Pull clock framework updates from Michael Turquette:
 "The changes to the common clock framework for 4.0 are mostly new clock
  drivers and updates to existing ones for feature enhancements and bug
  fixes.

  There is more churn than usual in the framework core due to the change
  to introduce per-user unique struct clk pointers in 4.0.  This caused
  several regressions to surface, some of which were sent as fixes to
  4.0.  New generic clock drivers were added for GPIO- and PWM-based
  clock controllers.

  Additionally the common clk-divider code recieved several fixes to the
  way it rounds rates"

* tag 'clk-for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: (91 commits)
  clk: check ->determine/round_rate() return value in clk_calc_new_rates
  clk: at91: usb: propagate rate modification to the parent clk
  clk: samsung: exynos4: Disable ARMCLK down feature on Exynos4210 SoC
  clk: don't use __initconst for non-const arrays
  clk: at91: change to using endian agnositc IO
  clk: clk-gpio-gate: Fix active low
  clk: Add PWM clock driver
  clk: Add clock driver for mb86s7x
  clk: pxa: pxa3xx: add missing os timer clock
  clk: tegra: Use the proper parent for plld_dsi
  clk: tegra: Use generic tegra_osc_clk_init() on Tegra114
  clk: tegra: Model oscillator as clock
  clk: tegra: Add peripheral registers for bank Y
  clk: tegra: Register the proper number of resets
  clk: tegra: Remove needless initializations
  clk: tegra: Use consistent indentation
  clk: tegra: Various whitespace cleanups
  clk: tegra: Enable HDA to HDMI clocks on Tegra124
  clk: tegra: Fix a bunch of sparse warnings
  clk: tegra: Fix typo tabel -> table
  ...

This reverts commit e2ac55b6.

Huang Ying reports that this causes a hang at boot with debugfs disabled.

It is true that the debugfs error checks are kind of confusing, and this
code certainly merits more cleanup and thinking about it, but there's
something wrong with the trivial "check not just for NULL, but for error
pointers too" patch.

Yes, with debugfs disabled, we will end up setting the o2hb_debug_dir
pointer variable to an error pointer (-ENODEV), and then continue as if
everything was fine.  But since debugfs is disabled, all the _users_ of
that pointer end up being compiled away, so even though the pointer can
not be dereferenced, that's still fine.

So it's confusing and somewhat questionable, but the "more correct"
error checks end up causing more trouble than they fix.
Reported-by: NHuang Ying <ying.huang@intel.com>
Acked-by: NAndrew Morton <akpm@linux-foundation.org>
Acked-by: NChengyu Song <csong84@gatech.edu>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Pull remoteproc update from Ohad Ben-Cohen:
 "Suman Anna is adding remoteproc support for processors not behind
  IOMMUs"

* tag 'remoteproc-4.1-next' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
  remoteproc: add IOMMU hardware capability flag

Pull misc kbuild updates:
 "This is the remaining part of kbuild stuff for v4.1-rc1:

   - One wew coccinelle script and a clarification of the proposed fix
     in bugon.coccinelle

   - CONFIG_KERNEL_LZ4 support for extract-ikconfig"

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  scripts/coccinelle/misc/bugon.cocci: update bug_on conversion warning
  scripts/extract-ikconfig: Support LZ4-compressed images.
  irqf_oneshot.cocci: add check of devm_request_threaded_irq()

Pull SCSI fixes from James Bottomley:
 "I'd like to say these were a set of regressions for the recent merge
  window code.  Unfortunately, they all predate the merge window code
  (stable cc'd).

  There are two fixes for data integrity (mostly only showing up on
  module removal), an mvsas crash with expander attached SATA devices
  which goes back to the dawn of the driver but is only just being
  picked up as sas expanders become a standard item in low end server
  hardware, an am53c974 one because the interrupt data isn't fully
  initialised before the line is and a megaraid_sas one because it uses
  smp_processor_id() to select MSI-X queues and that now triggers a
  WARN_ON()"

* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  mvsas: fix panic on expander attached SATA devices
  am53c974: Fix crash during modprobe
  megaraid_sas: use raw_smp_processor_id()
  sd: Fix missing ATO tag check
  sd: Unregister integrity profile

Pull fbdev updates from Tomi Valkeinen:
 "Small fixes and improvements to various fbdev drivers"

