The cache hierarchy can be identified through Cache Level ID(CLIDR)
architected system register. However in some cases it will provide
only the number of cache levels that are integrated into the processor
itself. In other words, it can't provide any information about the
caches that are external and/or transparent.

Some platforms require to export the information about all such external
caches to the userspace applications via the sysfs interface.

This patch adds support to override the cache levels using device tree
to take such external non-architected caches into account.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Tested-by: Tan Xiaojun <tanxiaojun@huawei.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

It is useful to have helper function just to get the number of cache
levels for a given logical cpu. We can obtain the same by just checking
the level at which the last cache is present. This patch adds support
to find the level of the last cache for a given cpu.

It will be used on ARM64 platform where the device tree provides the
information for the additional non-architected/transparent/external
last level caches that are not integrated with the processors.

Cc: Mark Rutland <mark.rutland@arm.com>
Suggested-by: Rob Herring <robh+dt@kernel.org>
Acked-by: Rob Herring <robh+dt@kernel.org>
Tested-by: Tan Xiaojun <tanxiaojun@huawei.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
[will: use u32 instead of int for cache_level]
Signed-off-by: Will Deacon <will.deacon@arm.com>

Definition of cpu ranges are hard to read if the cpu variant is not
zero. Provide MIDR_CPU_VAR_REV() macro to describe the full hardware
revision of a cpu including variant and (minor) revision.

Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Cosmetic change to use phys_addr_t instead of unsigned long for the
return value of __pa_symbol().

Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Miles Chen <miles.chen@mediatek.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

ARM v8.1 extensions include support for rounding double multiply
add/subtract instructions to the A64 SIMD instructions set. Let
the userspace know about it via a HWCAP bit.

Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

This patch adds support for DMA_ATTR_SKIP_CPU_SYNC attribute for
dma_{un}map_{page,sg} functions family to swiotlb.

DMA_ATTR_SKIP_CPU_SYNC allows platform code to skip synchronization of
the CPU cache for the given buffer assuming that it has been already
transferred to 'device' domain.

Ported from IOMMU .{un}map_{sg,page} ops.

Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Will Deacon <will.deacon@arm.com>

x86 has an option CONFIG_DEBUG_VIRTUAL to do additional checks
on virt_to_phys calls. The goal is to catch users who are calling
virt_to_phys on non-linear addresses immediately. This inclues callers
using virt_to_phys on image addresses instead of __pa_symbol. As features
such as CONFIG_VMAP_STACK get enabled for arm64, this becomes increasingly
important. Add checks to catch bad virt_to_phys usage.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

__pa_symbol is technically the marcro that should be used for kernel
symbols. Switch to this as a pre-requisite for DEBUG_VIRTUAL which
will do bounds checking.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

virt_to_pfn lacks a cast at the top level. Don't rely on __virt_to_phys
and explicitly cast to unsigned long.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Several macros for various x_to_y exist outside the bounds of an
__ASSEMBLY__ guard. Move them in preparation for support for
CONFIG_DEBUG_VIRTUAL.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Merge core DEBUG_VIRTUAL changes from Laura Abbott. Later arm and arm64
support depends on these.

* aarch64/for-next/debug-virtual:
  drivers: firmware: psci: Use __pa_symbol for kernel symbol
  mm/usercopy: Switch to using lm_alias
  mm/kasan: Switch to using __pa_symbol and lm_alias
  kexec: Switch to __pa_symbol
  mm: Introduce lm_alias
  mm/cma: Cleanup highmem check
  lib/Kconfig.debug: Add ARCH_HAS_DEBUG_VIRTUAL

Documentation for the infrastructure to expose CPU feature
register by emulating MRS.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Dave Martin <dave.martin@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

This patch adds the hook for emulating MRS instruction to
export the 'user visible' value of supported system registers.
We emulate only the following id space for system registers:

 Op0=3, Op1=0, CRn=0, CRm=[0, 4-7]

The rest will fall back to SIGILL. This capability is also
advertised via a new HWCAP_CPUID.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
[will: add missing static keyword to enable_mrs_emulation]
Signed-off-by: Will Deacon <will.deacon@arm.com>

__pa_symbol is technically the macro that should be used for kernel
symbols. Switch to this as a pre-requisite for DEBUG_VIRTUAL which
will do bounds checking.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

The usercopy checking code currently calls __va(__pa(...)) to check for
aliases on symbols. Switch to using lm_alias instead.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

__pa_symbol is the correct API to find the physical address of symbols.
Switch to it to allow for debugging APIs to work correctly. Other
functions such as p*d_populate may call __pa internally. Ensure that the
address passed is in the linear region by calling lm_alias.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

__pa_symbol is the correct api to get the physical address of kernel
symbols. Switch to it to allow for better debug checking.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Certain architectures may have the kernel image mapped separately to
alias the linear map. Introduce a macro lm_alias to translate a kernel
image symbol into its linear alias. This is used in part with work to
add CONFIG_DEBUG_VIRTUAL support for arm64.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

