This patch tries to reduce the amount of cmpxchg calls in the writer
failed path by checking the counter value first before issuing the
instruction.  If ->count is not set to RWSEM_WAITING_BIAS then there is
no point wasting a cmpxchg call.

Furthermore, Michel states "I suppose it helps due to the case where
someone else steals the lock while we're trying to acquire
sem->wait_lock."

Two very different workloads and machines were used to see how this
patch improves throughput: pgbench on a quad-core laptop and aim7 on a
large 8 socket box with 80 cores.

Some results comparing Michel's fast-path write lock stealing
(tps-rwsem) on a quad-core laptop running pgbench:

  | db_size | clients  |  tps-rwsem     |   tps-patch  |
  +---------+----------+----------------+--------------+
  | 160 MB   |       1 |           6906 |         9153 | + 32.5
  | 160 MB   |       2 |          15931 |        22487 | + 41.1%
  | 160 MB   |       4 |          33021 |        32503 |
  | 160 MB   |       8 |          34626 |        34695 |
  | 160 MB   |      16 |          33098 |        34003 |
  | 160 MB   |      20 |          31343 |        31440 |
  | 160 MB   |      30 |          28961 |        28987 |
  | 160 MB   |      40 |          26902 |        26970 |
  | 160 MB   |      50 |          25760 |        25810 |
  ------------------------------------------------------
  | 1.6 GB   |       1 |           7729 |         7537 |
  | 1.6 GB   |       2 |          19009 |        23508 | + 23.7%
  | 1.6 GB   |       4 |          33185 |        32666 |
  | 1.6 GB   |       8 |          34550 |        34318 |
  | 1.6 GB   |      16 |          33079 |        32689 |
  | 1.6 GB   |      20 |          31494 |        31702 |
  | 1.6 GB   |      30 |          28535 |        28755 |
  | 1.6 GB   |      40 |          27054 |        27017 |
  | 1.6 GB   |      50 |          25591 |        25560 |
  ------------------------------------------------------
  | 7.6 GB   |       1 |           6224 |         7469 | + 20.0%
  | 7.6 GB   |       2 |          13611 |        12778 |
  | 7.6 GB   |       4 |          33108 |        32927 |
  | 7.6 GB   |       8 |          34712 |        34878 |
  | 7.6 GB   |      16 |          32895 |        33003 |
  | 7.6 GB   |      20 |          31689 |        31974 |
  | 7.6 GB   |      30 |          29003 |        28806 |
  | 7.6 GB   |      40 |          26683 |        26976 |
  | 7.6 GB   |      50 |          25925 |        25652 |
  ------------------------------------------------------

For the aim7 worloads, they overall improved on top of Michel's
patchset.  For full graphs on how the rwsem series plus this patch
behaves on a large 8 socket machine against a vanilla kernel:

  http://stgolabs.net/rwsem-aim7-results.tar.gz



Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

 - make warning smp-safe
 - result of atomic _unless_zero functions should be checked by caller
   to avoid use-after-free error
 - trivial whitespace fix.

Link: https://lkml.org/lkml/2013/4/12/391



Tested: compile x86, boot machine and run xfstests
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
[ Removed line-break, changed to use WARN_ON_ONCE()  - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull more vfs updates from Al Viro:
 "A couple of fixes + getting rid of __blkdev_put() return value"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  proc: Use PDE attribute setting accessor functions
  make blkdev_put() return void
  block_device_operations->release() should return void
  mtd_blktrans_ops->release() should return void
  hfs: SMP race on directory close()

Pull parisc updates from Helge Deller:
 "Main fixes and updates in this patch series are:
   - we faced kernel stack corruptions because of multiple delivery of
     interrupts
   - added kernel stack overflow checks
   - added possibility to use dedicated stacks for irq processing
   - initial support for page sizes > 4k
   - more information in /proc/interrupts (e.g.  TLB flushes and number
     of IPI calls)
   - documented how the parisc gateway page works
   - and of course quite some other smaller cleanups and fixes."

* 'parisc-for-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: tlb flush counting fix for SMP and UP
  parisc: more irq statistics in /proc/interrupts
  parisc: implement irq stacks
  parisc: add kernel stack overflow check
  parisc: only re-enable interrupts if we need to schedule or deliver signals when returning to userspace
  parisc: implement atomic64_dec_if_positive()
  parisc: use long branch in fork_like macro
  parisc: fix NATIVE set up in build
  parisc: document the parisc gateway page
  parisc: fix partly 16/64k PAGE_SIZE boot
  parisc: Provide default implementation for dma_{alloc, free}_attrs
  parisc: fix whitespace errors in arch/parisc/kernel/traps.c
  parisc: remove the second argument of kmap_atomic

Pull ARM Xen SMP updates from Stefano Stabellini:
 "This contains a bunch of Xen/ARM specific changes, including some
  fixes, SMP support for Xen on ARM, and moving the xenvm machine from
  mach-vexpress to mach-virt.

