Should use acpi_device pointer directly instead of use handle and
get the device pointer again later.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Make acpi_bus_trim() work in analogy with acpi_bus_scan() and carry
out two passes such that ACPI drivers will be detached from device
nodes being removed in the first pass and the device nodes themselves
will be removed in the second pass.

For this purpose split the driver unregistration out of
acpi_bus_remove() into a new routine, acpi_bus_device_detach(), that
will be executed by acpi_bus_trim() in the additional first pass as
a post-order callback.

This is necessary, because some ACPI drivers' .remove() routines
unregister struct device objects associated with the ACPI device
nodes being removed and that needs to happen while the ACPI
device nodes are still around (for example, in case they need to be
used for power management or similar things at that time).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

The current acpi_bus_trim() implementation is not really
straightforward and may be simplified significantly by using
acpi_walk_namespace() with acpi_bus_remove() as a post-order
callback.

Observe that acpi_bus_remove(), as called by acpi_bus_trim(), cannot
actually fail, because its first argument is guaranteed not to be
NULL thanks to the acpi_bus_get_device() check in acpi_bus_trim(),
so simply move the acpi_bus_get_device() check to acpi_bus_remove()
and use acpi_walk_namespace() to execute it for every device under
start->handle as a post-order callback.  The, run it directly for
start->handle itself.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

All callers of acpi_bus_trim() pass 1 (true) as the second argument
of it, so remove that argument entirely and change acpi_bus_trim()
to always behave as though it were 1.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

Drop the second argument of acpi_device_unregister(), type, which is
not used by that function.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

The ops field in struct acpi_device is not used anywhere, so remove
it.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

With commit f2a33cde55a03 "ACPI: Drop ACPI device .bind() and .unbind()
callbacks", acpi_op_bind and acpi_op_unbind are not used any more. So
remove them.
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Since device_attach() returns 1 on success (a driver has been bound
to the device), the check against its return value in
acpi_bus_device_attach() should modified to take that into accout.
Make it so.

[rjw: Subject and changelog.]
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Commit 805d410 (ACPI: Separate adding ACPI device objects from
probing ACPI drivers) introduced an ACPI power resources management
regression, because it didn't ensure that the power resources
driver bind to the struct acpi_device objects corresponding
to power resources as soon as they were created.  As a result,
ACPI power management routines may attempt to access power resource
objects before they are ready to use.

To fix this problem, tell the acpi_add_single_object() in
acpi_bus_check_add() to probe the driver for objects of type
ACPI_BUS_TYPE_POWER.  This fix has been verified to work on
HP nx6325 where the problem was first observed.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Toshi Kani <toshi.kani@hp.com>

This member is never initialized and never referenced, so remove it.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Drop the .bind() and .unbind() that have no more users from
struct acpi_device_ops and remove all of the code referring to
them from drivers/acpi/scan.c.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Move the code related to _PRT setup and removal and to power
resources from acpi_pci_bind() and acpi_pci_unbind() to the .setup()
and .cleanup() callbacks in acpi_pci_bus and remove acpi_pci_bind()
and acpi_pci_unbind() that have no purpose any more.  Accordingly,
remove the code related to device .bind() and .unbind() operations
from the ACPI PCI root bridge driver.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Currently, the ACPI wakeup capability of PCI devices is set up
in two different places, partially in acpi_pci_bind() where
runtime wakeup is initialized and partially in
platform_pci_wakeup_init(), where system wakeup is initialized.
The cleanup is only done in acpi_pci_unbind() and it only covers
runtime wakeup.

Use the new .setup() and .cleanup() callbacks in struct acpi_bus_type
to consolidate that code and do the setup and the cleanup each in one
place.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Add two new callbacks,.setup() and .cleanup(), struct acpi_bus_type
and modify acpi_platform_notify() to call .setup() after executing
acpi_bind_one() successfully and acpi_platform_notify_remove() to
call .cleanup() before running acpi_unbind_one().  This will allow
the users of struct acpi_bus_type, PCI in particular, to specify
operations to be executed right after the given device has been
associated with a companion struct acpi_device and right before
it's going to be detached from that companion, respectively.

