The huge page size is 4M on non-PAE host, but 2M page size is used in
transparent_hugepage_adjust(), so the page we get after adjust the
mapping level is not the head page, the BUG_ON() will be triggered
Signed-off-by: NXiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

The generic version is both easier to support and more correct.
Signed-off-by: NJonas Bonn <jonas@southpole.se>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This is much the same as for SPARC except that we can do the find_zero()
function more efficiently using the count-leading-zeroes instructions.
Tested on 32-bit and 64-bit PowerPC.
Signed-off-by: NPaul Mackerras <paulus@samba.org>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The aligned_byte_mask() definition is wrong for 32-bit big-endian
machines: the "7-(n)" part of the definition assumes a long is 8
bytes.  This fixes it by using BITS_PER_LONG - 8 instead of 8*7.
Tested on 32-bit and 64-bit PowerPC.
Signed-off-by: NPaul Mackerras <paulus@samba.org>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This makes <asm/word-at-a-time.h> actually live up to its promise of
allowing architectures to help tune the string functions that do their
work a word at a time.

David had already taken the x86 strncpy_from_user() function, modified
it to work on sparc, and then done the extra work to make it generically
useful.  This then expands on that work by making x86 use that generic
version, completing the circle.

But more importantly, it fixes up the word-at-a-time interfaces so that
it's now easy to also support things like strnlen_user(), and pretty
much most random string functions.

David reports that it all works fine on sparc, and Jonas Bonn reported
that an earlier version of this worked on OpenRISC too.  It's pretty
easy for architectures to add support for this and just replace their
private versions with the generic code.

* generic-string-functions:
  sparc: use the new generic strnlen_user() function
  x86: use the new generic strnlen_user() function
  lib: add generic strnlen_user() function
  word-at-a-time: make the interfaces truly generic
  x86: use generic strncpy_from_user routine

Pull i2c-embedded changes from Wolfram Sang:
 "Major changes:

   - lots of devicetree additions for existing drivers.  I tried hard to
     make sure the bindings are proper.  In more complicated cases, I
     requested acks from people having more experience with them than
     me.  That took a bit of extra time and also some time went into
     discussions with developers about what bindings are and what not.
     I have the feeling that the workflow with bindings should be
     improved to scale better.  I will spend some more thought on
     this...

   - i2c-muxes are succesfully used meanwhile, so we dropped
     EXPERIMENTAL for them and renamed the drivers to a standard pattern
     to match the rest of the subsystem.  They can also be used with
     devicetree now.

   - ixp2000 was removed since the whole platform goes away.

   - cleanups (strlcpy instead of strcpy, NULL instead of 0)

   - The rest is typical driver fixes I assume.

  All patches have been in linux-next at least since v3.4-rc6."

Fixed up trivial conflict in arch/arm/mach-lpc32xx/common.c due to the
same patch already having come in through the arm/soc trees, with
additional patches on top of it.

* 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux: (35 commits)
  i2c: davinci: Free requested IRQ in remove
  i2c: ocores: register OF i2c devices
  i2c: tegra: notify transfer-complete after clearing status.
  I2C: xiic: Add OF binding support
  i2c: Rename last mux driver to standard pattern
  i2c: tegra: fix 10bit address configuration
  i2c: muxes: rename first set of drivers to a standard pattern
  of/i2c: implement of_find_i2c_adapter_by_node
  i2c: implement i2c_verify_adapter
  i2c-s3c2410: Add HDMIPHY quirk for S3C2440
  i2c-s3c2410: Rework device type handling
  i2c: muxes are not EXPERIMENTAL anymore
  i2c/of: Automatically populate i2c mux busses from device tree data.
  i2c: Add a struct device * parameter to i2c_add_mux_adapter()
  of/i2c: call i2c_verify_client from of_find_i2c_device_by_node
  i2c: designware: Add clk_{un}prepare() support
  i2c: designware: add PM support
  i2c: ixp2000: remove driver
  i2c: pnx: add device tree support
  i2c: imx: don't use strcpy but strlcpy
  ...

Pull sweeping late_initcall cleanup for arm-soc from Olof Johansson:
 "This is a patch series from Shawn Guo that moves from individual
  late_initcalls() to using a member in the machine structure to invoke
  a platform's late initcalls.

