Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (73 commits)
  net: Fix typo in net/core/sock.c.
  ppp: Do not free not yet unregistered net device.
  netfilter: xt_iprange: module aliases for xt_iprange
  netfilter: ctnetlink: dump conntrack ID in event messages
  irda: Fix a misalign access issue. (v2)
  sctp: Fix use of uninitialized pointer
  cipso: Relax too much careful cipso hash function.
  tcp FRTO: work-around inorder receivers
  tcp FRTO: Fix fallback to conventional recovery
  New maintainer for Intel ethernet adapters
  DM9000: Use delayed work to update MII PHY state
  DM9000: Update and fix driver debugging messages
  DM9000: Add __devinit and __devexit attributes to probe and remove
  sky2: fix simple define thinko
  [netdrvr] sfc: sfc: Add self-test support
  [netdrvr] sfc: Increment rx_reset when reported as driver event
  [netdrvr] sfc: Remove unused macro EFX_XAUI_RETRAIN_MAX
  [netdrvr] sfc: Fix code formatting
  [netdrvr] sfc: Remove kernel-doc comments for removed members of struct efx_nic
  [netdrvr] sfc: Remove garbage from comment
  ...

fix SMP data race in pagetable setup vs walking

There is a possible data race in the page table walking code. After the split
ptlock patches, it actually seems to have been introduced to the core code, but
even before that I think it would have impacted some architectures (powerpc
and sparc64, at least, walk the page tables without taking locks eg. see
find_linux_pte()).

The race is as follows:
The pte page is allocated, zeroed, and its struct page gets its spinlock
initialized. The mm-wide ptl is then taken, and then the pte page is inserted
into the pagetables.

At this point, the spinlock is not guaranteed to have ordered the previous
stores to initialize the pte page with the subsequent store to put it in the
page tables. So another Linux page table walker might be walking down (without
any locks, because we have split-leaf-ptls), and find that new pte we've
inserted. It might try to take the spinlock before the store from the other
CPU initializes it. And subsequently it might read a pte_t out before stores
from the other CPU have cleared the memory.

There are also similar races in higher levels of the page tables. They
obviously don't involve the spinlock, but could see uninitialized memory.

Arch code and hardware pagetable walkers that walk the pagetables without
locks could see similar uninitialized memory problems, regardless of whether
split ptes are enabled or not.

I prefer to put the barriers in core code, because that's where the higher
level logic happens, but the page table accessors are per-arch, and open-coding
them everywhere I don't think is an option. I'll put the read-side barriers
in alpha arch code for now (other architectures perform data-dependent loads
in order).

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

read_barrier_depends arch fixlets

read_barrie_depends has always been a noop (not a compiler barrier) on all
architectures except SMP alpha. This brings UP alpha and frv into line with all
other architectures, and fixes incorrect documentation.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

net: Fix typo in net/core/sock.c.

In sock_queue_rcv_skb()  (net/core/sock.c) it should be:
"Cast sk->rcvbuf ..." instead of: "Cast skb->rcvbuf ..."

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ppp: Do not free not yet unregistered net device.

An error path in ppp_create_interface() lacks one and may
BUG in free_netdev() checking for proper dev->reg_state.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: xt_iprange: module aliases for xt_iprange

Using iptables 1.3.8 with kernel 2.6.25, rules which include '-m
iprange' don't automatically pull in xt_iprange module.  Below patch
adds module aliases to fix that.  Patch against latest -git, but seems
like a good candidate for -stable also.

Signed-off-by: Phil Oester <kernel@linuxace.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: ctnetlink: dump conntrack ID in event messages

Conntrack ID is not put (anymore ?) in event messages. This causes
current ulogd2 code to fail because it uses the ID to build a hash in
userspace. This hash is used to be able to output the starting time of
a connection.

Conntrack ID can be used in userspace application to maintain an easy
match between kernel connections list and userspace one. It may worth
to add it if there is no performance related issue.

[ Patrick: it was never included in events, but really should be ]

Signed-off-by: Eric Leblond <eric@inl.fr>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

irda: Fix a misalign access issue. (v2)

Replace u16ho with put/get_unaligned functions

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

sctp: Fix use of uninitialized pointer

Introduced by c4492586 (sctp: Add address type check while process
paramaters of ASCONF chunk):

net/sctp/sm_make_chunk.c: In function 'sctp_process_asconf':
net/sctp/sm_make_chunk.c:2828: warning: 'addr_param' may be used uninitialized in this function
net/sctp/sm_make_chunk.c:2828: note: 'addr_param' was declared here

Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cipso: Relax too much careful cipso hash function.

The cipso_v4_cache is allocated to contain CIPSO_V4_CACHE_BUCKETS
buckets. The CIPSO_V4_CACHE_BUCKETS = 1 << CIPSO_V4_CACHE_BUCKETBITS,
where CIPSO_V4_CACHE_BUCKETBITS = 7.

