Current tmio_mmc driver is using tmio_mmc_data for driver/platform
specific data/callback, and it is needed for tmio_mmc_host_probe()
function. Because of this style, include/linux/mfd/tmio.h header has
tmio driver/framework specific data which is not needed from platform.
This patch adds new tmio_mmc_host_alloc/free() as cleanup preparation.
tmio driver specific data/callback will be implemented in tmio_mmc_host,
and platform specific data/callback will be implemented in tmio_mmc_data
in this cleanup.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
This patch defines a quirk to disable the block count
for single block transactions.
It is a preparation and will be used by Fujitsu
SDHCI controller f_sdh30 driver.
Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
This patch defines a quirk for tuning work
around for some sdhci host controller. It sets
both SDHCI_CTRL_EXEC_TUNING and SDHCI_CTRL_TUNED_CLK
for tuning.
It is a preparation and will be used by Fujitsu
SDHCI controller f_sdh30 driver.
Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Pull networking fixes from David Miller:
1) Socket addresses returned in the error queue need to be fully
initialized before being passed on to userspace, fix from Willem de
Bruijn.
2) Interrupt handling fixes to davinci_emac driver from Tony Lindgren.
3) Fix races between receive packet steering and cpu hotplug, from Eric
Dumazet.
4) Allowing netlink sockets to subscribe to unknown multicast groups
leads to crashes, don't allow it. From Johannes Berg.
5) One to many socket races in SCTP fixed by Daniel Borkmann.
6) Put in a guard against the mis-use of ipv6 atomic fragments, from
Hagen Paul Pfeifer.
7) Fix promisc mode and ethtool crashes in sh_eth driver, from Ben
Hutchings.
8) NULL deref and double kfree fix in sxgbe driver from Girish K.S and
Byungho An.
9) cfg80211 deadlock fix from Arik Nemtsov.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (36 commits)
s2io: use snprintf() as a safety feature
r8152: remove sram_read
r8152: remove generic_ocp_read before writing
bgmac: activate irqs only if there is nothing to poll
bgmac: register napi before the device
sh_eth: Fix ethtool operation crash when net device is down
sh_eth: Fix promiscuous mode on chips without TSU
ipv6: stop sending PTB packets for MTU < 1280
net: sctp: fix race for one-to-many sockets in sendmsg's auto associate
genetlink: synchronize socket closing and family removal
genetlink: disallow subscribing to unknown mcast groups
genetlink: document parallel_ops
net: rps: fix cpu unplug
net: davinci_emac: Add support for emac on dm816x
net: davinci_emac: Fix ioremap for devices with MDIO within the EMAC address space
net: davinci_emac: Fix incomplete code for getting the phy from device tree
net: davinci_emac: Free clock after checking the frequency
net: davinci_emac: Fix runtime pm calls for davinci_emac
net: davinci_emac: Fix hangs with interrupts
ip: zero sockaddr returned on error queue
...
As documented in Documentation/kernel-doc-nano-HOWTO.txt lines
terminated with a colon are treated as headings.
The current layout of the documentation when compiling the kernel
crypto API DocBook documentation is messed up by by treating some lines
as headings. The patch removes colons from comments that shall not be
treated as headings.
Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The current implementation of the libahci allows using multiple PHYs
but not multiple regulators. This patch adds the support of multiple
regulators. Until now it was mandatory to have a PHY under a subnode,
now a port subnode can contain either a regulator or a PHY (or both).
In order to be able to asociate a port with a regulator the port are
now a platform device in the device tree case.
There was only one driver which used directly the regulator field of
the ahci_host_priv structure. To preserve the bisectability the change
in the ahci_imx driver was done in the same patch.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Ronny reports: https://bugzilla.kernel.org/show_bug.cgi?id=87101
"Since commit 8a4aeec8d "libata/ahci: accommodate tag ordered
controllers" the access to the harddisk on the first SATA-port is
failing on its first access. The access to the harddisk on the
second port is working normal.
When reverting the above commit, access to both harddisks is working
fine again."
