[PATCH] libata: kill sht->max_sectors

The previous dev->max_sectors patch made sht->max_sectors meaningless.
Kill all initializations of sht->max_sectors.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: add per-device max_sectors

If a low level driver wants to control max_sectors, it had to adjust
ap->host->max_sectors and set ATA_DFLAG_LOCK_SECTORS to tell
ata_scsi_slave_config not to override the limit.  This is not only
cumbersome but also incorrect for hosts which support more than one
devices per port.

This patch adds per-device ->max_sectors.  If the field is unset
(zero), libata core layer will adjust ->max_sectors according to
default rules.  If the field is set, libata honors the setting.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: move cdb_len for host to device

cdb_len is per-device property.  Sharing cdb_len on ap results in
inaccurate configuration on revalidation and hotplugging.  This patch
makes cdb_len per-device.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: make ata_dev_knobble() per-device

ata_dev_knobble() unconditionally used the first device of the port to
determine whether a device is bridged or not.  This causes bridge
limit to be incorrectly applied or unapplied for hosts with slave
devices (e.g. ata_piix).

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: don't do EDD handling if ->probe_reset is used

EDD is never used with ->probe_reset.  Don't handle EDD special case
in ata_dev_identify if ->probe_reset is in use.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: make ata_dump_id() take @id instead of @dev

Make ata_dump_id() take @id instead of @dev.  This is preparation for
splitting ata_dev_identify().

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: separate out ata_id_major_version()

Separate out ATA major version calculation from ata_dev_identify()
into ata_id_major_version().  It's preparation for splitting
ata_dev_identify().

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: separate out ata_id_n_sectors()

Separate out n_sectors calculation into ata_id_n_sectors() from
ata_dev_identify().  This will be used by revalidation.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: use ata_dev_id_c_string()

Use ata_dev_id_c_string()

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: implement ata_dev_id_c_string()

ata_dev_id_c_string() reads ATA string from the specified offset of
the given IDENTIFY PAGE and puts it in the specified buffer in trimmed
and NULL-terminated form.  The caller must supply a buffer which is
one byte larger than the maximum size of the target ID string.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] ata_piix: kill spurious assignment in piix_sata_probe()

In piix_sata_probe(), mask gets assigned unnecessarily at the
beginning of the function.  Kill the assignment.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: allow ->probe_reset to return ATA_DEV_UNKNOWN

This patch makes ata_bus_probe() normalize classes[] returned by
->probe_reset such that ->probe_reset can return ATA_DEV_UNKNOWN.
This eases implementation of ->probe_reset's which don't directly use
ata_drive_probe_reset().

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: kill assert() macro

libata assert() now has no user left.  Kill it.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: convert assert(xxx)'s in low-level drivers to WARN_ON(!xxx)'s

This patch converts all assert(xxx)'s in low-level drivers to
WARN_ON(!xxx)'s.  After this patch, there is no in-kernel user of the
libata assert() macro.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: convert assert(X)'s in libata core layer to WARN_ON(!X)'s

In an effort to kill libata-specific assert() and use generic
WARN_ON(), this patch converts all assert(X)'s in libata core layer to
WARN_ON(!X)'s.  Most conversions are straight-forward logical negation
exception for the followings.

* In libata-core.c:ata_fill_sg(),
  assert(qc->n_elem > 0) is converted to WARN_ON(qc->n_elem == 0) because
  qc->n_elem is unsigned and unsigned <= 0 is weird.

* In libata-scsi.c:ata_gen_ata_desc/fixed_sense(),
  assert(NULL != qc->ap->ops->tf_read) is converted to
  WARN_ON(qc->ap->ops->tf_read == NULL), as there are no other users of
  'constant cond var' style in libata.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] ahci: convert to new reset mechanism

Convert ahci ->phy_reset to new reset mechanism.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] ahci: make ahci_fill_cmd_slot() take *pp instead of *ap

Make ahci_fill_cmd_slot() take struct ahci_port_priv *pp instead of
struct ata_port *ap as suggested by Jeff Garzik.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: inline ata_qc_complete()

This patch inlines ata_qc_complete() and uninlines __ata_qc_complete()
as suggested by Jeff Garzik.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

Merge branch 'master'

Merge master.kernel.org:/home/rmk/linux-2.6-serial

Merge master.kernel.org:/home/rmk/linux-2.6-arm

Merge git://git./linux/kernel/git/paulus/powerpc-merge

[ARM] 3326/1: H1940 - Control latches

Patch from Ben Dooks

Define the bits for the two board control latches
that control various items on the H1940 iPAQ.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[SERIAL] Remove incorrect code from ioc4 serial driver

Serial drivers in general should not write uart_info->flags - they're
private to serial_core.  Serial drivers have no need to fiddle with
tty->alt_speed, nor manipulate TTY_IO_ERROR in tty->flags.  Fix the
ioc4 serial driver for both these points by simply removing the
offending code.