  The non-Xen files that are touched are arch/arm/Kconfig, to select
  ARM_PSCI on XEN, and arch/arm/boot/dts/Makefile, to build the xenvm
  DTB if CONFIG_ARCH_VIRT.

  Highlights:

   - Move xenvm to mach-virt.

   - Implement SMP support in Xen on ARM.

   - Add support for machine reboot and power off via Xen hypercalls"

* tag '3.9-rc3-smp-6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/sstabellini/xen:
  xen/arm: remove duplicated include from enlighten.c
  xen/arm: use sched_op hypercalls for machine reboot and power off
  xenvm: add a simple PSCI node and a second cpu
  xen/arm: XEN selects ARM_PSCI
  xen: move the xenvm machine to mach-virt
  xen/arm: SMP support
  xen/arm: implement HYPERVISOR_vcpu_op
  xen/arm: actually pass a non-NULL percpu pointer to request_percpu_irq

Fix up build error on UP and show correctly number of function call
(ipi) irqs.

Signed-off-by: Helge Deller <deller@gmx.de>

Pull remoteproc update from Ohad Ben-Cohen:

 - Some refactoring, cleanups and small improvements from Sjur
   Brændeland.  The improvements are mainly about better supporting
   varios virtio properties (such as virtio's config space, status and
   features).  I now see that I messed up while commiting one of Sjur's
   patches and erroneously put myself as the author, as well as letting
   a nasty typo sneak in.  I will not fix this in order to avoid
   rebasing the patches.  Sjur - sorry!

 - A new remoteproc driver for OMAP-L13x (technically a DaVinci
   platform) from Robert Tivy.

 - Extend OMAP support to OMAP5 as well, from Vincent Stehlé.

 - Fix Kconfig VIRTUALIZATION dependency, from Suman Anna (a
   non-critical fix which arrived late during the rc cycle).

* tag 'remoteproc-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
  remoteproc: fix kconfig dependencies for VIRTIO
  remoteproc/davinci: add a remoteproc driver for OMAP-L13x DSP
  remoteproc: support default firmware name in rproc_alloc()
  remoteproc/omap: support OMAP5 too
  remoteproc: set vring addresses in resource table
  remoteproc: support virtio config space.
  remoteproc: perserve resource table data
  remoteproc: calculate max_notifyid by counting vrings
  remoteproc: code cleanup of resource parsing
  remoteproc: parse STE-firmware and find resource table address
  remoteproc: add find_loaded_rsc_table firmware ops
  remoteproc: refactor rproc_elf_find_rsc_table()

Pull rpmsg changes from Ohad Ben-Cohen:
 "A small pull request consisting of:

   - Make rpmsg process all pending messages instead of just one, from
     Robert Tivy

   - Fix Kconfig dependency on VIRTUALIZATION, from Suman.

     Note: this was submitted late during the 3.9 rc cycle and it seemed
     appropriate to wait with it for the merge window.

   - Belated addition of an rpmsg entry to the MAINTAINERS file.  People
     seem to look for this"

* tag 'rpmsg-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/rpmsg:
  rpmsg: fix kconfig dependencies for VIRTIO
  MAINTAINERS: add rpmsg entry
  rpmsg: process _all_ pending messages in rpmsg_recv_done

Pullhwspinlock update from Ohad Ben-Cohen:
 "A single patch from Vincent extending OMAP's hwspinlock support to
  OMAP5"

* tag 'hwspinlock-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/hwspinlock:
  hwspinlock/omap: support OMAP5 as well

Pull libata maintainership change from Tejun Heo.

Tejun is taking over from Jeff, after many many years. Thanks Jeff.

* 'for-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
  libata: change maintainer

Add framework and initial values for more fine grained statistics in
/proc/interrupts.

Signed-off-by: Helge Deller <deller@gmx.de>

Default kernel stack size on parisc is 16k.  During tests we found that the
kernel stack can easily grow beyond 13k, which leaves 3k left for irq
processing.

This patch adds the possibility to activate an additional stack of 16k per CPU
which is being used during irq processing.  This implementation does not yet
uses this irq stack for the irq bh handler.

The assembler code for call_on_stack was heavily cleaned up by John
David Anglin.

CC: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>

Add the CONFIG_DEBUG_STACKOVERFLOW config option to enable checks to
detect kernel stack overflows.

Stack overflows can not be detected reliable since we do not want to
introduce too much overhead.

Instead, during irq processing in do_cpu_irq_mask() we check kernel
stack usage of the interrupted kernel process. Kernel threads can be
easily detected by checking the value of space register 7 (sr7) which
is zero when running inside the kernel.

