Now that there are different variants of pt_regs for userspace and
kernel, the uapi for the BPF_PROG_TYPE_PERF_EVENT program type must be
changed by exporting the user_pt_regs structure instead of the pt_regs
structure that is in-kernel only.
Fixes: 002af9391b ("powerpc: Split user/kernel definitions of struct pt_regs")
Signed-off-by: Sandipan Das <sandipan@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently we face a severe problem on Braswell that manifests as invalid
ppGTT accesses. The code tries to maintain the PDP (page directory
pointers) inside the context in two ways, direct write into the context
and a pipelined LRI update. The direct write into the context is
fundamentally racy as it is unserialised with any access (read or write)
the GPU is doing. By asserting that Braswell is not used with vGPU
(currently an unsupported platform) we can eliminate the dangerous
direct write into the context image and solely use the pipelined update.
However, the LRI of the PDP fouls up the GPU, causing it to freeze and
take out the machine with "forcewake ack timeouts". This seems possible
to workaround by preventing the GPU from sleeping (via means of
disabling the power-state management interface, i.e. forcing each ring
to remain awake) around the update. Equally, it seems an EMIT_INVALIDATE
before the LRI is sufficient to prevent the forcewake errors.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108656
References: https://bugs.freedesktop.org/show_bug.cgi?id=108714
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181207090213.14352-3-chris@chris-wilson.co.uk
Gen11 fails to deliver wrt global observation point on
tail/entry updates and we sometimes see old entry.
Use clflush to forcibly evict our possibly stale copy
of the cacheline in hopes that we get fresh one from gpu.
Obviously there is something amiss in the coherency protocol so
this can be consired as a workaround until real cause
is found.
The working hardware will do the evict without our cue anyways,
so the cost in there should be ameliorated by that fact.
v2: for next pass, s/flush/evict, add reset (Chris)
References: https://bugzilla.freedesktop.org/show_bug.cgi?id=108315
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20181205134612.24822-1-mika.kuoppala@linux.intel.com
Technically the display-hub driver could access registers via the
specified region, though it practice it will do so via the display
controllers' register regions.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Technically the display-hub driver could access registers via the
specified region, though it practice it will do so via the display
controllers' register regions.
Signed-off-by: Thierry Reding <treding@nvidia.com>
This reverts commit 77cab92a2c.
Heiko reports that this breaks building of strace, which isn't ok.
Reported-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Rafael Aquini <aquini@redhat.com>
Cc: William Kucharski <william.kucharski@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
- spaces before tabs,
- spaces at the end of lines,
- multiple blank lines,
- blank lines before EXPORT_SYMBOL,
can all go.
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
posix_unblock_lock() is not specific to posix locks, and behaves
nearly identically to locks_delete_block() - the former returning a
status while the later doesn't.
So discard posix_unblock_lock() and use locks_delete_block() instead,
after giving that function an appropriate return value.
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
When we find an existing lock which conflicts with a request,
and the request wants to wait, we currently add the request
to a list. When the lock is removed, the whole list is woken.
This can cause the thundering-herd problem.
To reduce the problem, we make use of the (new) fact that
a pending request can itself have a list of blocked requests.
When we find a conflict, we look through the existing blocked requests.
If any one of them blocks the new request, the new request is attached
below that request, otherwise it is added to the list of blocked
requests, which are now known to be mutually non-conflicting.
This way, when the lock is released, only a set of non-conflicting
locks will be woken, the rest can stay asleep.
If the lock request cannot be granted and the request needs to be
requeued, all the other requests it blocks will then be woken
To make this more concrete:
If you have a many-core machine, and have many threads all wanting to
briefly lock a give file (udev is known to do this), you can get quite
poor performance.
When one thread releases a lock, it wakes up all other threads that
are waiting (classic thundering-herd) - one will get the lock and the
others go to sleep.
When you have few cores, this is not very noticeable: by the time the
4th or 5th thread gets enough CPU time to try to claim the lock, the
earlier threads have claimed it, done what was needed, and released.
So with few cores, many of the threads don't end up contending.
With 50+ cores, lost of threads can get the CPU at the same time,
and the contention can easily be measured.
This patchset creates a tree of pending lock requests in which siblings
don't conflict and each lock request does conflict with its parent.
When a lock is released, only requests which don't conflict with each
other a woken.
Testing shows that lock-acquisitions-per-second is now fairly stable
even as the number of contending process goes to 1000. Without this
patch, locks-per-second drops off steeply after a few 10s of
processes.
There is a small cost to this extra complexity.
