As reported by smatch:
drivers/media/v4l2-core/videobuf2-core.c:2415 __vb2_init_fileio() warn: possible memory leak of 'fileio'
While here, avoid the usage of sizeof(struct foo_struct).
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The handling of q->plane_sizes was wrong in vb2_core_create_bufs().
The q->plane_sizes array was global and it was overwritten by create_bufs.
So if reqbufs was called with e.g. size 100000 then q->plane_sizes[0] would
be set to 100000. If create_bufs was called afterwards with size 200000,
then q->plane_sizes[0] would be overwritten with the new value. Calling
create_bufs again for size 100000 would cause an error since 100000 is now
less than q->plane_sizes[0].
This patch fixes this problem by 1) removing q->plane_sizes and using the
vb->planes[].length field instead, and 2) by introducing a min_length field
in struct vb2_plane. This field is set to the plane size as returned by
the queue_setup op and is the minimum required plane size. So user pointers
or dmabufs should all be at least this size.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Ensure that the offsets are correct before buf_init() is called.
As a consequence the __setup_offsets() function now sets up the
offsets for the given buffer instead of for all new buffers.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Fill in q->bufs[vb->index] before the call to buf_init: it makes
sense that this is initialized correctly.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Rather than setting up the lengths at the end, set them up when
the vb2_buffer is allocated. This also ensures that buf_init()
sees the right length values.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
This ops can never fail, so make these void functions.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
This module is no longer V4L2 specific, so update the module description
accordingly.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Move things related with vb2 file I/O and vb2_thread without doing any
functional changes. After that, videobuf2-internal.h is removed because
it is not necessary any more.
Signed-off-by: Junghak Sung <jh1009.sung@samsung.com>
Signed-off-by: Geunyoung Kim <nenggun.kim@samsung.com>
Acked-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Add copy_timestamp to struct vb2_queue as a flag set if vb2-core should
copy timestamps.
Signed-off-by: Junghak Sung <jh1009.sung@samsung.com>
Signed-off-by: Geunyoung Kim <nenggun.kim@samsung.com>
Acked-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The queue_setup callback has a void pointer that is just for V4L2
and is the pointer to the v4l2_format struct that was passed to
VIDIOC_CREATE_BUFS. The idea was that drivers would use the information
from that struct to buffers suitable for the requested format.
After the vb2 split series this pointer is now a void pointer,
which is ugly, and the reality is that all existing drivers will
effectively just look at the sizeimage field of v4l2_format.
To make this more generic the queue_setup callback is changed:
the void pointer is dropped, instead if the *num_planes argument
is 0, then use the current format size, if it is non-zero, then
it contains the number of requested planes and the sizes array
contains the requested sizes. If either is unsupported, then return
-EINVAL, otherwise use the requested size(s).
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Prepare to divide videobuf2
- Separate vb2 trace events from v4l2 trace event.
- Make wrapper functions that will move to v4l2-side.
- Make vb2_core_* functions that will remain in core-side.
- Add a callback function table for buffer operation which makes vb2-core
to be able to invoke a v4l2-side functions.
- Rename internal functions as vb2_*.
Signed-off-by: Junghak Sung <jh1009.sung@samsung.com>
Signed-off-by: Geunyoung Kim <nenggun.kim@samsung.com>
Acked-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Remove v4l2 stuff - v4l2_buf, v4l2_plane - from struct vb2_buffer.
Add new member variables - bytesused, length, offset, userptr, fd,
data_offset - to struct vb2_plane in order to cover all information
of v4l2_plane.
struct vb2_plane {
<snip>
unsigned int bytesused;
unsigned int length;
union {
unsigned int offset;
unsigned long userptr;
int fd;
} m;
unsigned int data_offset;
}
Replace v4l2_buf with new member variables - index, type, memory - which
are common fields for buffer management.
struct vb2_buffer {
<snip>
unsigned int index;
unsigned int type;
unsigned int memory;
unsigned int num_planes;
struct vb2_plane planes[VIDEO_MAX_PLANES];
<snip>
};
v4l2 specific fields - flags, field, timestamp, timecode,
sequence - are moved to vb2_v4l2_buffer in videobuf2-v4l2.c
struct vb2_v4l2_buffer {
struct vb2_buffer vb2_buf;
__u32 flags;
__u32 field;
struct timeval timestamp;
struct v4l2_timecode timecode;
__u32 sequence;
};
Signed-off-by: Junghak Sung <jh1009.sung@samsung.com>
Signed-off-by: Geunyoung Kim <nenggun.kim@samsung.com>
Acked-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Make videobuf2-v4l2 as a wrapper of videobuf2-core for v4l2-use.