Acked-by: pfg@sgi.com
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[PATCH] Fix s390 build failure.

arch/s390/kernel/compat_signal.c:199: error: conflicting types for 'do_sigaction'
include/linux/sched.h:1115: error: previous declaration of 'do_sigaction' was here

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Updates for page migration

This adds some additional comments in order to help others figure out how
exactly the code works.  And fix a variable name.

Also swap_page does need to ignore all reference bits when unmapping a
page.  Otherwise we may have to repeatedly unmap a frequently touched page.
So change the try_to_unmap parameter to 1.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] tty buffering stall fix

Prevent stalled processing of received data when a driver allocates tty
buffer space but does not immediately follow the allocation with more data
and a call to schedule receive tty processing.  (example: hvc_console) This
bug was introduced by the first locking patch for the new tty buffering.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] slab: Avoid deadlock at kmem_cache_create/kmem_cache_destroy

Prevents deadlock situation between
kmem_cache_create()/kmem_cache_destory(), and kmem_cache_create() /cpu
hotplug.  The locking order probably got moved over time.

Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Signed-off-by: Shai Fultheim <shai@scalex86.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] shmdt cannot detach not-alined shm segment cleanly.

sys_shmdt() can manage shm segments which are covered by multiple vmas.  (This
can happen when a user uses mprotect() after shmat().)

This works well if shm is aligned to PAGE_SIZE, but if not, the last
segment cannot be detached.  It is because a comparison in sys_shmdt()

(vma->vm_end - addr) < size
addr == return address of shmat()
size == shmsize, argments to shmget()

size should be aligned to PAGE_SIZE before being compared with vma->vm_end,
which is aligned.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Hugh Dickins <hugh@veritas.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] prevent recursive panic from softlockup watchdog

When panic_timeout is zero, suppress triggering a nested panic due to soft
lockup detection.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86: don't initialise cpu_possible_map to all ones

Initialising cpu_possible_map to all-ones with CONFIG_HOTPLUG_CPU means that

a) All for_each_cpu() loops will iterate across all NR_CPUS CPUs, rather
   than over possible ones.  That can be quite expensive.

b) Soon we'll be allocating per-cpu areas only for possible CPUs.  So with
   CPU_MASK_ALL, we'll be wasting memory.

I also switched voyager over to not use CPU_MASK_ALL in the non-CPU-hotplug
case.  Should be OK..

I note that parisc is also using CPU_MASK_ALL.  Suggest that it stop doing
that.

Cc: James Bottomley <James.Bottomley@steeleye.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Paul Jackson <pj@sgi.com>
Cc: Ashok Raj <ashok.raj@intel.com>
Cc: Zwane Mwaikambo <zwane@linuxpower.ca>
Cc: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fix building external modules on ppc32

We are setting up sources for building external modules like this:

  /usr/src/linux-obj> # create a .config file
  /usr/src/linux-obj> make -C /usr/src/linux O=$PWD oldconfig
  /usr/src/linux-obj> make -C /usr/src/linux O=$PWD prepare
  /usr/src/linux-obj> make -C /usr/src/linux O=$PWD scripts
  /usr/src/linux-obj> make -C /usr/src/linux O=$PWD clean

After that, external modules can be built with:

  /usr/src/module> make -C /usr/src/linux-obj M=$PWD

This fails for ppc32 because the `make clean' removes the
arch/powerpc/include directory.  This should be done in archmrproper
instead of in archclean.

Signed-off-by: Andreas Gruenbacher <agruen@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] remove bogus comment from init/main.c

Remove bogus comment from init function which could lead to the assumption
that cpu_possible_map is setup in smp_prepare_cpus().

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] kexec: fix in free initrd when overlapped with crashkernel region

It is possible that the reserved crashkernel region can be overlapped with
initrd since the bootloader sets the initrd location.  When the initrd
region is freed, the second kernel memory will not be contiguous.  The
Kexec_load can cause an oops since there is no contiguous memory to write
the second kernel or this memory could be used in the first kernel itself
and may not be part of the dump.  For example, on powerpc, the initrd is
located at 36MB and the crashkernel starts at 32MB.  The kexec_load caused
panic since writing into non-allocated memory (after 36MB).  We could see
the similar issue even on other archs.

