that applications like sshd can take advantage
of extended subject for IPv6.  It should be noted that sshd has not
been modified.

16 years agoMFC: 1.7
marius [Sat, 10 Nov 2007 12:34:28 +0000 (12:34 +0000)]
MFC: 1.7

In _pthread_key_create() ensure that libthr is initialized. This
fixes a NULL-dereference of curthread when libstdc+ initializes
the exception handling globals on archs we can't use GNU TLS due
to lack of support in binutils 2.15 (i.e. arm and sparc64), yet,
thus making threaded C++ programs compiled with GCC 4.2.1 work
on these archs.

16 years agoMFC rev 1.29 phys_pager.c
remko [Sat, 10 Nov 2007 11:23:01 +0000 (11:23 +0000)]
MFC rev 1.29 phys_pager.c

  Correct a copy and paste'o in phys_pager.c, we are talking about phys here
  and not about devices.

  PR:           93755
  Approved by:  imp (mentor, implicit when re-assigning the ticket to me).

This includes another fix that is not part of CURRENT and RELENG_7 because
that is not part of the code there anymore.

Approved by:    re (kensmith), imp (mentor)

16 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Sat, 10 Nov 2007 10:21:30 +0000 (10:21 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

16 years agoMFC: Sync the -HEAD state of ata(4) back to RELENG_6, except interrupt filtering,
delphij [Fri, 9 Nov 2007 09:58:08 +0000 (09:58 +0000)]
MFC: Sync the -HEAD state of ata(4) back to RELENG_6, except interrupt filtering,
CAM MPSAFE, BIO_FLUSH and alpha support removal, which are not suitable for
RELENG_6.

This makes RELENG_6 to support the same set of ATA devices as RELENG_7 and -CURRENT
did.

Reviewed by: sos

16 years agoMFC: Don't look for GPT primary and secondary tables on a disk unless we
jhb [Fri, 9 Nov 2007 02:48:03 +0000 (02:48 +0000)]
MFC: Don't look for GPT primary and secondary tables on a disk unless we
have a valid PMBR.

16 years agoMFC changes to X11 distributions based on xorg-7.3 layout.
kensmith [Thu, 8 Nov 2007 23:32:26 +0000 (23:32 +0000)]
MFC changes to X11 distributions based on xorg-7.3 layout.
Versions:
        dist.c v1.250
        dist.h v1.68
        menus.c v1.426

Approved by:    re (bmah)

16 years agoMFC: Update copyright attribution.
jhb [Thu, 8 Nov 2007 21:33:49 +0000 (21:33 +0000)]
MFC: Update copyright attribution.

16 years agoMFC: Use the smaller cgbase() macro in ufsread.c if UFS_SMALL_CGBASE is
jhb [Thu, 8 Nov 2007 21:31:53 +0000 (21:31 +0000)]
MFC: Use the smaller cgbase() macro in ufsread.c if UFS_SMALL_CGBASE is
defined.

16 years agoMFC: improve HPET table probe and give it a lower priority as a timer,
njl [Thu, 8 Nov 2007 21:22:20 +0000 (21:22 +0000)]
MFC: improve HPET table probe and give it a lower priority as a timer,
fix divide by zero if battery status invalid

Approved by: re

16 years agoMFC: Split agp_generic_detach() up into two routines: agp_free_cdev() and
jhb [Thu, 8 Nov 2007 20:50:13 +0000 (20:50 +0000)]
MFC: Split agp_generic_detach() up into two routines: agp_free_cdev() and
agp_free_res().

16 years agoMFC: Don't always re-route IRQs on resume.
jhb [Thu, 8 Nov 2007 20:24:53 +0000 (20:24 +0000)]
MFC: Don't always re-route IRQs on resume.

16 years agoMFC: Rework the read/write support in the bios disk driver some to cut down
jhb [Thu, 8 Nov 2007 20:13:04 +0000 (20:13 +0000)]
MFC: Rework the read/write support in the bios disk driver some to cut down
on duplicated code and support 64-bit LBAs for GPT.  This also changes the
code to always use EDD on devices that support it which matches the rest of
the x86 bootstrap.