The main motivation is to be able to get rid of acpi_pci_bind()
and acpi_pci_unbind(), which are horrible horrible stuff.  [In short,
there are three problems with them: The way they populate the .bind()
and .unbind() callbacks of ACPI devices is rather less than
straightforward, they require special hotplug-specific paths to be
present in the ACPI namespace scanning code and by the time
acpi_pci_unbind() is called the PCI device object in question may
not exist any more.]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

The callers of acpi_bus_add() usually assume that if it has
succeeded, then a struct acpi_device object has been attached to
the handle passed as the first argument.  Unfortunately, however,
this assumption is wrong, because acpi_bus_scan(), and acpi_bus_add()
too as a result, may return a pointer to a different struct
acpi_device object on success (it may be an object corresponding to
one of the descendant ACPI nodes in the namespace scope below that
handle).

For this reason, the callers of acpi_bus_add() who care about
whether or not a struct acpi_device object has been created for
its first argument need to check that using acpi_bus_get_device()
anyway, so the second argument of acpi_bus_add() is not really
useful for them.  The same observation applies to acpi_bus_scan()
executed directly from acpi_scan_init().

Therefore modify the relevant callers of acpi_bus_add() to check the
existence of the struct acpi_device in question with the help of
acpi_bus_get_device() and drop the no longer necessary second
argument of acpi_bus_add().  Accordingly, modify acpi_scan_init() to
use acpi_bus_get_device() to get acpi_root and drop the no longer
needed second argument of acpi_bus_scan().
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

After the removal of the second argument of acpi_bus_scan() there is
no difference between the ACPI_BUS_ADD_MATCH and ACPI_BUS_ADD_START
add types, so the add_type field in struct acpi_device may be
replaced with a single flag.  Do that calling the flag match_driver.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

After the removal of acpi_start_single_object() and acpi_bus_start()
the second argument of acpi_bus_scan() is not necessary any more,
so drop it and update acpi_bus_check_add() accordingly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Notice that acpi_bus_add() uses only 2 of its 4 arguments and
redefine its header to match the body.  Update all of its callers as
necessary and observe that this leads to quite a number of removed
lines of code (Linus will like that).

Add a kerneldoc comment documenting acpi_bus_add() and wonder how
its callers make wrong assumptions about the second argument (make
note to self to take care of that later).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

The ACPI PCI root bridge driver was the only ACPI driver implementing
the .start() callback, which isn't used by any ACPI drivers any more
now.

For this reason, acpi_start_single_object() has no purpose any more,
so remove it and all references to it.  Also remove
acpi_bus_start_device(), whose only purpose was to call
acpi_start_single_object().

Moreover, since after the removal of acpi_bus_start_device() the
only purpose of acpi_bus_start() remains to call
acpi_update_all_gpes(), move that into acpi_bus_add() and drop
acpi_bus_start() too, remove its header from acpi_bus.h and
update all of its former users accordingly.

This change was previously proposed in a different from by
Yinghai Lu.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Move the code from the ACPI PCI root bridge's .start() callback
routine, acpi_pci_root_start(), directly into acpi_pci_root_add()
and drop acpi_pci_root_start().

It is safe to do that, because it is now always guaranteed that
when struct pci_dev objects are created, their companion struct
acpi_device objects are already present, so it is not necessary to
wait for them to be created before calling pci_bus_add_devices().