One possibility is to move the initrd outside of crashkernel region.  But,
the initrd region will be freed anyway before the system is up.  This patch
fixes this issue and frees only regions that are not part of crashkernel
memory in case overlaps.

Signed-off-by: Haren Myneni <haren@us.ibm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] wrong firmware location in IPW2100 Kconfig entry

Firmware should go into /lib/firmware, not /etc/firmware.

Found by Alejandro Bonilla.

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] sched: remove smpnice

I don't think the code is quite ready, which is why I asked for Peter's
additions to also be merged before I acked it (although it turned out that
it still isn't quite ready with his additions either).

Basically I have had similar observations to Suresh in that it does not
play nicely with the rest of the balancing infrastructure (and raised
similar concerns in my review).

The samples (group of 4) I got for "maximum recorded imbalance" on a 2x2
SMP+HT Xeon are as follows:

            | Following boot | hackbench 20        | hackbench 40
 -----------+----------------+---------------------+---------------------
 2.6.16-rc2 | 30,37,100,112  | 5600,5530,6020,6090 | 6390,7090,8760,8470
 +nosmpnice |  3, 2,  4,  2  |   28, 150, 294, 132 |  348, 348, 294, 347

Hackbench raw performance is down around 15% with smpnice (but that in
itself isn't a huge deal because it is just a benchmark).  However, the
samples show that the imbalance passed into move_tasks is increased by
about a factor of 10-30.  I think this would also go some way to explaining
latency blips turning up in the balancing code (though I haven't actually
measured that).

We'll probably have to revert this in the SUSE kernel.

Cc: "Siddha, Suresh B" <suresh.b.siddha@intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Peter Williams <pwil3058@bigpond.net.au>
Cc: "Martin J. Bligh" <mbligh@aracnet.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge /pub/scm/linux/kernel/git/davem/net-2.6

[PATCH] sata_sil24: add hardreset

Now that libata is smart enough to handle both soft and hard resets,
add hardreset method.  Note that sil24 hardreset doesn't supply
signature; still, the new reset mechanism can make good use of it.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sata_sil24: convert to new reset mechanism

Convert sata_sil24 ->phy_reset to new reset mechanism.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sata_sil: convert to new reset mechanism

Convert sata_sil to use new reset mechanism.  sata_sil is fairly
generic and can directly use std routine.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: kill SError clearing in sata_std_hardreset().

Don't clear SError in sata_std_hardreset().  This makes hardreset act
identically to ->phy_reset register-wise.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: make new reset act identical to ->phy_reset register-wise

This patch makes std component operations act identical to ->phy_reset
register-wise except for SError clearing on sata_std_hardreset.

Note that if a driver only implements/uses hardreset, it should not
use ata_std_probeinit() to avoid extra sata_phy_resume() and
ata_busy_sleep() compared to ->phy_reset.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

Merge branch 'master'

[PATCH] ahci: separate out ahci_fill_cmd_slot()

Separate out ahci_fill_cmd_slot() from ahci_qc_prep().
ahci_fill_cmd_slot() can later be used to issue non-standard commands.
(e.g. softreset)

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: kill NULL qc handling from ->eng_timeout callbacks

->eng_timeout cannot be invoked with NULL qc anymore.  Add an
assertion in ata_scsi_error() and kill NULL qc handling from all
->eng_timeout callbacks.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: use ata_scsi_timed_out()

Make all libata low level drivers use ata_scsi_timed_out().

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: implement ata_scsi_timed_out()

Implement ata_scsi_timed_out(), to be used as
scsi_host_template->eh_timed_out callback for all libata drivers.
Without this function, the following race exists.

If a qc completes after SCSI timer expires but before libata EH kicks
in, the qc gets completed but the scsicmd still gets passed to libata
EH resulting in ->eng_timeout invocation with NULL qc, which none is
handling properly.

This patch makes sure that scmd and qc share the same lifetime.
Original idea from Jeff Garzik <jgarzik@pobox.com>.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: add ATA_QCFLAG_EH_SCHEDULED

Add ATA_QCFLAG_EH_SCHEDULED.  If this flag is set, the qc is owned by
EH and normal completion path is not allowed to finish it.  This patch
doesn't actually use this flag.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] powerpc: Add CONFIG_DEFAULT_UIMAGE for embedded boards

Embedded boards that u-boot require a kernel image in the uImage format.
This allows a given board to specify it wants a uImage built by default.