  This cleanup is a step in the move towards multiplatform kernels since
  it would reduce the need to check for compatible platforms in each and
  every initcall."

Fix up trivial conflicts in arch/arm/mach-{exynos/mach-universal_c210.c,
imx/mach-cpuimx51.c, omap2/board-generic.c} due to changes nearby (and,
in the case of cpuimx51.c the board support being deleted)

* tag 'cleanup-initcall' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: ux500: use machine specific hook for late init
  ARM: tegra: use machine specific hook for late init
  ARM: shmobile: use machine specific hook for late init
  ARM: sa1100: use machine specific hook for late init
  ARM: s3c64xx: use machine specific hook for late init
  ARM: prima2: use machine specific hook for late init
  ARM: pnx4008: use machine specific hook for late init
  ARM: omap2: use machine specific hook for late init
  ARM: omap1: use machine specific hook for late init
  ARM: msm: use machine specific hook for late init
  ARM: imx: use machine specific hook for late init
  ARM: exynos: use machine specific hook for late init
  ARM: ep93xx: use machine specific hook for late init
  ARM: davinci: use machine specific hook for late init
  ARM: provide a late_initcall hook for platform initialization

Pull arm-soc: soc specific changes (part 2) from Olof Johansson:
 "This adds support for the spear13xx platform, which has first been
  under review a long time ago and finally been completed after generic
  spear work has gone into the clock, dt and pinctrl branches.

  Also a number of updates for the samsung socs are part of this branch."

Fix up trivial conflicts in drivers/gpio/gpio-samsung.c that look much
worse than they are: the exonys5 init code was refactored in commit
fd454997 ("gpio: samsung: refactor gpiolib init for exynos4/5"), and
then commit f10590c9 ("ARM: EXYNOS: add GPC4 bank instance") added a
new gpio chip define and did tiny updates to the init code.

So the conflict diff looks like hell, but it's actually a fairly simple
change.

* tag 'soc2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (34 commits)
  ARM: exynos: fix building with CONFIG_OF disabled
  ARM: EXYNOS: Add AUXDATA for i2c controllers
  ARM: dts: Update device tree source files for EXYNOS5250
  ARM: EXYNOS: Add device tree support for interrupt combiner
  ARM: EXYNOS: Add irq_domain support for interrupt combiner
  ARM: EXYNOS: Remove a new bus_type instance for EXYNOS5
  ARM: EXYNOS: update irqs for EXYNOS5250 SoC
  ARM: EXYNOS: Add pre-divider and fout mux clocks for bpll and mpll
  ARM: EXYNOS: add GPC4 bank instance
  ARM: EXYNOS: Redefine IRQ_MCT_L0,1 definition
  ARM: EXYNOS: Modify the GIC physical address for static io-mapping
  ARM: EXYNOS: Add watchdog timer clock instance
  pinctrl: SPEAr1310: Fix pin numbers for clcd_high_res
  SPEAr: Update MAINTAINERS and Documentation
  SPEAr13xx: Add defconfig
  SPEAr13xx: Add compilation support
  SPEAr13xx: Add dts and dtsi files
  pinctrl: Add SPEAr13xx pinctrl drivers
  pinctrl: SPEAr: Create macro for declaring GPIO PINS
  SPEAr13xx: Add common clock framework support
  ...

Pull arm-soc device tree conversions (part 2) from Olof Johansson:
 "These continue the device tree work from part 1, this set is for the
  tegra, mxs and imx platforms, all of which have dependencies on clock
  or pinctrl changes submitted earlier."

