Merge branch 'irq/gic-4.5' of git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms into irq/core

Pull the GIC related updates from Marc Zyngier:

 "Not a lot this time (what a relief!), but an interesting series from
  Linus Walleij coming out of his work converting the ARM RealView
  platforms to DT, and a couple of mundane fixes."

Merge branch 'irq/wire-msi-bridge' of git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms into irq/core

Pull the MSI wire bridge implementation from Marc Zyngier along with
the first user of it. This is infrastructure to support a wired
interrupt to MSI interrupt brigde. The first user is mbigen found in
Hisilicon ARM SoCs.

Add the interrupt controller chip operation functions of mbigen chip.
Signed-off-by: Ma Jun <majun258@huawei.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

For peripheral devices which connect to mbigen,mbigen is a interrupt
controller. So, we create irq domain for each mbigen device and add
mbigen irq domain into irq hierarchy structure.
Signed-off-by: Ma Jun <majun258@huawei.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

Mbigen means Message Based Interrupt Generator(MBIGEN).

Its a kind of interrupt controller that collects
the interrupts from external devices and generate msi interrupt.
Mbigen is applied to reduce the number of wire connected interrupts.

As the peripherals increasing, the interrupts lines needed is
increasing much, especially on the Arm64 server SOC.

Therefore, the interrupt pin in GIC is not enough to cover so
many peripherals.

Mbigen is designed to fix this problem.

Mbigen chip locates in ITS or outside of ITS.

Mbigen chip hardware structure shows as below:

		mbigen chip
|---------------------|-------------------|
mgn_node0	  mgn_node1		mgn_node2
 |		 |-------|		|-------|------|
dev1		dev1    dev2		dev1   dev3   dev4

Each mbigen chip contains several mbigen nodes.

External devices can connect to mbigen node through wire connecting way.

Because a mbigen node only can support 128 interrupt maximum, depends
on the interrupt lines number of devices, a device can connects to one
more mbigen nodes.

Also, several different devices can connect to a same mbigen node.

When devices triggered interrupt,mbigen chip detects and collects
the interrupts and generates the MBI interrupts by writing the ITS
Translator register.

To simplify mbigen driver,I used a new conception--mbigen device.
Each mbigen device is initialized as a platform device.

Mbigen device presents the parts(register, pin definition etc.) in
mbigen chip corresponding to a peripheral device.

So from software view, the structure likes below

	            mbigen chip
     |---------------------|-----------------|
mbigen device1       mbigen device2  mbigen device3
      |                   |                |
     dev1                dev2             dev3
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Ma Jun <majun258@huawei.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

Add the mbigen msi interrupt controller bindings document.

This patch based on Mark Rutland's patch
https://lkml.org/lkml/2015/7/23/558Signed-off-by: Ma Jun <majun258@huawei.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

There is currently a hack in the GIC driver making it possible
to pass the number of GIC instances from the platform-specific
include files and thus override the variable MAX_GIC_NR.

With multiplatform deployments, this will not work as we need
to get rid of the platform-specific include files.

It turns out that this feature is only used by the RealView
platform which has a cascaded GIC. So move the configuration
to Kconfig and bump to 2 instances if we're building for the
RealView. The include file hacks can then be removed.

Tested on the ARM PB11MPCore with its cascaded GIC.
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The GIC has no such thing as interrupt 1020: the last valid ID is
1019, and the range 1020-1023 is reserved - 1023 indicating that
no interrupt is pending. So let's make sure we don't try to handle
this ID.

This bug has been in since the initial GIC code was introduced in
8ad68bbf ("[ARM] Add support for ARM RealView board").
Reported-by: Eric Auger <eric.auger@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

On the error path, the v2m drivers drops the refcount on the parent
node instead of doing it on the node that generated the error.
Humph...
Reported-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

Instead of having the irqchip being a static struct, make it part
of the per-instance data so we can assign it a dynamic name. This
has the usable side effect of displaying the GIC with an instance
number as GIC0, GIC1 ... GICn in /proc/interrupts, which is helpful
when debugging cascaded GICs, such as on the ARM PB11MPCore.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The ARM RealView PB11MPCore reference design has some special
bits in a system controller register to set up the GIC in one
of three modes: legacy, new with DCC, new without DCC. The
register is also used to enable FIQ.