Since THREAD_SIZE is 16k and PAGE_SIZE is 4k, reduce the alignment of
the init thread to the lower value (PAGE_SIZE) in the kernel
vmlinux.ld.S linker script.

Signed-off-by: Helge Deller <deller@gmx.de>

arch/arm/mach-msm/last_radio_log.c: In function 'msm_init_last_radio_log':
arch/arm/mach-msm/last_radio_log.c:69:7: error: dereferencing pointer to incomplete type

arch/cris/kernel/profile.c: In function 'init_cris_profile':
arch/cris/kernel/profile.c:79:8: error: dereferencing pointer to incomplete type

Use proc_set_size(), cfr. commit 271a15ea


("proc: Supply PDE attribute setting accessor functions")

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

parisc: only re-enable interrupts if we need to schedule or deliver signals when returning to userspace

Helge and I have found that we have a kernel stack overflow problem
which causes a variety of random failures.
Currently, we re-enable interrupts when returning from an external
interrupt incase we need to schedule or delivery
signals.  As a result, a potentially unlimited number of interrupts
can occur while we are running on the kernel
stack.  It is very limited in space (currently, 16k).  This change
defers enabling interrupts until we have
actually decided to schedule or delivery signals.  This only occurs
when we about to return to userspace.  This
limits the number of interrupts on the kernel stack to one.  In other
cases, interrupts remain disabled until the
final return from interrupt (rfi).

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>

Pull late ARM Exynos multiplatform changes from Arnd Bergmann:
 "These continue the multiplatform support for exynos, adding support
  for building most of the essential drivers (clocksource, clk, irqchip)
  when combined with other platforms.  As a result, it should become
  really easy to add full multiplatform exynos support in 3.11, although
  we don't yet enable it for 3.10.

  The changes were not included in the earlier multiplatform series in
  order to avoid clashes with the other Exynos updates.

  This also includes work from Tomasz Figa to fix the pwm clocksource
  code on Exynos, which is not strictly required for multiplatform, but
  related to the other patches in this set and needed as a bug fix for
  at least one board."

* tag 'multiplatform-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (22 commits)
  ARM: dts: exynops4210: really add universal_c210 dts
  ARM: dts: exynos4210: Add basic dts file for universal_c210 board
  ARM: dts: exynos4: Add node for PWM device
  ARM: SAMSUNG: Do not register legacy timer interrupts on Exynos
  clocksource: samsung_pwm_timer: Work around rounding errors in clockevents core
  clocksource: samsung_pwm_timer: Correct programming of clock events
  clocksource: samsung_pwm_timer: Use proper clockevents max_delta
  clocksource: samsung_pwm_timer: Add support for non-DT platforms
  clocksource: samsung_pwm_timer: Drop unused samsung_pwm struct
  clocksource: samsung_pwm_timer: Keep all driver data in a structure
  clocksource: samsung_pwm_timer: Make PWM spinlock global
  clocksource: samsung_pwm_timer: Let platforms select the driver
  Documentation: Add device tree bindings for Samsung PWM timers
  clocksource: add samsung pwm timer driver
  irqchip: exynos: look up irq using irq_find_mapping
  irqchip: exynos: pass irq_base from platform
  irqchip: exynos: localize irq lookup for ATAGS
  irqchip: exynos: allocate combiner_data dynamically
  irqchip: exynos: pass max combiner number to combiner_init
  ARM: exynos: add missing properties for combiner IRQs
  ...

Pull ARM SoC late cleanups from Arnd Bergmann:
 "These are cleanups and smaller changes that either depend on earlier
  feature branches or came in late during the development cycle.  We
  normally try to get all cleanups early, so these are the exceptions:

   - A follow-up on the clocksource reworks, hopefully the last time we
     need to merge clocksource subsystem changes through arm-soc.

     A first set of patches was part of the original 3.10 arm-soc
     cleanup series because of interdependencies with timer drivers now
     moved out of arch/arm.

   - Migrating the SPEAr13xx platform away from using auxdata for DMA
     channel descriptions towards using information in device tree,
     based on the earlier SPEAr multiplatform series

   - A few follow-ups on the Atmel SAMA5 support and other changes for
     Atmel at91 based on the larger at91 reworks.

   - Moving the armada irqchip implementation to drivers/irqchip

   - Several OMAP cleanups following up on the larger series already
     merged in 3.10."

