IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
This patch convert dprintk(info) to pci_info (or pci_dbg if the message
is not important).
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Fix a lot of style issue found by checkpatch.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Fix some typo in doc.
[hverkuil: The TV encoder are -> The TV encoder is]
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
This reverts commit 8dce4b265a5357731058f69645840dabc718c687.
The revert is slighly modified:
- Documentation/media/v4l-drivers/index.rst is ,ot restored since removed from tree
- drivers/staging/media/Makefile is not restored since the zoran driver
is not compilable yet.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Add proper control type to the recently introduced
V4L2_CID_MPEG_VIDEO_FRAME_SKIP_MODE control. This will forward it to
v4l2_ctrl_new_std_menu() not v4l2_ctrl_new_std(), what causes the
failure. This fixes the following warning during driver initialization:
s5p_mfc_enc_ctrls_setup:2671: Adding control (18) failed
s5p_mfc_open:811: Failed to setup mfc controls
Fixes: ef56b3ee4f54 ("media: s5p-mfc: Use standard frame skip mode control")
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Neither Vlc nor Gstreamer likes the PES_scrambling_control bits.
In the case of GST, this can be seen with:
$ GST_DEBUG=2 LANG=C gst-play-1.0 pcm_audio.ts
...
0:00:00.097973439 12308 0x55f7ddd155e0 WARN pesparser pesparse.c:411:mpegts_parse_pes_header: Wrong '0x10' marker before PES_scrambling_control (0x40)
0:00:00.097987026 12308 0x55f7ddd155e0 WARN tsdemux tsdemux.c:2314:gst_ts_demux_parse_pes_header: Error parsing PES header. pid: 0x111 stream_type: 0x6
...
So, change, it. After such change, the stream now plays
fine with Vlc, Gstreamer, ffmpeg - and with programs
that use such libraries, like Kaffeine.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
When compared with a stream generated via ffmpeg, it can be
noticed that the PES doesn't contain any PCR info. That could
cause problems with userspace decoding. So, rewrite the
logic that fills the adaptation info, in order to allow it
to add PCR frames without breaking frame alignment.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Instead of passing struct pes_ts_header_write_args fields as
function parameters, just pass a pointer to the struct.
That would allow adding more args without needing to change
the function prototype.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
In order to better test userspace applications, add an encoding
for the strings. The original channel string is also too big.
As we're just playing the first compasses of the 5th Beethoven
Symphony, change the service name to just "Beethoven". Also,
add a provider's name. For such purpose, let's point it to
the Linux media community website.
With such changes, the SDT table now looks more like one got
from a real digital TV channel:
SDT
| table_id 0x42
| section_length 44
| one 2
| zero 1
| syntax 1
| transport_stream_id 1860
| current_next 1
| version 0
| one2 3
| section_number 0
| last_section_number 0
| network_id 65281
| reserved 255
|\
|- service 0x0880
| EIT schedule 0
| EIT present following 0
| free CA mode 0
| running status 4
| descriptor length 27
| 0x48: service_descriptor
| service type 1
| provider 'LinuxTV.org'
| name 'Beethoven'
|_ 1 services
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
There's no reason to use static vars to store PSI version
numbers.
Also, currently, version numbers are starting with 0x01,
because there's a code being called that increases it to
1 for all table initializer code, as the code may support
dynamic changes at the PS tables on some future.
So, let's just initialize them to 0x1f, in order for the
versions to be reported as starting from 0.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
This field should point to the network ID, and has different
ranges for cable, terrestrial or satellite. It also has
an special range for temporary private usage.
For now, let's use the temporary private one. Once the
Network Information Table (NIT) gets added, this should be
reviewed.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Changeset 870e350d4e39 ("media: vidtv: get rid of ENDIAN_BITFIELD nonsense")
was incomplete. There are still some wrong endannes logic at
the driver. Get rid of them.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
There's no need to have a check when setting timing.past_jiffies,
as we can simply do the initialization earlier at vidtv_mux_init().
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Using struct|enum :c:`type` here sounds a bad idea, as it may be
documenting data types that were already removed. Also, it makes
the file harder to read.
As automarkup.py should be able to handle those cases, let's
just get rid of the markup.