Approved by: re (kensmith)

16 years agoMFC: Don't test the APIC flag in the cpuid features for amd64 to see if a
jhb [Thu, 8 Nov 2007 20:09:45 +0000 (20:09 +0000)]
MFC: Don't test the APIC flag in the cpuid features for amd64 to see if a
local APIC is present or not.

16 years agoMFC
alc [Thu, 8 Nov 2007 18:02:27 +0000 (18:02 +0000)]
MFC
  Add comments explaining why all stores updating a non-kernel page table
  must be globally performed before calling any of the TLB invalidation
  functions.

  With one exception, on amd64, this requirement was already met.  Fix this
  one case.  Also, as a clarification, change an existing atomic op into a
  release.

16 years agoMerge audit_trigger.c:1.5 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 16:40:01 +0000 (16:40 +0000)]
Merge audit_trigger.c:1.5 from HEAD to RELENG_6:

  Close a very narrow race that might cause a trigger allocation to be
  leaked if a trigger is delivered as the trigger device is closed.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit.c:1.29, audit.h:1.12, audit_arg.c:1.12, audit_bsm.c:1.17,
rwatson [Thu, 8 Nov 2007 16:38:49 +0000 (16:38 +0000)]
Merge audit.c:1.29, audit.h:1.12, audit_arg.c:1.12, audit_bsm.c:1.17,
audit_bsm_klib.c:1.7, audit_pipe.c:1.11, audit_private.h:1.16,
audit_worker.c:1.16 from HEAD to RELENG_6:

  Clean up audit comments--formatting, spelling, etc.

16 years agoMerge audit.c:1.28, audit_private.h:1.15, audit_worker.c:1.15 from HEAD to
rwatson [Thu, 8 Nov 2007 16:32:29 +0000 (16:32 +0000)]
Merge audit.c:1.28, audit_private.h:1.15, audit_worker.c:1.15 from HEAD to
RELENG_6:

  Remove AUDIT_PRINTF() debugging statements and definition; clean up or
  remove associated comments.

  Slip audit_file_rotate_wait assignment in audit_rotate_vnode() before
  the drop of the global audit mutex.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit.h:1.11, audit_syscalls.c:1.16 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 16:26:06 +0000 (16:26 +0000)]
Merge audit.h:1.11, audit_syscalls.c:1.16 from HEAD to RELENG_6:

  Remove unused !AUDIT audit_proc_*() prototypes: unlike in Mac OS X, we
  don't define or use these functions if AUDIT isn't configured.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit_syscalls.c:1.15 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 16:23:03 +0000 (16:23 +0000)]
Merge audit_syscalls.c:1.15 from HEAD to RELENG_6:

  Consistent white space after .'s in comments.

16 years agoMerge audit.c:1.26 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:49:00 +0000 (15:49 +0000)]
Merge audit.c:1.26 from HEAD to RELENG_6:

  No need to force __inline__ of currecord(), as the compiler will usefully
  inline it when needed already, and the symbol is also required outside of
  audit.c.  This silences a new gcc warning on the topic of using __inline__
  instead of __inline.

16 years agoMerge audit_worker.c:1.14 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:41:19 +0000 (15:41 +0000)]
Merge audit_worker.c:1.14 from HEAD to RELENG_6:

  Short name of kernel audit worker thread from "audit_worker" to "audit".

16 years agoMerge audit_ioctl.h:1.5 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:36:18 +0000 (15:36 +0000)]
Merge audit_ioctl.h:1.5 from HEAD to RELENG_6:

  Don't expose #ifdef NOTYET parts to userspace via audit_ioctl.h, just
  remove them, since the functionlity they are associated with isn't there
  yet.

16 years agoMege audit_arg.c:1.9 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:33:10 +0000 (15:33 +0000)]
Mege audit_arg.c:1.9 from HEAD to RELENG_6:

  When returning early from audit_arg_file() due to so->so_pcb being NULL
  (due to an early reset or the like), remember to unlock the socket lock.
  This will not occur in 7-CURRENT, but could in theory occur in 6-STABLE.

16 years agoMerge audit_arg.c:1.8, audit_bsm_klib.c:1.5 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:31:48 +0000 (15:31 +0000)]
Merge audit_arg.c:1.8, audit_bsm_klib.c:1.5 from HEAD to RELENG_6:

  Add a witness sleep warning to canon_path(), which invokes vput() and hence
  may perform an unbounded sleep.  Remove an XXX comment suggesting that one
  be added.