* tag 'cleanup-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (50 commits)
  ARM: OMAP4: change the device names in usb_bind_phy
  ARM: OMAP2+: Fix mismerge for timer.c between ff931c82 and da4a686a
  ARM: SPEAr: conditionalize SMP code
  ARM: arch_timer: Silence debug preempt warnings
  ARM: OMAP: remove unused variable
  serial: amba-pl011: fix !CONFIG_DMA_ENGINE case
  ata: arasan: remove the need for platform_data
  ARM: at91/sama5d34ek.dts: remove not needed compatibility string
  ARM: at91: dts: add MCI DMA support
  ARM: at91: dts: add i2c dma support
  ARM: at91: dts: set #dma-cells to the correct value
  ARM: at91: suspend both memory controllers on at91sam9263
  irqchip: armada-370-xp: slightly cleanup irq controller driver
  irqchip: armada-370-xp: move IRQ handler to avoid forward declaration
  irqchip: move IRQ driver for Armada 370/XP
  ARM: mvebu: move L2 cache initialization in init_early()
  devtree: add binding documentation for sp804
  ARM: integrator-cp: convert use CLKSRC_OF for timer init
  ARM: versatile: use OF init for sp804 timer
  ARM: versatile: add versatile dtbs to dtbs target
  ...

Pull ARM SoC device tree updates (part 2) from Arnd Bergmann:
 "These are mostly new device tree bindings for existing drivers, as
  well as changes to the device tree source files to add support for
  those devices, and a couple of new boards, most notably Samsung's
  Exynos5 based Chromebook.

  The changes depend on earlier platform specific updates and touch the
  usual platforms: omap, exynos, tegra, mxs, mvebu and davinci."

* tag 'dt-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (169 commits)
  ARM: exynos: dts: cros5250: add EC device
  ARM: dts: Add sbs-battery for exynos5250-snow
  ARM: dts: Add i2c-arbitrator bus for exynos5250-snow
  ARM: dts: add mshc controller node for Exynos4x12 SoCs
  ARM: dts: Add chip-id controller node on Exynos4/5 SoC
  ARM: EXYNOS: Create virtual I/O mapping for Chip-ID controller using device tree
  ARM: davinci: da850-evm: add SPI flash support
  ARM: davinci: da850: override SPI DT node device name
  ARM: davinci: da850: add SPI1 DT node
  spi/davinci: add DT binding documentation
  spi/davinci: no wildcards in DT compatible property
  ARM: dts: mvebu: Convert mvebu device tree files to 64 bits
  ARM: dts: mvebu: introduce internal-regs node
  ARM: dts: mvebu: Convert all the mvebu files to use the range property
  ARM: dts: mvebu: move all peripherals inside soc
  ARM: dts: mvebu: fix cpus section indentation
  ARM: davinci: da850: add EHRPWM & ECAP DT node
  ARM/dts: OMAP3: fix pinctrl-single configuration
  ARM: dts: Add OMAP3430 SDP NOR flash memory binding
  ARM: dts: Add NOR flash bindings for OMAP2420 H4
  ...

Pull ARM SoC platform updates (part 3) from Arnd Bergmann:
 "This is the third and smallest of the SoC specific updates.  Changes
  include:

   - SMP support for the Xilinx zynq platform
   - Smaller imx changes
   - LPAE support for mvebu
   - Moving the orion5x, kirkwood, dove and mvebu platforms to a common
     "mbus" driver for their internal devices.

  It would be good to get feedback on the location of the "mbus" driver.
  Since this is used on multiple platforms may potentially get shared
  with other architectures (powerpc and arm64), it was moved to
  drivers/bus/.  We expect other similar drivers to get moved to the
  same place in order to avoid creating more top-level directories under
  drivers/ or cluttering up the messy drivers/misc/ even more."

* tag 'soc-for-linus-3' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (50 commits)
  ARM: imx: reset_controller may be disabled
  ARM: mvebu: Align the internal registers virtual base to support LPAE
  ARM: mvebu: Limit the DMA zone when LPAE is selected
  arm: plat-orion: remove addr-map code
  arm: mach-mv78xx0: convert to use the mvebu-mbus driver
  arm: mach-orion5x: convert to use mvebu-mbus driver
  arm: mach-dove: convert to use mvebu-mbus driver
  arm: mach-kirkwood: convert to use mvebu-mbus driver
  arm: mach-mvebu: convert to use mvebu-mbus driver
  ARM i.MX53: set CLK_SET_RATE_PARENT flag on the tve_ext_sel clock
  ARM i.MX53: tve_di clock is not part of the CCM, but of TVE
  ARM i.MX53: make tve_ext_sel propagate rate change to PLL
  ARM i.MX53: Remove unused tve_gate clkdev entry
  ARM i.MX5: Remove tve_sel clock from i.MX53 clock tree
  ARM: i.MX5: Add PATA and SRTC clocks
  ARM: imx: do not bring up unavailable cores
  ARM: imx: add initial imx6dl support
  ARM: imx1: mm: add call to mxc_device_init
  ARM: imx_v4_v5_defconfig: Add CONFIG_GPIO_SYSFS
  ARM: imx_v6_v7_defconfig: Select CONFIG_PERF_EVENTS
  ...