6b101e2a

 ("mm/CMA: fix boot regression due to physical address of
high_memory") added checks to use __pa_nodebug on x86 since
CONFIG_DEBUG_VIRTUAL complains about high_memory not being linearlly
mapped. arm64 is now getting support for CONFIG_DEBUG_VIRTUAL as well.
Rather than add an explosion of arches to the #ifdef, switch to an
alternate method to calculate the physical start of highmem using
the page before highmem starts. This avoids the need for the #ifdef and
extra __pa_nodebug calls.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

DEBUG_VIRTUAL currently depends on DEBUG_KERNEL && X86. arm64 is getting
the same support. Rather than add a list of architectures, switch this
to ARCH_HAS_DEBUG_VIRTUAL and let architectures select it as
appropriate.

Acked-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Track the user visible fields of a CPU feature register. This will be
used for exposing the value to the userspace. All the user visible
fields of a feature register will be passed on as it is, while the
others would be filled with their respective safe value.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Add a helper to extract the register field from a given
instruction.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Define helper macros to extract op0, op1, CRn, CRm & op2
for a given sys_reg id. While at it remove the explicit
masking only used for Op0.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Document the rules for choosing the safe value for different types
of features.

Cc: Dave Martin <dave.martin@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

This patch does the following clean ups :

1) All undescribed fields of a register are now treated as 'strict'
   with a safe value of 0. Hence we could leave an empty table for
   describing registers which are RAZ.

2) ID_AA64DFR1_EL1 is RAZ and should use the table for RAZ register.

3) ftr_generic32 is used to represent a register with a 32bit feature
   value. Rename this to ftr_singl32 to make it more obvious. Since
   we don't have a 64bit singe feature register, kill ftr_generic.

Based on a patch by Mark Rutland.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

We currently have some RAZ fields described explicitly in our
arm64_ftr_bits arrays. These are inconsistently commented, grouped,
and/or applied, and maintaining these is error-prone.

Luckily, we don't need these at all. We'll never need to inspect RAZ
fields to determine feature support, and init_cpu_ftr_reg() will ensure
that any bits without a corresponding arm64_ftr_bits entry are treated
as RES0 with strict matching requirements. In check_update_ftr_reg()
we'll then compare these bits from the relevant cpuinfo_arm64
structures, and need not store them in a arm64_ftr_reg.

This patch removes the unnecessary arm64_ftr_bits entries for RES0 bits.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Any fields not defined in an arm64_ftr_bits entry are propagated to the
system-wide register value in init_cpu_ftr_reg(), and while we require
that these strictly match for the sanity checks, we don't update them in
update_cpu_ftr_reg().

Generally, the lack of an arm64_ftr_bits entry indicates that the bits
are currently RES0 (as is the case for the upper 32 bits of all
supposedly 32-bit registers).

A better default would be to use zero for the system-wide value of
unallocated bits, making all register checking consistent, and allowing
for subsequent simplifications to the arm64_ftr_bits arrays.

This patch updates init_cpu_ftr_reg() to treat unallocated bits as RES0
for the purpose of the system-wide safe value. These bits will still be
sanity checked with strict match requirements, as is currently the case.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

The statistical profiling extension (SPE) is an optional feature of
ARMv8.1 and is unlikely to be supported by all of the CPUs in a
heterogeneous system.

This patch updates the cpufeature checks so that such systems are not
tainted as unsupported.

Acked-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Perf already supports multiple PMU instances for heterogeneous systems,
so there's no need to be strict in the cpufeature checking, particularly
as the PMU extension is optional in the architecture.

Acked-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Since its introduction, the UAO enable call was broken, and useless.
commit 2a6dcb2b

 ("arm64: cpufeature: Schedule enable() calls instead
of calling them via IPI"), fixed the framework so that these calls
are scheduled, so that they can modify PSTATE.

Now it is just useless. Remove it. UAO is enabled by the code patching
which causes get_user() and friends to use the 'ldtr' family of
instructions. This relies on the PSTATE.UAO bit being set to match
addr_limit, which we do in uao_thread_switch() called via __switch_to().

All that is needed to enable UAO is patch the code, and call schedule().
__apply_alternatives_multi_stop() calls stop_machine() when it modifies
the kernel text to enable the alternatives, (including the UAO code in
uao_thread_switch()). Once stop_machine() has finished __switch_to() is
called to reschedule the original task, this causes PSTATE.UAO to be set
appropriately. An explicit enable() call is not needed.

Reported-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: James Morse <james.morse@arm.com>

In commit 23c8a500

 ("arm64: kernel: use ordinary return/argument
register for el2_setup()"), we stopped using w20 as a global stash of
the boot mode flag, and instead pass this around in w0 as a function
parameter.