This change was previously proposed in a different form by
Yinghai Lu.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

If acpi_bus_check_add() is called for a handle already having an
existing struct acpi_device object attached, it is not necessary to
check the type and status of the device correspondig to it, so
change the ordering of acpi_bus_check_add() to avoid that.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Notice that one member of struct acpi_bus_ops, acpi_op_add, is not
used anywhere any more and the relationship between its remaining
members, acpi_op_match and acpi_op_start, is such that it doesn't
make sense to set the latter without setting the former at the same
time.  Therefore, replace struct acpi_bus_ops with new a enum type,
enum acpi_bus_add_type, with three values, ACPI_BUS_ADD_BASIC,
ACPI_BUS_ADD_MATCH, ACPI_BUS_ADD_START, corresponding to
both acpi_op_match and acpi_op_start unset, acpi_op_match set and
acpi_op_start unset, and both acpi_op_match and acpi_op_start set,
respectively.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Objects of type struct acpi_bus_ops are currently used to pass
information between different parts of the ACPI namespace scanning
code, sometimes in quite convoluted ways.  It turns out that that
is not necessary in some cases, so simplify the code by reducing
the utilization of struct acpi_bus_ops objects where clearly
possible.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

The current ACPI namespace scanning code suggests that acpi_bus_add()
and acpi_bus_start() share some code.  In fact, however, they are
completely different code paths (except for the initial checks), so
refactor the code to make that distinction visibly clear.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Instead of running acpi_pci_root_init() from a separate subsys
initcall, call it directly from acpi_scan_init() before scanning the
ACPI namespace for the first time, so that the PCI root bridge
driver's .add() routine, acpi_pci_root_start(), is always run
before binding ACPI drivers or attaching "companion" device objects
to struct acpi_device objects below the root bridge's device node in
the ACPI namespace.

The first, simpler reason for doing this is that it makes the
situation during boot more similar to the situation during hotplug,
in which the ACPI PCI root bridge driver is always present.

The second reason is that acpi_pci_root_init() causes struct pci_dev
objects to be created for all PCI devices below the bridge and
these objects may be necessary for whatever is done with the other
ACPI device nodes in that namespace scope.  For example, devices
created by acpi_create_platform_device() sometimes may need to be
added to the device hierarchy as children of PCI bridges.  For this
purpose, however, the struct pci_dev objects representing those
bridges need to exist before the platform devices in question are
registered.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Split the ACPI namespace scanning for devices into two passes, such
that struct acpi_device objects are registerd in the first pass
without probing ACPI drivers and the drivers are probed against them
directly in the second pass.

There are two main reasons for doing that.

First, the ACPI PCI root bridge driver's .add() routine,
acpi_pci_root_add(), causes struct pci_dev objects to be created for
all PCI devices under the given root bridge.  Usually, there are
corresponding ACPI device nodes in the ACPI namespace for some of
those devices and therefore there should be "companion" struct
acpi_device objects to attach those struct pci_dev objects to.  These
struct acpi_device objects should exist when the corresponding
struct pci_dev objects are created, but that is only guaranteed
during boot and not during hotplug.  This leads to a number of
functional differences between the boot and the hotplug cases which
are not strictly necessary and make the code more complicated.

For example, this forces the ACPI PCI root bridge driver to defer the
registration of the just created struct pci_dev objects and to use a
special .start() callback routine, acpi_pci_root_start(), to make
sure that all of the "companion" struct acpi_device objects will be
present at PCI devices registration time during hotplug.

If those differences can be eliminated, we will be able to
consolidate the boot and hotplug code paths for the enumeration and
registration of PCI devices and to reduce the complexity of that
code quite a bit.

The second reason is that, in general, it should be possible to
resolve conflicts of resources assigned by the BIOS to different
devices represented by ACPI namespace nodes before any drivers bind
to them and before they are attached to "companion" objects
representing physical devices (such as struct pci_dev).  However, for
this purpose we first need to enumerate all ACPI device nodes in the
given namespace scope.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Toshi Kani <toshi.kani@hp.com>

Pull LED fix from Bryan Wu.