* tag 'fbdev-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux: (24 commits)
  omapdss: extend pm notifier to handle hibernation
  OMAPDSS: Correct video ports description file path in DT binding doc
  OMAPDSS: disable VT switch
  fbdev: sh_mobile_lcdc: Fix destruction of uninitialized mutex
  video: fbdev: sh_mobile_lcdcfb: Fix ROP3 sysfs attribute parsing
  fbdev: pm3fb: cleanup some confusing indenting
  hyperv: hyperv_fb: match wait_for_completion_timeout return type
  video: fbdev: use msecs_to_jiffies for time conversions
  fbdev: via/via_clock: fix sparse warning
  video: fbdev: make of_device_id array const
  fbdev: sm501fb: use memset_io
  OMAPDSS: workaround for MFLAG + NV12 issue
  OMAPDSS: Add support for MFLAG
  OMAPDSS: setup default fifo thresholds
  OMAPDSS: DISPC: lock access to DISPC_CONTROL & DISPC_CONFIG
  OMAPDSS: DISPC: fix div by zero issue in overlay scaling
  OMAPDSS: DISPC: change sync_pclk_edge default value
  OMAPDSS: change signal_level & signal_edge enum values
  OMAPDSS: DISPC: explicit handling for sync and de levels
  OMAPDSS: DISPC: remove OMAPDSS_DRIVE_SIG_OPPOSITE_EDGES
  ...

Pull drm updates from Dave Airlie:
 "Highlights:

  Core:
   - Virtual GEM layer merged, this has been around for a long time, and
     it provides a software backed device that allows userspace to use
     it as a GEM shared memory handler.  This makes it a lot easier to
     do certain things when you have no GPU but still have to deal with
     DRI expectations.
   - atomic helper updates.
   - framebuffer modifier interface added.
   - i2c over auxch displayport fixes.
   - fb width/height confusion fixes.
   - new driver for ps8622/ps8625 bridge chips
   - lots of new panels

  i915:
   - more plane atomic conversion
   - vGPU guest support for XenGT
   - Skylake workarounds and fixes
   - Y-tiling support
   - work on dynamic pagetable allocation
   - EU count report param for gen9+
   - CHV fixes (no longer prelim)
   - remove ilk rc6
   - frontbuffer tracking for fbc
   - Displayport link rate refactoring
   - sprite colorkey refactor

  radeon:
   - Displayport MST support (not enabled by default)
   - non-ATOM native hw auxch support (DCE5+)
   - output csc support
   - new queries for userspace debug support
   - new VCE packet

  nouveau:
   - gk20a iommu support
   - gm107 graphics support
   - more gm20x bringup (waiting on signed nvidia fw).

  amdkfd:
   - multiple kgd instance support
   - use 64-bit time accessors

  msm:
   - stolen memory support
   - DSI and dual-DSI support
   - snapdragon 410 support

  exynos:
   - cleanups for atomic and pageflip

  imx-drm:
   - more media-bus formats
   - TV output prep
   - drm panel support

  tegra:
   - hw vblank counter using host1x syncpoints

  omap:
   - universal plane support
   - prep work for atomic modesetting

  rcar-du:
   - ported to atomic modesetting

  atmel-hlcdc:
   - ported to atomic modesetting
   - added suspend/resume support

  sti:
   - ported to atomic modesetting

  dwhdmi:
   - more compliant audio support
   - update rockchip phy support

  tda998x:
   - DT probing for attached crtcs
   - simplified EDID reading

  rockchip:
   - fixes

  adv7511:
   - fixes"

* 'drm-next-merged' of git://people.freedesktop.org/~airlied/linux: (689 commits)
  media-bus: Fixup RGB444_1X12, RGB565_1X16, and YUV8_1X24 media bus format
  drm/i915: Dont enable CS_PARSER_ERROR interrupts at all
  drm/i915: Move drm_framebuffer_unreference out of struct_mutex for takeover
  drm: fix trivial typo mistake
  drm: Make integer overflow checking cover universal cursor updates (v2)
  drm/nouveau/bios: fix fetching from acpi on certain systems
  drm/nouveau/gr/gm206: initial init+ctx code
  drm/nouveau/ce/gm206: enable support via gm204 code
  drm/nouveau/fifo/gm206: enable support via gm204 code
  drm/nouveau/gr/gm204: initial init+ctx code
  drm/nouveau: support for buffer moves via MaxwellDmaCopyA
  drm/nouveau/ce/gm204: initial support
  drm/nouveau: add support for gm20x fifo channels
  drm/nouveau/fifo/gm204: initial support
  drm/nouveau/gr/gk104-: prevent reading non-existent regs in intr handler
  drm/nouveau/gr/gm107: very slightly demagic part of attrib cb setup
  drm/nouveau/gr/gk104-: correct crop/zrop num_active_fbps setting
  drm/nouveau/gr/gf100-: add symbolic names for classes
  drm/nouveau/gr/gm107: support tpc "strand" ctxsw in gpccs ucode
  drm/nouveau/gr/gf100-: support mmio access with gpc offset from gpccs ucode
  ...

