GitHub/mt8127/android_kernel_alcatel_ttab.git
16 years ago[IPV4] FIB_HASH : Avoid unecessary loop in fn_hash_dump_zone()
Eric Dumazet [Fri, 18 Jan 2008 12:30:21 +0000 (04:30 -0800)]
[IPV4] FIB_HASH : Avoid unecessary loop in fn_hash_dump_zone()

I noticed "ip route list" was slower than "cat /proc/net/route" on a
machine with a full Internet routing table (214392 entries : Special
thanks to Robert ;) )

This is similar to problem reported in commit
d8c9283089287341c85a0a69de32c2287a990e71 ("[IPV4] ROUTE: ip_rt_dump()
is unecessary slow")

Fix is to avoid scanning the begining of fz_hash table, but directly
seek to the right offset.

Before patch :

time ip route >/tmp/ROUTE

real    0m1.285s
user    0m0.712s
sys     0m0.436s

After patch

# time ip route >/tmp/ROUTE

real    0m0.835s
user    0m0.692s
sys     0m0.124s

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[NET]: Fix interrupt semaphore corruption in Intel drivers.
David S. Miller [Fri, 18 Jan 2008 12:21:39 +0000 (04:21 -0800)]
[NET]: Fix interrupt semaphore corruption in Intel drivers.

Several of the Intel ethernet drivers keep an atomic counter used to
manage when to actually hit the hardware with a disable or an enable.

The way the net_rx_work() breakout logic works during a pending
napi_disable() is that it simply unschedules the poll even if it
still has work.

This can potentially leave interrupts disabled, but that is OK
because all of the drivers are about to disable interrupts
anyways in all such code paths that do a napi_disable().

Unfortunately, this trips up the semaphore used here in the Intel
drivers.  If you hit this case, when you try to bring the interface
back up it won't enable interrupts.  A reload of the driver module
fixes it of course.

So what we do is make sure all the sequences now go:

napi_disable();
atomic_set(&adapter->irq_sem, 0);
*_irq_disable();

which makes sure the counter is always in the correct state.

Reported by Robert Olsson.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV4] fib_trie: fix duplicated route issue
Joonwoo Park [Fri, 18 Jan 2008 11:45:18 +0000 (03:45 -0800)]
[IPV4] fib_trie: fix duplicated route issue

http://bugzilla.kernel.org/show_bug.cgi?id=9493

The fib allows making identical routes with 'ip route replace'.
This patch makes the fib return -EEXIST if replacement would cause duplication.

Signed-off-by: Joonwoo Park <joonwpark81@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV4] fib_hash: fix duplicated route issue
Joonwoo Park [Fri, 18 Jan 2008 11:44:48 +0000 (03:44 -0800)]
[IPV4] fib_hash: fix duplicated route issue

http://bugzilla.kernel.org/show_bug.cgi?id=9493

The fib allows making identical routes with 'ip route replace'.
This patch makes the fib return -EEXIST if replacement would cause duplication.

Signed-off-by: Joonwoo Park <joonwpark81@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV6]: Mischecked tw match in __inet6_check_established.
Pavel Emelyanov [Fri, 18 Jan 2008 11:27:12 +0000 (03:27 -0800)]
[IPV6]: Mischecked tw match in __inet6_check_established.

When looking for a conflicting connection the !sk->sk_bound_dev_if
check is performed only for live sockets, but not for timewait-ed.

This is not the case for ipv4, for __inet6_lookup_established in
both ipv4 and ipv6 and for other places that check for tw-s.

Was this missed accidentally? If so, then this patch fixes it and
besides makes use if the dif variable declared in the function.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agorfkill: call rfkill_led_trigger_unregister() on error
Eric Paris [Sun, 13 Jan 2008 21:20:56 +0000 (16:20 -0500)]
rfkill: call rfkill_led_trigger_unregister() on error

Code inspection turned up that error cases in rfkill_register() do not
call rfkill_led_trigger_unregister() even though we have already
registered.

Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoRemove bogus duplicate CONFIG_LGUEST_GUEST entry.
Rusty Russell [Sun, 20 Jan 2008 00:54:18 +0000 (11:54 +1100)]
Remove bogus duplicate CONFIG_LGUEST_GUEST entry.

It was moved to arch/x86/lguest/Kconfig, but I lost the deletion part in a
patch suffle.  My confused one-liner "fix" to turn it on is also reverted:
84f7466ee20cc094aa38617abfa2f3834871f054

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoarch: Ignore arch/i386 and arch/x86_64
Josef 'Jeff' Sipek [Sat, 19 Jan 2008 21:04:42 +0000 (16:04 -0500)]
arch: Ignore arch/i386 and arch/x86_64

The i386 and x86_64 arch directories contain nothing but a generated symlink
to arch/x86/boot/bzImage when a tree a built.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Anvin <hpa@zytor.com>
Signed-off-by: Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopata_pdc202xx_old: Fix crashes with ATAPI
Alan Cox [Sat, 19 Jan 2008 15:51:26 +0000 (15:51 +0000)]
pata_pdc202xx_old: Fix crashes with ATAPI

The PDC202xx older devices do not support ATAPI DMA via the usual
interfaces. What documentation I have isn't sufficient to support DMA and
it isn't clear if the Windows drivers do this or it is possible at all.
(Neither do the drivers/ide old drivers)

So turn it ATAPI DMA off, these are disk optimised controllers.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Sat, 19 Jan 2008 19:01:23 +0000 (11:01 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/ipath: Fix receiving UD messages with immediate data

16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Fri, 18 Jan 2008 22:06:44 +0000 (14:06 -0800)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (31 commits)
  Replace cpmac fix
  dl2k: the rest
  dl2k: MSCR, MSSR, ESR, PHY_SCR fixes
  dl2k: BMSR fixes
  dl2k: ANAR, ANLPAR fixes
  dl2k: BMCR_t fixes
  3c574, 3c515 bitfields abuse
  sbni endian fixes
  wan/lmc bitfields fixes
  dscc4 endian fixes
  S2io: Fixed synchronization between scheduling of napi with card reset and close
  atl1: fix frame length bug
  Documentation: add a guideline for hard_start_xmit method
  Revert "sky2: remove check for PCI wakeup setting from BIOS"
  e1000e Kconfig: remove ref to nonexistant docs
  bonding: Don't hold lock when calling rtnl_unlock
  bonding: fix lock ordering for rtnl and bonding_rwsem
  bonding: Fix up parameter parsing
  bonding: release slaves when master removed via sysfs
  bonding: fix locking during alb failover and slave removal
  ...

