Adds helpers to do SMC and HVC based on ARM SMC Calling Convention.
CONFIG_HAVE_ARM_SMCCC is enabled for architectures that may support the
SMC or HVC instruction. It's the responsibility of the caller to know if
the SMC instruction is supported by the platform.
This patch doesn't provide an implementation of the declared functions.
Later patches will bring in implementations and set
CONFIG_HAVE_ARM_SMCCC for ARM and ARM64 respectively.
Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Similar to memdup_user(), except that allocated buffer is one byte
longer and '\0' is stored after the copied data.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Document the multi-queue/ring feature in terms of XenStore keys to be written by
the backend and by the frontend.
Signed-off-by: Bob Liu <bob.liu@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
ConfigFS lacked binary attributes up until now. This patch
introduces support for binary attributes in a somewhat similar
manner of sysfs binary attributes albeit with changes that
fit the configfs usage model.
Problems that configfs binary attributes fix are everything that
requires a binary blob as part of the configuration of a resource,
such as bitstream loading for FPGAs, DTBs for dynamically created
devices etc.
Look at Documentation/filesystems/configfs/configfs.txt for internals
and howto use them.
This patch is against linux-next as of today that contains
Christoph's configfs rework.
Signed-off-by: Pantelis Antoniou <pantelis.antoniou@konsulko.com>
[hch: folded a fix from Geert Uytterhoeven <geert+renesas@glider.be>]
[hch: a few tiny updates based on review feedback]
Signed-off-by: Christoph Hellwig <hch@lst.de>
This implements UEFI kernel support for 32-bit ARM, based on the existing
arm64 support and existing generic early ioremap support. It is based on
commit f7d9248942 ("arm64/efi: refactor EFI init and runtime code for
reuse by 32-bit ARM"), which was pulled from the arm64 repo [1] as branch
'aarch64/efi'
[1] git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git
The qtree_fmt_operations structures are never modified, so declare them as
const.
Done with the help of Coccinelle.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Jan Kara <jack@suse.cz>
Complementing devm_led_classdev_register add a managed version of
led_trigger_register.
I omit a managed version of led_classdev_unregister as the equivalent
devm_led_classdev_unregister isn't used in the kernel as of today.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Now the core implements the work queue, remove it from the drivers,
and switch to using brightness_set_blocking op.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Antonio Ospite <ao2@ao2.it>
Reviewed-by: Mark Brown <broonie@kernel.org>
This patch removes SET_BRIGHTNESS_ASYNC and SET_BRIGHTNESS_SYNC flags.
led_set_brightness() now calls led_set_brightness_nosleep() instead of
choosing between sync and async op basing on the flags defined by the
driver.
From now on, if a user wants to make sure that brightness will be set
synchronously, they have to use led_set_brightness_sync() API. It is now
being made publicly available since it has become apparent that it is
a caller who should decide whether brightness is to be set in
a synchronous or an asynchronous way.
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
This patch makes LED core capable of setting brightness for drivers
that implement brightness_set_blocking op. It removes from LED class
drivers responsibility for using work queues on their own.
In order to achieve this set_brightness_delayed callback is being
modified to directly call one of available ops for brightness setting.
led_set_brightness_async() function didn't set brightness in an
asynchronous way in all cases. It was mistakenly assuming that all
LED subsystem drivers used work queue in their brightness_set op,
whereas only half of them did that. Since it has no users now,
it is being removed.
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
The initial purpose of brightness_set_sync op, introduced along with
the LED flash class extension, was to add a means for setting torch LED
brightness as soon as possible, which couldn't have been guaranteed by
brightness_set op. This patch renames the op to brightness_set_blocking,
which describes its purpose in a more generic way. It is beneficial
in view of the prospective changes in the LED core, aiming at removing
the need for using work queues in LED class drivers that can sleep
or use delays while setting brightness.
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Pavel Machek <pavel@ucw.cz>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
This patch adds LED_BLINK_BRIGHTNESS_CHANGE flag to indicate that blink
brightness has changed, and LED_BLINK_DISABLE flag to indicate that
blinking deactivation has been requested. In order to use the flags
led_timer_function and set_brightness_delayed callbacks as well as
led_set_brightness() function are being modified. The main goal of these
modifications is to prepare set_brightness_work for extension of the
scope of its responsibilities.