At 20 processes running a particular test on 72 cores, the lock
acquisitions per second drops from 1.8 million to 1.4 million with
this patch. For 100 processes, this patch still provides 1.4 million
while without this patch there are about 700,000.
Reported-and-tested-by: Martin Wilck <mwilck@suse.de>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
posix_locks_conflict() and flock_locks_conflict() both return int.
leases_conflict() returns bool.
This inconsistency will cause problems for the next patch if not
fixed.
So change posix_locks_conflict() and flock_locks_conflict() to return
bool.
Also change the locks_conflict() helper.
And convert some
return (foo);
to
return foo;
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Now that requests can block other requests, we
need to be careful to always clean up those blocked
requests.
Any time that we wait for a request, we might have
other requests attached, and when we stop waiting,
we must clean them up.
If the lock was granted, the requests might have been
moved to the new lock, though when merged with a
pre-exiting lock, this might not happen.
In all cases we don't want blocked locks to remain
attached, so we remove them to be safe.
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: J. Bruce Fields <bfields@redhat.com>
Tested-by: syzbot+a4a3d526b4157113ec6a@syzkaller.appspotmail.com
Tested-by: kernel test robot <rong.a.chen@intel.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
For a lot of use cases we need 64bit sequence numbers. Currently drivers
overload the dma_fence structure to store the additional bits.
Stop doing that and make the sequence number in the dma_fence always
64bit.
For compatibility with hardware which can do only 32bit sequences the
comparisons in __dma_fence_is_later only takes the lower 32bits as significant
when the upper 32bits are all zero.
v2: change the logic in __dma_fence_is_later
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Link: https://patchwork.freedesktop.org/patch/266927/
On H3 (ES1.x, ES2.0) and M3-W (ES1.0, ES1.1) the clock setting for HS400
needs a quirk to function properly. The reason for the quirk is that
there are two settings which produces same divider value for the SDn
clock. On the effected boards the one currently selected results in
HS400 not working.
This change uses the same method as the Gen2 CPG driver and simply
ignores the first clock setting as this is the offending one when
selecting the settings. Which of the two possible settings is used have
no effect for SDR104.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Document the known use cases of the different clock settings. This is
useful as different SoC and ES versions use different settings to do
the same thing as there is more than one combination to achieve the
same SDn clock speed.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
The driver tries to figure out which state a SD clock is in when the
clock is registered, instead of setting a known state. This can be
problematic for two reasons.
1. If the clock driver can't figure out the state of the clock,
registration of the clock fails, and setting of a known state by a
clock user is not possible.
2. The state of the clock depends on if and how the bootloader
configured it. The driver only checks that the rate is known, not if
the clock is stopped or not for example.
Fix this by setting a known state and making sure the clock is stopped.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
This patch will enable headset button for new Chrome platform.
Signed-off-by: Kailang Yang <kailang@realtek.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Extend some structs to add the support for jack button changes.
Now snd_hda_jack_add_kctl() receives two more arguments: the jack type
and the jack keymaps. Both are optional, and when zero are passed,
the function behaves just like before.
For reporting button state changes, you'd need to update
jack->button_state bits accordingly, typically in the jack callback.
Then the value OR'ed with button_state and the jack plug state is
passed to snd_jack_report().
Note that currently the code assumes only the one-shot button events,
i.e. it tries to send the button release soon after sending the button
event. If a driver really supports the button release handling by
itself, we may need to introduce some flag to control this behavior in
future.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For allowing the callee to evaluate the associated jack information
and the unsolicited event data, add the new fields to
hda_jack_callback. They can be used, for example, to retrieve the
headset button state in the callback.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Use DEFINE_SHOW_ATTRIBUTE macro to simplify the code.
Signed-off-by: Yangtao Li <tiny.windzz@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Use DEFINE_SHOW_ATTRIBUTE macro to simplify the code.
Signed-off-by: Yangtao Li <tiny.windzz@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
There is a plan to build the kernel with -Wimplicit-fallthrough and
these places in the code produced warnings. Fix them up.
Signed-off-by: Mathieu Malaterre <malat@debian.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Subsequent generations of Tegra, such as Tegra194, contain a completely
different set of GPIOs. In order to clarify that the Tegra186 defines
are indeed specific to Tegra186, change the prefix from TEGRA_ to
TEGRA186_.
Note that for now we need to keep the old definitions in place to avoid
breaking compilation in file that use this header. Once all users have
been converted to use the new defines, the old ones can be removed.