The bucket-selection function for this hash is calculated like this:

  bkt = hash & (CIPSO_V4_CACHE_BUCKETBITS - 1);
                                     ^^^

i.e. picking only 4 buckets of possible 128 :)

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Paul Moore <paul.moore@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: user_regset_view table fix for ia32 on 64-bit
  x86: arch/x86/mm/pat.c - fix warning
  x86: fix csum_partial() export
  x86: early_init_centaur(): use set_cpu_cap()
  x86: fix app crashes after SMP resume
  x86: wakeup.lds.S - section ordering fix
  x86: [VOYAGER] fix duplicate phys_cpu_present_map symbol
  x86/pci: fix broken ISA DMA

Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] qla1280: Fix queue depth problem
  [SCSI] aha152x: Fix oops on module removal
  [SCSI] aha152x: fix init suspiciously returned 1, it should follow 0/-E convention
  [SCSI] libiscsi regression in 2.6.25: fix setting of recv timer
  [SCSI] libiscsi regression in 2.6.25: fix nop timer handling
  [SCSI] gdth: fix Error: Driver 'gdth' is already registered, aborting...
  [SCSI] gdth: fix timer handling

Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  ACPI/PCI: another multiple _OSC memory leak fix
  x86/PCI: X86_PAT & mprotect
  PCI: enable nv_msi_ht_cap_quirk for ALi bridges
  PCI: Make the intel-iommu_wait_op macro work when jiffies are not running
  ACPI/PCI: handle multiple _OSC
  ACPI/PCI: handle multiple _OSC
  x86/PCI: fix broken ISA DMA
  PCI ACPI: fix uninitialized variable in __pci_osc_support_set

x86: user_regset_view table fix for ia32 on 64-bit

The user_regset_view table for the 32-bit regsets on the 64-bit build had
the wrong sizes for the FP regsets.  This bug had no user-visible effect
(just on kernel modules using the user_regset interfaces and the like).
But the fix is trivial and risk-free.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

x86: arch/x86/mm/pat.c - fix warning

fix this warning:

 arch/x86/mm/pat.c: In function `phys_mem_access_prot_allowed':
 arch/x86/mm/pat.c:558: warning: long long unsigned int format, long
 unsigned int arg (arg 6)
 arch/x86/mm/pat.c: In function `map_devmem':
 arch/x86/mm/pat.c:580: warning: long long unsigned int format, long
 unsigned int arg (arg 6)

Signed-off-by: D Pranith Kumar <bobby.prani@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

x86: fix csum_partial() export

Fix this symbol export problem:

    Building modules, stage 2.
    MODPOST 193 modules
    ERROR: "csum_partial" [fs/reiserfs/reiserfs.ko] undefined!
    make[1]: *** [__modpost] Error 1
    make: *** [modules] Error 2

This is due to a known weakness of symbol exports: if a symbol's
only in-core user is an EXPORT_SYMBOL from a lib-y section, the
symbol is not linked in.

The solution is to move the export to x8664_ksyms_64.c - but the real
solution would be to fix kbuild.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

x86: early_init_centaur(): use set_cpu_cap()

arch/x86/kernel/setup_64.c:954: warning: passing argument 2 of 'set_bit' from incompatible pointer type

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

x86: fix app crashes after SMP resume

After resume on a 2cpu laptop, kernel builds collapse with a sed hang,
sh or make segfault (often on 20295564), real-time signal to cc1 etc.

Several hurdles to jump, but a manually-assisted bisect led to -rc1's
d2bcbad5f3ad38a1c09861bca7e252dde7bb8259 x86: do not zap_low_mappings
in __smp_prepare_cpus.  Though the low mappings were removed at bootup,
they were left behind (with Global flags helping to keep them in TLB)
after resume or cpu online, causing the crashes seen.

Reinstate zap_low_mappings (with local __flush_tlb_all) for each cpu_up
on x86_32.  This used to be serialized by smp_commenced_mask: that's now
gone, but a low_mappings flag will do.  No need for native_smp_cpus_done
to repeat the zap: let mem_init zap BSP's low mappings just like on UP.

(In passing, fix error code from native_cpu_up: do_boot_cpu returns a
variety of diagnostic values, Dprintk what it says but convert to -EIO.
And save_pg_dir separately before zap_low_mappings: doesn't matter now,
but zapping twice in succession wiped out resume's swsusp_pg_dir.)

That worked well on the duo and one quad, but wouldn't boot 3rd or 4th
cpu on P4 Xeon, oopsing just after unlock_ipi_call_lock.  The TLB flush
IPI now being sent reveals a long-standing bug: the booting cpu has its
APIC readied in smp_callin at the top of start_secondary, but isn't put
into the cpu_online_map until just before that unlock_ipi_call_lock.

So native_smp_call_function_mask to online cpus would send_IPI_allbutself,
including the cpu just coming up, though it has been excluded from the
count to wait for: by the time it handles the IPI, the call data on
native_smp_call_function_mask's stack may well have been overwritten.

So fall back to send_IPI_mask while cpu_online_map does not match
cpu_callout_map: perhaps there's a better APICological fix to be
made at the start_secondary end, but I wouldn't know that.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[SCSI] qla1280: Fix queue depth problem

The qla1280 driver was ANDing the output value of mailbox register
0 with (1 << target-number) to determine whether to enable queueing
on the target in question.