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Fri, 18 Jan 2008 22:06:19 +0000 (14:06 -0800)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] clarify watchdog operation in documentation
  [WATCHDOG] Revert "Stop looking for device as soon as one is found"

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86
Linus Torvalds [Fri, 18 Jan 2008 22:06:08 +0000 (14:06 -0800)]
Merge git://git./linux/kernel/git/x86/linux-2.6-x86

* git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86:
  x86: add support for the latest Intel processors to Oprofile

16 years agoSelecting LGUEST should turn on Guest support, as in 2.6.23.
Rusty Russell [Fri, 18 Jan 2008 20:02:29 +0000 (07:02 +1100)]
Selecting LGUEST should turn on Guest support, as in 2.6.23.

There's currently no way to turn on Lguest guest support; the planned
Kconfig virtualization reorg didn't get into 2.6.25.

This was unnoticed because if you already had CONFIG_LGUEST_GUEST=y in
your config, it worked.  Too bad about new users...

Also, the Kconfig help was wrong now the virtio drivers are merged.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agox86: add support for the latest Intel processors to Oprofile
Arjan van de Ven [Fri, 18 Jan 2008 21:49:33 +0000 (22:49 +0100)]
x86: add support for the latest Intel processors to Oprofile

The latest Intel processors (the 45nm ones) have a model number of 23
(old ones had 15); they're otherwise compatible on the oprofile side.
This patch adds the new model number to the oprofile code.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years ago[WATCHDOG] clarify watchdog operation in documentation
Andrew Dyer [Tue, 8 Jan 2008 20:40:37 +0000 (14:40 -0600)]
[WATCHDOG] clarify watchdog operation in documentation

It was not clear what the difference is/was between the
nowayout feature and the Magic Close feature.

Signed-off-by: "Andrew Dyer" <amdyer@gmail.com>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
16 years ago[WATCHDOG] Revert "Stop looking for device as soon as one is found"
Wim Van Sebroeck [Fri, 18 Jan 2008 21:01:34 +0000 (21:01 +0000)]
[WATCHDOG] Revert "Stop looking for device as soon as one is found"

This reverts commit 3ff6eb4a2fe5757cbe7c5d57c8eb60ab0775f2f0.

the !found check in the for loop allready made sure that only one
device was found.

Signed-Off-By: Pádraig Brady <P@draigBrady.com>
Signed-Off-By: Wim Van Sebroeck <wim@iguana.be>
16 years agoReplace cpmac fix
Matteo Croce [Sat, 12 Jan 2008 18:05:23 +0000 (19:05 +0100)]
Replace cpmac fix

Please apply this patch since i reverted by mistake
the commit 4e3ab47a547616e583c7a5458beced6aa34c8ef3
in 6cd043d99dcf5d252fcc682958541f449113f7b3

Signed-off-by: Matteo Croce <technoboy85@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodl2k: the rest
Al Viro [Sun, 13 Jan 2008 14:18:25 +0000 (14:18 +0000)]
dl2k: the rest

remove an unused union-with-bitfield of the same sort,
add missing conversions in debugging printk

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodl2k: MSCR, MSSR, ESR, PHY_SCR fixes
Al Viro [Sun, 13 Jan 2008 14:18:15 +0000 (14:18 +0000)]
dl2k: MSCR, MSSR, ESR, PHY_SCR fixes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodl2k: BMSR fixes
Al Viro [Sun, 13 Jan 2008 14:18:05 +0000 (14:18 +0000)]
dl2k: BMSR fixes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodl2k: ANAR, ANLPAR fixes
Al Viro [Sun, 13 Jan 2008 14:17:55 +0000 (14:17 +0000)]
dl2k: ANAR, ANLPAR fixes

same story, different registers...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodl2k: BMCR_t fixes
Al Viro [Sun, 13 Jan 2008 14:17:45 +0000 (14:17 +0000)]
dl2k: BMCR_t fixes

broken use of bitfields; FUBAR on big-endian (and not valid C,
strictly speaking).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years ago3c574, 3c515 bitfields abuse
Al Viro [Sun, 13 Jan 2008 14:17:35 +0000 (14:17 +0000)]
3c574, 3c515 bitfields abuse

wn3_config is shared by these cards; the way we deal with it is both bad C
(union abuse) and broken on big-endian.  For 3c515 it's less serious (ISA
cards are quite rare outside of little-endian boxen), but 3c574 is a pcmcia
one and that'd better be endian-independent...  Fix is the same in both
cases.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agosbni endian fixes
Al Viro [Sun, 13 Jan 2008 14:17:25 +0000 (14:17 +0000)]
sbni endian fixes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agowan/lmc bitfields fixes
Al Viro [Sun, 13 Jan 2008 14:17:15 +0000 (14:17 +0000)]
wan/lmc bitfields fixes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agodscc4 endian fixes
Al Viro [Sun, 13 Jan 2008 14:17:05 +0000 (14:17 +0000)]
dscc4 endian fixes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoS2io: Fixed synchronization between scheduling of napi with card reset and close
Sreenivasa Honnur [Tue, 15 Jan 2008 01:23:04 +0000 (20:23 -0500)]
S2io: Fixed synchronization between scheduling of napi with card reset and close

- Fixed synchronization between scheduling of napi with card reset and close
  by moving the enabling and disabling of napi to card up and card down
  functions respectively instead of open and close.

Signed-off-by: Surjit Reang <surjit.reang@neterion.com>
Signed-off-by: Ramkrishna Vepa <ram.vepa@neterion.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoatl1: fix frame length bug
Jay Cliburn [Tue, 15 Jan 2008 01:56:41 +0000 (19:56 -0600)]
atl1: fix frame length bug

The driver sets up the hardware to accept a frame with max length
equal to MTU + Ethernet header + FCS + VLAN tag, but we neglect to
add the VLAN tag size to the ingress buffer.  When a VLAN-tagged
frame arrives, the hardware passes it, but bad things happen
because the buffer is too small.  This patch fixes that.

Thanks to David Harris for reporting the bug and testing the fix.

Tested-by: David Harris <david.harris@cpni-inc.com>
Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoDocumentation: add a guideline for hard_start_xmit method
Matti Linnanvuori [Tue, 15 Jan 2008 14:25:27 +0000 (06:25 -0800)]
Documentation: add a guideline for hard_start_xmit method

Add a guideline not to modify SKBs.

Signed-off-by: Matti Linnanvuori <mattilinnanvuori@yahoo.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoRevert "sky2: remove check for PCI wakeup setting from BIOS"
Stephen Hemminger [Tue, 15 Jan 2008 19:29:29 +0000 (11:29 -0800)]
Revert "sky2: remove check for PCI wakeup setting from BIOS"

This reverts commit 84cd2dfb04d23a961c5f537baa243fa54d0987ac.