While not required by automarkup, in order to make easier for
grep, both type and names are now at the same line.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
There are several places there where struct was used twice,
on multiple lines, like:
struct
struct ``v4l2_zoomcap``
That's probably done during the DocBook to ReSt conversion.
Probably the cases where "struct struct" were at the same line
was addressed back then, but the multi-line cases are still
there.
Get rid of them.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
As reported by Sphinx:
./Documentation/driver-api/media/v4l2-subdev:490: ./include/media/v4l2-subdev.h:384: WARNING: Unparseable C cross-reference: 'struct'
Invalid C declaration: Expected identifier in nested name, got keyword: struct [error at 6]
struct
------^
The markup there is wrong:
&struct &v4l2_input -> &struct v4l2_input
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The new C domain code on Sphinx 3 is a lot more pedantic.
It only accepts real functions declared as c:function.
So, declarations like this are not valid:
.. c:function::
int (*adap_enable)(struct cec_adapter *adap, bool enable);
Also, no blank lines are allowed after ".. c:function:", and
continuation lines should be like:
.. c:function:
int (void foo, \
int bar);
Change the logic there, in order to avoid lots of warnings
when built with Sphinx 3.x.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
There are two kernel-doc markups for include/media/v4l2-async.h,
one at v4l2-async.rst and another one at v4l2-subdev.rst.
Sphinx 3.x checks it and complains for duplicated symbols.
So, get rid of one of them.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The same markup are there twice, causing warnings with
Sphinx 3.x. Drop the second copy.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
There are several :c:type: definitions there, in order to
do cross-references with the driver's documentation.
Those are broken when docs are built with Sphinx 3.x, as
it would require :c:struct: instead.
For Sphinx < 3.x, the automarkup.py extension is able to do the
replacement already, and a future improvement on it should make
it also work with Sphinx 3.x.
So, get rid of the usage of the :c:type: macro there.
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Sphinx 3.x is pedantic with glossary entries:
Documentation/userspace-api/media/glossary.rst:17: WARNING: term device driver not found in case sensitive match.made a reference to Device Driver instead.
Documentation/userspace-api/media/glossary.rst:59: WARNING: term media hardware not found in case sensitive match.made a reference to Media Hardware instead.
Documentation/userspace-api/media/glossary.rst:59: WARNING: term IP block not found in case sensitive match.made a reference to IP Block instead.
Documentation/userspace-api/media/glossary.rst:64: WARNING: term hardware component not found in case sensitive match.made a reference to Hardware Component instead.
Documentation/userspace-api/media/glossary.rst:64: WARNING: term ip block not found in case sensitive match.made a reference to IP Block instead.
Documentation/userspace-api/media/glossary.rst:70: WARNING: term peripheral not found in case sensitive match.made a reference to Peripheral instead.
Documentation/userspace-api/media/glossary.rst:130: WARNING: term V4L2 hardware not found in case sensitive match.made a reference to V4L2 Hardware instead.
Documentation/userspace-api/media/glossary.rst:151: WARNING: term hardware peripheral not found in case sensitive match.made a reference to Hardware Peripheral instead.
Documentation/userspace-api/media/glossary.rst:183: WARNING: term device node not found in case sensitive match.made a reference to Device Node instead.
Documentation/userspace-api/media/glossary.rst:191: WARNING: term bridge driver not found in case sensitive match.made a reference to Bridge Driver instead.
While it works with case-insensitive entires, it complains.
Let's fix it, in order to cleanup the warnings. Also, I won't
doubt that a later change on Sphinx will end breaking support
for it.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Documentation/driver-api/media/camera-sensor.rst:123: WARNING: Inline literal start-string without end-string.
There's a missing blank space over there.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
- Unbreak the magic 'search the timer interrupt' logic in IO/APIC code
which got wreckaged when the core interrupt code made the state
tracking logic stricter. That caused the interrupt line to stay masked
after switching from IO/APIC to PIC delivery mode, which obviously
prevents interrupts from being delivered.
- Make run_on_irqstack_code() typesafe. The function argument is a void
pointer which is then casted to 'void (*fun)(void *). This breaks
Control Flow Integrity checking in clang. Use proper helper functions
for the three variants reuqired.
-----BEGIN PGP SIGNATURE-----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=BcVu
-----END PGP SIGNATURE-----
Merge tag 'x86-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
"Two fixes for the x86 interrupt code:
- Unbreak the magic 'search the timer interrupt' logic in IO/APIC
code which got wreckaged when the core interrupt code made the
state tracking logic stricter.