  Obtained from:  TrustedBSD Project

16 years agoMege audit_syscalls.c:1.10 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:30:01 +0000 (15:30 +0000)]
Mege audit_syscalls.c:1.10 from HEAD to RELENG_6:

  Use p_cansee() to check that a target process for an audit state
  manipulation is visible to the subject process.  Remove XXX comments
  suggesting this.

  Convert one XXX on a difference from Darwin into a note: it's not a
  bug, it's a feature.

  Obtained from:  TrustedBSD Project

16 years agoMege audit_syscalls.c:1.11 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:17:42 +0000 (15:17 +0000)]
Mege audit_syscalls.c:1.11 from HEAD to RELENG_6:

  Remove two XXX comments that no longer apply.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit.c:1.23, audit_bsm_token.c:1.9, audit_trigger.c:1.4, and
rwatson [Thu, 8 Nov 2007 15:16:19 +0000 (15:16 +0000)]
Merge audit.c:1.23, audit_bsm_token.c:1.9, audit_trigger.c:1.4, and
audit_worker.c:1.13 from HEAD to RELENG_6:

  Update a number of comments:

  - Replace XXX with Note: in several cases where observations are made about
    future functionality rather than problems or bugs.

  - Remove an XXX comment about byte order and au_to_ip() -- IP headers must
    be submitted in network byte order.  Add a comment to this effect.

  - Mention that we don't implement select/poll for /dev/audit.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit.c:1.21 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:11:11 +0000 (15:11 +0000)]
Merge audit.c:1.21 from HEAD to RELENG_6:

  Trim some no longer XXX comments.
  Remove some commented out debugging printfs.

  Obtained from:  TrustedBSD Project

16 years agoMerge audit_syscalls.c:1.7 from HEAD to RELENG_6:
rwatson [Thu, 8 Nov 2007 15:01:35 +0000 (15:01 +0000)]
Merge audit_syscalls.c:1.7 from HEAD to RELENG_6:

  Audit path argument when changing audit trails.

  Call NDFREE(), which while not currently strictly necessary, isn't a
  bad idea.

  Obtained from:  TrustedBSD Project

16 years agoMFC
emax [Wed, 7 Nov 2007 17:42:45 +0000 (17:42 +0000)]
MFC

Allow RFCOMM servers to bind to a ''wildcard'' RFCOMM channel
zero (0). Actual RFCOMM channel will be assigned after listen(2)
call is done on a RFCOMM socket bound to a ''wildcard'' RFCOMM
channel zero (0).

Address locking issues in ng_btsocket_rfcomm_bind()

Submitted by: Heiko Wundram (Beenic) < wundram at beenic dot net >
Approved by: re (kensmith)

16 years agoMFC: Catch up to current state of ports tree
kris [Wed, 7 Nov 2007 13:57:08 +0000 (13:57 +0000)]
MFC: Catch up to current state of ports tree

Approved by:  re (kensmith)

16 years agoMFC rev. 1.12:
kib [Wed, 7 Nov 2007 13:38:45 +0000 (13:38 +0000)]
MFC rev. 1.12:
Fix various memory leaks.

16 years agoThis delta addresses some users experiencing watchdogs under load with
jfv [Wed, 7 Nov 2007 01:33:28 +0000 (01:33 +0000)]
This delta addresses some users experiencing watchdogs under load with
the last driver version. I suspect sheduling to be at the bottom of it,
but this version fixes it by:
- Making FAST interrupt handling off by default, if you want
to use it #define EM_FAST_IRQ in the Makefile.
- Add dual locks, one for the TX subsystem, the other CORE
for things like init and interface managment.
- Add cleaner MSI/X setup code.
- If using FAST interrupts there is now a mechanism in the
watchdog routine to kick off TX cleanups if they have not
been scheduled.

I think everyone will be happier with this change.

16 years agoMFC 1.68: remove smp_rendezvous() during shutdown
njl [Tue, 6 Nov 2007 20:13:51 +0000 (20:13 +0000)]
MFC 1.68: remove smp_rendezvous() during shutdown

Approved by:    re

16 years agoMFC 1.17:
simon [Tue, 6 Nov 2007 18:22:05 +0000 (18:22 +0000)]
MFC 1.17:

  If the INDEX file cannot be opened, print the actual filename / path
  pkg_version tried to open instead of just "INDEX" to make the actual
  problem more clear (e.g. missing INDEX-8).