Some BIOS's break if Wake On Lan is enabled, and the machine
can't boot. Better to have some user's have to call ethtool to
enable WOL than to break a single user's boot.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoMerge branch 'fixes-jgarzik' of git://git.kernel.org/pub/scm/linux/kernel/git/linvill...
Jeff Garzik [Fri, 18 Jan 2008 19:40:47 +0000 (14:40 -0500)]
Merge branch 'fixes-jgarzik' of git://git./linux/kernel/git/linville/wireless-2.6 into upstream-fixes

16 years agoe1000e Kconfig: remove ref to nonexistant docs
Jason Uhlenkott [Thu, 17 Jan 2008 07:03:17 +0000 (23:03 -0800)]
e1000e Kconfig: remove ref to nonexistant docs

There is no Documentation/networking/e1000e.txt.

Signed-off-by: Jason Uhlenkott <jasonuhl@jasonuhl.org>
Cc: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoMerge branch 'ipg-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/romieu...
Jeff Garzik [Fri, 18 Jan 2008 19:39:26 +0000 (14:39 -0500)]
Merge branch 'ipg-fixes' of git://git./linux/kernel/git/romieu/netdev-2.6 into upstream-fixes

16 years agobonding: Don't hold lock when calling rtnl_unlock
Jay Vosburgh [Fri, 18 Jan 2008 00:25:03 +0000 (16:25 -0800)]
bonding: Don't hold lock when calling rtnl_unlock

Change bond_mii_monitor to not hold any locks when calling rtnl_unlock,
as rtnl_unlock can sleep (when acquring another mutex in netdev_run_todo).

Bug reported by Makito SHIOKAWA <mshiokawa@miraclelinux.com>, who
included a different patch.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: fix lock ordering for rtnl and bonding_rwsem
Jay Vosburgh [Fri, 18 Jan 2008 00:25:02 +0000 (16:25 -0800)]
bonding: fix lock ordering for rtnl and bonding_rwsem

Fix the handling of rtnl and the bonding_rwsem to always be acquired
in a consistent order (rtnl, then bonding_rwsem).

The existing code sometimes acquired them in this order, and sometimes
in the opposite order, which opens a window for deadlock between ifenslave
and sysfs.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: Fix up parameter parsing
Jay Vosburgh [Fri, 18 Jan 2008 00:25:01 +0000 (16:25 -0800)]
bonding: Fix up parameter parsing

A recent change to add an additional hash policy modified
bond_parse_parm, but it now does not correctly match parameters passed in
via sysfs.

Rewrote bond_parse_parm to handle (a) parameter matches that
are substrings of one another and (b) user input with whitespace (e.g.,
sysfs input often has a trailing newline).

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: release slaves when master removed via sysfs
Jay Vosburgh [Fri, 18 Jan 2008 00:25:00 +0000 (16:25 -0800)]
bonding: release slaves when master removed via sysfs

Add a call to bond_release_all in the bonding netdev event
handler for the master.  This releases the slaves for the case of, e.g.,
"echo -bond0 > /sys/class/net/bonding_masters", which otherwise will spin
forever waiting for references to be released.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: fix locking during alb failover and slave removal
Jay Vosburgh [Fri, 18 Jan 2008 00:24:59 +0000 (16:24 -0800)]
bonding: fix locking during alb failover and slave removal

alb_fasten_mac_swap (actually rlb_teach_disabled_mac_on_primary)
requries RTNL and no other locks.  This could cause dev_set_promiscuity
and/or dev_set_mac_address to be called with improper locking.

Changed callers to hold only RTNL during calls to alb_fasten_mac_swap
or functions calling it.  Updated header comments in affected functions to
reflect proper reality of locking requirements.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: fix ASSERT_RTNL that produces spurious warnings
Jay Vosburgh [Fri, 18 Jan 2008 00:24:58 +0000 (16:24 -0800)]
bonding: fix ASSERT_RTNL that produces spurious warnings

Move an ASSERT_RTNL down to where we should hold only RTNL;
the existing check produces spurious warnings because we hold additional
locks at _bh, tripping a debug warning in spin_lock_mutex().

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agobonding: fix locking in sysfs primary/active selection
Jay Vosburgh [Fri, 18 Jan 2008 00:24:57 +0000 (16:24 -0800)]
bonding: fix locking in sysfs primary/active selection

Fix the functions that store the primary and active slave
options via sysfs to hold the correct locks in the correct order.

The bond_change_active_slave and bond_select_active_slave
functions both require rtnl, bond->lock for read and curr_slave_lock for
write_bh, and no other locks.  This is so that the lower level
mode-specific functions (notably for balance-alb mode) can release locks
down to just rtnl in order to call, e.g., dev_set_mac_address with the
locks it expects (rtnl only).

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Thu, 17 Jan 2008 23:50:19 +0000 (15:50 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: ALPS - fix sync loss on Acer Aspire 5720ZG
  Input: psmouse - fix input_dev leak in lifebook driver
  Input: psmouse - fix potential memory leak in psmouse_connect()
  Input: usbtouchscreen - fix buffer overflow, make more egalax work
  Input: mousedev - handle mice that use absolute coordinates

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Thu, 17 Jan 2008 23:49:59 +0000 (15:49 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  [NIU]: Fix 1G PHY link state handling.
  [NET]: Fix TX timeout regression in Intel drivers.

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 17 Jan 2008 23:49:49 +0000 (15:49 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Fix hypervisor TLB operation error reporting.

16 years agofix wrong sized spinlock flags argument
Daniel Walker [Thu, 17 Jan 2008 20:52:05 +0000 (12:52 -0800)]
fix wrong sized spinlock flags argument

Correct wrong sized spinlock flags, form int to unsigned long.

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoFix unbalanced helper_lock in kernel/kmod.c
Nigel Cunningham [Thu, 17 Jan 2008 23:21:21 +0000 (15:21 -0800)]
Fix unbalanced helper_lock in kernel/kmod.c

call_usermodehelper_exec() has an exit path that can leave the
helper_lock() call at the top of the routine unbalanced.  The attached
patch fixes this issue.

Signed-off-by: Nigel Cunningham <nigel@tuxonice.net>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoRevert "local_t Documentation update"
Li Zefan [Thu, 17 Jan 2008 23:21:20 +0000 (15:21 -0800)]
Revert "local_t Documentation update"

This reverts commit e1265205c0ee3919c3f2c750662630154c8faab2.