Pull ARM SoC platform updates (part 2) from Arnd Bergmann:
 "These patches are all for Renesas shmobile, and depend on the earlier
  pinctrl updates.  Remarkably, this adds support for three new SoCs:
  r8a73a4, r8a73a4 and r8a7778.  The bulk of the code added for these is
  for pinctrl (using the new subsystem) and for clocks (not yet using
  the common clock subsystem).  The latter will have to get converted in
  one of the upcoming releases, but shmobile is not ready for that yet.

  The series also contains Renesas shmobile board changes, adding one
  board file for each of the three new SoCs.  These boards are using a
  mix of classic and device-tree based probing, as there is still a lot
  of infrastructure in shmobile that has not been converted to DT yet.
  Once those are resolved to the degree that no board specific setup
  code is needed, they can get folded into the respective SoC setup files."

* tag 'soc-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (78 commits)
  ARM: shmobile: use r8a7790 timer setup code on Lager
  ARM: shmobile: force enable of r8a7790 arch timer
  ARM: shmobile: Add second I/O range for r8a7790 PFC
  ARM: shmobile: bockw: enable network settings on bootargs
  ARM: shmobile: bockw: add SMSC ethernet support
  ARM: shmobile: R8A7778: add Ether support
  ARM: shmobile: bockw: enable SMSC ethernet on defconfig
  ARM: shmobile: r8a7778: add r8a7778_init_irq_extpin()
  ARM: shmobile: r8a7778: remove pointless PLATFORM_INFO()
  ARM: shmobile: mackerel: clean up MMCIF vs. SDHI1 selection
  ARM: shmobile: mackerel: add interrupt names for SDHI0
  ARM: shmobile: mackerel: switch SDHI and MMCIF interfaces to slot-gpio
  ARM: shmobile: mackerel: remove OCR masks, where regulators are used
  ARM: shmobile: mackerel: SDHI resources do not have to be numbered
  ARM: shmobile: Initial r8a7790 Lager board support
  ARM: shmobile: APE6EVM LAN9220 support
  ARM: shmobile: APE6EVM PFC support
  ARM: shmobile: APE6EVM base support
  ARM: shmobile: kzm9g-reference: add ethernet support
  ARM: shmobile: add R-Car M1A Bock-W platform support
  ...

Jeff is leaving for something more interesting and I'm inheriting the
maintainership of libata.  Thanks a lot for the good work and have
fun, Jeff!

v2: The original path forgot to update git tree URL.  Updated.
    Spotted by Sergei.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Jeff Garzik <jgarzik@pobox.com>
Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>

Pull more vhost fixes from Michael Tsirkin:
 "This fixes some minor issues in the patches that have been merged.

  We also finally drop the workaround disabling event_idx for scsi: it
  was always questionable, and now we know it's not needed.

  There's also a memory leak fix"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vhost-scsi: Enable VIRTIO_RING_F_EVENT_IDX
  vhost: drop virtio_net.h dependency
  vhost-net: Cleanup vhost_ubuf and vhost_zcopy
  vhost: Remove vhost_enable_zcopy in vhost.h
  vhost: Remove comments for hdr in vhost.h
  vhost: Move VHOST_NET_FEATURES to net.c
  vhost-net: Free ubuf when vhost_dev_set_owner fails
  vhost: Export vhost_dev_set_owner

Pull fuse updates from Miklos Szeredi:
 "This contains two patchsets from Maxim Patlasov.

  The first reworks the request throttling so that only async requests
  are throttled.  Wakeup of waiting async requests is also optimized.

  The second series adds support for async processing of direct IO which
  optimizes direct IO and enables the use of the AIO userspace
  interface."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse:
  fuse: add flag to turn on async direct IO
  fuse: truncate file if async dio failed
  fuse: optimize short direct reads
  fuse: enable asynchronous processing direct IO
  fuse: make fuse_direct_io() aware about AIO
  fuse: add support of async IO
  fuse: move fuse_release_user_pages() up
  fuse: optimize wake_up
  fuse: implement exclusive wakeup for blocked_waitq
  fuse: skip blocking on allocations of synchronous requests
  fuse: add flag fc->initialized
  fuse: make request allocations for background processing explicit

Pull powerpc updates from Benjamin Herrenschmidt:
 "Here are a few more powerpc bits that I would like in 3.10.

  Mostly remaining bolts & screw tightening of power8 support such as
  actually exposing the new features via the previously added AT_HWCAP2,
  and a few fixes, some of them for problems exposed recently like
  irqdomain warnings or sysfs access permission issues, some exposed by
  power8 hardware.