Maintain tag ordered submission as the default, but allow sata_sil24 to
continue with the old behavior.
Cc: <stable@vger.kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Reported-by: Ronny Hegewald <Ronny.Hegewald@online.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Systems may contain heterogeneous IOMMUs supporting differing minimum
page sizes, which may also not be common with the CPU page size.
Thus it is practical to have an explicit notion of IOVA granularity
to simplify handling of mapping and allocation constraints.
As an initial step, move the IOVA page granularity from an implicit
compile-time constant to a per-domain property so we can make use
of it in IOVA domain context at runtime. To keep the abstraction tidy,
extend the little API of inline iova_* helpers to parallel some of the
equivalent PAGE_* macros.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
To share the IOVA allocator with other architectures, it needs to
accommodate more general aperture restrictions; move the lower limit
from a compile-time constant to a runtime domain property to allow
IOVA domains with different requirements to co-exist.
Also reword the slightly unclear description of alloc_iova since we're
touching it anyway.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
In order to share the IOVA allocator with other architectures, break
the unnecssary dependency on the Intel IOMMU driver and move the
remaining IOVA internals to iova.c
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Commit 927609d622 ("kernel: tighten rules for ACCESS ONCE") results in
sparse warnings like "Using plain integer as NULL pointer" - Let's add a
type cast to the dummy assignment.
To avoid warnings lik "sparse: warning: cast to restricted __hc32" we also
use __force on that cast.
Fixes: 927609d622 ("kernel: tighten rules for ACCESS ONCE")
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Now that all non-scalar users of ACCESS_ONCE have been converted
to READ_ONCE or ASSIGN once, lets tighten ACCESS_ONCE to only
work on scalar types.
This variant was proposed by Alexei Starovoitov.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
One of the reasons omap_hsmmc doesn't use the slot-gpio library
is that it has some non-standard functionality in the card-detect
interrupt service routine.
To make it possible for omap_hsmmc (and maybe others) to be converted
to use slot-gpio, add 'mmc_gpio_request_cd_isr' which provide an
alternate isr to be register by the slot-gpio code.
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Always check if the card is alive after a successful reset. This allows
us to remove mmc_hw_reset_check(), leaving mmc_hw_reset() as the only
card reset interface.
Signed-off-by: Johan Rudholm <johanru@axis.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
This patch is based on the patches by Per Forlin, Tony Lin and Ryan QIAN.
This patch complete the API 'post_req' and 'pre_req' in sdhci host side,
Test Env:
1. i.MX6Q-SABREAUTO board, CPU @ 996MHz, use ADMA in uSDHC controller.
2. Test command:
$ echo 1 > /proc/sys/vm/drop_caches
write to sd card:
$ dd if=/dev/zero of=/dev/mmcblk0 bs=1M count=2000 conv=fsync
read the sd card:
$ dd if=/dev/mmcblk0 of=/dev/null bs=1M count=2000
3. TOSHIBA 16GB SD3.0 card, running at 4 bit, SDR104 @ 198MHZ
Performance with and without this patch:
-------------------------------------------------
| | read speed | write speed |
|------------------------------------------------
| with this patch | ~76.7 MB/s | ~23.3 MB/s |
|------------------------------------------------
|without this patch | ~60.5 MB/s | ~22.5 MB/s |
-------------------------------------------------
4. SanDisk 8GB SD3.0 card, running at 4 bit, DDR50 @ 50MHZ
Performance with and without this patch:
-------------------------------------------------
| | read speed | write speed |
|------------------------------------------------
| with this patch | ~40.5 MB/s | ~15.6 MB/s |
|------------------------------------------------
|without this patch | ~36.1 MB/s | ~14.1 MB/s |
-------------------------------------------------
5. Kingston 8GB SD2.0 card, running at 4 bit, High-speed @ 50MHZ
Performance with and without this patch:
-------------------------------------------------
| | read speed | write speed |
|------------------------------------------------
| with this patch | ~22.7 MB/s | ~8.2 MB/s |
|------------------------------------------------
|without this patch | ~21.3 MB/s | ~8.0 MB/s |
-------------------------------------------------
6. About eMMC, Sandisk 8GB eMMC on i.MX6DL-sabresd board, CPU @ 792MHZ,
eMMC running at 8 bit, DDR52 @ 52MHZ.