But mailbox register 0 has the status code for the mailbox command
(in this case, Set Target Parameters).  Potential values are:
/*
 * ISP mailbox command complete status codes
 */

So clearly that is in error.  I can't think what the author of that
line was looking for in a mailbox register, so I just eliminated the
AND.  flag is used later in the function, and I think that the later
usage was also wrong, though it was used to set values that aren't
used.  Oh well, an overhaul of this driver is not what I want to do
now -- just a bugfix.

After the fix, I found that my disks were getting a queue depth of
255, which is far too many.  Most SCSI disks are limited to 32 or
64.  In any case, there's no point, queueing up a bunch of commands
to the adapter that will just result in queue full or starve other
targets from being issued commands due to running out of internal
memory.  So I dropped default queue depth to 32 (from which 1 is
subtracted elsewhere, giving net of 31).

I tested with a Seagate ST336753LC, and results look good, so
I'm satisfied with this patch.

Signed-off-by: Jeremy Higdon <jeremy@sgi.com>
Acked-by: Jes Sorensen <jes@sgi.com>
Cc: Stable Tree <stable@kernel.org>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>

ACPI/PCI: another multiple _OSC memory leak fix

The acpi_query_osc() function can be called for the ACPI object that
doesn't have _OSC method. In this case, acpi_get_osc_data() would
allocate a useless memory region. To avoid this, we need to check the
existence of _OSC before calling acpi_get_osc_data() in acpi_query_osc().

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

x86/PCI: X86_PAT & mprotect

Some versions of X used the mprotect workaround to change caching type from UC
to WB, so that it can then use mtrr to program WC for that region [1].  Change
the mmap of pci space through /sys or /proc interfaces from UC to UC_MINUS.
With this change, X will not need to use mprotect workaround to get WC type
since the MTRR mapping type will be honored.

The bug in mprotect that clobbers PAT bits is fixed in a follow on patch. So,
this X workaround will stop working as well.

Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

PCI: enable nv_msi_ht_cap_quirk for ALi bridges

This applies the NVidia MSI enabled flag for HT capable devices quirk
to ALi bridges as well.

As described in more detail in http://bugzilla.kernel.org/show_bug.cgi?id=10667
this is required for my board which is using an nForce 3 250Gb chipset with an
ALi M1695 northbridge.

It fixes a regression introduced in 2.6.24 that made the internal NIC of the
board unusable (MSI initialisation of the NIC but disabled MSI on the
northbridge devices.

Signed-off-by: Björn Krombholz <fox.box@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

PCI: Make the intel-iommu_wait_op macro work when jiffies are not running

The following patch changes the intel-iommu.c code to use the TSC
instead of jiffies for detecting bad DMAR functionality.  Some systems
with bad bios's have been seen to hang in early boot spinning in the
IOMMU_WAIT_IO macro.  This patch will replace the infinite loop with a call to
panic.

Signed-off-by: Mark Gross <mgross@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

ACPI/PCI: handle multiple _OSC

The pci_osc_control_set() function can be called for the ACPI object
that doesn't have _OSC method. In this case, acpi_get_osc_data() would
allocate a useless memory region. To avoid this, we need to check the
existence of _OSC before calling acpi_get_osc_data(). Here is a patch
to fix this problem in pci_osc_control_set.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

ACPI/PCI: handle multiple _OSC

There is an IA64 system here which have two pci root bridges with _OSC.
One _OSC disables SHPC control bit but the other not. Below patch makes
_OSC data per-device instead of one global, otherwise linux takes both
root bridges don't support SHPC.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

x86/PCI: fix broken ISA DMA

Rene Herman reported:

> commit 8779f2fc3b84ebb6c5181fb13d702e9944c16069
>
> "x86: don't try to allocate from DMA zone at first"
>
> breaks all of ISA DMA. Or all of ALSA ISA DMA at least. All
> ISA soundcards are silent following that commit -- no error
> messages, everything appears fine, just silence.

That patch is buggy. We had an implicit assumption that
dev = NULL for ISA devices that require 24bit DMA.

The recent work on x86 dma_alloc_coherent() breaks the ISA DMA buffer
allocation, which is represented by "dev = NULL" and requires 24bit
DMA implicitly.

Bisected-by: Rene Herman <rene.herman@keyaccess.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

PCI ACPI: fix uninitialized variable in __pci_osc_support_set

Fix uninitialized variable in __pci_osc_support_set().