Since the platform will not boot unless this register is set
up to "new with DCC" mode, we need a special quirk to be
compiled-in for the RealView platforms.

If we find the right compatible string on the GIC TestChip,
we enable this quirk by looking up the system controller and
enabling the special bits.

We depend on the CONFIG_REALVIEW_DT Kconfig symbol as the old
boardfile code has the same fix hardcoded, and this is only
needed for the attempts to modernize the RealView code using
device tree.

After fixing this, the PB11MPCore boots with device tree
only.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The GIC bindings for the ARM11MPCore need to differentiate between
the GIC on the Test Chip and the one on the evaluation baseboard.
Split the binding in two and define new compatible-strings.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: devicetree@vger.kernel.org
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

We almost have all the needed bits requiredable to create a irq domain
on top of a MSI domain.

For this, we enable a few things:
- the virq is stored in the msi_desc
- device, msi_alloc_info and domain-specific data
  are stored in the platform_priv_data structure
- we introduce a new API for platform-msi:

  /* Create a MSI-based domain */
  struct irq_domain *
  platform_msi_create_device_domain(struct device *dev,
                                    unsigned int nvec,
                                    irq_write_msi_msg_t write_msi_msg,
                                    const struct irq_domain_ops *ops,
                                    void *host_data);

  /* Allocate MSIs in an MSI domain */
  int platform_msi_domain_alloc(struct irq_domain *domain,
				unsigned int virq,
				unsigned int nr_irqs);

  /* Free MSIs from an MSI domain */
  void platform_msi_domain_free(struct irq_domain *domain,
				unsigned int virq,
				unsigned int nvec);

  /* Obtain the host data passed to platform_msi_create_device_domain */
  void *platform_msi_get_host_data(struct irq_domain *domain);

platform_msi_create_device_domain() is a hybrid of irqdomain creation
and interrupt allocation, creating a domain backed by the MSIs associated
to a device. IRQs can then be allocated in that domain using
platform_msi_domain_alloc().

This now allows a wired irq to MSI bridge to be created.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

To be able to allocate interrupts from the MSI layer down,
add a new msi_domain_populate_irqs entry point.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The .prepare callbacks are so far only called from msi_domain_alloc_irqs.
In order to reuse that code, split that code and create a
msi_domain_prepare_irqs function that the existing code can call into.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

We are soon going to need the MSI layer to call into the domain
allocators. Instead of open coding this, make the standard
irq_domain_alloc_irqs_recursive function available to the MSI
layer.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

As we're going to have multiple paths to allocate/free the
platform-msi private data, factor this out into separate
utility functions.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

MSIs for a given device are normally all allocated in one go.
Make sure the internal code can allocate them one at a time
if required.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The new VMD device driver needs to iterate over a list of
"demultiplexing" interrupts. Protecting that list with a lock is not
possible because the list is also required in code pathes which hold
irq descriptor lock. Therefor the demultiplexing interrupt handler
would create a lock inversion scenario if it calls a demux handler
with the list protection lock held.

A solution for this is to free the irq descriptor via RCU, so the
list can be walked with rcu read lock held.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Keith Busch <keith.busch@intel.com>

Jan Stancek reported that I wrecked things for him by fixing things for
Vladimir :/

His report was due to an UNINTERRUPTIBLE wait getting -EINTR, which
should not be possible, however my previous patch made this possible by
unconditionally checking signal_pending().

We cannot use current->state as was done previously, because the
instruction after the store to that variable it can be changed.  We must
instead pass the initial state along and use that.

Fixes: 68985633 ("sched/wait: Fix signal handling in bit wait helpers")
Reported-by: Jan Stancek <jstancek@redhat.com>
Reported-by: Chris Mason <clm@fb.com>
Tested-by: Jan Stancek <jstancek@redhat.com>
Tested-by: Vladimir Murzin <vladimir.murzin@arm.com>
Tested-by: Chris Mason <clm@fb.com>
Reviewed-by: Paul Turner <pjt@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: tglx@linutronix.de
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: hpa@zytor.com
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull NFS client bugfix from Trond Myklebust:
 "SUNRPC: Fix a NFSv4.1 callback channel regression"

* tag 'nfs-for-4.4-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  SUNRPC: Fix callback channel

Pull timer fixlets from Thomas Gleixner:
 "Two trivial fixes which add missing header fileas and forward
  declarations so the code will compile even when the magic include
  chains are different"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3: Add missing include for barrier.h
  irqchip/gic-v3: Add missing struct device_node declaration