Signed-off-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
The of_get_named_gpio_flags() function does nothing other than returning
an error when CONFIG_OF_GPIO is disabled, but that causes spurious
warnings about possible use of uninitialized variables in any code that
does not check the of_get_named_gpio_flags() return value before trying
to use the flags:
drivers/input/misc/rotary_encoder.c: In function 'rotary_encoder_probe':
drivers/input/misc/rotary_encoder.c:223:28: warning: 'flags' may be used uninitialized in this function [-Wmaybe-uninitialized]
drivers/power/bq24735-charger.c: In function 'bq24735_charger_probe':
drivers/power/bq24735-charger.c:227:12: warning: 'flags' may be used uninitialized in this function [-Wmaybe-uninitialized]
drivers/power/sbs-battery.c: In function 'sbs_probe':
drivers/power/sbs-battery.c:782:17: warning: 'gpio_flags' may be used uninitialized in this function [-Wmaybe-uninitialized]
This changes the behavior of the inline helper to set the flags to zero
when OF_GPIO is disabled, to avoid the warnings. In all cases I've
encountered, we don't actually get to the place that uses the flags
if CONFIG_OF is disabled because we won't enter the DT parser code.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The drm_dp_mst_topology_cbs structures are never modified, so declare them
as const.
Done with the help of Coccinelle.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rather than just being able to turn DAX on and off via a mount
option, some applications may only want to enable DAX for certain
performance critical files in a filesystem.
This patch introduces a new inode flag to enable DAX in the v3 inode
di_flags2 field. It adds support for setting and clearing flags in
the di_flags2 field via the XFS_IOC_FSSETXATTR ioctl, and sets the
S_DAX inode flag appropriately when it is seen.
When this flag is set on a directory, it acts as an "inherit flag".
That is, inodes created in the directory will automatically inherit
the on-disk inode DAX flag, enabling administrators to set up
directory heirarchies that automatically use DAX. Setting this flag
on an empty root directory will make the entire filesystem use DAX
by default.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Hoist the ioctl definitions for the XFS_IOC_FS[SG]SETXATTR API from
fs/xfs/libxfs/xfs_fs.h to include/uapi/linux/fs.h so that the ioctls
can be used by all filesystems, not just XFS. This enables
(initially) ext4 to use the ioctl to set project IDs on inodes.
Based-on-patch-from: Li Xi <lixi@ddn.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
You can use this to duplicate packets and inject them at the egress path
of the specified interface. This duplication allows you to inspect
traffic from the dummy or any other interface dedicated to this purpose.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch allows you to invert the ratelimit matching criteria, so you
can match packets over the ratelimit. This is required to support what
hashlimit does.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The powercap_zone_ops and powercap_zone_constraint_ops structures are never
modified, so declare them as const.
Most of the actual changes adjust indentation to accomodate the const
keyword.
Done with the help of Coccinelle.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Hoist the btrfs EXTENT_SAME ioctl up to the VFS and make the name
more systematic (FIDEDUPERANGE).
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Notice that the boost_supported field in struct cpufreq_driver is
redundant, because the driver's ->set_boost callback may be left
unset if "boost" is not supported. Moreover, the only driver
populating the ->set_boost callback is acpi_cpufreq, so make it
avoid populating that callback if "boost" is not supported, rework
the core to check ->set_boost instead of boost_supported to
verify "boost" support and drop boost_supported which isn't
used any more.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
cpufreq_boost_supported() is not used outside of cpufreq.c, so make
it static.
While at it, refactor it as a one-liner (which it really is).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
ACPICA commit 071eff738c59eda1792ac24b3b688b61691d7e7c
Execute any module-level code after each ACPI table (DSDT or SSDT)
is loaded into the namespace (rather than after all AML tables have
been loaded). This matches the behavior of other ACPI
implementations and is required to support BIOS code that
depends on this behavior.
Link: https://github.com/acpica/acpica/commit/071eff73
Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Lv Zheng <lv.zheng@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
ACPICA commit 05492c4a7d1e106eb871a5e1357ed564d25740e5
1) Make the timer value output optional
2) Allow empty lines via null string or simple newline
3) No need for the object type name for strings and integers
4) Miscellaneous cleanup of output
Link: https://github.com/acpica/acpica/commit/05492c4a
Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Lv Zheng <lv.zheng@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
ACPICA commit e4743959b59ad93eab7310adf756adc930be0ddb
This reverts commit 8e7a8753827660c3dd1f571f3185610402b756f0.
The _SUB method was found to be problematic for this interface
because some implementations use control methods. Therefore,
it is being removed.
Operations cannot be used because this interface is called
during the device discovery scan and the region handlers are
not fully installed at that time.
Link: https://github.com/acpica/acpica/commit/e4743959
Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Lv Zheng <lv.zheng@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The patch reduces source code differences between the Linux kernel and the
ACPICA upstream so that the linuxized ACPICA 20151218 release can be
applied with reduced human intervention.