And replace videobuf2-core.h with videobuf2-v4l2.h.
This renaming change should be accompanied by the modifications
of all device drivers that include videobuf2-core.h.
It can be done with just running this shell script.
replace()
{
str1=$1
str2=$2
dir=$3
for file in $(find $dir -name *.h -o -name *.c -o -name Makefile)
do
echo $file
sed "s/$str1/$str2/g" $file > $file.out
mv $file.out $file
done
}
replace "videobuf2-core" "videobuf2-v4l2" "include/media/"
replace "videobuf2-core" "videobuf2-v4l2" "drivers/media/"
replace "videobuf2-core" "videobuf2-v4l2" "drivers/usb/gadget/"
replace "videobuf2-core" "videobuf2-v4l2" "drivers/staging/media/"
Signed-off-by: Junghak Sung <jh1009.sung@samsung.com>
Signed-off-by: Geunyoung Kim <nenggun.kim@samsung.com>
Acked-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=TEFw
-----END PGP SIGNATURE-----
Merge tag 'media/v4.3-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media
Pull media updates from Mauro Carvalho Chehab:
"A series of patches that move part of the code used to allocate memory
from the media subsystem to the mm subsystem"
[ The mm parts have been acked by VM people, and the series was
apparently in -mm for a while - Linus ]
* tag 'media/v4.3-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
[media] drm/exynos: Convert g2d_userptr_get_dma_addr() to use get_vaddr_frames()
[media] media: vb2: Remove unused functions
[media] media: vb2: Convert vb2_dc_get_userptr() to use frame vector
[media] media: vb2: Convert vb2_vmalloc_get_userptr() to use frame vector
[media] media: vb2: Convert vb2_dma_sg_get_userptr() to use frame vector
[media] vb2: Provide helpers for mapping virtual addresses
[media] media: omap_vout: Convert omap_vout_uservirt_to_phys() to use get_vaddr_pfns()
[media] mm: Provide new get_vaddr_frames() helper
[media] vb2: Push mmap_sem down to memops
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=cIed
-----END PGP SIGNATURE-----
Merge tag 'media/v4.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media
Pull media updates from Mauro Carvalho Chehab:
- new DVB frontend drivers: ascot2e, cxd2841er, horus3a, lnbh25
- new HDMI capture driver: tc358743
- new driver for NetUP DVB new boards (netup_unidvb)
- IR support for DVBSky cards (smipcie-ir)
- Coda driver has gain macroblock tiling support
- Renesas R-Car gains JPEG codec driver
- new DVB platform driver for STi boards: c8sectpfe
- added documentation for the media core kABI to device-drivers DocBook
- lots of driver fixups, cleanups and improvements
* tag 'media/v4.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (297 commits)
[media] c8sectpfe: Remove select on undefined LIBELF_32
[media] i2c: fix platform_no_drv_owner.cocci warnings
[media] cx231xx: Use wake_up_interruptible() instead of wake_up_interruptible_nr()
[media] tc358743: only queue subdev notifications if devnode is set
[media] tc358743: add missing Kconfig dependency/select
[media] c8sectpfe: Use %pad to print 'dma_addr_t'
[media] DocBook media: Fix typo "the the" in xml files
[media] tc358743: make reset gpio optional
[media] tc358743: set direction of reset gpio using devm_gpiod_get
[media] dvbdev: document most of the functions/data structs
[media] dvb_frontend.h: document the struct dvb_frontend
[media] dvb-frontend.h: document struct dtv_frontend_properties
[media] dvb-frontend.h: document struct dvb_frontend_ops
[media] dvb: Use DVBFE_ALGO_HW where applicable
[media] dvb_frontend.h: document struct analog_demod_ops
[media] dvb_frontend.h: Document struct dvb_tuner_ops
[media] Docbook: Document struct analog_parameters
[media] dvb_frontend.h: get rid of dvbfe_modcod
[media] add documentation for struct dvb_tuner_info
[media] dvb_frontend: document dvb_frontend_tune_settings
...
