Impact: improve wakeup affinity on NUMA systems, tweak SMP systems

Given the fixes+tweaks to the wakeup-buddy code, re-tweak the domain
balancing defaults on NUMA and SMP systems.

Turn on SD_WAKE_AFFINE which was off on x86 NUMA - there's no reason
why we would not want to have wakeup affinity across nodes as well.
(we already do this in the standard NUMA template.)

lat_ctx on a NUMA box is particularly happy about this change:

before:

 |   phoenix:~/l> ./lat_ctx -s 0 2
 |   "size=0k ovr=2.60
 |   2 5.70

after:

 |   phoenix:~/l> ./lat_ctx -s 0 2
 |   "size=0k ovr=2.65
 |   2 2.07

a 2.75x speedup.

pipe-test is similarly happy about it too:

 |  phoenix:~/sched-tests> ./pipe-test
 |   18.26 usecs/loop.
 |   14.70 usecs/loop.
 |   14.38 usecs/loop.
 |   10.55 usecs/loop.              # +WAKE_AFFINE on domain0+domain1
 |   8.63 usecs/loop.
 |   8.59 usecs/loop.
 |   9.03 usecs/loop.
 |   8.94 usecs/loop.
 |   8.96 usecs/loop.
 |   8.63 usecs/loop.

Also:

 - disable SD_BALANCE_NEWIDLE on NUMA and SMP domains (keep it for siblings)
 - enable SD_WAKE_BALANCE on SMP domains

Sysbench+postgresql improves all around the board, quite significantly:

           .28-rc3-11474e2c  .28-rc3-11474e2c-tune
-------------------------------------------------
    1:             571              688    +17.08%
    2:            1236             1206    -2.55%
    4:            2381             2642    +9.89%
    8:            4958             5164    +3.99%
   16:            9580             9574    -0.07%
   32:            7128             8118    +12.20%
   64:            7342             8266    +11.18%
  128:            7342             8064    +8.95%
  256:            7519             7884    +4.62%
  512:            7350             7731    +4.93%
-------------------------------------------------
  SUM:           55412            59341    +6.62%

So it's a win both for the runup portion, the peak area and the tail.
Signed-off-by: NIngo Molnar <mingo@elte.hu>

libata always uses PIO for ATAPI commands when the number of bytes to
transfer isn't multiple of 16 but quantum DAT72 chokes on odd bytes
PIO transfers.  Implement a horkage to skip the mod16 check and apply
it to the quantum device.

This is reported by John Clark in the following thread.

  http://thread.gmane.org/gmane.linux.ide/34748Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: John Clark <clarkjc@runbox.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Impact: cleanup

clean up ifdefs: change #ifdef CONFIG_X86_32/64 to
CONFIG_HAVE_ATOMIC_IOMAP.

flip around the #ifdef sections to clean up the structure.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This will let userland know when to submit its batchbuffers, before they get
too big to fit in the aperture.
Signed-off-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDave Airlie <airlied@redhat.com>

s/user/used/
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Impact: add new generic io_map_*() APIs

Graphics devices have large PCI apertures which would consume a significant
fraction of a 32-bit address space if mapped during driver initialization.
Using ioremap at runtime is impractical as it is too slow.

This new set of interfaces uses atomic mappings on 32-bit processors and a
large static mapping on 64-bit processors to provide reasonable 32-bit
performance and optimal 64-bit performance.

The current implementation sits atop the io_map_atomic fixmap-based
mechanism for 32-bit processors.

This includes some editorial suggestions from Randy Dunlap for
Documentation/io-mapping.txt
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: introduce new APIs, separate kmap code from CONFIG_HIGHMEM

This takes the code used for CONFIG_HIGHMEM memory mappings except that
it's designed for dynamic IO resource mapping.

These fixmaps are available even with CONFIG_HIGHMEM turned off.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

netns ops which are registered with register_pernet_gen_device() are
shutdown strictly before those which are registered with
register_pernet_subsys(). Sometimes this leads to opposite (read: buggy)
shutdown ordering between two modules.