Performance with and without this patch:
-------------------------------------------------
| | read speed | write speed |
|------------------------------------------------
| with this patch | ~37.3 MB/s | ~10.5 MB/s |
|------------------------------------------------
|without this patch | ~33.4 MB/s | ~10.5 MB/s |
-------------------------------------------------
Signed-off-by: Haibo Chen <haibo.chen@freescale.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Since previous patches removed the need for the tuning block patterns
to be exported, let's move them close to the mmc_send_tuning() API.
Those are now intended to be used only by the mmc core.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
By moving the allocation of the slot-gpio data into mmc_alloc_host(),
we can remove the slot-gpio internal calls to mmc_gpio_alloc().
This means mmc_gpio_alloc() has now only one caller left, which
consequence allow us to simplify and remove some of the slot-gpio code.
Additionally, this makes the slot-gpio mutex redundant, so let's remove
it.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The slot-gpio uses the devm*_ managed functions. Still it provide APIs
to explicitly free requested CD/WP GPIOs, but these API isn't being
used.
Therefore let's simplify slot-gpio by removing these unused APIs. If it
later turns out we need some of them, we can always consider to restore
the code.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The only function of these 'prepare' and 'complete' is to
disable the 'card detect' irq during suspend.
The commit which added this,
commit a48ce884d5
mmc: omap_hsmmc: Introduce omap_hsmmc_prepare/complete
justified it by the need to avoid the registration of new devices
during suspend.
However mmc_pm_notify will set ->rescan_disable in the 'prepare'
stage and clear it in the 'complete' stage, so no card detection
will actually happen.
Also the interrupt will be disabled before final suspend as part
of common suspend processing.
So this disabling of the interrupt is unnecessary, and interferes
with a transition to using common code for card-detect management.
Cc: Felipe Balbi <balbi@ti.com>
Cc: Venkatraman S <svenkatr@ti.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
We're running into cases where our enabling of the SDIO interrupt in
dw_mmc doesn't actually take effect. Specifically, adding patch like
this:
+++ b/drivers/mmc/host/dw_mmc.c
@@ -1076,6 +1076,9 @@ static void dw_mci_enable_sdio_irq(struct mmc_host *mmc, int enb)
mci_writel(host, INTMASK,
(int_mask | SDMMC_INT_SDIO(slot->id)));
+ int_mask = mci_readl(host, INTMASK);
+ if (!(int_mask & SDMMC_INT_SDIO(slot->id)))
+ dev_err(&mmc->class_dev, "failed to enable sdio irq\n");
} else {
...actually triggers the error message. That's because the
dw_mci_enable_sdio_irq() unsafely does a read-modify-write of the
INTMASK register.
We can't just use the standard host->lock since that lock is not irq
safe and mmc_signal_sdio_irq() (called from interrupt context) calls
dw_mci_enable_sdio_irq(). Add a new irq-safe lock to protect INTMASK.
An alternate solution to this is to punt mmc_signal_sdio_irq() to the
tasklet and then protect INTMASK modifications by the standard host
lock. This seemed like a bit more of a high-latency change.
Reported-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
This patch benefits from newly introduced switchdev notifier and uses it
to propagate fdb learn events from rocker driver to bridge. That avoids
direct function calls and possible use by other listeners (ovs).
Suggested-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The serirq gpio pin is used only as interrupt. After driver initialization,
the serirq signal is always used through interrupt and never with gpio
kernel API.
The irq can then be initialized during the platform_data definition within the client->irq pin.
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
In order to clean big buffers in st33zp24_platform_data structure,
replace with tpm_stm_dev for driver internal usage.
As only one buffer is really necessary replace with buf field.
In the mean time move tpm_i2c_stm_st33.h to include/linux/platform_data.