It's a duplicate commit of commit 74beb9db77930be476b267ec8518a642f39a04bf,
resulting in a duplicate section.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofix radeonfb regression with Xpress 200m 5955
Alex [Thu, 17 Jan 2008 23:21:18 +0000 (15:21 -0800)]
fix radeonfb regression with Xpress 200m 5955

Fix http://bugzilla.kernel.org/show_bug.cgi?id=9762

Framebuffer is ok only with default parameters only (it is 1280x800-8@60).  If
parameters are video=radeonfb:1280x800-32@60 then xres, yres and xres_virtual
are ok but yres_virtual is 1024.  It can be corrected by fbset utility so I
think it can be corrected in the driver code also.

Steps to reproduce: video=radeonfb:1280x800-32@60 or
video=radeonfb:1280x800-16@60

Add 1280x800 mode into modedb

Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago#ifdef very expensive debug check in page fault path
Carsten Otte [Thu, 17 Jan 2008 23:21:17 +0000 (15:21 -0800)]
#ifdef very expensive debug check in page fault path

This patch puts #ifdef CONFIG_DEBUG_VM around a check in vm_normal_page
that verifies that a pfn is valid.  This patch increases performance of the
page fault microbenchmark in lmbench by 13% and overall dbench performance
by 7% on s390x.  pfn_valid() is an expensive operation on s390 that needs a
high double digit amount of CPU cycles.  Nick Piggin suggested that
pfn_valid() involves an array lookup on systems with sparsemem, and
therefore is an expensive operation there too.

The check looks like a clear debug thing to me, it should never trigger on
regular kernels.  And if a pte is created for an invalid pfn, we'll find
out once the memory gets accessed later on anyway.  Please consider
inclusion of this patch into mm.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Acked-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoalpha: fix conversion from denormal float to double
Ivan Kokshaysky [Thu, 17 Jan 2008 23:21:13 +0000 (15:21 -0800)]
alpha: fix conversion from denormal float to double

The trap handler does properly update the fraction,
but not the exponent...

Thanks to Paolo Bonzini for the bug report and the testcase.

Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Paolo Bonzini <bonzini@gnu.org>
Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agojbd: do not try lock_acquire after handle made invalid
Jonas Bonn [Thu, 17 Jan 2008 23:21:13 +0000 (15:21 -0800)]
jbd: do not try lock_acquire after handle made invalid

This likely fixes the oops in __lock_acquire reported as:

http://www.kerneloops.org/raw.php?rawid=2753&msgid=
http://www.kerneloops.org/raw.php?rawid=2749&msgid=

In these reported oopses, start_this_handle is returning -EROFS.

Signed-off-by: Jonas Bonn <jonas.bonn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: fix section mismatch warning in page_alloc.c
Sam Ravnborg [Thu, 17 Jan 2008 23:21:12 +0000 (15:21 -0800)]
mm: fix section mismatch warning in page_alloc.c

With CONFIG_HOTPLUG=n and CONFIG_HOTPLUG_CPU=y we saw
following warning:
WARNING: mm/built-in.o(.text+0x6864): Section mismatch: reference to .init.text: (between 'process_zones' and 'pageset_cpuup_callback')

The culprit was zone_batchsize() which were annotated __devinit but used
from process_zones() which is annotated __cpuinit.  zone_batchsize() are
used from another function annotated __meminit so the only valid option is
to drop the annotation of zone_batchsize() so we know it is always valid to
use it.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoCRIS v10: vmlinux.lds.S: ix kernel oops on boot and use common defines
Jesper Nilsson [Thu, 17 Jan 2008 23:21:11 +0000 (15:21 -0800)]
CRIS v10: vmlinux.lds.S: ix kernel oops on boot and use common defines

- Move alignment to page size of init data outside ifdef for BLK_DEV_INITRD.
  The reservation up to page size of memory after init data was previously
  not done if BLK_DEV_INITRD was undefined.
  This caused a kernel oops when init memory pages were freed after startup,
  data placed in the same page as the last init memory would also be freed
  and reused, with disastrous results.

- Use macros for initcalls and .text sections.

- Replace hardcoded page size constant with PAGE_SIZE define.

- Change include/asm-cris/page.h to use the _AC macro to instead
  of testing __ASSEMBLY__.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Mikael Starvik <mikael.starvik@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopnpacpi: print resource shortage message only once (more)
Len Brown [Thu, 17 Jan 2008 23:21:10 +0000 (15:21 -0800)]
pnpacpi: print resource shortage message only once (more)

Wups, previous patch was ineffective in 2 cases.

http://bugzilla.kernel.org/show_bug.cgi?id=9535

Signed-off-by: Len Brown <len.brown@intel.com>
Reported-by: "Hartkopp, Oliver (K-EFE/E)" <oliver.hartkopp@volkswagen.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohfs: fix coverity-found null deref
Eric Sandeen [Thu, 17 Jan 2008 23:21:09 +0000 (15:21 -0800)]
hfs: fix coverity-found null deref

Fix potential null deref introduced by commit
cf0594625083111ae522496dc1c256f7476939c2
http://bugzilla.kernel.org/show_bug.cgi?id=9748

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Cc: Roman Zippel <zippel@linux-m68k.org>
Reported-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpufreq: Initialise default governor before use
Johannes Weiner [Thu, 17 Jan 2008 23:21:08 +0000 (15:21 -0800)]
cpufreq: Initialise default governor before use

When the cpufreq driver starts up at boot time, it calls into the default
governor which might not be initialised yet.  This hurts when the
governor's worker function relies on memory that is not yet set up by its
init function.

This migrates all governors from module_init() to fs_initcall() when being
the default, as was already done in cpufreq_performance when it was the
only possible choice.  The performance governor is always initialized early
because it might be used as fallback even when not being the default.

Fixes at least one actual oops where ondemand is the default governor and
cpufreq_governor_dbs() uses the uninitialised kondemand_wq work-queue
during boot-time.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Cc: Dave Jones <davej@codemonkey.org.uk>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoipg: fix Tx completion irq request
Francois Romieu [Thu, 10 Jan 2008 22:53:15 +0000 (23:53 +0100)]
ipg: fix Tx completion irq request

The current logic will only request an ack for the first pending
packet. No irq is triggered as soon as the CPU submits a few
packets a bit quickly.  Let's request an irq for every packet
instead.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
16 years agoipg: fix queue stop condition in the xmit handler
Francois Romieu [Thu, 10 Jan 2008 22:45:05 +0000 (23:45 +0100)]
ipg: fix queue stop condition in the xmit handler

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
16 years agoipg: plug Tx completion leak
Francois Romieu [Thu, 10 Jan 2008 22:40:59 +0000 (23:40 +0100)]
ipg: plug Tx completion leak

The Tx skb release could not free more than one skb per call.
Add it to the fact that the xmit handler does not check for
a queue full condition and you have a recipe to leak quickly.