If the ACPI namespace doesn't have any device object corresponding to
the specified hid, 'retval' in __pci_osc_support_set() is not changed
by the acpi_query_osc() callback. Since 'retval' is not initizlized in
the current implementation, the contents of 'retval' is undefined in
this case. This causes a mis-handling of ctrlset_buf[OSC_SUPPORT_TYPE]
and will cause an unexpected result in the subsequent
pci_osc_control_set() call as a result.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  [ALSA] ASoC: Fix wrong enum count for jack_function in N810 machine driver
  [ALSA] ASoC: build fix for snd_soc_info_bool_ext
  [ALSA] ASoC: Fix TLV320AIC3X mono line output interconnect
  [ALSA] soc - fsl_ssi.c fix "BUG: scheduling while atomic"
  [ALSA] emux midi synthesizer doesn't honor SOFT_PEDAL-release event

eCryptFS: fix imbalanced mutex locking

Fix imbalanced calls for mutex lock/unlock on ecryptfs_daemon_hash_mux
Revealed by Ingo Molnar: http://lkml.org/lkml/2008/5/7/260

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

vt/fbcon: fix background color on line feed

Another addendum to commit c9e587abfdec2c2aaa55fab83bcb4972e2f84f9b
("vt: fix background color on line feed").

fbcon still was not doing the right thing (read: continued to do old
behavior).  fbcon_clear() seems to clear the new line (e.g.  where your new
prompt appears after doing echo -en "\e[42mfoo\n"), while scr_memsetw clears
the previous one only (where "foo" appears).  So just temporarily set the
video_erase_char to the scrl_erase_char so that fbcon_clear does the right
thing.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

vt/fbcon: update scrl_erase_char after 256/512-glyph font switch

Addendum to commit c9e587abfdec2c2aaa55fab83bcb4972e2f84f9b ("vt: fix
background color on line feed").

vc->vc_scrl_erase_char was not updated when fbcon switches between
256- and 512-glyph fonts.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

rtc: m41t80: include <linux/kernel.h> for printk()

The driver uses printk(), but does not include <linux/kernel.h> -- add it.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Alexander Bigga <ab@mycable.de>
Cc: Wim Van Sebroeck <wim@iguana.be>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

RTC/watchdog: M41T80: fix a potential use of unitialized data

Watchdog handlers within the driver make use of "save_client" -- make sure it
has been initalized before the handlers are registered.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Alexander Bigga <ab@mycable.de>
Cc: Wim Van Sebroeck <wim@iguana.be>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/char/synclink_gt.c: don't return an uninitialised local

drivers/char/synclink_gt.c: In function 'put_char':
drivers/char/synclink_gt.c:919: warning: 'ret' may be used uninitialized in this function

The compiler speaketh truth.

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

Fix misuses of bdevname()

bdevname() fills the buffer that it is given as a parameter, so calling
strcpy() or snprintf() on the returned value is redundant (and probably not
guaranteed to work - I don't think strcpy and snprintf support overlapping
buffers.)

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Stephen Tweedie <sct@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fuse: add flag to turn on big writes

Prior to 2.6.26 fuse only supported single page write requests.  In theory all
fuse filesystem should be able support bigger than 4k writes, as there's
nothing in the API to prevent it.  Unfortunately there's a known case in
NTFS-3G where big writes cause filesystem corruption.  There could also be
other filesystems, where the lack of testing with big write requests would
result in bugs.

To prevent such problems on a kernel upgrade, disable big writes by default,
but let filesystems set a flag to turn it on.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Cc: Szabolcs Szakacsits <szaka@ntfs-3g.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: fix documentation

Correct the cgroups documentation to reflect the correct file names.

Signed-off-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Cc: Sudhir Kumar <skumar@linux.vnet.ibm.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: fix possible panic when CONFIG_MM_OWNER=y

When mm destruction happens, we should pass mm_update_next_owner() the old mm.
 But unfortunately new mm is passed in exec_mmap().

Thus, kernel panic is possible when a multi-threaded process uses exec().

Also, the owner member comment description is wrong.  mm->owner does not
necessarily point to the thread group leader.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: "Paul Menage" <menage@google.com>
Cc: "KAMEZAWA Hiroyuki" <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

spi_mpc83xx: much improved driver

The current driver may cause glitches on SPI CLK line since one must disable
the SPI controller before changing any HW settings.  Fix this by implementing
a local spi_transfer function that won't change speed and/or word size while
CS is active.

While doing that heavy lifting a few other issues were addressed too:
 - Make word size 16 and 32 work too.
 - Honor bits_per_word and speed_hz in spi transaction.
 - Optimize the common path.

This also stops using the "bitbang" framework (except for a few constants).

[Roel Kluin <12o3l@tiscali.nl>: "irq" needs to be signed]
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cpumask: remove bitmap_scnprintf_len and cpumask_scnprintf_len

They aren't used.  They were briefly used as part of some other patches to
provide an alternative format for displaying some /proc and /sys cpumasks.
They probably should have been removed when those other patches were dropped,
in favor of a different solution.

Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: "Mike Travis" <travis@sgi.com>
Cc: "Bert Wesarg" <bert.wesarg@googlemail.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: do not let CONFIG_FB_DEFERRED_IO default to y

CONFIG_FB_DEFERRED_IO can not be turned off, while it's already selected
automatically by the drivers that need it.