Pull IOMMU updates from Joerg Roedel:
 "Not much this time, but the changes include:

   - moving domain allocation into the iommu drivers to prepare for the
     introduction of default domains for devices

   - fixing the IO page-table code in the AMD IOMMU driver to correctly
     encode large page sizes

   - extension of the PCI support in the ARM-SMMU driver

   - various fixes and cleanups"

* tag 'iommu-updates-v4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (34 commits)
  iommu/amd: Correctly encode huge pages in iommu page tables
  iommu/amd: Optimize amd_iommu_iova_to_phys for new fetch_pte interface
  iommu/amd: Optimize alloc_new_range for new fetch_pte interface
  iommu/amd: Optimize iommu_unmap_page for new fetch_pte interface
  iommu/amd: Return the pte page-size in fetch_pte
  iommu/amd: Add support for contiguous dma allocator
  iommu/amd: Don't allocate with __GFP_ZERO in alloc_coherent
  iommu/amd: Ignore BUS_NOTIFY_UNBOUND_DRIVER event
  iommu/amd: Use BUS_NOTIFY_REMOVED_DEVICE
  iommu/tegra: smmu: Compute PFN mask at runtime
  iommu/tegra: gart: Set aperture at domain initialization time
  iommu/tegra: Setup aperture
  iommu: Remove domain_init and domain_free iommu_ops
  iommu/fsl: Make use of domain_alloc and domain_free
  iommu/rockchip: Make use of domain_alloc and domain_free
  iommu/ipmmu-vmsa: Make use of domain_alloc and domain_free
  iommu/shmobile: Make use of domain_alloc and domain_free
  iommu/msm: Make use of domain_alloc and domain_free
  iommu/tegra-gart: Make use of domain_alloc and domain_free
  iommu/tegra-smmu: Make use of domain_alloc and domain_free
  ...

Pull final removal of deprecated cpus_* cpumask functions from Rusty Russell:
 "This is the final removal (after several years!) of the obsolete
  cpus_* functions, prompted by their mis-use in staging.

  With these function removed, all cpu functions should only iterate to
  nr_cpu_ids, so we finally only allocate that many bits when cpumasks
  are allocated offstack"

* tag 'cpumask-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux: (25 commits)
  cpumask: remove __first_cpu / __next_cpu
  cpumask: resurrect CPU_MASK_CPU0
  linux/cpumask.h: add typechecking to cpumask_test_cpu
  cpumask: only allocate nr_cpumask_bits.
  Fix weird uses of num_online_cpus().
  cpumask: remove deprecated functions.
  mips: fix obsolete cpumask_of_cpu usage.
  x86: fix more deprecated cpu function usage.
  ia64: remove deprecated cpus_ usage.
  powerpc: fix deprecated CPU_MASK_CPU0 usage.
  CPU_MASK_ALL/CPU_MASK_NONE: remove from deprecated region.
  staging/lustre/o2iblnd: Don't use cpus_weight
  staging/lustre/libcfs: replace deprecated cpus_ calls with cpumask_
  staging/lustre/ptlrpc: Do not use deprecated cpus_* functions
  blackfin: fix up obsolete cpu function usage.
  parisc: fix up obsolete cpu function usage.
  tile: fix up obsolete cpu function usage.
  arm64: fix up obsolete cpu function usage.
  mips: fix up obsolete cpu function usage.
  x86: fix up obsolete cpu function usage.
  ...

Pull more s390 updates from Martin Schwidefsky:
 "The big thing in this second merge for s390 is the new eBPF JIT from
  Michael which replaces the old 32-bit backend.