Let's release every pending Tx descriptor which has been given
back to the host CPU by the network controller. The xmit handler
suggests that it is done through the IPG_TFC_TFDDONE bit.

Remove the former "curr" computing: it does not produce anything
usable in its current form.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
16 years agoipg: balance locking in irq handler
Francois Romieu [Thu, 10 Jan 2008 22:25:30 +0000 (23:25 +0100)]
ipg: balance locking in irq handler

Spotted-by: <linux@horizon.com>
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
16 years agoInput: ALPS - fix sync loss on Acer Aspire 5720ZG
Dmitry Torokhov [Thu, 17 Jan 2008 17:01:58 +0000 (12:01 -0500)]
Input: ALPS - fix sync loss on Acer Aspire 5720ZG

The recently added support for Dell Volstro 1400 was causing protocol
synchronization errors on Acer Aspire 5720ZG, fix it.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years agoInput: psmouse - fix input_dev leak in lifebook driver
Andres Salomon [Thu, 17 Jan 2008 17:01:51 +0000 (12:01 -0500)]
Input: psmouse - fix input_dev leak in lifebook driver

The lifebook driver may register a second input device, but it never
unregisters it.  This fixes that.

Signed-off-by: Andres Salomon <dilinger@debian.org>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years agoInput: psmouse - fix potential memory leak in psmouse_connect()
Andres Salomon [Thu, 17 Jan 2008 17:01:30 +0000 (12:01 -0500)]
Input: psmouse - fix potential memory leak in psmouse_connect()

If we successfully call input_register_device() in psmouse_connect()
but sysfs_create_group() fails, we'll enter the error path without
ever having called input_unregister_device() potentially leaking
memory.

Signed-off-by: Andres Salomon <dilinger@debian.org>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years agoInput: usbtouchscreen - fix buffer overflow, make more egalax work
Daniel Ritz [Thu, 17 Jan 2008 17:01:18 +0000 (12:01 -0500)]
Input: usbtouchscreen - fix buffer overflow, make more egalax work

Fix a buffer overflow in mutli-packet handling code.  The overflow can
only happen with eGalax devices and is even there very unlikely (only
non-report packet are affected any only when truncated after the first
byte).

Also changes the mutli-packet handling code not to drop unknown packets,
but rather just drop one byte.  This allows synchronizing on report packets
in the data stream.  It's required for some egalax devices to work at all.

Also remove the pointless 'flags' member of the device struct and set the
version number to 0.6, plus some minor cleanups.

[akpm@linux-foundation.org: coding-style fixes]

Signed-off-by: Daniel Ritz <daniel.ritz@gmx.ch>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years agoInput: mousedev - handle mice that use absolute coordinates
Micah Parrish [Thu, 17 Jan 2008 17:01:04 +0000 (12:01 -0500)]
Input: mousedev - handle mice that use absolute coordinates

Devices like the HP Integrated Remote Console Virtual Mouse, which are
standard equipment on all Proliant and Integrity servers, produce
absolute coordinates instead of relative coordinates.  This is done to
synchronize the position of the mouse cursor on the client desktop
with the mouse cursor position on the server.  Mousedev is not
designed to pass those absolute events directly to X, but it can
translate them into relative movements.  It currently does this for
tablet like devices and touchpads.  This patch merely tells it to also
include a device with ABS_X, ABS_Y, and mouse buttons in its list of
devices to process input for.

This patch enables the mouse pointer to move when using the remote
console.

Signed-off-by: Micah Parrish <micah.parrish@hp.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years ago[SPARC64]: Fix hypervisor TLB operation error reporting.
David S. Miller [Thu, 17 Jan 2008 09:32:09 +0000 (01:32 -0800)]
[SPARC64]: Fix hypervisor TLB operation error reporting.

1) Trap level wasn't being passed down properly, we need to
   move it from %l4 into the correct outgoing arg register.

2) Although the TPC often provides the most direct clue, we
   have the caller PC so we should provide that as well.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[NIU]: Fix 1G PHY link state handling.
David S. Miller [Wed, 16 Jan 2008 06:50:08 +0000 (22:50 -0800)]
[NIU]: Fix 1G PHY link state handling.

The code in link_status_1g() computes the active speed
and duplex but does not update the link config state
with those values.

As a result the link speed is not reported correctly
and the XIF is not reprogrammed properly on link up
events.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[NET]: Fix TX timeout regression in Intel drivers.
David S. Miller [Wed, 16 Jan 2008 06:43:24 +0000 (22:43 -0800)]
[NET]: Fix TX timeout regression in Intel drivers.

This fixes a regression added by changeset
53e52c729cc169db82a6105fac7a166e10c2ec36 ("[NET]: Make ->poll()
breakout consistent in Intel ethernet drivers.")

As pointed out by Jesse Brandeburg, for three of the drivers edited
above there is breakout logic in the *_clean_tx_irq() code to prevent
running TX reclaim forever.  If this occurs, we have to elide NAPI
poll completion or else those TX events will never be serviced.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
16 years agoIB/ipath: Fix receiving UD messages with immediate data
Ralph Campbell [Tue, 15 Jan 2008 23:58:13 +0000 (15:58 -0800)]
IB/ipath: Fix receiving UD messages with immediate data

This fixes a small bug in ipath_ud_rcv()'s handling of UD messages
with immediate data.  We need to test whether immediate data is
present and update the header size accordingly *before* testing the
packet size from the header against the actual received length.
Otherwise the wrong header size will be used and all messages with
immediate data will be dropped.

This bug keeps MVAPICH-UD and HP MPI from working at all on ipath devices.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
16 years agoMerge branch 'v2.6.24-rc7-lockdep' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 16 Jan 2008 17:55:49 +0000 (09:55 -0800)]
Merge branch 'v2.6.24-rc7-lockdep' of git://git./linux/kernel/git/peterz/linux-2.6-lockdep

* 'v2.6.24-rc7-lockdep' of git://git.kernel.org/pub/scm/linux/kernel/git/peterz/linux-2.6-lockdep:
  lockdep: more hardirq annotations for notify_die()
  lockdep: fix workqueue creation API lockdep interaction
  lockdep: fix internal double unlock during self-test

16 years agosysfs: fix bugs in sysfs_rename/move_dir()
Tejun Heo [Wed, 16 Jan 2008 03:10:53 +0000 (12:10 +0900)]
sysfs: fix bugs in sysfs_rename/move_dir()

sysfs_rename/move_dir() have the following bugs.