Also note that this is only a naming change and doesn't affect device
tree ABI.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Jon Hunter <jonathanh@nvidia.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Commit 78d3a92edb ("gpiolib-acpi: Register GpioInt ACPI event handlers
from a late_initcall") deferred the entire acpi_gpiochip_request_interrupt
call for each event resource.
This means it also delays the gpiochip_request_own_desc(..., "ACPI:Event")
call. This is a problem if some AML code reads the GPIO pin before we
run the deferred acpi_gpiochip_request_interrupt, because in that case
acpi_gpio_adr_space_handler() will already have called
gpiochip_request_own_desc(..., "ACPI:OpRegion") causing the call from
acpi_gpiochip_request_interrupt to fail with -EBUSY and we will fail to
register an event handler.
acpi_gpio_adr_space_handler is prepared for acpi_gpiochip_request_interrupt
already having claimed the pin, but the other way around does not work.
One example of a problem this causes, is the event handler for the OTG
ID pin on a Prowise PT301 tablet not registering, keeping the port stuck
in whatever mode it was in during boot and e.g. only allowing charging
after a reboot.
This commit fixes this by only deferring the request_irq call and the
initial run of edge-triggered IRQs instead of deferring all of
acpi_gpiochip_request_interrupt.
Cc: stable@vger.kernel.org
Fixes: 78d3a92edb ("gpiolib-acpi: Register GpioInt ACPI event ...")
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
We try to force NAND controller drivers to properly separate the NAND
controller object from the NAND chip one, so let's deprecate the dummy
controller object embedded in nand_chip to encourage them to create
their own instance.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Replace the license text by an SPDX tag.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Reviewed-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The dummy controller is kept around to support old drivers. Let's
patch this one and declare our own nand_controller instance.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Reviewed-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Mimic what has been done in the core and avoid passing mtd_info object
internally.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Reviewed-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
From LPC18xx and LPC43xx User Manuals the GPIO controller consists of
the following weakly connected blocks:
* GPIO pin interrupt block at 0x40087000,
* GPIO GROUP0 interrupt block at 0x40088000,
* GPIO GROUP1 interrupt block at 0x40089000,
* GPIO port block at 0x400F4000.
While all 4 sub-controller blocks have their own I/O addresses, moreover
all 3 interrupt blocks are APB0 peripherals and high-speed GPIO block is
an AHB slave, according to the hardware manual interrupt controllers and
GPIO controller block are seen as a single device, all 4 sub-controllers
have the shared reset signal RGU #28 and the same shared clock to access
registers CLK_Mx_GPIO on CCU1.
The change adds descriptions of the currently missing interrupt controller
blocks found on GPIO controller, new added properties are 'reg-names',
'resets', 'interrupt-controller' and '#interrupt-cells', also the example
is updated to reflect the changes in device tree binding description.
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
checkpatch reports a bunch of coding style issues. Let's fix them
all in one step.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Add an SPDX GPL-2.0 tag and update MODULE_LICENSE() to match the
license text.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The dummy controller is kept around to support old drivers. Let's
patch this one and declare our own nand_controller instance.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
nand_to_fsmc() is used almost everywhere except in
fsmc_setup_data_interface() where nand_get_controller_data() is used
instead. Make that consistent and drop the nand_set_controller_data()
call in the probe path.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The kernel-doc describing struct fsmc_nand_data is not in sync with
the struct itself. Add missing entries and drop invalid ones.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Mimic what has been done in the core and stop passing mtd_info
objects to internal functions.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
When adjusting sack block sequence numbers, skb_make_writable() gets
called to make sure tcp options are all in the linear area, and buffer
is not shared.
This can cause tcp header pointer to get reallocated, so we must
reaload it to avoid memory corruption.
This bug pre-dates git history.
Reported-by: Neel Mehta <nmehta@google.com>
Reported-by: Shane Huntley <shuntley@google.com>
Reported-by: Heather Adkins <argv@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The change adds support of LPC18xx/LPC43xx GPIO pin interrupt controller
block within SoC GPIO controller. The new interrupt controller driver
allows to configure and capture edge or level interrupts on 8 arbitrary
selectedinput GPIO pins, and lift the signals to be reported as NVIC rising
edge interrupts. Configuration of a particular GPIO pin to serve as
interrupt and its mapping to an interrupt on NVIC is done by SCU pin
controller, for more details see description of 'nxp,gpio-pin-interrupt'
device tree property of a GPIO pin [1].