16 years agoMFC:
benjsc [Tue, 6 Nov 2007 04:05:38 +0000 (04:05 +0000)]
MFC:

Regen:

* Hart:         rev 652 of pcidevs.txt (2007-09-24)
* Boemler:      vendors.txt (2007-10-25)

Approved by:  mlaier (mentor)
Approved by: re (kensmith)

16 years agoMFC of un-neede singl-threading code.
julian [Tue, 6 Nov 2007 03:01:32 +0000 (03:01 +0000)]
MFC of un-neede singl-threading code.

16 years agoMFC:
ariff [Tue, 6 Nov 2007 02:08:25 +0000 (02:08 +0000)]
MFC:

  Revisions:
  1.21     src/sys/dev/sound/pci/atiixp.c
  1.4      src/sys/dev/sound/pci/atiixp.h

Add support for trimmed down version of ATI SB600 AC97 audio
controller.

URL: http://www.ipc2u.de/catalog/P/PE/36486.html
PR: kern/117813
Submitted by: Klaus Mayr

16 years agoFix locking issue in ng_btsocket_l2cap_ctloutput()
emax [Mon, 5 Nov 2007 17:51:14 +0000 (17:51 +0000)]
Fix locking issue in ng_btsocket_l2cap_ctloutput()

Submitted by: Heiko Wundram (Beenic) < wundram at beenic dot net >
Approved by: re (kensmith)

16 years agoMFC: Remove mpte optimization from pmap_enter_quick().
ups [Mon, 5 Nov 2007 16:18:00 +0000 (16:18 +0000)]
MFC:  Remove mpte optimization from pmap_enter_quick().
There is a race with the current locking scheme and removing
it should have no measurable performance impact.
This fixes page faults leading to panics in pmap_enter_quick_locked()
on amd64/i386.

Requested by: alc@

16 years agoMFC:
phk [Mon, 5 Nov 2007 12:42:26 +0000 (12:42 +0000)]
MFC:
bring if_sis driver up to -current

16 years agoMFC:
phk [Mon, 5 Nov 2007 11:49:44 +0000 (11:49 +0000)]
MFC:
bus_alloc_resources() convenience API

16 years agoMFC: 1.30
keramida [Mon, 5 Nov 2007 11:26:41 +0000 (11:26 +0000)]
MFC: 1.30

  mtree should use getcwd() instead of getwd().  getcwd() is safer,
  because it includes checks for the length of the buffer it uses.

16 years agoMFC:
phk [Mon, 5 Nov 2007 11:17:39 +0000 (11:17 +0000)]
MFC:
Bring if_vr (Via Rhine) driver up to -current level.

16 years agoMFC miidevs rev 1.47 to RELENG_6
yongari [Mon, 5 Nov 2007 02:09:31 +0000 (02:09 +0000)]
MFC miidevs rev 1.47 to RELENG_6
  Add a newer RTL8211B(L) PHY.

16 years agoMFC rgephy.c rev 1.17-1.18, rgephyreg.h rev 1.3 to RELENG_6
yongari [Mon, 5 Nov 2007 02:06:27 +0000 (02:06 +0000)]
MFC rgephy.c rev 1.17-1.18, rgephyreg.h rev 1.3 to RELENG_6
rgephy.c rev 1.17, rgephyreg.h rev 1.3
  Add support for RealTek RTL8211B(L) PHY. It's based on the patch
  made by Michael Eisele and the patch was slightly modified by me.
  With this change several NVIDIA ethernet controllers(e.g. MCP61)
  works.

  RTL8211B(L) is RealTek's new gigabit PHY. The PHY has several
  features including crossover correction, polarity correction as
  well as supporting triple speed(10/100/1000bps). Data transfer
  between MAC and PHY is via RGMII for 1000baseT, MII for
  10baseT/100baseTX.
  Unfortunately, RealTek used the same model number for RTL8211B(L)
  PHY so there is no way to discriminate between RTL8211B(L) and its
  predecessors. ATM RTL8211B uses revision number 2 so checking the
  revision number seems to be only way to identify it.

rgephyreg.c rev 1.18
  Setting sc->mii_anegticks to MII_ANEGTICKS_GIGE in rgephy_attach()
  is redundant. mii_phy_add_media() already takes care of that.