 - On dentry lookup failure, kfree() is called on ERR_PTR() value.
 - sysfs_move_dir() has an extra dput() on success path.

Fix them.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosysfs: make sysfs_lookup() return ERR_PTR(-ENOENT) on failed lookup
Tejun Heo [Wed, 16 Jan 2008 03:06:14 +0000 (12:06 +0900)]
sysfs: make sysfs_lookup() return ERR_PTR(-ENOENT) on failed lookup

sysfs tries to keep dcache a strict subset of sysfs_dirent tree by
shooting down dentries when a node is removed, that is, no negative
dentry for sysfs.  However, the lookup function returned NULL and thus
created negative dentries when the target node didn't exist.

Make sysfs_lookup() return ERR_PTR(-ENOENT) on lookup failure.  This
fixes the NULL dereference bug in sysfs_get_dentry() discovered by
bluetooth rfcomm device moving around.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agort2x00: Fix ieee80211 payload alignment
Ivo van Doorn [Fri, 11 Jan 2008 19:53:07 +0000 (20:53 +0100)]
rt2x00: Fix ieee80211 payload alignment

As Johannes Berg indicated, the NET_IP_ALIGN doesn't
need to be used for ieee80211 frames. This means we
can simplify the alignment calculation to just
use the result of the header size modulus 4 as frame
alignment.

Furthermore we shouldn't use NET_IP_ALIGN in rt2x00usb
because it could be 0 on some architectures and we absolutely
need to have 2 bytes reserved for possible aligning.

Signed-off-by: Ivo van Doorn<IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: fix use-after-free rfkill bug
Stefano Brivio [Sun, 13 Jan 2008 17:30:14 +0000 (18:30 +0100)]
b43: fix use-after-free rfkill bug

Fix rfkill code which caused a use-after-free bug.

Signed-off-by: Stefano Brivio <stefano.brivio@polimi.it>
Acked-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoipw2200: fix typo in kerneldoc
Stefano Brivio [Sat, 12 Jan 2008 22:12:26 +0000 (23:12 +0100)]
ipw2200: fix typo in kerneldoc

Fix a typo in kerneldoc for ipw2200.

Signed-off-by: Stefano Brivio <stefano.brivio@polimi.it>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agowireless/libertas support for 88w8385 sdio older revision
Marc Pignat [Fri, 11 Jan 2008 15:12:28 +0000 (16:12 +0100)]
wireless/libertas support for 88w8385 sdio older revision

Identifiaction of another revision of 88w8385 in sdio mode.

Signed-off-by: Marc Pignat <marc.pignat@hevs.ch>
Acked-by: Pierre Ossman <drzeus@drzeus.cx>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agohostap: section mismatch warning
Randy Dunlap [Thu, 10 Jan 2008 22:31:30 +0000 (14:31 -0800)]
hostap: section mismatch warning

Fix section mismatch by changing variable name to match one of the
whitelisted (allowable) names for pointing into init data:

WARNING: vmlinux.o(.data+0xce618): Section mismatch: reference to .init.data:prism2_plx_id_table (between 'prism2_plx_drv_id' and 'dev_info')

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolockdep: more hardirq annotations for notify_die()
Peter Zijlstra [Wed, 16 Jan 2008 08:51:59 +0000 (09:51 +0100)]
lockdep: more hardirq annotations for notify_die()
On Sat, 2007-12-29 at 18:06 +0100, Marcin Slusarz wrote:
> Hi
> Today I've got this (while i was upgrading my gentoo box):
>
> WARNING: at kernel/lockdep.c:2658 check_flags()
> Pid: 21680, comm: conftest Not tainted 2.6.24-rc6 #63
>
> Call Trace:
>  [<ffffffff80253457>] check_flags+0x1c7/0x1d0
>  [<ffffffff80257217>] lock_acquire+0x57/0xc0
>  [<ffffffff8024d5c0>] __atomic_notifier_call_chain+0x60/0xd0
>  [<ffffffff8024d641>] atomic_notifier_call_chain+0x11/0x20
>  [<ffffffff8024d67e>] notify_die+0x2e/0x30
>  [<ffffffff8020da0a>] do_divide_error+0x5a/0xa0
>  [<ffffffff80522bdd>] trace_hardirqs_on_thunk+0x35/0x3a
>  [<ffffffff80255b89>] trace_hardirqs_on+0xd9/0x180
>  [<ffffffff80522bdd>] trace_hardirqs_on_thunk+0x35/0x3a
>  [<ffffffff80523c2d>] error_exit+0x0/0xa9
>
> possible reason: unannotated irqs-off.
> irq event stamp: 4693
> hardirqs last  enabled at (4693): [<ffffffff80522bdd>] trace_hardirqs_on_thunk+0x35/0x3a
> hardirqs last disabled at (4692): [<ffffffff80522c17>] trace_hardirqs_off_thunk+0x35/0x37
> softirqs last  enabled at (3546): [<ffffffff80238343>] __do_softirq+0xb3/0xd0
> softirqs last disabled at (3521): [<ffffffff8020c97c>] call_softirq+0x1c/0x30

more early fixups for notify_die()..

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agolockdep: fix workqueue creation API lockdep interaction
Johannes Berg [Wed, 16 Jan 2008 08:51:58 +0000 (09:51 +0100)]
lockdep: fix workqueue creation API lockdep interaction
Dave Young reported warnings from lockdep that the workqueue API
can sometimes try to register lockdep classes with the same key
but different names. This is not permitted in lockdep.

Unfortunately, I was unaware of that restriction when I wrote
the code to debug workqueue problems with lockdep and used the
workqueue name as the lockdep class name. This can obviously
lead to the problem if the workqueue name is dynamic.