Pull timer fix from Thomas Gleixner:
 "A single fix to unbreak a clocksource driver which has more than 32bit
  counter width"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource: Mmio: remove artificial 32bit limitation

Pull fpga driver fixes from Greg KH:
 "Only two small fpga driver fixes here, both have been in linux-next
  for a while, and resolve some reported issues"

* tag 'char-misc-4.4-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  fpga manager: Fix firmware resource leak on error
  fpga manager: remove label

Pull staging driver fixes from Greg KH:
 "Here are a few staging and IIO driver fixes for 4.4-rc5.

  All of them resolve reported problems and have been in linux-next for
  a while.  Nothing major here, just small fixes where needed"

* tag 'staging-4.4-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: lustre: echo_copy.._lsm() dereferences userland pointers directly
  iio: adc: spmi-vadc: add missing of_node_put
  iio: fix some warning messages
  iio: light: apds9960: correct ->last_busy count
  iio: lidar: return -EINVAL on invalid signal
  staging: iio: dummy: complete IIO events delivery to userspace

Pull USB driver fixes from Greg KH:
 "Here are a number of small USB fixes for 4.4-rc5.  All of them have
  been in linux-next.  The majority are gadget and phy issues, with a
  few new quirks and device ids added as well"

* tag 'usb-4.4-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (32 commits)
  USB: add quirk for devices with broken LPM
  xhci: fix usb2 resume timing and races.
  usb: musb: fail with error when no DMA controller set
  usb: gadget: uvc: fix permissions of configfs attributes
  usb: musb: core: Fix pm runtime for deferred probe
  usb: phy: msm: fix a possible NULL dereference
  USB: host: ohci-at91: fix a crash in ohci_hcd_at91_overcurrent_irq
  usb: Quiet down false peer failure messages
  usb: xhci: fix config fail of FS hub behind a HS hub with MTT
  xhci: Fix memory leak in xhci_pme_acpi_rtd3_enable()
  usb: Use the USB_SS_MULT() macro to decode burst multiplier for log message
  USB: whci-hcd: add check for dma mapping error
  usb: core : hub: Fix BOS 'NULL pointer' kernel panic
  USB: quirks: Apply ALWAYS_POLL to all ELAN devices
  usb-storage: Fix scsi-sd failure "Invalid field in cdb" for USB adapter JMicron
  USB: quirks: Fix another ELAN touchscreen
  usb: dwc3: gadget: don't prestart interrupt endpoints
  USB: serial: Another Infineon flash loader USB ID
  USB: cdc_acm: Ignore Infineon Flash Loader utility
  USB: cp210x: Remove CP2110 ID from compatibility list
  ...

Pull ARM SoC fixes from Arnd Bergmann:
 "Here are a bunch of small bug fixes for various ARM platforms, nothing
  really sticks out this week, most of either fixes bugs in code that
  was just added in 4.4, or that has been broken for many years without
  anyone noticing.

  at91/sama5d2:
   - fix sama5de hardware setup of sd/mmc interface
   - proper selection of pinctrl drivers.  PIO4 is necessary for sama5d2

  berlin:
   - fix incorrect clock input for SDIO

  exynos:
   - Fix potential NULL pointer dereference in Exynos PMU driver.

  imx:
   - Fix vf610 SAI clock configuration bug which is discovered by the
     newly added master mode support in SAI audio driver.
   - Fix buggy L2 cache latency values in vf610 device trees, which may
     cause system hang when cpu runs at a higher frequency.

  ixp4xx:
   - fix prototypes for readl/writel functions

  ls2080a:
   - use little-endian register access for GPIO and SDHCI

  omap:
   - Fix clock source for ARM TWD and global timers on am437x
   - Always select REGULATOR_FIXED_VOLTAGE for omap2+ instead of when
     MACH_OMAP3_PANDORA is selected
   - Fix SPI DMA handles for dm816x as only some were mapped
   - Fix up mbox cells for dm816x to make mailbox usable

  pxa:
   - use PWM lookup table for all ezx machines

  s3c24xx:
   - Remove incorrect __init annotation from s3c24xx cpufreq driver
     structures.

  versatile:
   - fix PCI IRQ mapping on Versatile PB"

* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ls2080a/dts: Add little endian property for GPIO IP block
  dt-bindings: define little-endian property for QorIQ GPIO
  ARM64: dts: ls2080a: fix eSDHC endianness
  ARM: dts: vf610: use reset values for L2 cache latencies
  ARM: pxa: use PWM lookup table for all machines
  ARM: dts: berlin: add 2nd clock for BG2Q sdhci0 and sdhci1
  ARM: dts: berlin: correct BG2Q's sdhci2 2nd clock
  ARM: dts: am4372: fix clock source for arm twd and global timers
  ARM: at91: fix pinctrl driver selection
  ARM: at91/dt: add always-on to 1.8V regulator
  ARM: dts: vf610: fix clock definition for SAI2
  ARM: imx: clk-vf610: fix SAI clock tree
  ARM: ixp4xx: fix read{b,w,l} return types
  irqchip/versatile-fpga: Fix PCI IRQ mapping on Versatile PB
  ARM: OMAP2+: enable REGULATOR_FIXED_VOLTAGE
  ARM: dts: add dm816x missing spi DT dma handles
  ARM: dts: add dm816x missing #mbox-cells
  cpufreq: s3c24xx: Do not mark s3c2410_plls_add as __init
  ARM: EXYNOS: Fix potential NULL pointer access in exynos_sys_powerdown_conf

Pull powerpc fixes from Michael Ellerman:
 - opal-irqchip: Fix double endian conversion from Alistair Popple
 - cxl: Set endianess of kernel contexts from Frederic Barrat
 - sbc8641: drop bogus PHY IRQ entries from DTS file from Paul Gortmaker
 - Revert "powerpc/eeh: Don't unfreeze PHB PE after reset" from Andrew
   Donnellan

* tag 'powerpc-4.4-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  Revert "powerpc/eeh: Don't unfreeze PHB PE after reset"
  powerpc/sbc8641: drop bogus PHY IRQ entries from DTS file
  cxl: Set endianess of kernel contexts
  powerpc/opal-irqchip: Fix double endian conversion

Merge misc fixes from Andrew Morton:
 "17 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  MIPS: fix DMA contiguous allocation
  sh64: fix __NR_fgetxattr
  ocfs2: fix SGID not inherited issue
  mm/oom_kill.c: avoid attempting to kill init sharing same memory
  drivers/base/memory.c: prohibit offlining of memory blocks with missing sections
  tmpfs: fix shmem_evict_inode() warnings on i_blocks
  mm/hugetlb.c: fix resv map memory leak for placeholder entries
  mm: hugetlb: call huge_pte_alloc() only if ptep is null
  kernel: remove stop_machine() Kconfig dependency
  mm: kmemleak: mark kmemleak_init prototype as __init
  mm: fix kerneldoc on mem_cgroup_replace_page
  osd fs: __r4w_get_page rely on PageUptodate for uptodate
  MAINTAINERS: make Vladimir co-maintainer of the memory controller
  mm, vmstat: allow WQ concurrency to discover memory reclaim doesn't make any progress
  mm: fix swapped Movable and Reclaimable in /proc/pagetypeinfo
  memcg: fix memory.high target
  mm: hugetlb: fix hugepage memory leak caused by wrong reserve count

Pull parisc fixes from Helge Deller:
 "Fix the boot crash on Mako machines with Huge Pages, prevent a panic
  with SATA controllers (and others) by correctly calculating the IOMMU
  space, hook up the mlock2 syscall and drop unneeded code in the parisc
  pci code"

* 'parisc-4.4-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Disable huge pages on Mako machines
  parisc: Wire up mlock2 syscall
  parisc: Remove unused pcibios_init_bus()
  parisc iommu: fix panic due to trying to allocate too large region

Pull block layer fixes from Jens Axboe:
 "A set of fixes for the current series.  This contains:

   - A bunch of fixes for lightnvm, should be the last round for this
     series.  From Matias and Wenwei.

   - A writeback detach inode fix from Ilya, also marked for stable.

   - A block (though it says SCSI) fix for an OOPS in SCSI runtime power
     management.