Although it's nice to have more compile-coverage, not being able to disable a
rarely used feature is annoying.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Jaya Kumar <jayakumar.lkml@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

atmel_lcdfb: fix pixclock divider calculation

Fix divider calculation and allow CLKVAL = 0 (divisor 2)

It was not possible to get the clock value 0 (divisor 2) because
the test "<=0" force the BYPASS bit to be activated instead.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Per Hedblom <per.hedblom@abem.se>
Cc: Roel Kluin <12o3l@tiscali.nl>
Cc: Jan Weber <jw022609@uni-greifswald.de>
Cc: Andrew Victor <linux@maxim.org.za>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

rtc: rtc_time_to_tm: use unsigned arithmetic

The input argument to rtc_time_to_tm() is unsigned as well as are members of
the output structure.  However signed arithmetic is used within for
calculations leading to incorrect results for input values outside the signed
positive range.  If this happens the time of day returned is out of range.

Found the problem when fiddling with the RTC and the driver where year was set
to an unexpectedly large value like 2070, e.g.:

rtc0: setting system clock to 2070-01-01 1193046:71582832:26 UTC (3155760954)

while it should be:

rtc0: setting system clock to 2070-01-01 00:15:54 UTC (3155760954)

Changing types to unsigned fixes the problem.

[akpm@linux-foundation.org: remove old-fashioned `register' keyword]
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Cc: Dmitri Vorobiev <dmitri.vorobiev@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68knommu: missing sections for linker script

Include the missing kcrctab and kcrctab_unused sections into the m68knommu
linker script.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init: don't lose initcall return values

There is an ability to lose an initcall return value if it happened with irq
disabled or imbalanced preemption (and if we debug initcall).

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Update Geode mailing list in MAINTAINERS

Update the Geode list location in the MAINTAINERS file.

Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Fix hfsplus oops on image without extents

Fix an oops with a corrupted hfs+ image.

See http://bugzilla.kernel.org/show_bug.cgi?id=10548 for details.

Problem is that we call hfs_btree_open() from hfsplus_fill_super() to set
HFSPLUS_SB(sb).[ext_tree|cat_tree] Both trees are still NULL at this moment.
If hfs_btree_open() fails for any reason it calls iput() on the page, which
gets to hfsplus_releasepage() which tries to access HFSPLUS_SB(sb).* which is
still NULL and oopses while dereferencing it.

[akpm@linux-foundation.org: build fix]
Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

parisc: fix DISCONTIGMEM compile breakage

PA-RISC to aid debugging prints out the zonelists setup by the system.  A
bad call to node_zonelist() breaks at compile-time.  This patch fixes it.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Grant Grundler <grundler@parisc-linux.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

PNP: set IRQ index in sysfs "set irq" interface

We have to set the ISAPNP register index when setting an IRQ via the sysfs
interface.  We already do it for IO, MEM, and DMA resources; I just missed the
IRQ one.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mn10300: replace deprecated "TOPDIR" with newer "srctree"

This would appear to be the last reference to TOPDIR in the entire tree, after
which i'm guessing that variable can be dropped.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Acked-by: David Howells <dhowells@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/video/pnx4008: eliminate double free

The function framebuffer_release just calls kfree, so calling kfree
subsequently on the same argument represents a double free.  The comments with
the definition of framebuffer_release in drivers/video/fbsysfs.c suggest that
a more elaborate definition of this function is planned, such that the
splitting up of framebuffer_release and kfree as done in the second instance
might someday make sense, but it does not make sense now.

This was found using the following semantic match.
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@@
expression E;
@@

* kfree(E);
...
* framebuffer_release(E);

@@
expression E;
@@

* framebuffer_release(E);
...
* kfree(E);
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Vitaly Wool <vitalywool@gmail.com>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Grigory Tolstolytkin <gtolstolytkin@ru.mvista.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/pdflush.c: merge the same code in two path

Signed-off-by: Denis Cheng <crquan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

adt7473: minor documentation update

Add a sentence about when fan speed increases to maximum.

Signed-off-by: Darrick J. Wong <djwong@us.ibm.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

capabilities: add bounding set to /proc/self/status

There is currently no way to query the bounding set of another task.  As there
appears to be no security reason not to, and as Michael Kerrisk points out the
following valid reasons to do so exist:

* consistency (I can see all of the other per-thread/process sets in
  /proc/.../status)

* debugging -- I could imagine that it would make the job of debugging an
  application that uses capabilities a little simpler.

this patch adds the bounding set to /proc/self/status right after the
effective set.

Signed-off-by: Serge E. Hallyn <serue@us.ibm.com>
Acked-by: Michael Kerrisk <mtk.manpages@gmail.com>
Acked-by: Andrew G. Morgan <morgan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

md: fix raid5 'repair' operations

commit bd2ab67030e9116f1e4aae1289220255412b37fd "md: close a livelock window
in handle_parity_checks5" introduced a bug in handling 'repair' operations.
After a repair operation completes we clear the state bits tracking this
operation.  However, they are cleared too early and this results in the code
deciding to re-run the parity check operation.  Since we have done the repair
in memory the second check does not find a mismatch and thus does not do a
writeback.