Add register_pernet_gen_subsys()/unregister_pernet_gen_subsys() for modules
which aren't elite enough for entry in struct net, and which can't use
register_pernet_gen_device(). PPTP conntracking module is such one.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove excess kernel-doc function parameters from networking header
& driver files:

Warning(include/net/sock.h:946): Excess function parameter or struct member 'sk' description in 'sk_filter_release'
Warning(include/linux/netdevice.h:1545): Excess function parameter or struct member 'cpu' description in 'netif_tx_lock'
Warning(drivers/net/wan/z85230.c:712): Excess function parameter or struct member 'regs' description in 'z8530_interrupt'
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

libata currently imposes a UDMA5 max transfer rate and 200 sector max
transfer size for SATA devices that sit behind a pata-sata bridge. Lots
of devices have known good bridges that don't need this limit applied.
The MTRON SSD disks are such devices. Transfer rates are increased by
20-30% with the restriction removed.

So add a "blacklist" entry for the MTRON devices, with a flag indicating
that the bridge is known good.
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

The init_phy() function attaches to the PHY, then configures the
SerDes<->TBI link (in SGMII mode).  The TBI is on the MDIO bus with the PHY
(sort of) and is accessed via the gianfar's MDIO registers, using the
functions gfar_local_mdio_read/write(), which don't do any locking.

The previously attached PHY will start a work-queue on a timer, and
probably an irq handler as well, which will talk to the PHY and thus use
the MDIO bus.  This uses phy_read/write(), which have locking, but not
against the gfar_local_mdio versions.

The result is that PHY code will try to use the MDIO bus at the same time
as the SerDes setup code, corrupting the transfers.

Setting up the SerDes before attaching to the PHY will insure that there is
no race between the SerDes code and *our* PHY, but doesn't fix everything.
Typically the PHYs for all gianfar devices are on the same MDIO bus, which
is associated with the first gianfar device.  This means that the first
gianfar's SerDes code could corrupt the MDIO transfers for a different
gianfar's PHY.

The lock used by phy_read/write() is contained in the mii_bus structure,
which is pointed to by the PHY.  This is difficult to access from the
gianfar drivers, as there is no link between a gianfar device and the
mii_bus which shares the same MDIO registers.  As far as the device layer
and drivers are concerned they are two unrelated devices (which happen to
share registers).

Generally all gianfar devices' PHYs will be on the bus associated with the
first gianfar.  But this might not be the case, so simply locking the
gianfar's PHY's mii bus might not lock the mii bus that the SerDes setup
code is going to use.

We solve this by having the code that creates the gianfar platform device
look in the device tree for an mdio device that shares the gianfar's
registers.  If one is found the ID of its platform device is saved in the
gianfar's platform data.

A new function in the gianfar mii code, gfar_get_miibus(), can use the bus
ID to search through the platform devices for a gianfar_mdio device with
the right ID.  The platform device's driver data is the mii_bus structure,
which the SerDes setup code can use to lock the current bus.
Signed-off-by: NTrent Piepho <tpiepho@freescale.com>
CC: Andy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Fix compile error below:

     LD      drivers/spi/built-in.o
     CC [M]  drivers/spi/spi_gpio.o
   In file included from drivers/spi/spi_gpio.c:26:
   include/linux/spi/spi_bitbang.h:23: error: field `work' has incomplete type
   make[2]: *** [drivers/spi/spi_gpio.o] Error 1
   make[1]: *** [drivers/spi] Error 2
   make: *** [drivers] Error 2
Signed-off-by: NFernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Junjiro R.  Okajima reported a problem where knfsd crashes if you are
using it to export shmemfs objects and run strict overcommit.  In this
situation the current->mm based modifier to the overcommit goes through a
NULL pointer.

We could simply check for NULL and skip the modifier but we've caught
other real bugs in the past from mm being NULL here - cases where we did
need a valid mm set up (eg the exec bug about a year ago).

To preserve the checks and get the logic we want shuffle the checking
around and add a new helper to the vm_ security wrappers

Also fix a current->mm reference in nommu that should use the passed mm

[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: fix build]
Reported-by: NJunjiro R. Okajima <hooanon05@yahoo.co.jp>
Acked-by: NJames Morris <jmorris@namei.org>
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Put the kernel-doc for might_sleep() _immediately_ before the macro
(no intervening lines).  Otherwise kernel-doc complains like so:

Warning(linux-2.6.27-rc3-git2//include/linux/kernel.h:129): No description found for parameter 'file'
Warning(linux-2.6.27-rc3-git2//include/linux/kernel.h:129): No description found for parameter 'line'

because kernel-doc is looking at the wrong function prototype (i.e.,
__might_sleep).  [Yes, I have a todo note to myself to check/warn for that
inconsistency in scripts/kernel-doc.]
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Cc: <Uwe.Kleine-Koenig@digi.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Nothing uses prepare_write or commit_write. Remove them from the tree
completely.