  The only change outside of arch/powerpc is a small one to irqdomain.c
  to allow silent failure to fix a problem on Cell where we get a dozen
  WARN_ON's tripping at boot for what is basically a normal case."

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc: Make hard_irq_disable() do the right thing vs. irq tracing
  powerpc/topology: Fix spurr attribute permission
  powerpc/pci: Support per-aperture memory offset
  powerpc/cell/iommu: Improve error message for missing node
  powerpc/cell/spufs: Fix status attribute permission
  irqdomain: Allow quiet failure mode
  powerpc/pnv: Fix "compatible" property for P8 PHB
  powerpc/pci: Don't add bogus empty resources to PHBs
  powerpc/powerpnv: Properly handle failure starting CPUs
  powerpc/cputable: Advertise support for ISEL/HTM/DSCR/TAR on POWER8
  powerpc/cputable: Advertise ISEL support on appropriate embedded processors
  powerpc/cputable: Advertise DSCR support on P7/P7+
  powerpc/cputable: Reserve bits in HWCAP2 for new features
  powerpc/pseries: Perform proper max_bus_speed detection
  powerpc/pseries: Force 32 bit MSIs for devices that require it
  powerpc/tm: Fix null pointer deference in flush_hash_page
  powerpc/powernv: Defer OPAL exception handler registration
  powerpc: Emulate non privileged DSCR read and write

Merge rwsem optimizations from Michel Lespinasse:
 "These patches extend Alex Shi's work (which added write lock stealing
  on the rwsem slow path) in order to provide rwsem write lock stealing
  on the fast path (that is, without taking the rwsem's wait_lock).

  I have unfortunately been unable to push this through -next before due
  to Ingo Molnar / David Howells / Peter Zijlstra being busy with other
  things.  However, this has gotten some attention from Rik van Riel and
  Davidlohr Bueso who both commented that they felt this was ready for
  v3.10, and Ingo Molnar has said that he was OK with me pushing
  directly to you.  So, here goes :)

  Davidlohr got the following test results from pgbench running on a
  quad-core laptop:

    | db_size | clients  |  tps-vanilla   |   tps-rwsem  |
    +---------+----------+----------------+--------------+
    | 160 MB   |       1 |           5803 |         6906 | + 19.0%
    | 160 MB   |       2 |          13092 |        15931 |
    | 160 MB   |       4 |          29412 |        33021 |
    | 160 MB   |       8 |          32448 |        34626 |
    | 160 MB   |      16 |          32758 |        33098 |
    | 160 MB   |      20 |          26940 |        31343 | + 16.3%
    | 160 MB   |      30 |          25147 |        28961 |
    | 160 MB   |      40 |          25484 |        26902 |
    | 160 MB   |      50 |          24528 |        25760 |
    ------------------------------------------------------
    | 1.6 GB   |       1 |           5733 |         7729 | + 34.8%
    | 1.6 GB   |       2 |           9411 |        19009 | + 101.9%
    | 1.6 GB   |       4 |          31818 |        33185 |
    | 1.6 GB   |       8 |          33700 |        34550 |
    | 1.6 GB   |      16 |          32751 |        33079 |
    | 1.6 GB   |      20 |          30919 |        31494 |
    | 1.6 GB   |      30 |          28540 |        28535 |
    | 1.6 GB   |      40 |          26380 |        27054 |
    | 1.6 GB   |      50 |          25241 |        25591 |
    ------------------------------------------------------
    | 7.6 GB   |       1 |           5779 |         6224 |
    | 7.6 GB   |       2 |          10897 |        13611 | + 24.9%
    | 7.6 GB   |       4 |          32683 |        33108 |
    | 7.6 GB   |       8 |          33968 |        34712 |
    | 7.6 GB   |      16 |          32287 |        32895 |
    | 7.6 GB   |      20 |          27770 |        31689 | + 14.1%
    | 7.6 GB   |      30 |          26739 |        29003 |
    | 7.6 GB   |      40 |          24901 |        26683 |
    | 7.6 GB   |      50 |          17115 |        25925 | + 51.5%
    ------------------------------------------------------

  (Davidlohr also has one additional patch which further improves
  throughput, though I will ask him to send it directly to you as I have
  suggested some minor changes)."