* 'fixes-for-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/linux-leds:
  leds: leds-gpio: set devm_gpio_request_one() flags param correctly

commit a99d76f9 leds: leds-gpio: use gpio_request_one

changed the leds-gpio driver to use gpio_request_one() instead
of gpio_request() + gpio_direction_output()

Unfortunately, it also made a semantic change that breaks the
leds-gpio driver.

The gpio_request_one() flags parameter was set to:

GPIOF_DIR_OUT | (led_dat->active_low ^ state)

Since GPIOF_DIR_OUT is 0, the final flags value will just be the
XOR'ed value of led_dat->active_low and state.

This value were used to distinguish between HIGH/LOW output initial
level and call gpio_direction_output() accordingly.

With this new semantic gpio_request_one() will take the flags value
of 1 as a configuration of input direction (GPIOF_DIR_IN) and will
call gpio_direction_input() instead of gpio_direction_output().

int gpio_request_one(unsigned gpio, unsigned long flags, const char *label)
{
..
	if (flags & GPIOF_DIR_IN)
		err = gpio_direction_input(gpio);
	else
		err = gpio_direction_output(gpio,
				(flags & GPIOF_INIT_HIGH) ? 1 : 0);
..
}

The right semantic is to evaluate led_dat->active_low ^ state and
set the output initial level explicitly.
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Reported-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Tested-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: Bryan Wu <cooloney@gmail.com>

Pull watchdog fixes from Wim Van Sebroeck:
 "This fixes some small errors in the new da9055 driver, eliminates a
  compiler warning and adds DT support for the twl4030_wdt driver (so
  that we can have multiple watchdogs with DT on the omap platforms)."

* git://www.linux-watchdog.org/linux-watchdog:
  watchdog: twl4030_wdt: add DT support
  watchdog: omap_wdt: eliminate unused variable and a compiler warning
  watchdog: da9055: Don't update wdt_dev->timeout in da9055_wdt_set_timeout error path
  watchdog: da9055: Fix invalid free of devm_ allocated data

Pull PCI updates from Bjorn Helgaas:
 "Some fixes for v3.8.  They include a fix for the new SR-IOV sysfs
  management support, an expanded quirk for Ricoh SD card readers, a
  Stratus DMI quirk fix, and a PME polling fix."

* tag '3.8-pci-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: Reduce Ricoh 0xe822 SD card reader base clock frequency to 50MHz
  PCI/PM: Do not suspend port if any subordinate device needs PME polling
  PCI: Add PCIe Link Capability link speed and width names
  PCI: Work around Stratus ftServer broken PCIe hierarchy (fix DMI check)
  PCI: Remove spurious error for sriov_numvfs store and simplify flow

Commit 56c176c9 ("UAPI: strip the _UAPI prefix from header guards
during header installation") strips the _UAPI prefix from header guards,
but only if there's a single space between the cpp directive and the
label.

Make it more flexible and able to handle tabs and multiple white space
characters.
Signed-off-by: David Howells <dhowell@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Empty files can get deleted by the patch program, so remove empty Kbuild
files and their links from the parent Kbuilds.
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge tag 'ecryptfs-3.8-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs

Pull ecryptfs fixes from Tyler Hicks:
 "Two self-explanatory fixes and a third patch which improves
  performance: when overwriting a full page in the eCryptfs page cache,
  skip reading in and decrypting the corresponding lower page."