Test results:
$ echo repair > /sys/block/md0/md/sync_action
$ cat /sys/block/md0/md/mismatch_cnt
51072
$ echo repair > /sys/block/md0/md/sync_action
$ cat /sys/block/md0/md/mismatch_cnt
0

(also fix incorrect indentation)

Cc: <stable@kernel.org>
Tested-by: George Spelvin <linux@horizon.com>
Acked-by: NeilBrown <neilb@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

tridentfb: improve clock setting accuracy

Improve clock calculation precision (to kHz from MHz) and removes parameter
field vclk from the tridentfb_par structure.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

tridentfb: remove misplaced enable_mmio()

Remove redundant enable_mmio() call as the mmio mode is enabled in the probe
function earlier.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mmc: make one-bit signed bitfields unsigned

Otherwise it can only take the values 0/-1 which doesn't seem to
have been intended.

drivers/mmc/host/sdhci.h:190:20: error: dubious one-bit signed bitfield

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Acked-by: Pierre Ossman <drzeus-list@drzeus.cx>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

quota: don't call sync_fs() from vfs_quota_off() when there's no quota turn off

Sometimes, vfs_quota_off() is called on a partially set up super block (for
example when fill_super() fails for some reason).  In such cases we cannot
call ->sync_fs() because it can Oops because of not properly filled in super
block.  So in case we find there's not quota to turn off, we just skip
everything and return which fixes the above problem.

[akpm@linux-foundation.org: fxi tpyo]
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ufs: remove unneeded ufs_put_inode prototype

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

make vmstat cpu-unplug safe

When accessing cpu_online_map, we should prevent dynamic changing
of cpu_online_map by get_online_cpus().

Unfortunately, all_vm_events() doesn't do that.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68knommu: ColdFire add support for kernel preemption (missing chunk)

As the subject says this patch adds the support for kernel preemption on
m68knommu Coldfire.  I thing the same changes could be applied to 68360 &
68328 but since I don't have the HW, I don't touch it.  Kconfig enables the
preemption item only on coldfire.

This is a missing chunk from Sebastian's original patch that I lost from the
first submission.

Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ecryptfs: clean up (un)lock_parent

dget(dentry->d_parent) --> dget_parent(dentry)

unlock_parent() is racy and unnecessary.  Replace single caller with
unlock_dir().

There are several other suspect uses of ->d_parent in ecryptfs...

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Cc: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/misc/sgi-xp: replace partid_t with a short

In preparation for supporting greater than 64 partitions replace partid_t by
short in drivers/misc/sgi-xp.

Signed-off-by: Dean Nelson <dcn@sgi.com>
Acked-by: Robin Holt <holt@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/misc/sgi-xp: clean up return values

Make XP return values more generic to XP and not so tied to XPC by changing
enum xpc_retval to xp_retval, along with changing return value prefixes from
xpc to xp.  Also, cleanup a comment block that referenced some of these return
values as well as the handling of BTE related return values.

Signed-off-by: Dean Nelson <dcn@sgi.com>
Acked-by: Robin Holt <holt@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

spi: pxa2xx_spi clock resume bugfix

There is a typo in pxa2xx_spi.c, comment says "Enable the SSP clock", code
says: clk_disable ...  so after resume, the SSP is dead.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Ned Forrester <nforrester@whoi.edu>
Cc: Stephen Street <stephen@streetfiresound.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: track and make up lost ticks

Alarm delivery could be noticably late in the !CONFIG_NOHZ case because lost
ticks weren't being taken into account.  This is now treated more carefully,
with the time between ticks being calculated and the appropriate number of
ticks delivered to the timekeeping system.

Cc: Nix <nix@esperi.org.uk>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: style fixes in the random driver

Give random.c a style workover while I'm changing it.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: random driver fixes

The random driver would essentially hang if the host's /dev/random returned
-EAGAIN.  There was a test of need_resched followed by a schedule inside the
loop, but that didn't help and it's the wrong way to work anyway.

The right way is to ask for an interrupt when there is input available from
the host and handle it then rather than polling.

Now, when the host's /dev/random returns -EAGAIN, the driver asks for a wakeup
when there's randomness available again and sleeps.  The interrupt routine
just wakes up whatever processes are sleeping on host_read_wait.

There is an atomic_t, host_sleep_count, which counts the number of processes
waiting for randomness.  When this reaches zero, the interrupt is disabled.

An added complication is that async I/O notification was only recently added
to /dev/random (by me), so essentially all hosts will lack it.  So, we use the
sigio workaround here, which is to have a separate thread poll on the
descriptor and send an interrupt when there is input on it.  This mechanism is
activated when a process gets -EAGAIN (activating this multiple times is
harmless, if a bit wasteful) and deactivated by the last process still
waiting.

The module name was changed from "random" to "hw_random" in order for udev to
recognize it.

The sigio workaround needed some changes.  sigio_broken was added for cases
when we know that async notification doesn't work.  This is now called from
maybe_sigio_broken, which deals with pts devices.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: physical memory shouldn't include initial stack

The top of physical memory should be below the initial process stack, not the
top of the address space, at least for as long as the stack isn't known to the
kernel VM system and appropriately reserved.