The pscode.c has already been out of sync for months, and it becomes more
and more difficult to merge pscode.c changes, so instead of update the
affected lines of pscode.c, this patch synchronizes entire pscode.c file.
Signed-off-by: Lv Zheng <lv.zheng@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Several drivers want to know if the acpi-video is generating key-presses
for brightness change hotkeys to avoid sending double key-events to
userspace for these. Currently these driver use this construct for this:
if (acpi_video_get_backlight_type() == acpi_backlight_vendor)
report_brightness_key_event();
This indirect way of detecting if acpi-video is active does not make the
code easier to understand, and in some cases it is wrong because just
because the preferred type != vendor does not mean that acpi-video is
actually listening for brightness events, e.g. there may be no acpi-video
bus on the system at all.
This commit adds a acpi_video_handles_brightness_key_presses() helper
function, making the code needing this functionality both easier to read
and more correct.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
gcc-4.4.4 has problems with initialization of anonymous unions:
drivers/mfd/intel-lpss-acpi.c:30: error: unknown field 'value' specified in initializer
work around this by crafting the initializers in a manner which the
compiler can handle.
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Pull networking fixes from David Miller:
1) Prevent XFRM per-cpu counter updates for one namespace from being
applied to another namespace. Fix from DanS treetman.
2) Fix RCU de-reference in iwl_mvm_get_key_sta_id(), from Johannes
Berg.
3) Remove ethernet header assumption in nft_do_chain_netdev(), from
Pablo Neira Ayuso.
4) Fix cpsw PHY ident with multiple slaves and fixed-phy, from Pascal
Speck.
5) Fix use after free in sixpack_close and mkiss_close.
6) Fix VXLAN fw assertion on bnx2x, from Yuval Mintz.
7) natsemi doesn't check for DMA mapping errors, from Alexey
Khoroshilov.
8) Fix inverted test in ip6addrlbl_get(), from ANdrey Ryabinin.
9) Missing initialization of needed_headroom in geneve tunnel driver,
from Paolo Abeni.
10) Fix conntrack template leak in openvswitch, from Joe Stringer.
11) Mission initialization of wq->flags in sock_alloc_inode(), from
Nicolai Stange.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (35 commits)
sctp: sctp should release assoc when sctp_make_abort_user return NULL in sctp_close
net, socket, socket_wq: fix missing initialization of flags
drivers: net: cpsw: fix error return code
openvswitch: Fix template leak in error cases.
sctp: label accepted/peeled off sockets
sctp: use GFP_USER for user-controlled kmalloc
qlcnic: fix a loop exit condition better
net: cdc_ncm: avoid changing RX/TX buffers on MTU changes
geneve: initialize needed_headroom
ipv6: honor ifindex in case we receive ll addresses in router advertisements
addrconf: always initialize sysctl table data
ipv6/addrlabel: fix ip6addrlbl_get()
switchdev: bridge: Pass ageing time as clock_t instead of jiffies
sh_eth: fix 16-bit descriptor field access endianness too
veth: don’t modify ip_summed; doing so treats packets with bad checksums as good.
net: usb: cdc_ncm: Adding Dell DW5813 LTE AT&T Mobile Broadband Card
net: usb: cdc_ncm: Adding Dell DW5812 LTE Verizon Mobile Broadband Card
natsemi: add checks for dma mapping errors
rhashtable: Kill harmless RCU warning in rhashtable_walk_init
openvswitch: correct encoding of set tunnel action attributes
...
If the layout segment is invalid, then we should not be adding more
write requests to the commit list. Instead, those writes should be
replayed after requesting a new layout.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Allow synchronous RPC calls to wait for pending RPC calls to finish,
but also allow asynchronous ones to just fire off another commit.
With this patch, the xfstests generic/074 test completes in 226s
instead of 242s
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
The flexfiles layout in particular, seems to want to poke around in the
O_DIRECT flags when retransmitting.
This patch sets up an interface to allow it to call back into O_DIRECT
to handle retransmission correctly. It also fixes a potential bug whereby
we could change the behaviour of O_DIRECT if an error is already pending.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Merge "Samsung DeviceTree updates and improvements for 4.5" from Krzysztof Kozlowski:
1. eMMC/SDIO minor fixes usage of bindings on Snow and Peach
Chromebooks.
2. Remove FIMD from Odroid XU3-family because on XU3 it cannot be used
yet and on XU3-Lite and XU4 it is not supported.