16 years agoMFC rgephy.c rev 1.16 to RELENG_6
yongari [Mon, 5 Nov 2007 02:03:09 +0000 (02:03 +0000)]
MFC rgephy.c rev 1.16 to RELENG_6
  o RTL8169S/8110S integrated PHY and RTL8211B are gigabit PHYs so set
    mii_anegticks to MII_ANEGTICKS_GIGE and use it. Previously it used
    to MII_ANEGTICKS which may not be enough to wait before retrying
    autonegotiation process at 1000bps.
  o Reset autonegotation timer if media option is not IFM_AUTO or we
    got a valid link.
  o Announce link loss right after it happends.
  o Autonegiation is retried every mii_anegticks seconds.
  o Report link state changes right after setting autonegotiation.

16 years agoAdd latest firmware
kmacy [Mon, 5 Nov 2007 01:29:41 +0000 (01:29 +0000)]
Add latest firmware

16 years agoUpdate to reflect new firmware version
kmacy [Mon, 5 Nov 2007 00:55:52 +0000 (00:55 +0000)]
Update to reflect new firmware version

16 years agoMFC version 1.13, update of IP address for l.root-servers.net
dougb [Sun, 4 Nov 2007 23:59:51 +0000 (23:59 +0000)]
MFC version 1.13, update of IP address for l.root-servers.net

Approved by:    re (kensmith)

16 years agoMFC revision 1.350, move where we audit the PID to ensure that it gets
csjp [Sun, 4 Nov 2007 18:22:25 +0000 (18:22 +0000)]
MFC revision 1.350, move where we audit the PID to ensure that it gets
audited even if it's invalid.

16 years agoMFC revision 1.13
csjp [Sun, 4 Nov 2007 16:52:33 +0000 (16:52 +0000)]
MFC revision 1.13

Make sure we are incrementing read counters

16 years agoMFC: 1.19 - mdoc warning fix: delete spurious empty line.
keramida [Sun, 4 Nov 2007 15:24:44 +0000 (15:24 +0000)]
MFC: 1.19 - mdoc warning fix: delete spurious empty line.

16 years agoMFC: 1.14 - mdoc warning fix: Remove a spurious .El request.
keramida [Sun, 4 Nov 2007 15:18:40 +0000 (15:18 +0000)]
MFC: 1.14 - mdoc warning fix: Remove a spurious .El request.

16 years agoMFC minherit.2 -r 1.22-1.24
keramida [Sun, 4 Nov 2007 15:12:04 +0000 (15:12 +0000)]
MFC minherit.2 -r 1.22-1.24

16 years agoMFC: mdconfig.8 -r 1.37
keramida [Sun, 4 Nov 2007 15:07:28 +0000 (15:07 +0000)]
MFC: mdconfig.8 -r 1.37

% Add an example which shows how mdconfig(8) can be used
% to mount an ISO 9660 CD image file.
%
% PR:             112691
% Submitted by:   Warren Block, wblock at wonkity.com

16 years agoMFC documentation for recent uscanner cleanup.
luigi [Sun, 4 Nov 2007 03:54:16 +0000 (03:54 +0000)]
MFC documentation for recent uscanner cleanup.

16 years agoMFC 2007-10-05 07:26:40 UTC
luigi [Sun, 4 Nov 2007 03:28:31 +0000 (03:28 +0000)]
MFC 2007-10-05 07:26:40 UTC
Add entries for Epson multifunction scanner/printer/card readers,
with all functions supported.
More details on related topics (SANE configuration, OCR, etc.)
at http://info.iet.unipi.it/~luigi/FreeBSD/dx5050.html

16 years agoMFC revision 1.12
csjp [Sat, 3 Nov 2007 17:15:27 +0000 (17:15 +0000)]
MFC revision 1.12

- Wakeup all threads sleeping on an audit pipe instead of just one
- Wakeup threads with a consistent priority

16 years agoMFC revision 1.182
csjp [Sat, 3 Nov 2007 17:13:16 +0000 (17:13 +0000)]
MFC revision 1.182