  The remaining commits are bug fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/pci: add locking for fmb access
  s390/pci: extract software counters from fmb
  s390/dasd: Fix unresumed device after suspend/resume having no paths
  s390/dasd: fix unresumed device after suspend/resume
  s390/dasd: fix inability to set a DASD device offline
  s390/mm: Fix memory hotplug for unaligned standby memory
  s390/bpf: Add s390x eBPF JIT compiler backend
  s390: Use bool function return values of true/false not 1/0

Pull m68k fixes from Greg Ungerer:
 "Nothing big, spelling fixes and fix/cleanup for ColdFire eth device setup"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68knommu: fix fec setup warning for ColdFire 5271 builds
  m68knommu: ColdFire 5271 only has a single FEC controller
  m68k: Fix trivial typos in comments

Yes, it should work, but it's a bad idea.  Not only did ARM64 not have
the 16-bit access code (there's a separate patch to add it), it's just
not a good atomic type.  Some architectures fundamentally don't do
atomic accesses in them (alpha), and it's not like it saves any space
here anyway because of structure packing issues.

We normally should aim for flags to be "unsigned int" or "unsigned
long".  And if space is at a premium, use a single byte (although that
causes problems on alpha again).  There might be very special cases
where a 16-byte entity is really wanted, but this is not one of them.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add handling of missed events in omap_dss_pm_notif which are
needed to support hibernation (suspend to disk).
Signed-off-by: NGrygorii Strashko <Grygorii.Strashko@linaro.org>
Signed-off-by: NTomi Valkeinen <tomi.valkeinen@ti.com>

The doc refers to Documentation/devicetree/bindings/video/video-ports.txt
which does not exist. The documentation seems to be outdated and wants to
refer to Documentation/devicetree/bindings/graph.txt instead.
Signed-off-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NTomi Valkeinen <tomi.valkeinen@ti.com>

We don't need VT switch when suspending/resuming, so disable it. This
speeds up suspend/resume.
Signed-off-by: NTomi Valkeinen <tomi.valkeinen@ti.com>
Cc: NeilBrown <neil@brown.name>

The merge is clean, but the arm build fails afterwards,
due to API changes in the regulator tree.

I've included the patch into the merge to fix the build.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Change the constant values for RGB444_1X12, RGB565_1X16, and YUV8_1X24 media
bus formats in anticipation of a merge conflict with the media tree, where
the old values are already taken by RBG888_1X24, RGB888_1X32_PADHI, and
VUY8_1X24, respectively.
Signed-off-by: NPhilipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Pull turbostat update from Len Brown:
 "Updates to the turbostat utility.

  Just one kernel dependency in this batch -- added a #define to
  msr-index.h"

* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
  tools/power turbostat: correct dumped pkg-cstate-limit value
  tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL
  tools/power turbostat: correct DRAM RAPL units on recent Xeon processors
  tools/power turbostat: Initial Skylake support
  tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile
  tools/power turbostat: modprobe msr, if needed
  tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2
  tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names
  x86 msr-index: define MSR_TURBO_RATIO_LIMIT,1,2
  tools/power turbostat: label base frequency
  tools/power turbostat: update PERF_LIMIT_REASONS decoding
  tools/power turbostat: simplify default output

Pull ext4 updates from Ted Ts'o:
 "A few bug fixes and add support for file-system level encryption in
  ext4"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (31 commits)
  ext4 crypto: enable encryption feature flag
  ext4 crypto: add symlink encryption
  ext4 crypto: enable filename encryption
  ext4 crypto: filename encryption modifications
  ext4 crypto: partial update to namei.c for fname crypto
  ext4 crypto: insert encrypted filenames into a leaf directory block
  ext4 crypto: teach ext4_htree_store_dirent() to store decrypted filenames
  ext4 crypto: filename encryption facilities
  ext4 crypto: implement the ext4 decryption read path
  ext4 crypto: implement the ext4 encryption write path
  ext4 crypto: inherit encryption policies on inode and directory create
  ext4 crypto: enforce context consistency
  ext4 crypto: add encryption key management facilities
  ext4 crypto: add ext4 encryption facilities
  ext4 crypto: add encryption policy and password salt support
  ext4 crypto: add encryption xattr support
  ext4 crypto: export ext4_empty_dir()
  ext4 crypto: add ext4 encryption Kconfig
  ext4 crypto: reserve codepoints used by the ext4 encryption feature
  ext4 crypto: add ext4_mpage_readpages()
  ...

The test_data_1_le[] array is a const array of const char *.  To avoid
dropping any const information, we need to use "const char * const *",
not just "const char **".

I'm not sure why the different test arrays end up having different
const'ness, but let's make the pointer we use to traverse them as const
as possible, since we modify neither the array of pointers _or_ the
pointers we find in the array.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This prevents a race between chown() and execve(), where chowning a
setuid-user binary to root would momentarily make the binary setuid
root.