Cc: "Christopher S. Aker" <caker@theshore.net>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: use PAGE_SIZE in linker scripts

This patch includes page.h header into linker scripts that allow us to
use PAGE_SIZE macro instead of numeric constant.

To be able to include page.h into linker scripts page.h is needed for
some modification - i.e.  we need to use __ASSEMBLY__ and _AC macro

[jdike@linux.intel.com - fixed conflict with as-layout.h]
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: use DIV_ROUND_UP

I just saw similar patches in the janitor kernel's list, and spotted place it
fits.

Signed-off-by: Jiri Olsa <olsajiri@gmail.com>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: fix CONFIG_RAW dependencies

Add the BLOCK dependency for RAW_DRIVER, to match what's in
drivers/char/Kconfig.  Also, while we're there, update the alleged
obsolesence of RAW_DRIVER since it doesn't seem to be going away any
time soon.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: use __SPIN_LOCK_UNLOCKED

From: Robert P. J. Day <rpjday@crashcourse.ca>

Use newer, non-deprecated __SPIN_LOCK_UNLOCKED macro.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: fix bad NTP interaction with clock

UML's supposed nanosecond clock interacts badly with NTP when NTP
decides that the clock has drifted ahead and needs to be slowed down.
Slowing down the clock is done by decrementing the cycle-to-nanosecond
multiplier, which is 1.  Decrementing that gives you 0 and time is
stopped.

This is fixed by switching to a microsecond clock, with a multiplier
of 1000.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: remove unused header

From: Adrian Bunk <bunk@kernel.org>

This patch removes the unused and broken (the normal asm/keyboard.h
files no longer exists) include/asm-um/keyboard.h

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: fix build when SLOB is enabled

Reintroduce uml_kmalloc for the benefit of UML libc code.  The
previous tactic of declaring __kmalloc so it could be called directly
from the libc side of the house turned out to be getting too intimate
with slab, and it doesn't work with slob.

So, the uml_kmalloc wrapper is back.  It calls kmalloc or whatever
that translates into, and libc code calls it.

kfree is left alone since that still works, leaving a somewhat
inconsistent API.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: fix errno return

Error returns are negative.

Signed-off-by: Johann Felix Soden <johfel@users.sourceforge.net>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: tidy ptrace interface

Tidy the ptrace interface code.  Removed a bunch of unused macros.
Started converting register sets from arrays of longs to structures.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: move hppfs_kern.c to hppfs.c

There's no reason for the _kern in hppfs_kern.c, so move it to hppfs.c.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: hppfs fixes

hppfs tidying and fixes noticed during hch's get_inode work -
      style fixes
      a copy_to_user got its return value checked
      hppfs_write no longer fiddles file->f_pos because it gets and
returns pos in its arguments
      hppfs_delete_inode dputs the underlyng procfs dentry stored in
its private data and mntputs the vfsmnt stashed in s_fs_info
      hppfs_put_super no longer needs to mntput the s_fs_info, so it
no longer needs to exist
      hppfs_readlink and hppfs_follow_link were doing a bunch of stuff
with a struct file which they didn't use
      there is now a ->permission which calls generic_permission
      get_inode was always returning 0 for some reason - it now
returns an inode if nothing bad happened

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: style fixes

A few random style fixes.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

uml: redo host capability detection and disabling

Redo how host capabilities are recorded at startup and disabled on the
command line.

There are now explicit variables saying what's been disabled by the
command line rather than the implicitness of the have_* variable being
zero.  The capability variables now start at zero and are set to one
as their capabilities are found to be present on the host.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[ALSA] ASoC: Fix wrong enum count for jack_function in N810 machine driver

Fix this typo and avoid similar errors by using ARRAY_SIZE macro.

Signed-off-by: Jarkko Nikula <jarkko.nikula@nokia.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

[ALSA] ASoC: build fix for snd_soc_info_bool_ext

I suspect that snd_ctl_boolean_mono should have been
snd_ctl_boolean_mono_info instead. This fixes the build for magician.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

[ALSA] ASoC: Fix TLV320AIC3X mono line output interconnect

There is no endpoint called MONOLOUT but MONO_LOUT.