Fix PID portion of stats in scenarios where the peer forks

16 years agoMFC revision 1.34
csjp [Sat, 3 Nov 2007 15:33:04 +0000 (15:33 +0000)]
MFC revision 1.34
Add support for processing extended (ip6) subject termids

16 years agoMFC lockf.c revision 1.17
csjp [Sat, 3 Nov 2007 15:31:38 +0000 (15:31 +0000)]
MFC lockf.c revision 1.17
MFC lockf.1 revision 1.19

16 years agoFix panic on kernel startup. This allowed me to successfully run a
wilko [Fri, 2 Nov 2007 21:40:04 +0000 (21:40 +0000)]
Fix panic on kernel startup.  This allowed me to successfully run a
6.3-PRE kernel.

Patch kindly provided by: jhb

16 years agoMFC 1.55: Fix another edge case in rman_manage_region() that could result
jhb [Fri, 2 Nov 2007 19:00:06 +0000 (19:00 +0000)]
MFC 1.55: Fix another edge case in rman_manage_region() that could result
in the the resource list getting missorted.

16 years agoMFC: Don't attach to the "IBM0057" ID.
jhb [Fri, 2 Nov 2007 18:54:50 +0000 (18:54 +0000)]
MFC: Don't attach to the "IBM0057" ID.

16 years agoMFC 1.28:
dwmalone [Fri, 2 Nov 2007 16:27:22 +0000 (16:27 +0000)]
MFC 1.28:
Make the type of the memory used by the BPF filter unsigned, so it
matches the BPF registers (which are the only thing that is assigned
to/from BPF memory). This is a pedantic change that shouldn't change
any behaviour.

PR: 115931
Submitted by: Matthew Luckie <mjl@luckie.org.nz>

16 years agoMFC 1.18: Skip requests use a minimum of 1 byte so that the
kientzle [Fri, 2 Nov 2007 08:44:11 +0000 (08:44 +0000)]
MFC 1.18: Skip requests use a minimum of 1 byte so that the
underlying I/O framework can avoid copying and so that we
avoid a 32-bit overflow issue that was causing bsdtar to
fail when skipping uncompressed tar entries that were an
exact multiple of 4GB.

16 years agoMFC revision 1.389
alc [Fri, 2 Nov 2007 06:33:20 +0000 (06:33 +0000)]
MFC revision 1.389
  Correct an error in vm_map_sync(), nee vm_map_clean(), that has existed
  since revision 1.1.  Specifically, neither traversal of the vm map checks
  whether the end of the vm map has been reached.  Consequently, the first
  traversal can wrap around and bogusly return an error.

16 years agoMFC: Reindent the read/write code of bd_realstrategy().
jhb [Thu, 1 Nov 2007 18:23:10 +0000 (18:23 +0000)]
MFC: Reindent the read/write code of bd_realstrategy().

16 years agoMFC: Slightly cleanup the 'bootdev' concept on x86.
jhb [Thu, 1 Nov 2007 18:20:37 +0000 (18:20 +0000)]
MFC: Slightly cleanup the 'bootdev' concept on x86.

16 years agoMFC e1000phy.c rev 1.19, e1000phyreg.h rev. 1.5 to RELENG_6
yongari [Thu, 1 Nov 2007 04:27:30 +0000 (04:27 +0000)]
MFC e1000phy.c rev 1.19, e1000phyreg.h rev. 1.5 to RELENG_6

  Add 88E1116/88E1116R PHY support code that takes the PHY out of
  power-down mode.

16 years agoMFC: etc/rc.d/ppp,v1.14 and etc/defaults/rc.conf,v1.319
emax [Wed, 31 Oct 2007 18:53:07 +0000 (18:53 +0000)]
MFC: etc/rc.d/ppp,v1.14 and etc/defaults/rc.conf,v1.319

Teach /etc/rc.d/ppp to start multiple instances of ppp.

ppp_profile variable can now contain multiple profiles.
Overrides for ppp mode and nat can go into ppp_$profile_mode
and ppp_$profile_nat variables respectively. If those are
not specified, defaults from ppp_mode and ppp_nat are used.

MFC: etc/rc.d/ppp v1.15

Teach /etc/rc.d/ppp how to start/stop individual instances
of ppp. This is an extension of previous commit.