Currently vb2 core acquires mmap_sem just around call to
__qbuf_userptr(). However since commit f035eb4e97 (videobuf2: fix
lockdep warning) it isn't necessary to acquire it so early as we no
longer have to drop queue mutex before acquiring mmap_sem. So push
acquisition of mmap_sem down into .get_userptr memop so that the
semaphore is acquired for a shorter time and it is clearer what it is
needed for.
Note that we also need mmap_sem in .put_userptr memop since that ends up
calling vb2_put_vma() which calls vma->vm_ops->close() which should be
called with mmap_sem held. However we didn't hold mmap_sem in some code
paths anyway (e.g. when called via vb2_ioctl_reqbufs() ->
__vb2_queue_free() -> vb2_dma_sg_put_userptr()) and getting mmap_sem in
put_userptr() introduces a lock inversion with queue->mmap_lock in the
above mentioned call path.
Luckily this whole locking mess will get resolved once we convert
videobuf2 core to the new mm helper which avoids the need for mmap_sem
in .put_userptr memop altogether.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Commit 77a3c6fd90 ("[media] vb2: Don't WARN when v4l2_buffer.bytesused
is 0 for multiplanar buffers") uses the __WARN() macro which isn't
defined when CONFIG_BUG isn't set. This introduces a compilation
breakage. Fix it by using WARN_ON() instead.
The commit was also broken in that it merged v1 of the patch while a new
v2 version had been submitted, reviewed and acked. Fix it by
incorporating the changes from v1 to v2.
Fixes: 77a3c6fd90 ("[media] vb2: Don't WARN when v4l2_buffer.bytesused is 0 for multiplanar buffers")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Add videobuf2 specific vb2_qbuf and vb2_dqbuf trace events that mirror the
v4l2_qbuf and v4l2_dqbuf trace events, only they include additional
information about queue fill state and are emitted right before the buffer
is enqueued in the driver or userspace is woken up. This allows to make
sense of the timeline of trace events in combination with others that might
be triggered by __enqueue_in_driver.
Also two new trace events vb2_buf_queue and vb2_buf_done are added,
allowing to trace the handover between videobuf2 framework and driver.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Buffers can be returned back to videobuf2 in driver's streamon handler. In
this case vb2_buffer_done() with buffer state VB2_BUF_STATE_QUEUED will
cause the driver's buf_queue vb2 operation to be called, queueing the same
buffer again only to be returned to videobuf2 using vb2_buffer_done() and so
on.
Add a new buffer state VB2_BUF_STATE_REQUEUEING which, when used as the
state argument to vb2_buffer_done(), will result in buffers queued to the
driver. Using VB2_BUF_STATE_QUEUED will leave the buffer to videobuf2, as it
was before "[media] vb2: allow requeuing buffers while streaming".
Fixes: ce0eff016f ("[media] vb2: allow requeuing buffers while streaming")
[mchehab@osg.samsung.com: fix warning: enumeration value 'VB2_BUF_STATE_REQUEUEING' not handled in switch]
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: stable@vger.kernel.org # for v4.1
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Commit f61bf13b6a ("[media] vb2: add allow_zero_bytesused flag to the
vb2_queue struct") added a WARN_ONCE to catch usage of a deprecated API
using a zero value for v4l2_buffer.bytesused.
However, the condition is checked incorrectly, as the v4L2_buffer
bytesused field is supposed to be ignored for multiplanar buffers. This
results in spurious warnings when using the multiplanar API.
Fix it by checking v4l2_buffer.bytesused for uniplanar buffers and
v4l2_plane.bytesused for multiplanar buffers.
Fixes: f61bf13b6a ("[media] vb2: add allow_zero_bytesused flag to the vb2_queue struct")
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Cc: stable@vger.kernel.org # for v4.0
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
vb2_buffer_done() already allows STATE_QUEUED, but currently only when not
streaming. It is useful to allow it while streaming as well, as this makes
it possible for drivers to requeue buffers while waiting for a stable
video signal.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
If the last buffer was dequeued from a capture queue, let poll return
immediately and let DQBUF return -EPIPE to signal there will no more
buffers to dequeue until STREAMOFF.
The driver signals the last buffer by setting the V4L2_BUF_FLAG_LAST.