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Christophe Ricard <christophe-h.ricard@st.com>
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
A secured user-space accessible pstore object. Writes
to /dev/pmsg0 are appended to the buffer, on reboot
the persistent contents are available in
/sys/fs/pstore/pmsg-ramoops-[ID].
One possible use is syslogd, or other daemon, can
write messages, then on reboot provides a means to
triage user-space activities leading up to a panic
as a companion to the pstore dmesg or console logs.
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
In addition to the problem Jeff Layton reported, I looked at the code
and reproduced the same warning by subscribing and removing the genl
family with a socket still open. This is a fairly tricky race which
originates in the fact that generic netlink allows the family to go
away while sockets are still open - unlike regular netlink which has
a module refcount for every open socket so in general this cannot be
triggered.
Trying to resolve this issue by the obvious locking isn't possible as
it will result in deadlocks between unregistration and group unbind
notification (which incidentally lockdep doesn't find due to the home
grown locking in the netlink table.)
To really resolve this, introduce a "closing socket" reference counter
(for generic netlink only, as it's the only affected family) in the
core netlink code and use that in generic netlink to wait for all the
sockets that are being closed at the same time as a generic netlink
family is removed.
This fixes the race that when a socket is closed, it will should call
the unbind, but if the family is removed at the same time the unbind
will not find it, leading to the warning. The real problem though is
that in this case the unbind could actually find a new family that is
registered to have a multicast group with the same ID, and call its
mcast_unbind() leading to confusing.
Also remove the warning since it would still trigger, but is now no
longer a problem.
This also moves the code in af_netlink.c to before unreferencing the
module to avoid having the same problem in the normal non-genl case.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This makes things a bit more efficient in the cifs and ceph lock
pushing code.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Now that we use standard list_heads for tracking leases, we can have
lm_change take a pointer to the lease to be modified instead of a
double pointer.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
We can now add a dedicated spinlock without expanding struct inode.
Change to using that to protect the various i_flctx lists.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Nothing uses it anymore. Also add a forward declaration for struct
file_lock to silence some compiler warnings that the removal triggers.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
The current scheme of using the i_flock list is really difficult to
manage. There is also a legitimate desire for a per-inode spinlock to
manage these lists that isn't the i_lock.
Start conversion to a new scheme to eventually replace the old i_flock
list with a new "file_lock_context" object.
We start by adding a new i_flctx to struct inode. For now, it lives in
parallel with i_flock list, but will eventually replace it. The idea is
to allocate a structure to sit in that pointer and act as a locus for
all things file locking.
We allocate a file_lock_context for an inode when the first lock is
added to it, and it's only freed when the inode is freed. We use the
i_lock to protect the assignment, but afterward it should mostly be
accessed locklessly.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
...that we can use to queue file_locks to per-ctx list_heads. Go ahead
and convert locks_delete_lock and locks_dispose_list to use it instead
of the fl_block list.
Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Pull USB fixes from Greg KH:
"Here is a bunch of USB fixes for 3.19-rc5.
Most of these are gadget driver fixes, along with the xhci driver fix
that we both reported having problems with, as well as some new device
ids and other tiny fixes.
All have been in linux-next with no problems"
* tag 'usb-3.19-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (43 commits)
usb: dwc3: gadget: Stop TRB preparation after limit is reached
usb: dwc3: gadget: Fix TRB preparation during SG
usb: phy: mv-usb: fix usb_phy build errors
usb: serial: handle -ENODEV quietly in generic_submit_read_urb
usb: serial: silence all non-critical read errors
USB: console: fix potential use after free
USB: console: fix uninitialised ldisc semaphore
usb: gadget: udc: atmel: fix possible oops when unloading module
usb: gadget: gadgetfs: fix an oops in ep_write()
usb: phy: Fix deferred probing
OHCI: add a quirk for ULi M5237 blocking on reset
uas: Add US_FL_NO_ATA_1X for 2 more Seagate disk enclosures
uas: Do not blacklist ASM1153 disk enclosures
usb: gadget: udc: avoid dereference before NULL check in ep_queue
usb: host: ehci-tegra: request deferred probe when failing to get phy
uas: disable UAS on Apricorn SATA dongles
uas: Add US_FL_NO_REPORT_OPCODES for JMicron JMS566 with usb-id 0bc2:a013
uas: Add US_FL_NO_ATA_1X for Seagate devices with usb-id 0bc2:a013
xhci: Add broken-streams quirk for Fresco Logic FL1000G xhci controllers
USB: EHCI: adjust error return code
...