This also fixes a warning at config time, as this symbol is referred
to in arch/powerpc/platforms/83xx/Kconfig.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] powerpc: Add FSL USB node to documentation

Updated the documentation to include the definition of the USB device
node format for Freescale SOC devices.

Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] powerpc: unshare system call registration

Registers system call for the powerpc architecture.

Signed-off-by: Janak Desai <janak@us.ibm.com>
Cc: Al Viro <viro@ftp.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

ppc: Use the system call table from arch/powerpc/kernel/systbl.S

With this, new system calls only have to be wired up in one place
for ARCH=ppc and ARCH=powerpc, rather than 2.

Signed-off-by: Paul Mackerras <paulus@samba.org>

[BRIDGE]: fix error handling for add interface to bridge

Refactor how the bridge code interacts with kobject system.
It should still use kobjects even if not using sysfs.
Fix the error unwind handling in br_add_if.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[BRIDGE]: netfilter handle RCU during removal

Bridge netfilter code needs to handle the case where device is
removed from bridge while packet in process. In these cases the
bridge_parent can become null while processing.

This should fix: http://bugzilla.kernel.org/show_bug.cgi?id=5803

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[BRIDGE]: fix for RCU and deadlock on device removal

Change Bridge receive path to correctly handle RCU removal of device
from bridge.  Also fixes deadlock between carrier_check and del_nbp.
This replaces the previous deleted flag fix.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

[TCP]: rcvbuf lock when tcp_moderate_rcvbuf enabled

The rcvbuf lock should probably be honored here.

Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: out of range array access

This patch fixes an out of range array access in irnet_irda.c.

Author: David Binderman <dcb314@hotmail.com>
Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: Set proper IrLAP device address length

This patch set IrDA's addr_len properly, i.e to 4 bytes, the size of the
IrLAP device address.

Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETLINK]: illegal use of pid in rtnetlink

When a netlink message is not related to a netlink socket,
it is issued by kernel socket with pid 0. Netlink "pid" has nothing
to do with current->pid. I called it incorrectly, if it was named "port",
the confusion would be avoided.

Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETLINK]: Fix a severe bug

netlink overrun was broken while improvement of netlink.
Destination socket is used in the place where it was meant to be source socket,
so that now overrun is never sent to user netlink sockets, when it should be,
and it even can be set on kernel socket, which results in complete deadlock
of rtnetlink.

Suggested fix is to restore status quo passing source socket as additional
argument to netlink_attachskb().

A little explanation: overrun is set on a socket, when it failed
to receive some message and sender of this messages does not or even
have no way to handle this error. This happens in two cases:
1. when kernel sends something. Kernel never retransmits and cannot
   wait for buffer space.
2. when user sends a broadcast and the message was not delivered
   to some recipients.

Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

Merge branch 'upstream-fixes' of /linux/kernel/git/jgarzik/libata-dev

[PATCH] do_sigaction: cleanup ->sa_mask manipulation

Clear unblockable signals beforehand.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] sys_signal: initialize ->sa_mask

Pointed out by Linus Torvalds.

sys_signal() forgets to initialize ->sa_mask.

( I suspect arch/ia64/ia32/ia32_signal.c:sys32_signal()
  also needs this fix )

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[SPARC64]: Fix sys_newfstatat syscall table entry for 64-bit.

The sparc64 64 bit syscall table seems to be broken as it has
compat_sys_newfstatat in its syscall table instead of sys_newfstatat.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull new-syscalls into release branch

[IA64] mca_drv: Add minstate validation

MCA driver can cause panic if kernel gets a state info with no minstate.
This patch adds minstate validation before handling it.

Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64] sys32_signal() forgets to initialize ->sa_mask

Pointed out by Oleg Nesterov <oleg@tv-sign.ru>, who in turn
got the hint from Linus.

Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] Small cleanup for misuse of list_for_each to list_for_each_safe.

Patch was suggested by Kenneth W. Chen here

Signed-off-by: Prarit Bhargava <prarit@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

[IA64-SGI] Hotplug driver related fix in the SN ia64 code.

Remove an erroneous kfree, and unlink the pcidev_info struct from the
pcidev_info list prior to free'ing the pcidev_info struct.

Signed-off-by: Prarit Bhargava <prarit@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

[ARM] 3314/1: S3C2400 - adds s3c2400.h

Patch from Lucas Correia Villa Real

This patch adds s3c2400.h, fixing the build for the 2410/2440
platforms.

