modinfo and kerneldoc disagreed on the meaning of this field.
Reported-by: Andrea Oliveri <oliveriandrea@gmail.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Add OTP to the string: "can't parse empty OTP/NVM section"
NVM usually refers to nvm_file while the problem can be in
the OTP.
Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The recent unified scan api change introduced issues
when stopping ongoing scans, since both regular and
sched scan now use same stopped notification.
When issuing a new scan right after a running one,
we get the "old" notification and handle it wrongly
as notification for the current scan.
Fix it by introducing a new function that make sure
we consume the pending notifications before issuing
a new scan.
Signed-off-by: Eliad Peller <eliad@wizery.com>
Reviewed-by: ArikX Nemtsov <arik@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
According to new requirements, the ACK / CTS kill mask is
not related to reduced TX power anymore. This allows to
remove the code that tracked reduced TX power enablement
across different interfaces.
The ACK / CTS kill mask is now fetch from a table. It
depends on the Activity grading (activity from BT) and on
the Look Up Table (LUT) type.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The Remain On Channel framework added to the firmare is
a bit like time events. It allows the driver to request
the firmware to be on a certain channel for a certain time.
Unlike the time events, the ROC infrastructure doesn't need
a MAC context in the firmware - it uses a generic context
called "auxiliary framework".
This is useful for any offchannel activity that is not bound
to a specific MAC.
The flow is synchronized much like with time events:
1) The driver receives an action frame from the wpa_supplicant
via nl80211 that requests to be sent offchannel.
2) The driver sends an Aux ROC command (0x53) to the firmware.
3) The firmware responds with the unique id of the time event.
4) When time event starts, the driver puts the frame in the
Aux queue.
Special care needs to be taken when the time events ends:
the queue needs to be cleaned-up.
Signed-off-by: Ariej Marjieh <ariej.marjieh@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
They are the same and nr_node_ids is provided by the memory subsystem.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
After the locking was moved up to the caller of the get_unbound_pool(),
out_unlock label doesn't need to do any unlock operation and the name
became bad, so we just remove this label, and the only usage-site
"goto out_unlock" is subsituted to "return pool".
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
In 75ccf5950f ("workqueue: prepare flush_workqueue() for dynamic
creation and destrucion of unbound pool_workqueues"), a comment
about the synchronization for the pwq in pwq_unbound_release_workfn()
was added. The comment claimed the flush_mutex wasn't strictly
necessary, it was correct in that time, due to the pwq was protected
by workqueue_lock.
But it is incorrect now since the wq->flush_mutex was renamed to
wq->mutex and workqueue_lock was removed, the wq->mutex is strictly
needed. But the comment was miss-updated when the synchronization
was changed.
This patch removes the incorrect comments and doesn't add any new
comment to explain why wq->mutex is needed here, which is definitely
obvious and wq->pwqs_node has "WQ" notation in its definition which is
better comment.
The old commit mentioned above also introduced a comment in link_pwq()
about the synchronization. This comment is also removed in this patch
since the whole link_pwq() is proteced by wq->mutex.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
In 51697d3939 ("workqueue: use generic attach/detach routine for
rescuers"), The rescuer detaches itself from the pool before put_pwq()
so that the put_unbound_pool() will not destroy the rescuer-attached
pool.
It is unnecessary. worker_detach_from_pool() can be used as the last
statement to access to the pool just like the regular workers,
put_unbound_pool() will wait for it to detach and then free the pool.
So we move the worker_detach_from_pool() down, make it coincide with
the regular workers.
tj: Minor description update.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Simply unfold the code of start_worker() into create_worker() and
remove the original start_worker() and create_and_start_worker().
The only trade-off is the introduced overhead that the pool->lock
is released and regrabbed after the newly worker is started.
The overhead is acceptible since the manager is slow path.
And because this new locking behavior, the newly created worker
may grab the lock earlier than the manager and go to process
work items. In this case, the recheck need_to_create_worker() may be
true as expected and the manager goes to restart which is the
correct behavior.
tj: Minor updates to description and comments.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Commit 6079d38 ("dmaengine: pl330: Remove useless xfer_cb indirection")
removed the __callback() function which created an unnecessary level of
indirection to execute the tranfer callback .xfer_cb
Unfortunately the commit also changed the semantics slightly since that
function used to check if the request was not NULL before attempting to
execute the callback function. Not checking this could lead to a kernel
NULL pointer dereference error.
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Acked-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
worker_set_flags() has only two callers, each specifying %true and
%false for @wakeup. Let's push the wake up to the caller and remove
@wakeup from worker_set_flags(). The caller can use the following
instead if wakeup is necessary:
worker_set_flags();
if (need_more_worker(pool))
wake_up_worker(pool);
This makes the code simpler. This patch doesn't introduce behavior
changes.
tj: Updated description and comments.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Peter Griffin <peter.griffin@linaro.org>
Acked-by: Lee Jones <lee.jones@linaro.org>
Acked-by: Maxime Coquelin <maxime.coquelin@st.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
Apparently fw/hw generates a corrupted QoS Control
Field in Qos NullFunc frames. The only way to
workaround this is to downgrade frames to
NullFunc. This should be okay since powersave is
done by fw/hw and these frames are only used for
CQM purposes (e.g. from hostapd to check if
station is still connected).
This doesn't fix any user visible bug that I know
of. It just prevents from sending out funky frames
on the air.
Reported-by: Janusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
It always bugged me how tid is computed and stored
in a temporary var before written to the control
buffer. It was confusing and it made it difficult
to work with tx helpers.