* tag 'ecryptfs-3.8-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
  fs/ecryptfs/crypto.c: make ecryptfs_encode_for_filename() static
  eCryptfs: fix to use list_for_each_entry_safe() when delete items
  eCryptfs: Avoid unnecessary disk read and data decryption during writing

Pull Ceph fixes from Sage Weil:
 "Two of Alex's patches deal with a race when reseting server
  connections for open RBD images, one demotes some non-fatal BUGs to
  WARNs, and my patch fixes a protocol feature bit failure path."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: fix protocol feature mismatch failure path
  libceph: WARN, don't BUG on unexpected connection states
  libceph: always reset osds when kicking
  libceph: move linger requests sooner in kick_requests()

Sasha was fuzzing with trinity and reported the following problem:

  BUG: sleeping function called from invalid context at kernel/mutex.c:269
  in_atomic(): 1, irqs_disabled(): 0, pid: 6361, name: trinity-main
  2 locks held by trinity-main/6361:
   #0:  (&mm->mmap_sem){++++++}, at: [<ffffffff810aa314>] __do_page_fault+0x1e4/0x4f0
   #1:  (&(&mm->page_table_lock)->rlock){+.+...}, at: [<ffffffff8122f017>] handle_pte_fault+0x3f7/0x6a0
  Pid: 6361, comm: trinity-main Tainted: G        W
  3.7.0-rc2-next-20121024-sasha-00001-gd95ef01-dirty #74
  Call Trace:
    __might_sleep+0x1c3/0x1e0
    mutex_lock_nested+0x29/0x50
    mpol_shared_policy_lookup+0x2e/0x90
    shmem_get_policy+0x2e/0x30
    get_vma_policy+0x5a/0xa0
    mpol_misplaced+0x41/0x1d0
    handle_pte_fault+0x465/0x6a0

This was triggered by a different version of automatic NUMA balancing
but in theory the current version is vunerable to the same problem.

do_numa_page
  -> numa_migrate_prep
    -> mpol_misplaced
      -> get_vma_policy
        -> shmem_get_policy

It's very unlikely this will happen as shared pages are not marked
pte_numa -- see the page_mapcount() check in change_pte_range() -- but
it is possible.

To address this, this patch restores sp->lock as originally implemented
by Kosaki Motohiro.  In the path where get_vma_policy() is called, it
should not be calling sp_alloc() so it is not necessary to treat the PTL
specially.
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull ext4 bug fixes from Ted Ts'o:
 "Various bug fixes for ext4.  Perhaps the most serious bug fixed is one
  which could cause file system corruptions when performing file punch
  operations."

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: avoid hang when mounting non-journal filesystems with orphan list
  ext4: lock i_mutex when truncating orphan inodes
  ext4: do not try to write superblock on ro remount w/o journal
  ext4: include journal blocks in df overhead calcs
  ext4: remove unaligned AIO warning printk
  ext4: fix an incorrect comment about i_mutex
  ext4: fix deadlock in journal_unmap_buffer()
  ext4: split off ext4_journalled_invalidatepage()
  jbd2: fix assertion failure in jbd2_journal_flush()
  ext4: check dioread_nolock on remount
  ext4: fix extent tree corruption caused by hole punch

Remove the unused argument (formerly no_context) from mpol_parse_str()
and from mpol_to_str().
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing.  Very nasty.  Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse.  "mpol=prefer" and "mpol=prefer:Node" are equally toxic.

Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.

mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect.  Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).

I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL.  I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

EPOLL_CTL_MOD sets the interest mask before calling f_op->poll() to
ensure events are not missed.  Since the modifications to the interest
mask are not protected by the same lock as ep_poll_callback, we need to
ensure the change is visible to other CPUs calling ep_poll_callback.

We also need to ensure f_op->poll() has an up-to-date view of past
events which occured before we modified the interest mask.  So this
barrier also pairs with the barrier in wq_has_sleeper().

This should guarantee either ep_poll_callback or f_op->poll() (or both)
will notice the readiness of a recently-ready/modified item.

This issue was encountered by Andreas Voellmy and Junchang(Jason) Wang in:
http://thread.gmane.org/gmane.linux.kernel/1408782/Signed-off-by: Eric Wong <normalperson@yhbt.net>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andreas Voellmy <andreas.voellmy@yale.edu>
Tested-by: "Junchang(Jason) Wang" <junchang.wang@yale.edu>
Cc: netdev@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>