   - Module init error path fixes for null_blk from Minfei"

* 'for-linus' of git://git.kernel.dk/linux-block:
  null_blk: Fix error path in module initialization
  lightnvm: do not compile in debugging by default
  lightnvm: prevent gennvm module unload on use
  lightnvm: fix media mgr registration
  lightnvm: replace req queue with nvmdev for lld
  lightnvm: comments on constants
  lightnvm: check mm before use
  lightnvm: refactor spin_unlock in gennvm_get_blk
  lightnvm: put blks when luns configure failed
  lightnvm: use flags in rrpc_get_blk
  block: detach bdev inode from its wb in __blkdev_put()
  SCSI: Fix NULL pointer dereference in runtime PM

Pull arm64 fixes from Catalin Marinas:

 - Update the linker script to use L1_CACHE_BYTES instead of hard-coded
   64.  We recently changed L1_CACHE_BYTES to 128

 - Improve race condition reporting on set_pte_at() and change the BUG
   to WARN_ONCE.  With hardware update of the accessed/dirty state, we
   need to ensure that set_pte_at() does not inadvertently override
   hardware updated state.  The patch also makes the checks ignore
   !pte_valid() new entries

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Improve error reporting on set_pte_at() checks
  arm64: update linker script to increased L1_CACHE_BYTES value

Recent changes to how GFP_ATOMIC is defined seems to have broken the
condition to use mips_alloc_from_contiguous() in
mips_dma_alloc_coherent().

I couldn't bottom out the exact change but I think it's this commit
d0164adc ("mm, page_alloc: distinguish between being unable to
sleep, unwilling to sleep and avoiding waking kswapd").

GFP_ATOMIC has multiple bits set and the check for !(gfp & GFP_ATOMIC)
isn't enough.

The reason behind this condition is to check whether we can potentially
do a sleeping memory allocation.  Use gfpflags_allow_blocking() instead
which should be more robust.
Signed-off-by: Qais Yousef <qais.yousef@imgtec.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

According to arch/sh/kernel/syscalls_64.S and common sense, __NR_fgetxattr
has to be defined to 259, but it doesn't.  Instead, it's defined to 269,
which is of course used by another syscall, __NR_sched_setaffinity in this
case.

This bug was found by strace test suite.
Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Commit 8f1eb487 ("ocfs2: fix umask ignored issue") introduced an
issue, SGID of sub dir was not inherited from its parents dir.  It is
because SGID is set into "inode->i_mode" in ocfs2_get_init_inode(), but
is overwritten by "mode" which don't have SGID set later.

Fixes: 8f1eb487 ("ocfs2: fix umask ignored issue")
Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Acked-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

It's possible that an oom killed victim shares an ->mm with the init
process and thus oom_kill_process() would end up trying to kill init as
well.

This has been shown in practice:

	Out of memory: Kill process 9134 (init) score 3 or sacrifice child
	Killed process 9134 (init) total-vm:1868kB, anon-rss:84kB, file-rss:572kB
	Kill process 1 (init) sharing same memory
	...
	Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000009

And this will result in a kernel panic.

If a process is forked by init and selected for oom kill while still
sharing init_mm, then it's likely this system is in a recoverable state.
However, it's better not to try to kill init and allow the machine to
panic due to unkillable processes.

[rientjes@google.com: rewrote changelog]
[akpm@linux-foundation.org: fix inverted test, per Ben]
Signed-off-by: Chen Jie <chenjie6@huawei.com>
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Li Zefan <lizefan@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Commit bdee237c ("x86: mm: Use 2GB memory block size on large-memory
x86-64 systems") and 982792c7 ("x86, mm: probe memory block size for
generic x86 64bit") introduced large block sizes for x86.  This made it
possible to have multiple sections per memory block where previously,
there was a only every one section per block.

Since blocks consist of contiguous ranges of section, there can be holes
in the blocks where sections are not present.  If one attempts to
offline such a block, a crash occurs since the code is not designed to
deal with this.

This patch is a quick fix to gaurd against the crash by not allowing
blocks with non-present sections to be offlined.

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=107781Signed-off-by: Seth Jennings <sjennings@variantweb.net>
Reported-by: Andrew Banman <abanman@sgi.com>
Cc: Daniel J Blueman <daniel@numascale.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Greg KH <greg@kroah.com>
Cc: Russ Anderson <rja@sgi.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Dmitry Vyukov provides a little program, autogenerated by syzkaller,
which races a fault on a mapping of a sparse memfd object, against
truncation of that object below the fault address: run repeatedly for a
few minutes, it reliably generates shmem_evict_inode()'s
WARN_ON(inode->i_blocks).