This patch solves the problem by always using a constant name
for the workqueue's lockdep class, namely either the constant
name that was passed in or a string consisting of the variable
name.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
16 years agolockdep: fix internal double unlock during self-test
Nick Piggin [Wed, 16 Jan 2008 08:51:58 +0000 (09:51 +0100)]
lockdep: fix internal double unlock during self-test
Lockdep, during self-test (when it was simulating double unlocks) was
sometimes unconditionally unlocking a spinlock when it had not been
locked. This won't work for ticket locks.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
16 years agoLinux 2.6.24-rc8
Linus Torvalds [Wed, 16 Jan 2008 04:22:48 +0000 (20:22 -0800)]
Linux 2.6.24-rc8

16 years ago[IA64] Fix unaligned handler for floating point instructions with base update
Luck, Tony [Mon, 14 Jan 2008 17:59:24 +0000 (09:59 -0800)]
[IA64] Fix unaligned handler for floating point instructions with base update

The compiler team did the hard work for this distilling a problem in
large fortran application which showed up when applied to a 290MB input
data set down to this instruction:

ldfd f34=[r17],-8

Which they noticed incremented r17 by 0x10 rather than decrementing it
by 8 when the value in r17 caused an unaligned data fault.  I tracked
it down to some bad instruction decoding in unaligned.c. The code
assumes that the 'x' bit can determine whether the instruction is
an "ldf" or "ldfp" ... which it is for opcode=6 (see table 4-29 on
page 3:302 of the SDM).  But for opcode=7 the 'x' bit is irrelevent,
all variants are "ldf" instructions (see table 4-36 on page 3:306).

Note also that interpreting the instruction as "ldfp" means that the
"paired" floating point register (f35 in the example here) will also
be corrupted.

Signed-off-by: Tony Luck <tony.luck@intel.com>
16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Tue, 15 Jan 2008 21:49:09 +0000 (13:49 -0800)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: correct handling of TSS DVD
  [libata] core checkpatch fix
  [libata] pata_bf54x: checkpatch fixes
  libata fixes for sparse-found problems

16 years agolibata: correct handling of TSS DVD
Alan Cox [Mon, 14 Jan 2008 09:06:40 +0000 (01:06 -0800)]
libata: correct handling of TSS DVD

Devices that misreport the validity bit for word 93 look like SATA.  If
they are on the blacklist then we must not test for SATA but assume 40 wire
in the 40 wire case (The TSSCorp reports 80 wire on SATA it seems!)

Signed-off-by: Alan Cox <alan@redhat.com>
Cc: Tejun Heo <htejun@gmail.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years ago[libata] core checkpatch fix
Andrew Morton [Thu, 10 Jan 2008 22:33:09 +0000 (14:33 -0800)]
[libata] core checkpatch fix

Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Tejun Heo <htejun@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years ago[libata] pata_bf54x: checkpatch fixes
Andrew Morton [Thu, 10 Jan 2008 22:33:08 +0000 (14:33 -0800)]
[libata] pata_bf54x: checkpatch fixes

WARNING: line over 80 characters
#36: FILE: drivers/ata/pata_bf54x.c:1512:
+ while (bfin_port_info[board_idx].udma_mask>0 && udma_fsclk[udma_mode] > fsclk) {

ERROR: need spaces around that '>' (ctx:VxV)
#36: FILE: drivers/ata/pata_bf54x.c:1512:
+ while (bfin_port_info[board_idx].udma_mask>0 && udma_fsclk[udma_mode] > fsclk) {
                                            ^

total: 1 errors, 1 warnings, 19 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

Please run checkpatch prior to sending patches

Cc: Jeff Garzik <jeff@garzik.org>
Cc: Sonic Zhang <sonic.zhang@analog.com>
Cc: Tejun Heo <htejun@gmail.com>
Cc: sonic zhang <sonic.adi@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agolibata fixes for sparse-found problems
Al Viro [Sat, 12 Jan 2008 14:16:14 +0000 (14:16 +0000)]
libata fixes for sparse-found problems

In pata_legacy and pata_winbond we've got bugs - cpu_to_le16() instead
of cpu_to_le32().  Fortunately, both affected suckers are VLB, thus
l-e-only, so we might get away with that unless we hit it with slop == 3
(hadn't checked if playing with badly aligned sg could trigger that).
Still buggy...  Moreover, pata_legacy, pata_winbond and pata_qdi forgot to
initialize pad on the write side of 32bit case in their ->data_xfer().
Hopefully the hardware does't care, but still, sending uninitialized
data to it...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
16 years agoFix Blackfin HARDWARE_PM support
Mathieu Desnoyers [Tue, 15 Jan 2008 17:42:02 +0000 (12:42 -0500)]
Fix Blackfin HARDWARE_PM support

This patch restores the blackfin Hardware Performance Monitor Profiling
support that was killed by the combining of instrumentation menus in
commit 09cadedbdc01f1a4bea1f427d4fb4642eaa19da9.

Since there seems to be no good reason to behave differently from other
architectures, it now automatically selects the hardware performance
counters whenever the profiling is activated.

mach-common/irqpanic.c: pm_overflow calls pm_overflow_handler which is
in oprofile/op_model_bf533.c.  I doubt that setting HARDWARE_PM as "m"
will work at all, since the pm_overflow_handler should be in the core
kernel image because it is called by irqpanic.c.

Therefore, I change HARDWARE_PM from a tristate to a bool.

The whole arch/$(ARCH)/oprofile/ is built depending on CONFIG_OPROFILE. Since
part of the HARDWARE_PM support files sits in this directory, it makes sense to
also depend on OPROFILE, not only PROFILING. Since OPROFILE already depends on
PROFILING, it is correct to only depend on OPROFILE only.

Thanks to Adrian Bunk for finding this bug and providing an initial
patch.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Adrian Bunk <adrian.bunk@movial.fi>
CC: Randy Dunlap <randy.dunlap@oracle.com>
CC: bryan.wu@analog.com
Acked-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoFix ARM profiling/instrumentation configuration
Linus Torvalds [Tue, 15 Jan 2008 17:23:51 +0000 (09:23 -0800)]
Fix ARM profiling/instrumentation configuration

Commit 09cadedbdc01f1a4bea1f427d4fb4642eaa19da9 ("Combine
instrumentation menus in kernel/Kconfig.instrumentation") broke ARM
profiling support, since ARM has some extra Kconfig options and doesn't
just use the common OPROFILE/KPROBES config options.

Rather than just revert the thing outright, or add ARM-specific
knowledge to the generic Kconfig.instrumentation file (where the only
and whole point was to be generic, not too architecture-specific), this
just makes ARM not use the generic version, since it doesn't suit it.

So create an arm-specific version of Kconfig.instrumentation instead,
and use that.

Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Russell King <rmk+lkml@arm.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86
Linus Torvalds [Tue, 15 Jan 2008 16:08:40 +0000 (08:08 -0800)]
Merge git://git./linux/kernel/git/x86/linux-2.6-x86

* git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86:
  x86: fix RTC_AIE with CONFIG_HPET_EMULATE_RTC
  x86: asm-x86/msr.h: pull in linux/types.h
  x86: fix boot crash on HIGHMEM4G && SPARSEMEM

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Tue, 15 Jan 2008 16:07:03 +0000 (08:07 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Fix build with SPARSEMEM_VMEMMAP disabled.
  [SPARC]: Make gettimeofday() monotonic again.