Fix up trivial conflicts due to nearby changes in
drivers/{gpio/gpio,i2c/busses/i2c}-mxs.c

* tag 'dt2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (73 commits)
  ARM: dt: tegra: invert status=disable vs status=okay
  ARM: dt: tegra: consistent basic property ordering
  ARM: dt: tegra: sort nodes based on bus order
  ARM: dt: tegra: remove duplicate device_type property
  ARM: dt: tegra: consistenly use lower-case for hex constants
  ARM: dt: tegra: format regs properties consistently
  ARM: dt: tegra: gpio comment cleanup
  ARM: dt: tegra: remove unnecessary unit addresses
  ARM: dt: tegra: whitespace cleanup
  ARM: dt: tegra cardhu: fix typo in SDHCI node name
  ARM: dt: tegra: cardhu: register core regulator tps62361
  ARM: dt: tegra30.dtsi: Add SMMU node
  ARM: dt: tegra20.dtsi: Add GART node
  ARM: dt: tegra30.dtsi: Add Memory Controller(MC) nodes
  ARM: dt: tegra20.dtsi: Add Memory Controller(MC) nodes
  ARM: dt: tegra: Add device tree support for AHB
  ARM: dts: enable audio support for imx28-evk
  ARM: dts: enable i2c device for imx28-evk
  i2c: mxs: add device tree probe support
  ARM: dts: enable mmc for imx28-evk
  ...

Pull arm-soc stmp-dev library code from Olof Johansson:
 "A number of devices are using a common register layout, this adds
  support code for it in lib/stmp_device.c so we do not need to
  duplicate it in each driver."

Fix up trivial conflicts in drivers/i2c/busses/i2c-mxs.c and
lib/Makefile

* tag 'stmp-dev' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  i2c: mxs: use global reset function
  lib: add support for stmp-style devices

Pull arm-soc clock driver changes from Olof Johansson:
 "The new clock subsystem was merged in linux-3.4 without any users,
  this now moves the first three platforms over to it: imx, mxs and
  spear.

  The series also contains the changes for the clock subsystem itself,
  since Mike preferred to have it together with the platforms that
  require these changes, in order to avoid interdependencies and
  conflicts."

Fix up trivial conflicts in arch/arm/mach-kirkwood/common.c (code
removed in one branch, added OF support in another) and
drivers/dma/imx-sdma.c (independent changes next to each other).

* tag 'clock' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (97 commits)
  clk: Fix CLK_SET_RATE_GATE flag validation in clk_set_rate().
  clk: Provide dummy clk_unregister()
  SPEAr: Update defconfigs
  SPEAr: Add SMI NOR partition info in dts files
  SPEAr: Switch to common clock framework
  SPEAr: Call clk_prepare() before calling clk_enable
  SPEAr: clk: Add General Purpose Timer Synthesizer clock
  SPEAr: clk: Add Fractional Synthesizer clock
  SPEAr: clk: Add Auxiliary Synthesizer clock
  SPEAr: clk: Add VCO-PLL Synthesizer clock
  SPEAr: Add DT bindings for SPEAr's timer
  ARM i.MX: remove now unused clock files
  ARM: i.MX6: implement clocks using common clock framework
  ARM i.MX35: implement clocks using common clock framework
  ARM i.MX5: implement clocks using common clock framework
  ARM: Kirkwood: Replace clock gating
  ARM: Orion: Audio: Add clk/clkdev support
  ARM: Orion: PCIE: Add support for clk
  ARM: Orion: XOR: Add support for clk
  ARM: Orion: CESA: Add support for clk
  ...

Pull arm-soc cleanups (part 2) from Olof Johansson:
 "More cleanups, continuing an earlier set with omap and samsung
  specific cleanups.  These could not go into the first set because they
  have dependencies on various other series that in turn depend on the
  first cleanups."

Fixed up conflicts in arch/arm/plat-omap/counter_32k.c due to commit
bd0493ea: "move read_{boot,persistent}_clock to the architecture
level" that changed how the persistent clocks were handled.  And trivial
conflicts in arch/arm/mach-omap1/common.h due to just independent
changes close to each other.