[akpm@linux-foundation.org: schedule simple_prepare_write() for unexporting]
Signed-off-by: NNick Piggin <npiggin@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- remove 'private' field from struct subsys
- remove cgroup_init_smp()
Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Acked-by: NPaul Menage <menage@google.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't duplicate the implementation of thaw_process().

[akpm@linux-foundation.org: make __thaw_process() static]
Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Acked-by: NMatt Helsley <matthltc@us.ibm.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

By default, non-privileged tasks can only mlock() a small amount of
memory to avoid a DoS attack by ordinary users.  The Linux kernel
defaulted to 32k (on a 4k page size system) to accommodate the needs of
gpg.

However, newer gpg2 needs 64k in various circumstances and otherwise
fails miserably, see bnc#329675.

Change the default to 64k, and make it more agnostic to PAGE_SIZE.
Signed-off-by: NKurt Garloff <garloff@suse.de>
Signed-off-by: NNick Piggin <npiggin@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When ASoC was converted to support full int width masks SOC_SINGLE_VALUE()
omitted the assignment of rshift, causing the control operatins to report
some mono controls as stereo. This happened to work some of the time due
to a confusion between shift and min in snd_soc_info_volsw().
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add rodata equivalents for assembly use, and fix the section attributes
used by __REFCONST.
Signed-off-by: NJan Beulich <jbeulich@novell.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

Change all users of header files to correct path.
Remove some unneeded headers for arch-v32.
Signed-off-by: NJesper Nilsson <jesper.nilsson@axis.com>

struct hid_device_id contains hidden padding which is bad for cross
compiling.  Make the padding explicit and consistent across
architectures.
Signed-off-by: NAndreas Schwab <schwab@suse.de>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

Reserve some space so we can add more data.
Signed-off-by: NAvi Kivity <avi@qumranet.com>

Every call of kvm_set_irq() should offer an irq_source_id, which is
allocated by kvm_request_irq_source_id(). Based on irq_source_id, we
identify the irq source and implement logical OR for shared level
interrupts.

The allocated irq_source_id can be freed by kvm_free_irq_source_id().

Currently, we support at most sizeof(unsigned long) different irq sources.

[Amit: - rebase to kvm.git HEAD
       - move definition of KVM_USERSPACE_IRQ_SOURCE_ID to common file
       - move kvm_request_irq_source_id to the update_irq ioctl]

[Xiantao: - Add kvm/ia64 stuff and make it work for kvm/ia64 guests]
Signed-off-by: NSheng Yang <sheng@linux.intel.com>
Signed-off-by: NAmit Shah <amit.shah@redhat.com>
Signed-off-by: NXiantao Zhang <xiantao.zhang@intel.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Fix mac80211.h kernel-doc: it had some extra parameters that were
no longer valid and incorrect format for a return value in 2 places.

Warning(lin2628-rc2//include/net/mac80211.h:1487): Excess function parameter or struct member 'control' description in 'ieee80211_beacon_get'
Warning(lin2628-rc2//include/net/mac80211.h:1596): Excess function parameter or struct member 'control' description in 'ieee80211_get_buffered_bc'
Warning(lin2628-rc2//include/net/mac80211.h:1632): Excess function parameter or struct member 'rc4key' description in 'ieee80211_get_tkip_key'
Warning(lin2628-rc2//include/net/mac80211.h:1735): Excess function parameter or struct member 'return' description in 'ieee80211_start_tx_ba_session'
Warning(lin2628-rc2//include/net/mac80211.h:1775): Excess function parameter or struct member 'return' description in 'ieee80211_stop_tx_ba_session'
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Acked-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Right now callers have to check whether scsi_host->bqt is already
set up, it's much cleaner to just have scsi_init_shared_tag_map()
does this check on its own.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Usbhid structure is allocated on start invoked only from probe
of some driver. When there is no driver, the structure is null
and causes null-dereference oopses.

Fix it by allocating the structure on probe and disconnect of
the device itself. Also make sure we won't race between start
and resume or stop and suspend respectively.