3. Remove deprecated since June 2013 samsung,exynos5-hdmi.
4. Add support for Pseudo Random Generator on Exynos4 (Trats2 for now).
This depends on new SSS clock.
5. Add rotator nodes for Exynos4 and Exynos5.
6. Switch DWC3_1 on Odroid XU3 and XU3-Lite to peripheral mode because
now it cannot be used as OTG.
7. Cleanup the G2D usage on Exynos4 and add it to a proper domain
in case of Exynos4210.
8. Put MDMA1 in proper domain on Exynos4210 as well.
9. Minor cleanups.
* tag 'samsung-dt-4.5-2' of git://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux: (21 commits)
ARM: dts: Unify G2D device node with other devices on exynos4
ARM: dts: Add power domain to G2D device on exynos4210
ARM: dts: MDMA1 device belongs to LCD0 power domain on exynos4210
ARM: dts: Remove unneeded GPIO include in exynos4412-odroidu3
ARM: dts: exynos4210-universal_c210: Disable DMA for UARTs
ARM: dts: Use peripheral mode for dwc3_1 on exynos5422-odroidxu3
ARM: dts: Add rotator node on exynos5420
ARM: dts: Add rotator node on exynos5250
ARM: dts: Fix power domain for sysmmu-rotator device on exynos4
ARM: dts: Add rotator nodes on exynos4
ARM: dts: Enable PRNG module on exynos4412-trats2
ARM: dts: Add PRNG module for exynos4
dt-bindings: remove deprecated compatible string from exynos-hdmi
ARM: dts: Remove fimd node from exynos5422-odroidxu3-common
ARM: dts: Mark eMMC as non-removable in exynos5250-snow-common
ARM: dts: Remove broken-cd from eMMC node in exynos5420-peach-pi
ARM: dts: Remove broken-cd from eMMC node in exynos5800-peach-pi
ARM: dts: Mark SDIO as non-removable in exynos5250-snow-common
ARM: dts: Mark SDIO as non-removable in exynos5420-peach-pit
ARM: dts: Mark SDIO as non-removable in exynos5800-peach-pi
...
Pull "BCM2835 drivers changes for 4.5" from Eric Anholt:
This pull request includes the bcm2835 changes for 4.5 targeting the
arm-soc next/drivers branch.
* tag 'bcm2835-drivers-next-2015-12-28' of http://github.com/anholt/linux:
ARM: bcm2835: add rpi power domain driver
dt-bindings: add rpi power domain driver bindings
ARM: bcm2835: Define two new packets from the latest firmware.
Pull "reworked dts changes for ti81xx devices and minimal
dra62x j5ec-evm support" from Tony Lindgren:
Add minimal device tree support for dra62x also known j5eco. It is
related to dm814x, just the clocks are a bit different and it has a
different set of integrated devices. And let's get some basic dm814x
and dra62x devices working as many of the devices are like on am33xx::
- pinctrl using the pinctrl defines as for am33xx
- Updated EDMA bindings with support for using exma_xbar
- MMC support for dm814x-evm, t410 and dra62x-j5eco-evm
- USB support for dm814x-evm, t410 and dra62x-j5eco-evm
This branch depends on an earlier omap-for-v4.5/81xx-fixes-signed
branch that has dm814x dts fixes interlaced with SoC related fixes to
keep things booting. The interlaced SoC and dts fixes were needed
because of issues with the device tree defined clocks that just
happened to work on bootloader timings for t410 earlier.
* tag 'omap-for-v4.5/81xx-dts-signed' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap: (21 commits)
ARM: dts: Add usb support for j5-eco evm
ARM: dts: Add usb support for hp t410
ARM: dts: Add usb support for dm814x-evm
ARM: dts: Add usb support for dm814x and dra62x
ARM: dts: Enable emmc on hp t410
ARM: dts: Add mmc support for dra62x j5-eco evm
ARM: dts: Add mmc support for dm8148-evm
ARM: dts: Add mmc device entries for dm814x
ARM: dts: Update edma bindings on dm814x to use edma_xbar
ARM: dts: Add pinctrl macros for dm814x
ARM: dts: Add minimal dra62x j5-eco evm support
ARM: dts: Add basic support for dra62x j5-eco SoC
ARM: OMAP2+: Remove useless check for legacy booting for dm814x
ARM: OMAP2+: Enable GPIO for dm814x
ARM: dts: Fix dm814x pinctrl address and mask
ARM: dts: Fix dm8148 control modules ranges
ARM: OMAP2+: Fix timer entries for dm814x
ARM: dts: Fix some mux and divider clocks to get dm814x-evm booting
ARM: OMAP2+: Add DPPLS clock manager for dm814x
clk: ti: Add few dm814x clock aliases
...