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 15 Jan 2008 16:06:47 +0000 (08:06 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  [TOKENRING]: rif_timer not initialized properly
  [NETFILTER]: bridge: fix double POST_ROUTING invocation
  [NETFILTER]: xt_helper: Do not bypass RCU
  [NETFILTER]: ip6t_eui64: Fixes calculation of Universal/Local bit
  [MACVLAN]: Prevent nesting macvlan devices
  [VLAN]: nested VLAN: fix lockdep's recursive locking warning
  [DECNET] ROUTE: fix rcu_dereference() uses in /proc/net/decnet_cache
  [BLUETOOTH]: Always send explicit hci_ll wake-up acks.
  [BLUETOOTH]: rfcomm tty BUG_ON() code fix
  [AX25] af_ax25: Possible circular locking.
  [AX25]: Kill user triggable printks.
  [IPV4] ROUTE: fix rcu_dereference() uses in /proc/net/rt_cache
  [NEIGH]: Fix race between neigh_parms_release and neightbl_fill_parms
  [NIU]: Support for Marvell PHY

16 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Tue, 15 Jan 2008 16:01:48 +0000 (08:01 -0800)]
Merge branch 'merge' of git://git./linux/kernel/git/paulus/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Fix boot failure on POWER6
  [POWERPC] Workaround for iommu page alignment

16 years agox86: fix RTC_AIE with CONFIG_HPET_EMULATE_RTC
Bernhard Walle [Tue, 15 Jan 2008 15:44:38 +0000 (16:44 +0100)]
x86: fix RTC_AIE with CONFIG_HPET_EMULATE_RTC

In the current code, RTC_AIE doesn't work if the RTC relies on
CONFIG_HPET_EMULATE_RTC because the code sets the RTC_AIE flag in
hpet_set_rtc_irq_bit().  The interrupt handles does accidentally check
for RTC_PIE and not RTC_AIE when comparing the time which was set in
hpet_set_alarm_time().

I now verified on a test system here that without the patch applied,
the attached test program fails on a system that has HPET with
2.6.24-rc7-default. That's not critical since I guess the problem has
been there for several kernel releases, but as the fix is quite
obvious.

Configuration is CONFIG_RTC=y and CONFIG_HPET_EMULATE_RTC=y.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: asm-x86/msr.h: pull in linux/types.h
Mike Frysinger [Tue, 15 Jan 2008 15:44:38 +0000 (16:44 +0100)]
x86: asm-x86/msr.h: pull in linux/types.h

Since the msr.h header uses types like __u32, it should pull in linux/types.h.

[ mingo@elte.hu: affects user-space that includes this header. We dont
  actually like user-space including raw kernel headers but it's a
  longstanding practice and it's easy for the kernel to be nice about
  this. ]

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: fix boot crash on HIGHMEM4G && SPARSEMEM
Ingo Molnar [Tue, 15 Jan 2008 15:44:37 +0000 (16:44 +0100)]
x86: fix boot crash on HIGHMEM4G && SPARSEMEM

Denys Fedoryshchenko reported a bootup crash when he upgraded
his system from 3GB to 4GB RAM:

   http://lkml.org/lkml/2008/1/7/9

the bug is due to HIGHMEM4G && SPARSEMEM kernels making pfn_to_page()
to return an invalid pointer when the pfn is in a memory hole. The
256 MB PCI aperture at the end of RAM was not mapped by sparsemem,
and hence the pfn was not valid. But set_highmem_pages_init() iterated
this range without checking the pfn's validity first.

this bug was probably present in the sparsemem code ever since sparsemem
has been introduced in v2.6.13. It was masked due to HIGHMEM64G using
larger memory regions in sparsemem_32.h:

 #ifdef CONFIG_X86_PAE
 #define SECTION_SIZE_BITS       30
 #define MAX_PHYSADDR_BITS       36
 #define MAX_PHYSMEM_BITS        36
 #else
 #define SECTION_SIZE_BITS       26
 #define MAX_PHYSADDR_BITS       32
 #define MAX_PHYSMEM_BITS        32
 #endif

which creates 1GB sparsemem regions instead of 64MB sparsemem regions.
So in practice we only ever created true sparsemem holes on x86 with
HIGHMEM4G - but that was rarely used by distros.

( btw., we could probably save 2MB of mem_map[]s on X86_PAE if we reduced
  the sparsemem region size to 256 MB. )

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
16 years ago[POWERPC] Fix boot failure on POWER6
Paul Mackerras [Tue, 15 Jan 2008 06:29:33 +0000 (17:29 +1100)]
[POWERPC] Fix boot failure on POWER6

Commit 473980a99316c0e788bca50996375a2815124ce1 added a call to clear
the SLB shadow buffer before registering it.  Unfortunately this means
that we clear out the entries that slb_initialize has previously set in
there.  On POWER6, the hypervisor uses the SLB shadow buffer when doing
partition switches, and that means that after the next partition switch,
each non-boot CPU has no SLB entries to map the kernel text and data,
which causes it to crash.

This fixes it by reverting most of 473980a9 and instead clearing the
3rd entry explicitly in slb_initialize.  This fixes the problem that
473980a9 was trying to solve, but without breaking POWER6.

Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Tue, 15 Jan 2008 05:26:08 +0000 (21:26 -0800)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: relocate sdev->manage_start_stop configuration
  sata_sil24: freeze on non-dev errors reported via CERR
  sata_sil24: fix stupid typo
  ata_piix: ignore ATA_DMA_ERR on vmware ich4

16 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 15 Jan 2008 05:25:39 +0000 (21:25 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] Cacheops.h: Fix typo.
  [MIPS] Cobalt: Qube1 has no serial port so don't use it
  [MIPS] Cobalt: Fix ethernet interrupts for RaQ1
  [MIPS] Kconfig fixes for BCM47XX platform

16 years agoRevert "writeback: introduce writeback_control.more_io to indicate more io"
Linus Torvalds [Tue, 15 Jan 2008 05:21:29 +0000 (21:21 -0800)]
Revert "writeback: introduce writeback_control.more_io to indicate more io"

This reverts commit 2e6883bdf49abd0e7f0d9b6297fc3be7ebb2250b, as
requested by Fengguang Wu.  It's not quite fully baked yet, and while
there are patches around to fix the problems it caused, they should get
more testing.  Says Fengguang: "I'll resend them both for -mm later on,
in a more complete patchset".

See

http://bugzilla.kernel.org/show_bug.cgi?id=9738

for some of this discussion.

Requested-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>