* tag 'cleanup2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (35 commits)
  ARM: SAMSUNG: merge plat-s5p into plat-samsung
  ARM: SAMSUNG: move options for common s5p into plat-samsung/Kconfig
  ARM: SAMSUNG: move setup code for s5p mfc and mipiphy into plat-samsung
  ARM: SAMSUNG: move platform device for s5p uart into plat-samsung
  ARM: SAMSUNG: move hr timer for common s5p into plat-samsung
  ARM: SAMSUNG: move pm part for common s5p into plat-samsung
  ARM: SAMSUNG: move interrupt part for common s5p into plat-samsung
  ARM: SAMSUNG: move clock part for common s5p into plat-samsung
  ARM: S3C24XX: Use common macro to define resources on dev-uart.c
  ARM: S3C24XX: move common clock init into common.c
  ARM: S3C24XX: move common power-management code to mach-s3c24xx
  ARM: S3C24XX: move plat-s3c24xx/dev-uart.c into common.c
  ARM: S3C24XX: move plat-s3c24xx/cpu.c
  ARM: OMAP2+: Kconfig: convert SOC_OMAPAM33XX to SOC_AM33XX
  ARM: OMAP2+: Kconfig: convert SOC_OMAPTI81XX to SOC_TI81XX
  GPMC: add ECC control definitions
  ARM: OMAP2+: dmtimer: remove redundant sysconfig context restore
  ARM: OMAP: AM35xx: convert 3517 detection/flags to AM35xx
  ARM: OMAP: AM35xx: remove redunant cpu_is checks for AM3505
  ARM: OMAP1: Pass dma request lines in platform data to MMC driver
  ...

Pull arm-soc driver specific updates from Olof Johansson:
 "These changes are specific to some driver that may be used by multiple
  boards or socs.  The most significant change in here is the move of
  the samsung iommu code from a platform specific in-kernel interface to
  the generic iommu subsystem."

Fix up trivial conflicts in arch/arm/mach-exynos/Kconfig

* tag 'drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (28 commits)
  mmc: dt: Consolidate DT bindings
  iommu/exynos: Add iommu driver for EXYNOS Platforms
  ARM: davinci: optimize the DMA ISR
  ARM: davinci: implement DEBUG_LL port choice
  ARM: tegra: Add SMMU enabler in AHB
  ARM: tegra: Add Tegra AHB driver
  Input: pxa27x_keypad add choice to set direct_key_mask
  Input: pxa27x_keypad direct key may be low active
  Input: pxa27x_keypad bug fix for direct_key_mask
  Input: pxa27x_keypad keep clock on as wakeup source
  ARM: dt: tegra: pinmux changes for USB ULPI
  ARM: tegra: add USB ULPI PHY reset GPIO to device tree
  ARM: tegra: don't hard-code USB ULPI PHY reset_gpio
  ARM: tegra: change pll_p_out4's rate to 24MHz
  ARM: tegra: fix pclk rate
  ARM: tegra: reparent sclk to pll_c_out1
  ARM: tegra: Add pllc clock init table
  ARM: dt: tegra cardhu: basic audio support
  ARM: dt: tegra30.dtsi: Add audio-related nodes
  ARM: tegra: add AUXDATA required for audio
  ...

Pull arm-soc defconfig updates from Olof Johansson:
 "For the first time, we have one branch that collects just updates to
  defconfig files, mostly for adapting to changes in other subsystems."

* tag 'defconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: PRIMA2: add prima2_defconfig for CSR SiRFprimaII
  ARM: tegra: update defconfig
  ARM: tegra: update defconfig
  ARM: imx_v6_v7_defconfig: Add SPI NOR support
  ARM: imx_v4_v5_defconfig: Let CONFIG_MACH_IMX27_DT be built by default

This throws away the sparc-specific functions in favor of the generic
optimized version.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This throws away the old x86-specific functions in favor of the generic
optimized version.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This adds a new generic optimized strnlen_user() function that uses the
<asm/word-at-a-time.h> infrastructure to portably do efficient string
handling.

In many ways, strnlen is much simpler than strncpy, and in particular we
can always pre-align the words we load from memory.  That means that all
the worries about alignment etc are a non-issue, so this one can easily
be used on any architecture.  You obviously do have to do the
appropriate word-at-a-time.h macros.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This changes the interfaces in <asm/word-at-a-time.h> to be a bit more
complicated, but a lot more generic.

In particular, it allows us to really do the operations efficiently on
both little-endian and big-endian machines, pretty much regardless of
machine details.  For example, if you can rely on a fast population
count instruction on your architecture, this will allow you to make your
optimized <asm/word-at-a-time.h> file with that.

NOTE! The "generic" version in include/asm-generic/word-at-a-time.h is
not truly generic, it actually only works on big-endian.  Why? Because
on little-endian the generic algorithms are wasteful, since you can
inevitably do better. The x86 implementation is an example of that.