* emailed patches from Michel Lespinasse <walken@google.com>:
  rwsem: no need for explicit signed longs
  x86 rwsem: avoid taking slow path when stealing write lock
  rwsem: do not block readers at head of queue if other readers are active
  rwsem: implement support for write lock stealing on the fastpath
  rwsem: simplify __rwsem_do_wake
  rwsem: skip initial trylock in rwsem_down_write_failed
  rwsem: avoid taking wait_lock in rwsem_down_write_failed
  rwsem: use cmpxchg for trying to steal write lock
  rwsem: more agressive lock stealing in rwsem_down_write_failed
  rwsem: simplify rwsem_down_write_failed
  rwsem: simplify rwsem_down_read_failed
  rwsem: move rwsem_down_failed_common code into rwsem_down_{read,write}_failed
  rwsem: shorter spinlocked section in rwsem_down_failed_common()
  rwsem: make the waiter type an enumeration rather than a bitmask

Pull slab changes from Pekka Enberg:
 "The bulk of the changes are more slab unification from Christoph.

  There's also few fixes from Aaron, Glauber, and Joonsoo thrown into
  the mix."

* 'slab/for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/linux: (24 commits)
  mm, slab_common: Fix bootstrap creation of kmalloc caches
  slab: Return NULL for oversized allocations
  mm: slab: Verify the nodeid passed to ____cache_alloc_node
  slub: tid must be retrieved from the percpu area of the current processor
  slub: Do not dereference NULL pointer in node_match
  slub: add 'likely' macro to inc_slabs_node()
  slub: correct to calculate num of acquired objects in get_partial_node()
  slub: correctly bootstrap boot caches
  mm/sl[au]b: correct allocation type check in kmalloc_slab()
  slab: Fixup CONFIG_PAGE_ALLOC/DEBUG_SLAB_LEAK sections
  slab: Handle ARCH_DMA_MINALIGN correctly
  slab: Common definition for kmem_cache_node
  slab: Rename list3/l3 to node
  slab: Common Kmalloc cache determination
  stat: Use size_t for sizes instead of unsigned
  slab: Common function to create the kmalloc array
  slab: Common definition for the array of kmalloc caches
  slab: Common constants for kmalloc boundaries
  slab: Rename nodelists to node
  slab: Common name for the per node structures
  ...

Pull misc kbuild updates from Michal Marek:
 "Non-critical kbuild changes:

   - make coccicheck improvements, but no new semantic patches this time

   - make rpm improvements

   - make tar-pkg change to include the architecture in the filename.

     This is a deliberate incompatibility, but nobody has complained so
     far and it is useful if you build for different architectures.  It
     also matches what the deb-pkg and rpm-pkg targets produce.

   - kbuild documentation fix"

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  rpm-pkg: Remove pointless set -e statements
  rpm-pkg: Always regenerate the specfile
  rpm-pkg: Do not write to the parent directory
  rpm-pkg: Do not package the whole source directory
  buildtar: Add ARCH to the archive name
  Coccinelle: Fix patch output when coccicheck is used with M= and C=
  Coccinelle: Add support to the SPFLAGS variable
  Coccinelle: Cleanup the setting of the FLAGS and OPTIONS variables
  Coccinelle: Restore coccicheck verbosity in ONLINE mode (C=1 or C=2)
  scripts/package/Makefile: compare objtree with srctree instead of test KBUILD_OUTPUT
  doc: change example to existing Makefile fragment
  scripts/tags.sh: Add magic for OFFSET and DEFINE

Pull kconfig updates from Michal Marek:
 - use pkg-config to detect curses libraries
 - clean up the way curses headers are searched
 - Some randconfig fixes, of which one had to be reverted
 - KCONFIG_SEED for randconfig debugging
 - memuconfig memory leak plugged
 - menuconfig > breadcrumbs > navigation
 - xconfig compilation fix
 - Other minor fixes

* 'kconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  kconfig: fix lists definition for C++
  Revert "kconfig: fix randomising choice entries in presence of KCONFIG_ALLCONFIG"
  kconfig: implement KCONFIG_PROBABILITY for randconfig
  kconfig: allow specifying the seed for randconfig
  kconfig: fix randomising choice entries in presence of KCONFIG_ALLCONFIG
  kconfig: do not override symbols already set
  kconfig: fix randconfig tristate detection
  kconfig/lxdialog: rationalise the include paths where to find {.n}curses{,w}.h
  menuconfig: Add "breadcrumbs" navigation aid
  menuconfig: Fix memory leak introduced by jump keys feature
  merge_config.sh: Avoid creating unnessary source softlinks
  kconfig: optionally use pkg-config to detect ncurses libs
  menuconfig: optionally use pkg-config to detect ncurses libs

Pull kbuild changes from Michal Marek:
 "Kbuild commits for v3.10-rc1:

   - Fix make mrproper after mod/file2alias rework
   - Fix ld-option Makefile function
   - Rewrite headers_install to shell to drop Perl dependency.

  There are some more patches I have to look at, so I might send another
  pull request later.  Or just queue them for 3.11."