Unfortunately, we missed a couple of comments, which still refer to the
old convention of using w20/x20.

This patch fixes up the comments to describe the code as it currently
works.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

A few printk calls in arm64 omit a trailing newline, even though there
is no subsequent KERN_CONT printk associated with them, and we actually
want a newline.

This can result in unrelated lines being appended, rather than appearing
on a new line. Additionally, timestamp prefixes may appear in-line. This
makes the logs harder to read than necessary.

Avoid this by adding a trailing newline.

These were found with a shortlist generated by:

$ git grep 'pr\(intk\|_.*\)(.*)' -- arch/arm64 | grep -v pr_fmt | grep -v '\\n"'

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
CC: James Morse <james.morse@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Function graph tracer shows negative time (wrap around) when tracing
__switch_to if the nosleep-time trace option is enabled.

Time compensation for nosleep-time is done by an ftrace probe on
sched_switch. This doesn't work well for the following events (with
letters representing timestamps):
A - sched switch probe called for task T switch out
B - __switch_to calltime is recorded
C - sched_switch probe called for task T switch in
D - __switch_to rettime is recorded

If C - A > D - B, then we end up over compensating for the time spent in
__switch_to giving rise to negative times in the trace output.

On x86, __switch_to is not traced if function graph tracer is enabled.
Do the same for arm64 as well.

Cc: Todd Kjos <tkjos@google.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Joel Fernandes <joelaf@google.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

Pull USB fixes from Greg KH:
 "Here are a bunch of USB fixes for 4.10-rc3. Yeah, it's a lot, an
  artifact of the holiday break I think.

  Lots of gadget and the usual XHCI fixups for reported issues (one day
  that driver will calm down...) Also included are a bunch of usb-serial
  driver fixes, and for good measure, a number of much-reported MUSB
  driver issues have finally been resolved.

  All of these have been in linux-next with no reported issues"

* tag 'usb-4.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (72 commits)
  USB: fix problems with duplicate endpoint addresses
  usb: ohci-at91: use descriptor-based gpio APIs correctly
  usb: storage: unusual_uas: Add JMicron JMS56x to unusual device
  usb: hub: Move hub_port_disable() to fix warning if PM is disabled
  usb: musb: blackfin: add bfin_fifo_offset in bfin_ops
  usb: musb: fix compilation warning on unused function
  usb: musb: Fix trying to free already-free IRQ 4
  usb: musb: dsps: implement clear_ep_rxintr() callback
  usb: musb: core: add clear_ep_rxintr() to musb_platform_ops
  USB: serial: ti_usb_3410_5052: fix NULL-deref at open
  USB: serial: spcp8x5: fix NULL-deref at open
  USB: serial: quatech2: fix sleep-while-atomic in close
  USB: serial: pl2303: fix NULL-deref at open
  USB: serial: oti6858: fix NULL-deref at open
  USB: serial: omninet: fix NULL-derefs at open and disconnect
  USB: serial: mos7840: fix misleading interrupt-URB comment
  USB: serial: mos7840: remove unused write URB
  USB: serial: mos7840: fix NULL-deref at open
  USB: serial: mos7720: remove obsolete port initialisation
  USB: serial: mos7720: fix parallel probe
  ...

Pull char/misc fixes from Greg KH:
 "Here are a few small char/misc driver fixes for 4.10-rc3.

  Two MEI driver fixes, and three NVMEM patches for reported issues, and
  a new Hyper-V driver MAINTAINER update. Nothing major at all, all have
  been in linux-next with no reported issues"

* tag 'char-misc-4.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  hyper-v: Add myself as additional MAINTAINER
  nvmem: fix nvmem_cell_read() return type doc
  nvmem: imx-ocotp: Fix wrong register size
  nvmem: qfprom: Allow single byte accesses for read/write
  mei: move write cb to completion on credentials failures
  mei: bus: fix mei_cldev_enable KDoc

Pull staging/IIO fixes from Greg KH:
 "Here are some staging and IIO driver fixes for 4.10-rc3.

  Most of these are minor IIO fixes of reported issues, along with one
  network driver fix to resolve an issue. And a MAINTAINERS update with
  a new mailing list. All of these, except the MAINTAINERS file update,
  have been in linux-next with no reported issues (the MAINTAINERS patch
  happened on Friday...)"

* tag 'staging-4.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  MAINTAINERS: add greybus subsystem mailing list
  staging: octeon: Call SET_NETDEV_DEV()
  iio: accel: st_accel: fix LIS3LV02 reading and scaling
  iio: common: st_sensors: fix channel data parsing
  iio: max44000: correct value in illuminance_integration_time_available
  iio: adc: TI_AM335X_ADC should depend on HAS_DMA
  iio: bmi160: Fix time needed to sleep after command execution
  iio: 104-quad-8: Fix active level mismatch for the preset enable option
  iio: 104-quad-8: Fix off-by-one errors when addressing IOR
  iio: 104-quad-8: Fix index control configuration

Several people report seeing warnings about inconsistent radix tree
nodes followed by crashes in the workingset code, which all looked like
use-after-free access from the shadow node shrinker.