(But there's nothing specific to memfd here, nor to the fstat which it
happened to use to generate the fault: though that looked suspicious,
since a shmem_recalc_inode() had been added there recently.  The same
problem can be reproduced with open+unlink in place of memfd_create, and
with fstatfs in place of fstat.)

v3.7 commit 0f3c42f5 ("tmpfs: change final i_blocks BUG to WARNING")
explains one cause of such a warning (a race with shmem_writepage to
swap), and possible solutions; but we never took it further, and this
syzkaller incident turns out to have a different cause.

shmem_getpage_gfp()'s error recovery, when a freshly allocated page is
then found to be beyond eof, looks plausible - decrementing the alloced
count that was just before incremented - but in fact can go wrong, if a
racing thread (the truncator, for example) gets its shmem_recalc_inode()
in just after our delete_from_page_cache().  delete_from_page_cache()
decrements nrpages, that shmem_recalc_inode() will balance the books by
decrementing alloced itself, then our decrement of alloced take it one
too low: leading to the WARNING when the object is finally evicted.

Once the new page has been exposed in the page cache,
shmem_getpage_gfp() must leave it to shmem_recalc_inode() itself to get
the accounting right in all cases (and not fall through from "trunc:" to
"decused:").  Adjust that error recovery block; and the reinitialization
of info and sbinfo can be removed too.

While we're here, fix shmem_writepage() to avoid the original issue: it
will be safe against a racing shmem_recalc_inode(), if it merely
increments swapped before the shmem_delete_from_page_cache() which
decrements nrpages (but it must then do its own shmem_recalc_inode()
before that, while still in balance, instead of after).  (Aside: why do
we shmem_recalc_inode() here in the swap path? Because its raison d'etre
is to cope with clean sparse shmem pages being reclaimed behind our
back: so here when swapping is a good place to look for that case.) But
I've not now managed to reproduce this bug, even without the patch.

I don't see why I didn't do that earlier: perhaps inhibited by the
preference to eliminate shmem_recalc_inode() altogether.  Driven by this
incident, I do now have a patch to do so at last; but still want to sit
on it for a bit, there's a couple of questions yet to be resolved.
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Dmitry Vyukov reported the following memory leak

unreferenced object 0xffff88002eaafd88 (size 32):
  comm "a.out", pid 5063, jiffies 4295774645 (age 15.810s)
  hex dump (first 32 bytes):
    28 e9 4e 63 00 88 ff ff 28 e9 4e 63 00 88 ff ff  (.Nc....(.Nc....
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
     kmalloc include/linux/slab.h:458
     region_chg+0x2d4/0x6b0 mm/hugetlb.c:398
     __vma_reservation_common+0x2c3/0x390 mm/hugetlb.c:1791
     vma_needs_reservation mm/hugetlb.c:1813
     alloc_huge_page+0x19e/0xc70 mm/hugetlb.c:1845
     hugetlb_no_page mm/hugetlb.c:3543
     hugetlb_fault+0x7a1/0x1250 mm/hugetlb.c:3717
     follow_hugetlb_page+0x339/0xc70 mm/hugetlb.c:3880
     __get_user_pages+0x542/0xf30 mm/gup.c:497
     populate_vma_page_range+0xde/0x110 mm/gup.c:919
     __mm_populate+0x1c7/0x310 mm/gup.c:969
     do_mlock+0x291/0x360 mm/mlock.c:637
     SYSC_mlock2 mm/mlock.c:658
     SyS_mlock2+0x4b/0x70 mm/mlock.c:648

Dmitry identified a potential memory leak in the routine region_chg,
where a region descriptor is not free'ed on an error path.

However, the root cause for the above memory leak resides in region_del.
In this specific case, a "placeholder" entry is created in region_chg.
The associated page allocation fails, and the placeholder entry is left
in the reserve map.  This is "by design" as the entry should be deleted
when the map is released.  The bug is in the region_del routine which is
used to delete entries within a specific range (and when the map is
released).  region_del did not handle the case where a placeholder entry
exactly matched the start of the range range to be deleted.  In this
case, the entry would not be deleted and leaked.  The fix is to take
these special placeholder entries into account in region_del.

The region_chg error path leak is also fixed.

Fixes: feba16e2 ("mm/hugetlb: add region_del() to delete a specific range of entries")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: <stable@vger.kernel.org>	[4.3+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>