While at it rename the qos workaround function as
it was misleading - it's not a workaround but
preparation for nwifi tx mode.
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
Now that the custom motion detection API in this driver has been
replaced with a standard API there is no reason anymore to keep it
in staging. So (finally!) move it to drivers/media/usb.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Now that the custom motion detection API has been replaced with a
standard API there is no reason anymore to keep it in staging.
So (finally!) move it to drivers/media/pci.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This adds a DT binding documentation for the MT6589 SoC from Mediatek.
Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
This adds a generic devicetree board file and a dtsi for boards
based on MT6589 SoCs from Mediatek.
Apart from the generic parts (gic, clocks) the only component
currently supported are the timers.
Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com>
Make of_device_id array const, because all OF functions handle it as const.
Signed-off-by: Kiran Padwal <kiran.padwal21@gmail.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Dell XPS 15 (1028:05fe) gives the bad feedback noise from analog
loopback line even if the channels are muted. The similar problem has
been seen on XPS 13, and it was fixed by simply disabling loopback.
The same fixup was confirmed to be applicable and fix the problem on
XPS 15, too.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=80821
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This commit is a supplement to my previous patch.
http://mailman.alsa-project.org/pipermail/alsa-devel/2014-July/079190.html
The special_clk_ctl_put() still returns 0 in error handling case. It should
return -EINVAL.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Similarly to workbuf_size and tempbuf_size, store iram_size in the
coda_devtype structure. This also decreases the IRAM used on i.MX6DL
to 128 KiB.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Similarly to the work buffer size, store the temporary buffer size in the
coda_devtype structure.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
We had the workbuf_size field since the beginning.
Use it to tighten the code a little bit.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This patch exports all auxiliary buffers, including SRAM, as debugfs binary
blobs for debugging purposes. It shows, for example, that psbuf currently
doesn't seem to be used at all on CODA7541, and that slicebuf and workbuf
usage is far from the maximum. It can also be used to validate SRAM size
allocation.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
When encoding into h.264, the input frame stride needs to be a multiple of 16.
During allocation of the input buffers, it may not be known yet whether the
encoder should create h.264 or not. Assume the worst and always use a frame
stride that is a multiple of 16.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
CODA7541 only supports encoding h.264 frames with width and height that are
multiples of the macroblock size.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Even though the CODA h.264 decoder always decodes complete macroblocks, we can
set the stride to the corresponding multiple of 16 and use a value smaller than
that as real width. Unfortunately the same doesn't work for height, as there
is no vertical linesperframe stride for discontiguous planar YUV frames.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
bytesperline is calculated in multiple places, store it in the coda_q_data
structure. This will be more useful later when adding JPEG support.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
On i.MX53 and i.MX6, the CODA VPU can be reset by the System Reset Controller.
We can use this to get out of dire situations, for example after a picture
run timeout.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Rename the per-context prescan_failed variable to hold, as this is what the
flag does: it temporarily keeps the coda from running until new data is fed
into the bitstream buffer or stop_streaming is called on the input side.
A prescan failure on i.MX5 is one possible reason to enter this state, another
one is a picture run timeout on i.MX6.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The coda h.264 decoder also counts PIC_RUNs where no frame was decoded but
a frame was rotated out / marked as ready to be displayed. This causes an
offset between the incoming encoded frame's sequence number and the decode
sequence number returned by the coda. This patch introduces a sequence
counter offset variable to keep track of the difference.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
If the CODA reports macroblock errors, also set the VB2_BUF_STATE_ERROR flag
to alert userspace.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The coda driver advertises timestamp_type V4L2_BUF_FLAG_TIMESTAMP_COPY on
both queues, so we have to copy timestamps from input v4l2 buffers to the
corresponding destination v4l2 buffers. Since the h.264 decoder can reorder
frames, a timestamp queue is needed to keep track of and assign the correct
timestamp to destination buffers.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
In case no further buffers are queued after the stop command, restart
job scheduling explicitly.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Some drivers might allow to decode remaining frames from an internal ringbuffer
after a decoder stop command. Allow those to call v4l2_m2m_try_schedule
directly.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Allow userspace to enable cyclic intra refresh by setting the number of
intra macroblocks per frame to a non-zero value.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
If the bitrate control is set, the encoder works in CBR mode, dynamically
changing the quantization parameters to achieve a constant bitrate.
With the min/max QP controls the quantization parameters can be limited
to a given range.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The driver uses the genalloc API, which doesn't have stubs in
case GENERIC_ALLOCATOR is disabled.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This adds a new function coda_check_firmware that does the firmware
version checks so that this can be done only once from coda_probe
instead of every time the runtime pm framework resumes the coda.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
This patch allows to use the runtime pm and generic pm domain frameworks
to completely gate power to the VPU if it is unused. This functionality
is available on i.MX6.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Use the mem2mem helpers introduced to get rid of some duplicated code.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Using the coda_mutex lock to serialize hardware access would cause
"INFO: possible circular locking dependency detected" lockdep warnings.
Since the possible locking paths are hard to follow, serialize hardware
access with a single workqueue thread. Ultimately the workqueue could
be converted to only do register setup and readout for per-command work
items.
Using the initialized context property, SEQ_END is only queued in
coda_release when needed.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The h.264 decoder produces capture frames that are a multiple of the macroblock
size (16 pixels). To inform userspace about invalid pixel data at the edges,
use the active and padded composing rectangles on the capture queue.
The cropping information is obtained from the h.264 sequence parameter set.
Reviewed-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