Dave Jones managed to reproduce the issue with a debug patch applied,
which confirmed that the radix tree shrinking indeed frees shadow nodes
while they are still linked to the shadow LRU:

  WARNING: CPU: 2 PID: 53 at lib/radix-tree.c:643 delete_node+0x1e4/0x200
  CPU: 2 PID: 53 Comm: kswapd0 Not tainted 4.10.0-rc2-think+ #3
  Call Trace:
     delete_node+0x1e4/0x200
     __radix_tree_delete_node+0xd/0x10
     shadow_lru_isolate+0xe6/0x220
     __list_lru_walk_one.isra.4+0x9b/0x190
     list_lru_walk_one+0x23/0x30
     scan_shadow_nodes+0x2e/0x40
     shrink_slab.part.44+0x23d/0x5d0
     shrink_node+0x22c/0x330
     kswapd+0x392/0x8f0

This is the WARN_ON_ONCE(!list_empty(&node->private_list)) placed in the
inlined radix_tree_shrink().

The problem is with 14b46879 ("mm: workingset: move shadow entry
tracking to radix tree exceptional tracking"), which passes an update
callback into the radix tree to link and unlink shadow leaf nodes when
tree entries change, but forgot to pass the callback when reclaiming a
shadow node.

While the reclaimed shadow node itself is unlinked by the shrinker, its
deletion from the tree can cause the left-most leaf node in the tree to
be shrunk.  If that happens to be a shadow node as well, we don't unlink
it from the LRU as we should.

Consider this tree, where the s are shadow entries:

       root->rnode
            |
       [0       n]
        |       |
     [s    ] [sssss]

Now the shadow node shrinker reclaims the rightmost leaf node through
the shadow node LRU:

       root->rnode
            |
       [0        ]
        |
    [s     ]

Because the parent of the deleted node is the first level below the
root and has only one child in the left-most slot, the intermediate
level is shrunk and the node containing the single shadow is put in
its place:

       root->rnode
            |
       [s        ]

The shrinker again sees a single left-most slot in a first level node
and thus decides to store the shadow in root->rnode directly and free
the node - which is a leaf node on the shadow node LRU.

  root->rnode
       |
       s

Without the update callback, the freed node remains on the shadow LRU,
where it causes later shrinker runs to crash.

Pass the node updater callback into __radix_tree_delete_node() in case
the deletion causes the left-most branch in the tree to collapse too.

Also add warnings when linked nodes are freed right away, rather than
wait for the use-after-free when the list is scanned much later.

Fixes: 14b46879

 ("mm: workingset: move shadow entry tracking to radix tree exceptional tracking")
Reported-by: Dave Chinner <david@fromorbit.com>
Reported-by: Hugh Dickins <hughd@google.com>
Reported-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-and-tested-by: Dave Jones <davej@codemonkey.org.uk>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Chris Leech <cleech@redhat.com>
Cc: Lee Duncan <lduncan@suse.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@linuxonhyperv.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

4.10-rc loadtest (even on x86, and even without THPCache) fails with
"fork: Cannot allocate memory" or some such; and /proc/meminfo shows
PageTables growing.

Commit 953c66c2 ("mm: THP page cache support for ppc64") that got
merged in rc1 removed the freeing of an unused preallocated pagetable
after do_fault_around() has called map_pages().

This is usually a good optimization, so that the followup doesn't have
to reallocate one; but it's not sufficient to shift the freeing into
alloc_set_pte(), since there are failure cases (most commonly
VM_FAULT_RETRY) which never reach finish_fault().

Check and free it at the outer level in do_fault(), then we don't need
to worry in alloc_set_pte(), and can restore that to how it was (I
cannot find any reason to pte_free() under lock as it was doing).

And fix a separate pagetable leak, or crash, introduced by the same
change, that could only show up on some ppc64: why does do_set_pmd()'s
failure case attempt to withdraw a pagetable when it never deposited
one, at the same time overwriting (so leaking) the vmf->prealloc_pte?
Residue of an earlier implementation, perhaps? Delete it.

Fixes: 953c66c2

 ("mm: THP page cache support for ppc64")
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull kbuild fix from Michal Marek:
 "The asm-prototypes.h file added in the last merge window results in
  invalid code with CONFIG_KMEMCHECK=y. The net result is that genksyms
  segfaults.

  This pull request fixes the header, the genksyms fix is in my kbuild
  branch for 4.11"

* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  asm-prototypes: Clear any CPP defines before declaring the functions