(The only truly non-generic part of the asm-generic implementation is
the "find_zero()" function, and you could make a little-endian version
of it.  And if the Kbuild infrastructure allowed us to pick a particular
header file, that would be lovely)

The <asm/word-at-a-time.h> functions are as follows:

 - WORD_AT_A_TIME_CONSTANTS: specific constants that the algorithm
   uses.

 - has_zero(): take a word, and determine if it has a zero byte in it.
   It gets the word, the pointer to the constant pool, and a pointer to
   an intermediate "data" field it can set.

   This is the "quick-and-dirty" zero tester: it's what is run inside
   the hot loops.

 - "prep_zero_mask()": take the word, the data that has_zero() produced,
   and the constant pool, and generate an *exact* mask of which byte had
   the first zero.  This is run directly *outside* the loop, and allows
   the "has_zero()" function to answer the "is there a zero byte"
   question without necessarily getting exactly *which* byte is the
   first one to contain a zero.

   If you do multiple byte lookups concurrently (eg "hash_name()", which
   looks for both NUL and '/' bytes), after you've done the prep_zero_mask()
   phase, the result of those can be or'ed together to get the "either
   or" case.

 - The result from "prep_zero_mask()" can then be fed into "find_zero()"
   (to find the byte offset of the first byte that was zero) or into
   "zero_bytemask()" (to find the bytemask of the bytes preceding the
   zero byte).

   The existence of zero_bytemask() is optional, and is not necessary
   for the normal string routines.  But dentry name hashing needs it, so
   if you enable DENTRY_WORD_AT_A_TIME you need to expose it.

This changes the generic strncpy_from_user() function and the dentry
hashing functions to use these modified word-at-a-time interfaces.  This
gets us back to the optimized state of the x86 strncpy that we lost in
the previous commit when moving over to the generic version.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The generic strncpy_from_user() is not really optimal, since it is
designed to work on both little-endian and big-endian.  And on
little-endian you can simplify much of the logic to find the first zero
byte, since little-endian arithmetic doesn't have to worry about the
carry bit propagating into earlier bytes (only later bytes, which we
don't care about).

But I have patches to make the generic routines use the architecture-
specific <asm/word-at-a-time.h> infrastructure, so that we can regain
the little-endian optimizations.  But before we do that, switch over to
the generic routines to make the patches each do just one well-defined
thing.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Pull x86/mce merge window patches from Tony Luck:
 "Including two that make error_context() checks less sucky"

* tag 'x86-mce-merge' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  x86/mce: Add instruction recovery signatures to mce-severity table
  x86/mce: Fix check for processor context when machine check was taken.
  MCE: Fix vm86 handling for 32bit mce handler
  x86/mce Add validation check before GHES error is recorded
  x86/mce: Avoid reading every machine check bank register twice.

Pull tile updates from Chris Metcalf:
 "These changes cover a range of new arch/tile features and
  optimizations.  They've been through LKML review and on linux-next for
  a month or so.  There's also one bug-fix that just missed 3.4, which
  I've marked for stable."

Fixed up trivial conflict in arch/tile/Kconfig (new added tile Kconfig
entries clashing with the generic timer/clockevents changes).

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  tile: default to tilegx_defconfig for ARCH=tile
  tile: fix bug where fls(0) was not returning 0
  arch/tile: mark TILEGX as not EXPERIMENTAL
  tile/mm/fault.c: Port OOM changes to handle_page_fault
  arch/tile: add descriptive text if the kernel reports a bad trap
  arch/tile: allow querying cpu module information from the hypervisor
  arch/tile: fix hardwall for tilegx and generalize for idn and ipi
  arch/tile: support multiple huge page sizes dynamically
  mm: add new arch_make_huge_pte() method for tile support
  arch/tile: support kexec() for tilegx
  arch/tile: support <asm/cachectl.h> header for cacheflush() syscall
  arch/tile: Allow tilegx to build with either 16K or 64K page size
  arch/tile: optimize get_user/put_user and friends
  arch/tile: support building big-endian kernel
  arch/tile: allow building Linux with transparent huge pages enabled
  arch/tile: use interrupt critical sections less

Now we have four copies of this code, Linus "suggested" it was about time
we stopped copying it and turned it into a helper.
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add device info into list before doing context mapping, because device
info will be used by iommu_enable_dev_iotlb(). Without it, ATS won't get
enabled as it should be.