Ethernet PHYs can maintain statistics, for example errors while idle
and receive errors. Add an ethtool mechanism to retrieve these
statistics, using the same model as MAC statistics.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull block fixes from Jens Axboe:
"Make the block layer great again.
Basically three amazing fixes in this pull request, split into 4
patches. Believe me, they should go into 4.4. Two of them fix a
regression, the third and last fixes an easy-to-trigger bug.
- Fix a bad irq enable through null_blk, for queue_mode=1 and using
timer completions. Add a block helper to restart a queue
asynchronously, and use that from null_blk. From me.
- Fix a performance issue in NVMe. Some devices (Intel Pxxxx) expose
a stripe boundary, and performance suffers if we cross it. We took
that into account for merging, but not for the newer splitting
code. Fix from Keith.
- Fix a kernel oops in lightnvm with multiple channels. From Matias"
* 'for-linus' of git://git.kernel.dk/linux-block:
lightnvm: wrong offset in bad blk lun calculation
null_blk: use async queue restart helper
block: add blk_start_queue_async()
block: Split bios on chunk boundaries
snd_soc_dapm_mutex_lock currently uses the un-nested call which can
cause lockdep warnings when called from control handlers (a relatively
common usage) and using modules. As creating the control causes a
potential mutex inversion with the handler, creating the control will
take the controls_rwsem under the dapm_mutex and accessing the control
will take the dapm_mutex under controls_rwsem.
All the users look like they want to be using the runtime class of the
lock anyway, so this patch just changes snd_soc_dapm_mutex_lock to use
the nested call, with the SND_SOC_DAPM_CLASS_RUNTIME class.
Fixes: f6d5e586b4 ("ASoC: dapm: Add helpers to lock/unlock DAPM mutex")
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
mod_zone_page_state() takes a "delta" integer argument. delta contains
the number of pages that should be added or subtracted from a struct
zone's vm_stat field.
If a zone is larger than 8TB this will cause overflows. E.g. for a
zone with a size slightly larger than 8TB the line
mod_zone_page_state(zone, NR_ALLOC_BATCH, zone->managed_pages);
in mm/page_alloc.c:free_area_init_core() will result in a negative
result for the NR_ALLOC_BATCH entry within the zone's vm_stat, since 8TB
contain 0x8xxxxxxx pages which will be sign extended to a negative
value.
Fix this by changing the delta argument to long type.
This could fix an early boot problem seen on s390, where we have a 9TB
system with only one node. ZONE_DMA contains 2GB and ZONE_NORMAL the
rest. The system is trying to allocate a GFP_DMA page but ZONE_DMA is
completely empty, so it tries to reclaim pages in an endless loop.
This was seen on a heavily patched 3.10 kernel. One possible
explaination seem to be the overflows caused by mod_zone_page_state().
Unfortunately I did not have the chance to verify that this patch
actually fixes the problem, since I don't have access to the system
right now. However the overflow problem does exist anyway.
Given the description that a system with slightly less than 8TB does
work, this seems to be a candidate for the observed problem.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
All variables required by the watchdog core to manage a watchdog are
currently stored in struct watchdog_device. The lifetime of those
variables is determined by the watchdog driver. However, the lifetime
of variables used by the watchdog core differs from the lifetime of
struct watchdog_device. To remedy this situation, watchdog drivers
can implement ref and unref callbacks, to be used by the watchdog
core to lock struct watchdog_device in memory.
While this solves the immediate problem, it depends on watchdog drivers
to actually implement the ref/unref callbacks. This is error prone,
often not implemented in the first place, or not implemented correctly.
To solve the problem without requiring driver support, split the variables
in struct watchdog_device into two data structures - one for variables
associated with the watchdog driver, one for variables associated with
the watchdog core. With this approach, the watchdog core can keep track
of its variable lifetime and no longer depends on ref/unref callbacks
in the driver. As a side effect, some of the variables originally in
struct watchdog_driver are now private to the watchdog core and no longer
visible in watchdog drivers.
As a side effect of the changes made, an ioctl will now always fail
with -ENODEV after a watchdog device was unregistered with the character
device still open. Previously, it would only fail with -ENODEV in some
situations. Also, ioctl operations are now atomic from driver perspective.
With this change, it is now guaranteed that the driver will not unregister
a watchdog between a timeout change and the subsequent ping.
The 'ref' and 'unref' callbacks in struct watchdog_driver are no longer
used and marked as deprecated.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