Pull NFS client bugfixes from Trond Myklebust:
"Highlights include:
- Stable fix for a NFSv3/lockd race
- Fixes for several NFSv4.1 client id trunking bugs
- Remove an incorrect test when checking for delegated opens"
* tag 'nfs-for-3.19-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
NFSv4: Remove incorrect check in can_open_delegated()
NFS: Ignore transport protocol when detecting server trunking
NFSv4/v4.1: Verify the client owner id during trunking detection
NFSv4: Cache the NFSv4/v4.1 client owner_id in the struct nfs_client
NFSv4.1: Fix client id trunking on Linux
LOCKD: Fix a race when initialising nlmsvc_timeout
Per the PCI Power Management spec r1.2, sec 3.2.4, a device that advertises
No_Soft_Reset == 0 in the PMCSR register (reported by lspci as "NoSoftRst-")
should perform an internal reset when transitioning from D3hot to D0 via
software control. Configuration context is lost and the device requires a
full reinitialization sequence.
Unfortunately the definition of "internal reset", beyond the application of
the configuration context, is largely left to the interpretation of the
specific device. Some devices don't seem to perform an "internal reset"
even if they report No_Soft_Reset == 0.
We still need to honor the PCI specification and restore PCI config context
in the event that we do a PM reset, so we don't cache and modify the
PCI_PM_CTRL_NO_SOFT_RESET bit for the device, but for interfaces where the
intention is to reset the device, like pci_reset_function(), we need a
mechanism to flag that PM reset (a D3hot->D0 transition) doesn't perform
any significant "internal reset" of the device.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Add pci_claim_bridge_resource() to claim a PCI-PCI bridge window. This is
like regular pci_claim_resource(), except that if we fail to claim the
window, we check to see if we can reduce the size of the window and try
again.
This is for scenarios like this:
pci_bus 0000:00: root bus resource [mem 0xc0000000-0xffffffff]
pci 0000:00:01.0: bridge window [mem 0xbdf00000-0xddefffff 64bit pref]
pci 0000:01:00.0: reg 0x10: [mem 0xc0000000-0xcfffffff pref]
The 00:01.0 window is illegal: it starts before the host bridge window, so
we have to assume the [0xbdf00000-0xbfffffff] region is inaccessible. We
can make it legal by clipping it to [mem 0xc0000000-0xddefffff 64bit pref].
Previously we discarded the 00:01.0 window and tried to reassign that part
of the hierarchy from scratch. That is a problem because Linux doesn't
always assign things optimally. For example, in this case, BIOS put the
01:00.0 device in a prefetchable window below 4GB, but after 5b28541552,
Linux puts the prefetchable window above 4GB where the 32-bit 01:00.0
device can't use it.
Clipping the 00:01.0 window is less intrusive than completely reassigning
things and is sufficient to let us use most of the BIOS configuration. Of
course, it's possible that devices below 00:01.0 will no longer fit. If
that's the case, we'll have to reassign things. But that's a separate
problem.
[bhelgaas: changelog, split into separate patch]
Link: https://bugzilla.kernel.org/show_bug.cgi?id=85491
Reported-by: Marek Kordik <kordikmarek@gmail.com>
Fixes: 5b28541552 ("PCI: Restrict 64-bit prefetchable bridge windows to 64-bit resources")
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.16+
Enable a mechanism for devices to quirk that they do not behave when
doing a PCI bus reset. We require a modest level of spec compliant
behavior in order to do a reset, for instance the device should come
out of reset without throwing errors and PCI config space should be
accessible after reset. This is too much to ask for some devices.
Link: http://lkml.kernel.org/r/20140923210318.498dacbd@dualc.maya.org
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.14+