ATS, while a dubious decision from a security point of view, can be very
important for performance.
Signed-off-by: NXudong Hao <xudong.hao@intel.com>
Signed-off-by: NXiantao Zhang <xiantao.zhang@intel.com>
Acked-by: NChris Wright <chrisw@sous-sol.org>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>
Cc: stable@kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is no "ARCH=tile" (just like there is no "ARCH=x86") so we need
to pick a default configuration, either tilepro or tilegx, when users
specify ARCH=tile.  We'll use tilegx, since that's our current chip.
Reported-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

This is because __builtin_clz(0) returns 64 for the "undefined" case
of 0, since the builtin just does a right-shift 32 and "clz" instruction.
So, use the alpha approach of casting to u32 and using __builtin_clzll().

Cc: stable@vger.kernel.org
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

Also create a TILEPRO config setting to use for #ifdefs where it
is cleaner to do so, and make the 64BIT setting depend directly
on the setting of TILEGX.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

Commit d065bd81
(mm: retry page fault when blocking on disk transfer) and
commit 37b23e05
(x86,mm: make pagefault killable)

The above commits introduced changes into the x86 pagefault handler
for making the page fault handler retryable as well as killable.

These changes reduce the mmap_sem hold time, which is crucial
during OOM killer invocation.

Port these changes to tile.
Signed-off-by: NKautuk Consul <consul.kautuk@gmail.com>
[cmetcalf@tilera.com: initialize "flags" after "write" updated.]
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

If the kernel unexpectedly takes a bad trap, it's convenient to
have it report the type of trap as part of the error.  This gives
customers a bit more context before they call up customer support.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

This just adds a few more attributes to the information Linux
can query from the hypervisor for the /sys/hypervisor/board/ directory,
providing part, serial#, revision#, and description for cpu modules
(as opposed to the board itself, or any mezzanine boards).
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

The hardwall drain code was not properly implemented for tilegx,
just tilepro, so you couldn't reliably restart an application that
made use of the udn.

In addition, the code was only applicable to the udn (user dynamic
network).  On tilegx there is a second user network that is available
(the "idn"), and there is support for having I/O shims deliver
user-level interrupts to applications ("ipi") which functions in a
very similar way to the inter-core permissions used for udn/idn.
So this change also generalizes the code from supporting just the udn
to supports udn/idn/ipi on tilegx.

By default we now use /dev/hardwall/{udn,idn,ipi} with separate
minor numbers for the three devices.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

This change adds support for a new "super" bit in the PTE, using the new
arch_make_huge_pte() method.  The Tilera hypervisor sees the bit set at a
given level of the page table and gangs together 4, 16, or 64 consecutive
pages from that level of the hierarchy to create a larger TLB entry.

One extra "super" page size can be specified at each of the three levels
of the page table hierarchy on tilegx, using the "hugepagesz" argument
on the boot command line.  A new hypervisor API is added to allow Linux
to tell the hypervisor how many PTEs to gang together at each level of
the page table.

To allow pre-allocating huge pages larger than the buddy allocator can
handle, this change modifies the Tilera bootmem support to put all of
memory on tilegx platforms into bootmem.

As part of this change I eliminate the vestigial CONFIG_HIGHPTE support,
which never worked anyway, and eliminate the hv_page_size() API in favor
of the standard vma_kernel_pagesize() API.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

The tile support for multiple-size huge pages requires tagging
the hugetlb PTE with a "super" bit for PTEs that are multiples of
the basic size of a pagetable span.  To set that bit properly
we need to tweak the PTe in make_huge_pte() based on the vma.