This patch was mostly written by Linus Torvalds.
Signed-off-by: NJann Horn <jann@thejh.net>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Commit 8053871d ("smp: Fix smp_call_function_single_async()
locking") fixed the locking for the asynchronous smp-call case, but in
the process of moving the lock handling around, one of the error cases
ended up not unlocking the call data at all.

This went unnoticed on x86, because this is a "caller is buggy" case,
where the caller is trying to call a non-existent CPU.  But apparently
ARM does that (at least under qemu-arm).  Bindly doing cross-cpu calls
to random CPU's that aren't even online seems a bit fishy, but the error
handling was clearly not correct.

Simply add the missing "csd_unlock()" to the error path.
Reported-and-tested-by: NGuenter Roeck <linux@roeck-us.net>
Analyzed-by: NRabin Vincent <rabin@rab.in>
Acked-by: NIngo Molnar <mingo@kernel.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

They were for use by the deprecated first_cpu() and next_cpu() wrappers,
but sparc used them directly.

They're now replaced by cpumask_first / cpumask_next.  And __next_cpu_nr
is completely obsolete.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Acked-by: NDavid S. Miller <davem@davemloft.net>

Pull sparc fixes from David Miller
 "Unfortunately, I brown paper bagged the generic iommu pool allocator
  by applying the wrong revision of the patch series.

  This reverts the bad one, and puts the right one in"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  iommu-common: Fix PARISC compile-time warnings
  sparc: Make LDC use common iommu poll management functions
  sparc: Make sparc64 use scalable lib/iommu-common.c functions
  Break up monolithic iommu table/lock into finer graularity pools and lock
  sparc: Revert generic IOMMU allocator.

Pull 9pfs updates from Eric Van Hensbergen:
 "Some accumulated cleanup patches for kerneldoc and unused variables as
  well as some lock bug fixes and adding privateport option for RDMA"

* tag 'for-linus-4.1-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  net/9p: add a privport option for RDMA transport.
  fs/9p: Initialize status in v9fs_file_do_lock.
  net/9p: Initialize opts->privport as it should be.
  net/9p: use memcpy() instead of snprintf() in p9_mount_tag_show()
  9p: use unsigned integers for nwqid/count
  9p: do not crash on unknown lock status code
  9p: fix error handling in v9fs_file_do_lock
  9p: remove unused variable in p9_fd_create()
  9p: kerneldoc warning fixes

Sowmini Varadhan says:

====================
Generic IOMMU pooled allocator

Investigation of network performance on Sparc shows a high
degree of locking contention in the IOMMU allocator, and it
was noticed that the PowerPC code has a better locking model.

This patch series tries to extract the generic parts of the
PowerPC code so that it can be shared across multiple PCI
devices and architectures.

v10: resend patchv9 without RFC tag, and a new mail Message-Id,
(previous non-RFC attempt did not show up on the patchwork queue?)

Full revision history below:
v2 changes:
  - incorporate David Miller editorial comments: sparc specific
    fields moved from iommu-common into sparc's iommu_64.h
  - make the npools value an input parameter, for the case when
    the iommu map size is not very large
  - cookie_to_index mapping, and optimizations for span-boundary
    check, for use case such as LDC.

v3: eliminate iommu_sparc, rearrange the ->demap indirection to
    be invoked under the pool lock.

v4: David Miller review changes:
  - s/IOMMU_ERROR_CODE/DMA_ERROR_CODE
  - page_table_map_base and page_table_shift are unsigned long, not u32.

v5: removed ->cookie_to_index and ->demap indirection from the
    iommu_tbl_ops The caller needs to call these functions as needed,
    before invoking the generic arena allocator functions.
    Added the "skip_span_boundary" argument to iommu_tbl_pool_init() for
    those callers like LDC which do no care about span boundary checks.

v6: removed iommu_tbl_ops, and instead pass the ->flush_all as
    an indirection to iommu_tbl_pool_init(); only invoke ->flush_all
    when there is no large_pool, based on the assumption that large-pool
    usage is infrequently encountered

v7: moved pool_hash initialization to lib/iommu-common.c and cleaned up
    code duplication from sun4v/sun4u/ldc.

v8: Addresses BenH comments with one exception: I've left the
    IOMMU_POOL_HASH as is, so that powerpc can tailor it to their
    convenience.  Discard trylock for simple spin_lock to acquire pool

v9: Addresses latest BenH comments: need_flush checks, add support
    for dma mask and align_order.

v10: resend without RFC tag, and new mail Message-Id.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes warnings due to
- no DMA_ERROR_CODE on PARISC,
- sizeof (unsigned long) == 4 bytes on PARISC.
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Note that this conversion is only being done to consolidate the
code and ensure that the common code provides the sufficient
abstraction. It is not expected to result in any noticeable
performance improvement, as there is typically one ldc_iommu
per vnet_port, and each one has 8k entries, with a typical
request for 1-4 pages.  Thus LDC uses npools == 1.
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In iperf experiments running linux as the Tx side (TCP client) with
10 threads results in a severe performance drop when TSO is disabled,
indicating a weakness in the software that can be avoided by using
the scalable IOMMU arena DMA allocation.