Submitted by: Yuri Kurenkov < y dot kurenkov at init dot ru >
Reviewed by: mtm
Approved by: re (kensmith)

16 years agoMFC: Print IRQs in decimal rather than hex.
jhb [Wed, 31 Oct 2007 16:17:41 +0000 (16:17 +0000)]
MFC: Print IRQs in decimal rather than hex.

16 years agoMFC: Actually support the new HT capability type from HT 2.00b.
jhb [Wed, 31 Oct 2007 16:16:19 +0000 (16:16 +0000)]
MFC: Actually support the new HT capability type from HT 2.00b.

16 years agoMFC: Don't destroy an ACPI device_t for a PCI device if the ACPI device_t
jhb [Wed, 31 Oct 2007 16:11:01 +0000 (16:11 +0000)]
MFC: Don't destroy an ACPI device_t for a PCI device if the ACPI device_t
is already attached to a driver.

16 years agoMFC to RELENG_6 dhclient RFC3442 support (the classless static route
emaste [Wed, 31 Oct 2007 14:30:20 +0000 (14:30 +0000)]
MFC to RELENG_6 dhclient RFC3442 support (the classless static route
option)

dhclient.conf 1.2

  Add a $FreeBSD$ tag missing after the original import.  Note that this
  file isn't actually installed; the one in src/etc is.

clparse.c 1.3
dhclient-script 1.15
dhclient.c 1.21
dhclient.conf 1.3
dhcp.h 1.3
tables.c 1.4

  Implement RFC3442, the Classless Static Route option.

  The original DHCP specification includes a route option but it
  supports only class-based routes.  RFC3442 adds support for
  specifying the netmask width for each static route.  A variable
  length encoding is used to minimize the size of this option.

  PR:             bin/99534
  Submitted by:   Andrey V. Elsukov <bu7cher@yandex.ru>
  Reviewed by:    brooks

dhclient-script 1.16

  The minimum size of an RFC3442 destination descriptor is five bytes,
  so correct test to -ge 5.  Without this change an RFC3442 encoded
  default route would be ignored.

16 years agoMFC:
phk [Wed, 31 Oct 2007 13:06:17 +0000 (13:06 +0000)]
MFC:
recognize AMD GeodeLX companion chip CS5536

16 years agoMFi386: revision 1.616.2.9
nyan [Wed, 31 Oct 2007 11:53:17 +0000 (11:53 +0000)]
MFi386: revision 1.616.2.9

  Don't abuse tf_err to pass the faulting virtual address to signal handlers.

16 years agoMFC:
phk [Wed, 31 Oct 2007 08:42:18 +0000 (08:42 +0000)]
MFC:
GeodeLX and advantech watchdogs
Soekris NET5501 recognition and /dev/led/error

16 years agoMFC (revision 1.74):
ariff [Wed, 31 Oct 2007 04:00:07 +0000 (04:00 +0000)]
MFC (revision 1.74):

Remap and virtualize mixer controls for HP nx6110 with
AD1981B AC97 codec, unifying master volume control.

  * Remap "phout" --> SOUND_MIXER_VOLUME (internal speakers)
  * Virtual "vol" --> { "phout", "ogain" (headphone) }

Tested by: Frederic Chardon

16 years agoMFC (revision 1.11):
ariff [Wed, 31 Oct 2007 03:59:24 +0000 (03:59 +0000)]
MFC (revision 1.11):

Enable headphone jack-sense for HP nx6100 with AD1981B AC'97 codec,
automuting internal speakers.

Tested by: Frederic Chardon

16 years agoMFC r1.180
thompsa [Wed, 31 Oct 2007 01:36:06 +0000 (01:36 +0000)]
MFC r1.180
 Do not attempt to load the kernel module when checking if an interface exists.

16 years agoMFC: Revision 1.117 of route.c as well as the fixes to close a race when
jhb [Tue, 30 Oct 2007 18:07:55 +0000 (18:07 +0000)]
MFC: Revision 1.117 of route.c as well as the fixes to close a race when
trying to lookup a gateway route in rt_check().

16 years agoMFC: Update copyright attribution.
jhb [Tue, 30 Oct 2007 18:01:44 +0000 (18:01 +0000)]
MFC: Update copyright attribution.