This change provides the API for a subsequent tile-specific
change to use.
Reviewed-by: NHillf Danton <dhillf@gmail.com>
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

We already had a syscall that did some dcache flushing, but it was
not used in practice.  Make it MIPS compatible instead so it can
do both the DCACHE and ICACHE actions.  We have code that wants to
be able to use the ICACHE flush mode from userspace so this change
enables that.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

This change introduces new flags for the hv_install_context()
API that passes a page table pointer to the hypervisor.  Clients
can explicitly request 4K, 16K, or 64K small pages when they
install a new context.  In practice, the page size is fixed at
kernel compile time and the same size is always requested every
time a new page table is installed.

The <hv/hypervisor.h> header changes so that it provides more abstract
macros for managing "page" things like PFNs and page tables.  For
example there is now a HV_DEFAULT_PAGE_SIZE_SMALL instead of the old
HV_PAGE_SIZE_SMALL.  The various PFN routines have been eliminated and
only PA- or PTFN-based ones remain (since PTFNs are always expressed
in fixed 2KB "page" size).  The page-table management macros are
renamed with a leading underscore and take page-size arguments with
the presumption that clients will use those macros in some single
place to provide the "real" macros they will use themselves.

I happened to notice the old hv_set_caching() API was totally broken
(it assumed 4KB pages) so I changed it so it would nominally work
correctly with other page sizes.

Tag modules with the page size so you can't load a module built with
a conflicting page size.  (And add a test for SMP while we're at it.)
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

Use direct load/store for the get_user/put_user.

Previously, we would call out to a helper routine that would do the
appropriate thing and then return, handling the possible exception
internally.  Now we inline the load or store, along with a "we succeeded"
indication in a register; if the load or store faults, we write a
"we failed" indication into the same register and then return to the
following instruction.  This is more efficient and gives us more compact
code, as well as being more in line with what other architectures do.

The special futex assembly source file for TILE-Gx also disappears in
this change; we just use the same inlining idiom there as well, putting
the appropriate atomic operations directly into futex_atomic_op_inuser()
(and thus into the FUTEX_WAIT function).

The underlying atomic copy_from_user, copy_to_user functions were
renamed using the (cryptic) x86 convention as copy_from_user_ll and
copy_to_user_ll.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

The toolchain supports big-endian mode now, so add support for building
the kernel to run big-endian as well.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

The change adds some infrastructure for managing tile pmd's more generally,
using pte_pmd() and pmd_pte() methods to translate pmd values to and
from ptes, since on TILEPro a pmd is really just a nested structure
holding a pgd (aka pte).  Several existing pmd methods are moved into
this framework, and a whole raft of additional pmd accessors are defined
that are used by the transparent hugepage framework.

The tile PTE now has a "client2" bit.  The bit is used to indicate a
transparent huge page is in the process of being split into subpages.

This change also fixes a generic bug where the return value of the
generic pmdp_splitting_flush() was incorrect.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

In general we want to avoid ever touching memory while within an
interrupt critical section, since the page fault path goes through
a different path from the hypervisor when in an interrupt critical
section, and we carefully decided with tilegx that we didn't need
to support this path in the kernel.  (On tilepro we did implement
that path as part of supporting atomic instructions in software.)

In practice we always need to touch the kernel stack, since that's
where we store the interrupt state before releasing the critical
section, but this change cleans up a few things.  The IRQ_ENABLE
macro is split up so that when we want to enable interrupts in a
deferred way (e.g. for cpu_idle or for interrupt return) we can
read the per-cpu enable mask before entering the critical section.
The cache-migration code is changed to use interrupt masking instead
of interrupt critical sections.  And, the interrupt-entry code is
changed so that we defer loading "tp" from per-cpu data until after
we have released the interrupt critical section.
Signed-off-by: NChris Metcalf <cmetcalf@tilera.com>

Pull dma-buf updates from Sumit Semwal:
 "Here's the first signed-tag pull request for dma-buf framework.  It
  includes the following key items:
   - mmap support
   - vmap support
   - related documentation updates

  These are needed by various drivers to allow mmap/vmap of dma-buf
  shared buffers.  Dave Airlie has some prime patches dependent on the
  vmap pull as well."

* tag 'tag-for-linus-3.5' of git://git.linaro.org/people/sumitsemwal/linux-dma-buf:
  dma-buf: add initial vmap documentation
  dma-buf: minor documentation fixes.
  dma-buf: add vmap interface
  dma-buf: mmap support