Signed-off-by: Jarkko Nikula <jarkko.nikula@nokia.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

[ALSA] soc - fsl_ssi.c fix "BUG: scheduling while atomic"

This patch fixes following bug caught with PREEMPT enabled:

root@b1:~# cat /dev/dsp > /dev/null
BUG: scheduling while atomic: cat/965/0x00000003
Call Trace:
[df165ce0] [c0008e84] show_stack+0x4c/0x1ac (unreliable)
[df165d20] [c001c18c] __schedule_bug+0x64/0x78
[df165d30] [c02b3344] schedule+0x2d8/0x334
[df165d70] [c02b3674] schedule_timeout+0x64/0xe4
[df165db0] [c002c05c] msleep+0x1c/0x34
[df165dc0] [c01f2fe0] fsl_ssi_trigger+0x130/0x144
[df165dd0] [c01ece54] soc_pcm_trigger+0x94/0xb8
[df165df0] [c01da764] snd_pcm_do_start+0x48/0x60
[df165e00] [c01da630] snd_pcm_action_single+0x4c/0xb4
[df165e20] [c01e0f50] snd_pcm_lib_read1+0x2a0/0x2d4
[df165e70] [c01ec274] snd_pcm_oss_read3+0xf0/0x13c
[df165eb0] [c01ec2e4] snd_pcm_oss_read2+0x24/0x4c
[df165ec0] [c01ec4ac] snd_pcm_oss_read+0x1a0/0x1f0
[df165ef0] [c0076478] vfs_read+0xb4/0x108
[df165f10] [c00768cc] sys_read+0x4c/0x90
[df165f40] [c00117a4] ret_from_syscall+0x0/0x38

Acked-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

[ALSA] emux midi synthesizer doesn't honor SOFT_PEDAL-release event

When the hardware wavetable synthesizer of an Creative SB Audigy or SB
Live! card (with emu10k chip) receives the MIDI SOFT_PEADAL-press event
(?? 67 127) the appropriate voice is attenuted. Unfortunately when the
pedal is released (event ?? 67 0) the voice does not get it's original
volume again.

Boolean MIDI controls should interpret 0..63 as false and 64..127 as true.
Thanks to Clemens Ladisch for review and correction.

Original patch from "Uwe Kraeger" <uwe_debbug@arcor.de>
Submitted to http://bugs.debian.org/474312

Signed-off-by: maximilian attems <max@stro.at>
Cc: uwe_debbug@arcor.de
Cc: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

tcp FRTO: work-around inorder receivers

If receiver consumes segments successfully only in-order, FRTO
fallback to conventional recovery produces RTO loop because
FRTO's forward transmissions will always get dropped and need to
be resent, yet by default they're not marked as lost (which are
the only segments we will retransmit in CA_Loss).

Price to pay about this is occassionally unnecessarily
retransmitting the forward transmission(s). SACK blocks help
a bit to avoid this, so it's mainly a concern for NewReno case
though SACK is not fully immune either.

This change has a side-effect of fixing SACKFRTO problem where
it didn't have snd_nxt of the RTO time available anymore when
fallback become necessary (this problem would have only occured
when RTO would occur for two or more segments and ECE arrives
in step 3; no need to figure out how to fix that unless the
TODO item of selective behavior is considered in future).

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Damon L. Chesser <damon@damtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp FRTO: Fix fallback to conventional recovery

It seems that commit 009a2e3e4ec ("[TCP] FRTO: Improve
interoperability with other undo_marker users") run into
another land-mine which caused fallback to conventional
recovery to break:

1. Cumulative ACK arrives after FRTO retransmission
2. tcp_try_to_open sees zero retrans_out, clears retrans_stamp
   which should be kept like in CA_Loss state it would be
3. undo_marker change allowed tcp_packet_delayed to return
   true because of the cleared retrans_stamp once FRTO is
   terminated causing LossUndo to occur, which means all loss
   markings FRTO made are reverted.

This means that the conventional recovery basically recovered
one loss per RTT, which is not that efficient. It was quite
unobvious that the undo_marker change broken something like
this, I had a quite long session to track it down because of
the non-intuitiviness of the bug (luckily I had a trivial
reproducer at hand and I was also able to learn to use kprobes
in the process as well :-)).

This together with the NewReno+FRTO fix and FRTO in-order
workaround this fixes Damon's problems, this and the first
mentioned are enough to fix Bugzilla #10063.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Sebastian Hyrwall <zibbe@cisko.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'linus' into x86/urgent

Merge branch 'upstream-davem' of /linux/kernel/git/jgarzik/netdev-2.6

New maintainer for Intel ethernet adapters

I'm handing over maintainership to Jeff Kirsher and moving on
to other Linux/Open Source work within Intel. Good luck to Jeff ;)

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

DM9000: Use delayed work to update MII PHY state

Periodically check the MII PHY status to ensure that the
network layer's link status is updated and the user informed
of any changes.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

DM9000: Update and fix driver debugging messages

There was a missing newline in a dev_dbg() message.  Values read
from/written into PHY registers might be for interest too, so I added
new dbg messages there.

Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

DM9000: Add __devinit and __devexit attributes to probe and remove

There were missing __dev* annotations for the dm9000_probe()
and dm9000_drv_remove() functions.

Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

sky2: fix simple define thinko

noticed while browsing code, apparent thinko.  compile tested only.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

[netdrvr] sfc: sfc: Add self-test support

Add a set of self-tests accessible thorugh ethtool.
Add hardware loopback and TX disable control code to support them.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

[netdrvr] sfc: Increment rx_reset when reported as driver event

An RX_RESET event can be reported either as a global or as a driver event.
We were counting only global events.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

[netdrvr] sfc: Remove unused macro EFX_XAUI_RETRAIN_MAX

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>