To reenable dequeuing on the capture queue, the driver must explicitly
call vb2_clear_last_buffer_queued. The last buffer queued flag is
cleared automatically during STREAMOFF.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Currently vb2 core acquires mmap_sem just around call to
__qbuf_userptr(). However since commit f035eb4e97 (videobuf2: fix
lockdep warning) it isn't necessary to acquire it so early as we no
longer have to drop queue mutex before acquiring mmap_sem. So push
acquisition of mmap_sem down into .get_userptr and .put_userptr memops
so that the semaphore is acquired for a shorter time and it is clearer
what it is needed for.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
My 'allmodconfig' build is _almost_ free of warnings, and most of the
remaining ones are for legacy drivers that just do bad things that I
can't find it in my black heart to care too much about. But this one
was just annoying me:
drivers/media/v4l2-core/videobuf2-core.c:3256:26: warning: unused variable ‘fileio’ [-Wunused-variable]
because commit 0e66100637 ("[media] vb2: fix 'UNBALANCED' warnings
when calling vb2_thread_stop()") removed all users of 'fileio' and
instead calls "__vb2_cleanup_fileio(q)" to clean up q->fileio. But the
now unused 'fileio' variable was left around.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* patchwork: (404 commits)
[media] uvcvideo: add support for VIDIOC_QUERY_EXT_CTRL
[media] uvcvideo: fix cropcap v4l2-compliance failure
[media] media: omap3isp: remove unused clkdev
[media] coda: Add tracing support
[media] coda: drop dma_sync_single_for_device in coda_bitstream_queue
[media] coda: fix fill bitstream errors in nonstreaming case
[media] coda: call SEQ_END when the first queue is stopped
[media] coda: fail to start streaming if userspace set invalid formats
[media] coda: remove duplicate error messages for buffer allocations
[media] coda: move parameter buffer in together with context buffer allocation
[media] coda: allocate bitstream buffer from REQBUFS, size depends on the format
[media] coda: allocate per-context buffers from REQBUFS
[media] coda: use strlcpy instead of snprintf
[media] coda: bitstream payload is unsigned
[media] coda: fix double call to debugfs_remove
[media] coda: check kasprintf return value in coda_open
[media] coda: bitrate can only be set in kbps steps
[media] v4l2-mem2mem: no need to initialize b in v4l2_m2m_next_buf and v4l2_m2m_buf_remove
[media] s5p-mfc: set allow_zero_bytesused flag for vb2_queue_init
[media] coda: set allow_zero_bytesused flag for vb2_queue_init
...
The vb2: fix bytesused == 0 handling (8a75ffb) patch changed the behavior
of __fill_vb2_buffer function, so that if bytesused is 0 it is set to the
size of the buffer. However, bytesused set to 0 is used by older codec
drivers as as indication used to mark the end of stream.
To keep backward compatibility, this patch adds a flag passed to the
vb2_queue_init function - allow_zero_bytesused. If the flag is set upon
initialization of the queue, the videobuf2 keeps the value of bytesused
intact in the OUTPUT queue and passes it to the driver.
Reported-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
This patch splits the io_flags member of vb2_queue into a bit field.
Instead of an enum with flags separate bit fields were introduced.
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Return -EINVAL if read() or write() is not supported by the queue. This
makes it possible to provide both vb2_fop_read and vb2_fop_write in a
struct v4l2_file_operations since the vb2_fop_* function will check if
the file operation is allowed.
A similar check exists in __vb2_init_fileio() which is called from
__vb2_perform_fileio(), but that check is only done if no file I/O is
active. So the sequence of read() followed by write() would be allowed,
which is obviously a bug.
In addition, vb2_fop_write/read should always return -EINVAL if the
operation is not allowed, and by putting the check in the lower levels
of the code it is possible that other error codes are returned (EBUSY
or ERESTARTSYS).
All these issues are avoided by just doing a quick explicit check.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Stopping the vb2 thread (as used by several DVB devices) can result
in an 'UNBALANCED' warning such as this:
vb2: counters for queue ffff880407ee9828: UNBALANCED!
vb2: setup: 1 start_streaming: 1 stop_streaming: 1
vb2: wait_prepare: 249333 wait_finish: 249334
This is due to a race condition between stopping the thread and
calling vb2_internal_streamoff(). While I have not been able to deduce
the exact mechanism how this race condition can produce this warning,
I can see that the way the stream is stopped is likely to lead to a
race somewhere.