16 years agoMFC pccarddevs v1.130
remko [Tue, 30 Oct 2007 10:21:03 +0000 (10:21 +0000)]
MFC pccarddevs v1.130
MFC pccard_cis_quirks.c v1.17

  Bring in some new Sierra Wireless drivers running under pccard.

  (AC710/AC750/AC850/A550/A555/A710/A750)

  PR: kern/110190
  Submitted by: Pascal Vizeli<pvizeli at yahoo dot de>
  Approved by: imp (mentor)

  MFC After: 1 week

Approved by: re (kensmith)
Approved by: imp (mentor)

16 years ago- Bump __FreeBSD_version for ctype ABI forward compatibility restored
rafan [Tue, 30 Oct 2007 04:44:10 +0000 (04:44 +0000)]
- Bump __FreeBSD_version for ctype ABI forward compatibility restored

Approved by: re (kensmith)

16 years agoMFC: Disable the coretemp on pc98.
nyan [Tue, 30 Oct 2007 02:56:31 +0000 (02:56 +0000)]
MFC: Disable the coretemp on pc98.

16 years agoRestore ctype(3) ABI forward compatibility in 6.x world. This was broken
rafan [Tue, 30 Oct 2007 01:30:42 +0000 (01:30 +0000)]
Restore ctype(3) ABI forward compatibility in 6.x world. This was broken
with last ctype(3) single-byte and multi-byte separation for fixing
operating on UTF-8 locale. To be specific, we introduced a new symbol in
libc. And this symbol is referred via inline functions here. So, you can
not run a binary built with this version of libc on an older system. To
restore the compatibility, make these functions built as non-inlined
form (see lib/libc/nomarco.c) and MFC rev 1.32 which drops the usage of
that symbol in question for __isctype(). As this may impact performance,
I only intend to fix in 6.x, but not 7 (not yet released) and 8 (it's HEAD).

Discussed on: cvs-src
Requested by: scottl
Reviewed by: ache, delphij
Approved by: re (kensmith)

16 years agoMFC: consolidate parsing of nfs root mount options in one place
sam [Tue, 30 Oct 2007 00:49:41 +0000 (00:49 +0000)]
MFC: consolidate parsing of nfs root mount options in one place
  and handle all options; this enables things like TCP mounts

16 years agounbreak build on arm: sade is like sysinstall, it cannot be built on an
sam [Mon, 29 Oct 2007 22:53:45 +0000 (22:53 +0000)]
unbreak build on arm: sade is like sysinstall, it cannot be built on an
architecture where libdisk is not supported

16 years agoMFC: remove some no-longer-needed _FREEFALL_CONFIG hacks.
peter [Mon, 29 Oct 2007 21:47:56 +0000 (21:47 +0000)]
MFC: remove some no-longer-needed _FREEFALL_CONFIG hacks.

16 years agoMFC to RELENG_6 part of kern_intr.c revision 1.142:
emaste [Mon, 29 Oct 2007 21:10:03 +0000 (21:10 +0000)]
MFC to RELENG_6 part of kern_intr.c revision 1.142:

  - Don't do the interrupt storm protection stuff for software interrupt
    handlers.

16 years agoMFC 1.62: Fix reading of files that use pax 'size' attribute
kientzle [Mon, 29 Oct 2007 03:29:21 +0000 (03:29 +0000)]
MFC 1.62:    Fix reading of files that use pax 'size' attribute
to store size.  In particular, the pax 'size' attribute is required
for any file over 8G.

This seems to fix tar -cf - | tar -xf - when there are files
over 8G in the heirarchy being copied.

16 years agoMFC: ng_ppp(4) MPSAFE.
bmah [Sun, 28 Oct 2007 21:44:01 +0000 (21:44 +0000)]
MFC:  ng_ppp(4) MPSAFE.

16 years agoMFC: 1.31
mtm [Sun, 28 Oct 2007 21:13:23 +0000 (21:13 +0000)]
MFC: 1.31
Allow adding a user(s) to additional groups in batch mode as well.

Submitted by: Eygene Ryabinkin <rea-fbsd@codelabs.ru>
      (modulo minor changes)

16 years agoMFC 1.166: make dev.ath.N.ledpin have an immediate effect
sam [Sun, 28 Oct 2007 18:14:57 +0000 (18:14 +0000)]
MFC 1.166: make dev.ath.N.ledpin have an immediate effect