Signed-off-by: Lucas Correia Villa Real <lucasvr@gobolinux.org>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

Revert "[PATCH] kconfig: detect if -lintl is needed when linking conf,mconf"

This reverts commit 5e375bc7d586e0df971734a5a5f1f080ffd89b68.

Kyle McMartin steps on his soap-box:

  "Sigh.  Can everyone please stop assuming gcc can output to /dev/null?
   On several platforms, ld tries to lseek in the output file, and fails
   if it can't."

Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[libata] Move PCI IDE BMDMA-related code to new file libata-bmdma.c.

Merge branch 'upstream-fixes'

Merge branch 'upstream-fixes'

[PATCH] libata: implement ata_std_probeinit()

This patch implements the off-the-shelf probeinit component operation.
Currently, all it does is waking up the PHY if it's a SATA port.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: add probeinit component operation to ata_drive_probe_reset()

This patch adds probeinit component operation to
ata_drive_probe_reset().  If present, this new operation is called
before performing any reset.  The operations's roll is to prepare @ap
for following probe-reset operations.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[libata scsi] build fix for ATA_FLAG_IN_EH change

[PATCH] libata: separate out sata_phy_resume() from sata_std_hardreset()

This patch separates out sata_phy_resume() from sata_std_hardreset().
The function will later be used by probeinit callback.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: fix ata_std_probe_reset() SATA detection

ap->cbl is initialized during postreset and thus unknown on entry to
ata_std_probe_reset().  This patch makes ata_std_probe_reset() use
ATA_FLAG_SATA flag instead of ap->cbl to detect SATA port.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: EH / pio tasks synchronization

This patch makes sure that pio tasks are flushed before proceeding
with EH.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] libata: implement ATA_FLAG_IN_EH port flag

ATA_FLAG_IN_EH flag is set on entry to EH and cleared on completion.
This patch just sets and clears the flag.  Following patches will
build normal qc execution / EH synchronization aroung this flag.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

Merge branch 'upstream-fixes'

Merge branch 'master'

Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

[IPV6]: Address autoconfiguration does not work after device down/up cycle

If you set network interface down and up again, the IPv6 address
autoconfiguration does not work. 'ip addr' shows that the link-local
address is in tentative state. We don't even react to periodical router
advertisements.

During NETDEV_DOWN we clear IF_READY, and we don't set it back in
NETDEV_UP. While starting to perform DAD on the link-local address, we
notice that the device is not in IF_READY, and we abort autoconfiguration
process (which would eventually send router solicitations).

Acked-by: Juha-Matti Tapio <jmtapio@verkkotelakka.net>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PATCH] x86-64: Add sys_unshare

Add unshare syscall for x86-64

ppoll/pselect are not ready yet, but add reservations.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[IA64] unshare system call registration for ia64

Registers system call for the ia64 architecture.

Reserves space for ppoll and pselect, and adds unshare at system
call number 1296.

Signed-off-by: Janak Desai <janak@us.ibm.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

Merge master.kernel.org:/home/rmk/linux-2.6-mmc

Merge master.kernel.org:/home/rmk/linux-2.6-serial

Merge master.kernel.org:/home/rmk/linux-2.6-arm

Manual conflict merge of arch/arm/Kconfig

[MMC] Remove extra character in AU1XXX MMC Kconfig entry

An obvious vi fat finger on my part.

Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[ARM] 3278/1: OMAP: 3/3 Fix low-level io init for omap2 boards

Patch from Tony Lindgren

This patch fixes the low-level IO init for omap2 boards.

Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[ARM] 3280/1: OMAP: 2/3 Fix low-level io init for omap1 boards

Patch from Tony Lindgren

This patch fixes the low-level IO init for omap1 boards.

Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[ARM] 3279/1: OMAP: 1/3 Fix low-level io init

Patch from Tony Lindgren

This patch adds the missing cache flushes to common low-level
init that are needed to access the IO region. These flushes
are normally done at the end of devicemaps_init(), but we
need to detect the OMAP core type early.

Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[ARM] 3299/1: S3C24XX - fix irq range on adc device

Patch from Ben Dooks

Change the IRQ resource range for the ADC device
to be two distinct IRQs

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[SERIAL] 8250_pci: add new PCI serial card support

This patch adds new PCI serial card support.

Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[SERIAL] ip22zilog: Whitespace cleanup.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[SERIAL] PPC32 CPM_UART: update to utilize the new TTY flip API

This replaces old direct usage of tty->flip stuff with relative flip API
calls.

Signed-off-by: Vitaly Bordug <vbordug@ru.mvista.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>