From LPC18xx and LPC43xx User Manuals the GPIO controller consists of
the following blocks:
* GPIO pin interrupt block at 0x40087000, this change adds its support,
* GPIO GROUP0 interrupt block at 0x40088000,
* GPIO GROUP1 interrupt block at 0x40089000,
* GPIO port block at 0x400F4000, it is supported by the original driver.
While all 4 sub-controller blocks have their own I/O addresses, moreover
all 3 interrupt blocks are APB0 peripherals and high-speed GPIO block is
an AHB slave, according to the hardware manual the GPIO controller is
seen as a single block, and 4 sub-controllers have the shared reset signal
RGU #28 and clock to register interface CLK_CPU_GPIO on CCU1.
Likely support of two GPIO group interrupt blocks won't be added in short
term, because the mechanism to mask several interrupt sources is not well
defined.
[1] Documentation/devicetree/bindings/pinctrl/nxp,lpc1850-scu.txt
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
This is a non-functional change, it simplifies multiple access to
'struct device' pointer derived from a platform device pointer,
the new local variable will also be used in the following changes.
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Replace GPLv2 header with the SPDX identifier.
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The error cases of mediatek_gpio_bank_probe() would go unnoticed (except
for the dev_err() messages). The probe function should return an error
if one of the banks failed to initialize properly indicated by
not returning non-0.
Fixes: 4ba9c3afda ("gpio: mt7621: Add a driver for MT7621")
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Sean Wang <sean.wang@kernel.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
kasprintf() may return NULL on failure of internal allocation thus the
assigned label is not safe if not explicitly checked. On error
mediatek_gpio_bank_probe() returns negative values so -ENOMEM in the
(unlikely) failure case should be fine here.
Fixes: 4ba9c3afda ("gpio: mt7621: Add a driver for MT7621")
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Reviewed-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Acked-by: Sean Wang <sean.wang@kernel.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
In nand_scan_ident(), the controller driver resets every NAND chip.
This is done by sending NAND_CMD_RESET. The Denali IP provides
another way to do the equivalent thing; if a bit is set in the
DEVICE_RESET register, the controller sends the RESET command to
the corresponding device. denali_reset_banks() uses it to reset
all devices beforehand.
This redundant reset sequence was needed to know the actual number
of chips before calling nand_scan_ident(); if DEVICE_RESET fails,
there is no chip in that chip select. Then, denali_reset_banks()
sets denali->max_banks to the number of detected chips.
As commit f486287d23 ("mtd: nand: denali: fix bank reset function
to detect the number of chips") explained, nand_scan_ident() issued
Set Features (0xEF) command to all CS lines, some of which may not be
connected with a chip. Then, the driver would wait for R/B# response,
which never happens.
This problem was solved by commit 107b7d6a7a ("mtd: rawnand: avoid
setting again the timings to mode 0 after a reset"). In the current
code, nand_setup_data_interface() is called from nand_scan_tail(),
which is invoked after the chip detection.
Now, we can really remove the redundant denali_nand_banks() by simply
passing the maximum number of chip selects supported by this IP
(typically 4 or 8) to nand_scan(). Let's leave all the chip detection
process to nand_scan_ident().
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The Denali NAND IP has no way to read out the current signal level
of the R/B# pin. Instead, denali_dev_ready() checks if the R/B#
transition has already happened. (The INTR__INT_ACT interrupt is
asserted at the rising edge of the R/B# pin.) It is not a correct
way to implement the ->dev_ready() hook.
In fact, it has a drawback; in the nand_scan_ident phase, the chip
detection iterates over maxchips until it fails to find a homogeneous
chip. For the last loop, nand_reset() fails if no chip is there.
If ->dev_ready hook exists, nand_command(_lp) calls nand_wait_ready()
after NAND_CMD_RESET. However, we know denali_dev_ready() never
returns 1 unless there exists a chip that toggles R/B# in that chip
select. Then, nand_wait_ready() just ends up with wasting 400 msec,
in the end, shows the "timeout while waiting for chip to become ready"
warning.
Let's remove the mis-implemented dev_ready hook, and fallback to
sending the NAND_CMD_STATUS and nand_wait_status_ready(), which
bails out more quickly.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Amstrad Delta NAND driver now uses GPIO API for data I/O so there is no
need to assign memory I/O resource to the device any longer. Drop it.
Signed-off-by: Janusz Krzysztofik <jmkrzyszt@gmail.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Don't readw()/writew() data directly from/to GPIO port which is under
control of gpio-omap driver, use GPIO consumer API instead.
The driver should now work with any 8-bit bidirectional GPIO port, not
only OMAP.
Signed-off-by: Janusz Krzysztofik <jmkrzyszt@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>