References: http://bugzilla.kernel.org/show_bug.cgi?id=11827Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: Andreas Schwab <schwab@suse.de>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

Impact: build fix

DIE_OOPS is now used in the generic trace handling code so it needs to
be defined for all architectures.  Define it in asm-generic so that it's
available to all by default and doesn't cause build errors for
architectures that rely on the generic implementation.
Signed-off-by: NJonas Bonn <jonas.bonn@gmail.com>
Acked-by: NSteven Rostedt <srostedt@redhat.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This fixes a potential error packet loop.
Signed-off-by: NRemi Denis-Courmont <remi.denis-courmont@nokia.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This fixes an OOPS in hard_header if a Phonet address is assigned to a
non-Phonet network interface.
Signed-off-by: NRemi Denis-Courmont <remi.denis-courmont@nokia.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Based on input from Andi Kleen:
share the platform detection code with ioat_dma and disable the channel in
dma engine only for specific platforms.
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NLen Brown <len.brown@intel.com>

libata currently has a pretty dumb ATA_MAX_QUEUE loop for finding
a free tag to use. Instead of fixing that up, convert libata to
using block layer tagging - gets rid of code in libata, and is also
much faster.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>
Acked-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We're getting a lot of storage drivers blamed for interrupt misrouting
issues.  This patch provides a standard way of reporting the problem
... and, if possible, correcting it.
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>

Add support for the OxSemi 'Tornado' devices.

Reformatted and reworked a bit by Alan Cox
Signed-off-by: NLee Howard <lee.howard@mainpine.com>
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- atomic operations which both modify the variable and return something imply
  full smp memory barriers before and after the memory operations involved
  (failing atomic_cmpxchg, atomic_add_unless, etc don't imply a barrier because
  they don't modify the target). See Documentation/atomic_ops.txt.
  So remove extra barriers and branches.

- All architectures support atomic_cmpxchg. This has no relation to
  __HAVE_ARCH_CMPXCHG. We can just take the atomic_cmpxchg path unconditionally

This reduces a simple single threaded fastpath lock+unlock test from 590 cycles
to 203 cycles on a ppc970 system.
Signed-off-by: NNick Piggin <npiggin@suse.de>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

page_cgroup_init() is called from mem_cgroup_init(). But at this
point, we cannot call alloc_bootmem().
(and this caused panic at boot.)

This patch moves page_cgroup_init() to init/main.c.

Time table is following:
==
  parse_args(). # we can trust mem_cgroup_subsys.disabled bit after this.
  ....
  cgroup_init_early()  # "early" init of cgroup.
  ....
  setup_arch()         # memmap is allocated.
  ...
  page_cgroup_init();
  mem_init();   # we cannot call alloc_bootmem after this.
  ....
  cgroup_init() # mem_cgroup is initialized.
==

Before page_cgroup_init(), mem_map must be initialized. So,
I added page_cgroup_init() to init/main.c directly.

(*) maybe this is not very clean but
    - cgroup_init_early() is too early
    - in cgroup_init(), we have to use vmalloc instead of alloc_bootmem().
    use of vmalloc area in x86-32 is important and we should avoid very large
    vmalloc() in x86-32. So, we want to use alloc_bootmem() and added page_cgroup_init()
    directly to init/main.c

[akpm@linux-foundation.org: remove unneeded/bad mem_cgroup_subsys declaration]
[akpm@linux-foundation.org: fix build]
Acked-by: NBalbir Singh <balbir@linux.vnet.ibm.com>
Tested-by: NBalbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When a checkpointing IO fails, current JBD code doesn't check the error
and continue journaling.  This means latest metadata can be lost from both
the journal and filesystem.

This patch leaves the failed metadata blocks in the journal space and
aborts journaling in the case of log_do_checkpoint().  To achieve this, we
need to do:

1. don't remove the failed buffer from the checkpoint list where in
   the case of __try_to_free_cp_buf() because it may be released or
   overwritten by a later transaction
2. log_do_checkpoint() is the last chance, remove the failed buffer
   from the checkpoint list and abort the journal
3. when checkpointing fails, don't update the journal super block to
   prevent the journaled contents from being cleaned.  For safety,
   don't update j_tail and j_tail_sequence either
4. when checkpointing fails, notify this error to the ext3 layer so
   that ext3 don't clear the needs_recovery flag, otherwise the
   journaled contents are ignored and cleaned in the recovery phase
5. if the recovery fails, keep the needs_recovery flag
6. prevent cleanup_journal_tail() from being called between
   __journal_drop_transaction() and journal_abort() (a race issue
   between journal_flush() and __log_wait_for_space()
Signed-off-by: NHidehiro Kawai <hidehiro.kawai.ez@hitachi.com>
Acked-by: NJan Kara <jack@suse.cz>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

In the case where procfs is disabled, create_proc_profile() does not
exist. Stub it in with the others.
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Now that everything was moved to their more or less expected places,
apply rm(1).
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>