That caused the interrupt line to stay masked after switching from
IO/APIC to PIC delivery mode, which obviously prevents interrupts
from being delivered.
- Make run_on_irqstack_code() typesafe. The function argument is a
void pointer which is then cast to 'void (*fun)(void *).
This breaks Control Flow Integrity checking in clang. Use proper
helper functions for the three variants reuqired"
* tag 'x86-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/ioapic: Unbreak check_timer()
x86/irq: Make run_on_irqstack_cond() typesafe
- Reset the TI/DM timer before enabling it instead of doing it the other
way round.
- Initialize the reload value for the GX6605s timer correctly so the
hardware counter starts at 0 again after overrun.
- Make error return value negative in the h8300 timer init function
-----BEGIN PGP SIGNATURE-----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=Htin
-----END PGP SIGNATURE-----
Merge tag 'timers-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer updates from Thomas Gleixner:
"A set of clocksource/clockevents updates:
- Reset the TI/DM timer before enabling it instead of doing it the
other way round.
- Initialize the reload value for the GX6605s timer correctly so the
hardware counter starts at 0 again after overrun.
- Make error return value negative in the h8300 timer init function"
* tag 'timers-urgent-2020-09-27' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
clocksource/drivers/timer-gx6605s: Fixup counter reload
clocksource/drivers/timer-ti-dm: Do reset before enable
clocksource/drivers/h8300_timer8: Fix wrong return value in h8300_8timer_init()
Pinned pages shouldn't be write-protected when fork() happens, because
follow up copy-on-write on these pages could cause the pinned pages to
be replaced by random newly allocated pages.
For huge PMDs, we split the huge pmd if pinning is detected. So that
future handling will be done by the PTE level (with our latest changes,
each of the small pages will be copied). We can achieve this by let
copy_huge_pmd() return -EAGAIN for pinned pages, so that we'll
fallthrough in copy_pmd_range() and finally land the next
copy_pte_range() call.
Huge PUDs will be even more special - so far it does not support
anonymous pages. But it can actually be done the same as the huge PMDs
even if the split huge PUDs means to erase the PUD entries. It'll
guarantee the follow up fault ins will remap the same pages in either
parent/child later.
This might not be the most efficient way, but it should be easy and
clean enough. It should be fine, since we're tackling with a very rare
case just to make sure userspaces that pinned some thps will still work
even without MADV_DONTFORK and after they fork()ed.
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This allows copy_pte_range() to do early cow if the pages were pinned on
the source mm.
Currently we don't have an accurate way to know whether a page is pinned
or not. The only thing we have is page_maybe_dma_pinned(). However
that's good enough for now. Especially, with the newly added
mm->has_pinned flag to make sure we won't affect processes that never
pinned any pages.
It would be easier if we can do GFP_KERNEL allocation within
copy_one_pte(). Unluckily, we can't because we're with the page table
locks held for both the parent and child processes. So the page
allocation needs to be done outside copy_one_pte().
Some trick is there in copy_present_pte(), majorly the wrprotect trick
to block concurrent fast-gup. Comments in the function should explain
better in place.
Oleg Nesterov reported a (probably harmless) bug during review that we
didn't reset entry.val properly in copy_pte_range() so that potentially
there's chance to call add_swap_count_continuation() multiple times on
the same swp entry. However that should be harmless since even if it
happens, the same function (add_swap_count_continuation()) will return
directly noticing that there're enough space for the swp counter. So
instead of a standalone stable patch, it is touched up in this patch
directly.
Link: https://lore.kernel.org/lkml/20200914143829.GA1424636@nvidia.com/
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This prepares for the future work to trigger early cow on pinned pages
during fork().
No functional change intended.
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(Commit message majorly collected from Jason Gunthorpe)
Reduce the chance of false positive from page_maybe_dma_pinned() by
keeping track if the mm_struct has ever been used with pin_user_pages().
This allows cases that might drive up the page ref_count to avoid any
penalty from handling dma_pinned pages.
Future work is planned, to provide a more sophisticated solution, likely
to turn it into a real counter. For now, make it atomic_t but use it as
a boolean for simplicity.
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Maxime is no more a BayLibre employee, and his e-mail address is now
invalid.