Baseline numbers before this patch:
   with default settings (TSO enabled) :    9-9.5 Gbps
   Disable TSO using ethtool- drops badly:  2-3 Gbps.

After this patch, iperf client with 10 threads, can give a
throughput of at least 8.5 Gbps, even when TSO is disabled.
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Investigation of multithreaded iperf experiments on an ethernet
interface show the iommu->lock as the hottest lock identified by
lockstat, with something of the order of  21M contentions out of
27M acquisitions, and an average wait time of 26 us for the lock.
This is not efficient. A more scalable design is to follow the ppc
model, where the iommu_map_table has multiple pools, each stretching
over a segment of the map, and with a separate lock for each pool.
This model allows for better parallelization of the iommu map search.

This patch adds the iommu range alloc/free function infrastructure.
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

I applied the wrong version of this patch series, V4 instead
of V10, due to a patchwork bundling snafu.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

HSW expanded MSR_PKG_CST_CONFIG_CONTROL.Package-C-State-Limit,
from bits[2:0] used by previous implementations, to [3:0].
The value 1000b is unlimited, and is used by BDW and SKL too.
Signed-off-by: NLen Brown <len.brown@intel.com>

turbostat --debug
...
CPUID(0x15): eax_crystal: 2 ebx_tsc: 100 ecx_crystal_hz: 0
TSC: 1200 MHz (24000000 Hz * 100 / 2 / 1000000)
Signed-off-by: NLen Brown <len.brown@intel.com>

While not yet documented in the Software Developer's Manual,
the data-sheet for modern Xeon states that DRAM RAPL ENERGY units
are fixed at 15.3 uJ, rather than being discovered via MSR.

Before this patch, DRAM energy on these products is over-stated by turbostat
because the RAPL units are 4x larger.

ref: "Xeon E5-2600 v3/E5-1600 v3 Datasheet Volume 2"
http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e5-v3-datasheet-vol-2.pdfSigned-off-by: NAndrey Semin <andrey.semin@intel.com>
Signed-off-by: NLen Brown <len.brown@intel.com>

Skylake adds some additional residency counters.

Skylake supports a different mix of RAPL registers
from any previous product.

In most other ways, Skylake is like Broadwell.
Signed-off-by: NLen Brown <len.brown@intel.com>

Since commit ee0778a3
("tools/power: turbostat: make Makefile a bit more capable")
turbostat's Makefile is using

  [...]
  BUILD_OUTPUT    := $(PWD)
  [...]

which obviously causes trouble when building "turbostat" with

  make -C /usr/src/linux/tools/power/x86/turbostat ARCH=x86 turbostat

because GNU make does not update nor guarantee that $PWD is set.

This patch changes the Makefile to use $CURDIR instead, which GNU make
guarantees to set and update (i.e. when using "make -C ...") and also
adds support for the O= option (see "make help" in your root of your
kernel source tree for more details).

Link: https://bugs.gentoo.org/show_bug.cgi?id=533918
Fixes: ee0778a3 ("tools/power: turbostat: make Makefile a bit more capable")
Signed-off-by: NThomas D. <whissi@whissi.de>
Cc: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: NLen Brown <len.brown@intel.com>

Some distros (Ubuntu) ship the msr driver as a module.
If turbosat is run as root on those systems, and discovers
that there is no /dev/cpu/cpu0/msr, it will now "modprobe msr"
for the user.

If not root, the modprobe attempt will fail, and turbostat will exit as before:

turbostat: no /dev/cpu/0/msr, Try "# modprobe msr" : No such file or directory
Signed-off-by: NLen Brown <len.brown@intel.com>

and up to 18 cores of turbo ratio limit
when using the turbostat --debug option.
Signed-off-by: NLen Brown <len.brown@intel.com>

s/MSR_NHM_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT/
s/MSR_IVT_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT1/

syntax only -- use the documented strings describing these registers.
Signed-off-by: NLen Brown <len.brown@intel.com>