* 'kbuild' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  Fix cleaning in scripts/mod
  headers_install.pl: convert to headers_install.sh
  kbuild: fix ld-option function

Searching for PPC_EFIKA results in a segmentation fault, and it's
because get_symbol_prop() returns NULL.

In this case CONFIG_PPC_EFIKA is defined in arch/powerpc/platforms/
52xx/Kconfig, so it won't be parsed if ARCH!=PPC, but menuconfig knows
this symbol when it parses sound/soc/fsl/Kconfig:

    config SND_MPC52xx_SOC_EFIKA
        tristate "SoC AC97 Audio support for bbplan Efika and STAC9766"
        depends on PPC_EFIKA

This bug was introduced by commit bcdedcc1

 ("menuconfig: print more
info for symbol without prompts").

Reported-and-tested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Li Zefan <lizefan@huawei.com>
Tested-by: Libo Chen <libo.chen@huawei.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

A scheduling while atomic bug was introduced recently (by commit
ce43a216

: "e1000e: cleanup USLEEP_RANGE checkpatch checks").

Revert the particular instance of usleep_range() which causes the bug.

Reported-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Change explicit "signed long" declarations into plain "long" as suggested
by Peter Hurley.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Reviewed-by: Michel Lespinasse <walken@google.com>
Signed-off-by: Michel Lespinasse <walken@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

modify __down_write[_nested] and __down_write_trylock to grab the write
lock whenever the active count is 0, even if there are queued waiters
(they must be writers pending wakeup, since the active count is 0).

Note that this is an optimization only; architectures without this
optimization will still work fine:

- __down_write() would take the slow path which would take the wait_lock
  and then try stealing the lock (as in the spinlocked rwsem implementation)

- __down_write_trylock() would fail, but callers must be ready to deal
  with that - since there are some writers pending wakeup, they could
  have raced with us and obtained the lock before we steal it.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

This change fixes a race condition where a reader might determine it
needs to block, but by the time it acquires the wait_lock the rwsem has
active readers and no queued waiters.

In this situation the reader can run in parallel with the existing
active readers; it does not need to block until the active readers
complete.

Thanks to Peter Hurley for noticing this possible race.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

When we decide to wake up readers, we must first grant them as many read
locks as necessary, and then actually wake up all these readers.  But in
order to know how many read shares to grant, we must first count the
readers at the head of the queue.  This might take a while if there are
many readers, and we want to be protected against a writer stealing the
lock while we're counting.  To that end, we grant the first reader lock
before counting how many more readers are queued.

We also require some adjustments to the wake_type semantics.

RWSEM_WAKE_NO_ACTIVE used to mean that we had found the count to be
RWSEM_WAITING_BIAS, in which case the rwsem was known to be free as
nobody could steal it while we hold the wait_lock.  This doesn't make
sense once we implement fastpath write lock stealing, so we now use
RWSEM_WAKE_ANY in that case.

Similarly, when rwsem_down_write_failed found that a read lock was
active, it would use RWSEM_WAKE_READ_OWNED which signalled that new
readers could be woken without checking first that the rwsem was
available.  We can't do that anymore since the existing readers might
release their read locks, and a writer could steal the lock before we
wake up additional readers.  So, we have to use a new RWSEM_WAKE_READERS
value to indicate we only want to wake readers, but we don't currently
hold any read lock.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

This is mostly for cleanup value:

- We don't need several gotos to handle the case where the first
  waiter is a writer. Two simple tests will do (and generate very
  similar code).

- In the remainder of the function, we know the first waiter is a reader,
  so we don't have to double check that. We can use do..while loops
  to iterate over the readers to wake (generates slightly better code).

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

We can skip the initial trylock in rwsem_down_write_failed() if there
are known active lockers already, thus saving one likely-to-fail
cmpxchg.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

In rwsem_down_write_failed(), if there are active locks after we wake up
(i.e.  the lock got stolen from us), skip taking the wait_lock and go
back to sleep immediately.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Using rwsem_atomic_update to try stealing the write lock forced us to
undo the adjustment in the failure path.  We can have simpler and faster
code by using cmpxchg instead.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Some small code simplifications can be achieved by doing more agressive
lock stealing:

- When rwsem_down_write_failed() notices that there are no active locks
  (and thus no thread to wake us if we decided to sleep), it used to wake
  the first queued process. However, stealing the lock is also sufficient
  to deal with this case, so we don't need this check anymore.

- In try_get_writer_sem(), we can steal the lock even when the first waiter
  is a reader. This is correct because the code path that wakes readers is
  protected by the wait_lock. As to the performance effects of this change,
  they are expected to be minimal: readers are still granted the lock
  (rather than having to acquire it themselves) when they reach the front
  of the wait queue, so we have essentially the same behavior as in
  rwsem-spinlock.

Signed-off-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>