I'll be happy to add him back using another e-mail address if he wants
to continue maintaining this driver.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The function 'rkisp1_stream_start' calls 'rkisp1_set_next_buf'
which access the buffers, so the call should be protected by
taking the cap->buf.lock.
After this patch, all locks are reviewed and commented so remove
the TODO item "review and comment every lock"
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
When locking, use either 'spin_lock' or 'spin_lock_irq'
according to the context. There is nowhere need to
use 'spin_lock_irqsave'.
Outside of irq context, always use 'spin_lock_irq'
to be on the safe side.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
In the function 'rkisp1_params_config_parameter' the lock
is taken while applying the default config. But the lock
only needs to protect the buffers list and the 'is_streaming'
field, so move the locking to lock only what is needed.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The field stats->is_streaming is written in 'start_streaming' callback
without the stats->lock protection.
The isr might run together with the callback so 'spin_lock_irq'
should be used.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The signal RKISP1_CIF_ISP_FRAME_IN is not used in the isr so
there is no need to enable it.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The signal RKISP1_CIF_ISP_FRAME is set when the ISP completes
outputting the frame to the next block in the pipeline.
In order to keep buffer synchronization we assume that the
RKISP1_CIF_ISP_V_START signal never arrives together with the
RKISP1_CIF_ISP_FRAME signal.
In case those signals arrive together then the code is not able to
tell if the RKISP1_CIF_ISP_FRAME signal relates to the frame of
the current v-start or the previous. This patch adds a WARN_ONCE
and a debugfs var to catch it.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The isp.frame_sequence is now read only from the irq handlers
that are all fired from the same interrupt, so there is not need
for atomic operation.
In addition, the frame seq incrementation is moved from
rkisp1_isp_queue_event_sof to rkisp1_isp_isr to make the code
clearer and the incorrect inline comment is removed.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The params isr is called when the ISP finishes processing a frame
(RKISP1_CIF_ISP_FRAME). Configurations performed in the params isr
will be applied on the next frame. Since frame_sequence is updated
on the vertical sync signal, we should use frame_sequence + 1 as
the vb.sequence of the params buffer to indicate to userspace on
which frame these parameters are being applied.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Currently, the first buffer queued in the params node is returned
immediately to userspace and a copy of it is saved in the field
'cur_params'. The copy is later used for the first configuration
when the stream is initiated by one of selfpath/mainpath capture nodes.
There are 3 problems with this implementation:
- The first params buffer is applied and returned to userspace even if
userspace never calls to streamon on the params node.
- If the first params buffer is queued after the stream started on the
params node then it will return to userspace but will never be used.
- The frame_sequence of the first buffer is set to -1 if the main/selfpath
did not start streaming.
A correct implementation is to apply the first params buffer when stream
is started from mainpath/selfpath and only if params is also streaming.
The patch adds a new function 'rkisp1_params_apply_params_cfg' which takes
a buffer from the buffers queue, apply it and returns it to userspace.
The function is called from the irq handler and when main/selfpath stream
starts - in the function 'rkisp1_params_config_parameter'
Also remove the fields 'cur_params', 'is_first_params' which are no
more needed.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Reported-by: kernel test robot <lkp@intel.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The cproc (color processing) configuration needs to know if
an image effect is configured. The code uses the image effect
in 'cur_params' which is the first params buffer queued in the stream.
This is the wrong place to read the value. The value should be
taken from the current params buffer.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Currently the code in the isr checks if the buffer list is not
empty before referencing a buffer and then check again if the
buffer is not NULL. Instead we can save one 'if' statement by
returning if the buffers list is empty.
Also remove non-helpful inline doc 'get one empty buffer'
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The code in '.stop_streaming' callback releases and acquire the lock
at each iteration when returning the buffers.
Holding the lock disables interrupts so it should be minimized.
To make the code cleaner and still minimize holding the lock,
the buffer list is first moved to a local list and
then can be iterated without the lock.
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Acked-by: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Documentation on how to call the subdev ops is currently in the middle of
synchronous and asynchronous registration. Move it to a dedicated
subsection after the registration methods.
Also move the final paragraph "The advantage of using v4l2_subdev..." to
the beginning of the new section as it has an introductory content.
[hverkuil: correct accidental deletion of the last part of a sentence]
Suggested-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