This patch simplifies how this is done by first ensuring that the
thread is completely stopped before cleaning up the vb2 queue. It
does that by setting threadio->stop to true, followed by a call to
vb2_queue_error() which will wake up the thread. The thread sees that
'stop' is true and it will exit.
The call to kthread_stop() waits until the thread has exited, and only
then is the queue cleaned up by calling __vb2_cleanup_fileio().
This is a much cleaner sequence and the warning has now disappeared.
Reported-by: Jurgen Kramer <gtmkramer@xs4all.nl>
Tested-by: Jurgen Kramer <gtmkramer@xs4all.nl>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: <stable@vger.kernel.org> # for v3.18 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The locking scheme inside the vb2 thread is unsafe when stopping the
thread. In particular kthread_stop was called *after* internal data
structures were cleaned up instead of doing that before. In addition,
internal vb2 functions were called after threadio->stop was set to
true and vb2_internal_streamoff was called. This is also not allowed.
All this led to a variety of race conditions and kernel warnings and/or
oopses.
Fixed by moving the kthread_stop call up before the cleanup takes
place, and by checking threadio->stop before calling internal vb2
queuing operations.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Cc: <stable@vger.kernel.org> # for v3.16 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The function releases the queue if the file being released is the queue
owner. The check reads the queue->owner field without taking the queue
lock, creating a race condition with functions that set the queue owner,
such as vb2_ioctl_reqbufs() for instance.
Fix this by moving the queue->owner check within the mutex protected
section.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The vb2_fop_poll() implementation tries to be clever on whether it needs
to lock the queue mutex by checking whether polling might start fileio.
The test requires reading the q->num_buffer field, which is racy if we
don't hold the queue mutex in the first place.
Remove the extra cleverness and just lock the mutex.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
This is needed for the next patch where the dma-sg alloc memop needs
to know the dma_dir.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Pawel Osciak <pawel@osciak.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The 'write' argument is very ambiguous. I first assumed that if it is 1,
then we're doing video output but instead it meant the reverse.
Since it is used to setup the dma_dir value anyway it is now replaced by
the correct dma_dir value which is unambiguous.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Pawel Osciak <pawel@osciak.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
The recent conversion of saa7134 to vb2 unconvered a poll() bug that
broke the teletext applications alevt and mtt. These applications
expect that calling poll() without having called VIDIOC_STREAMON will
cause poll() to return POLLERR. That did not happen in vb2.
This patch fixes that behavior. It also fixes what should happen when
poll() is called when STREAMON is called but no buffers have been
queued. In that case poll() will also return POLLERR, but only for
capture queues since output queues will always return POLLOUT
anyway in that situation.
This brings the vb2 behavior in line with the old videobuf behavior.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
It's also invalid when plane_no is equal to vb->num_planes
Signed-off-by: Zhaowei Yuan <zhaowei.yuan@samsung.com>
Cc: stable@vger.kernel.org # for v3.7 and up
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Commit bd994ddb2a (vb2: Fix stream start and
buffer completion race) broke the buffer state check in vb2_buffer_done.
So accept all three possible states there since I can no longer tell the
difference between vb2_buffer_done called from start_streaming or from
elsewhere.
Instead add a WARN_ON at the end of start_streaming that will check whether
any buffers were added to the done list, since that implies that the wrong
state was used as well.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: stable@vger.kernel.org # for v3.15 and up
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Recently WARN_ON() calls have been added to warn if the driver is not
properly returning buffers to vb2 in start_streaming (if it fails) or
stop_streaming(). Add comments before those WARN_ON calls that refer
to the videobuf2-core.h header that explains what drivers are supposed
to do in these situations. That should help point developers in the
right direction if they see these warnings.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Pawel Osciak <pawel@osciak.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
(Changes since v2: dropped local variable as suggested by Laurent)
Commit f035eb4e97 (videobuf2: fix lockdep warning)
unfortunately removed the mmap_sem lock that is needed around the call to
__qbuf_userptr. Amazingly nobody noticed this (especially me as the author)
until Jan Kara pointed this out to me.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Jan Kara <jack@suse.cz>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
If this is a multiplanar buf_type and the plane we want to read has a
non-zero data_offset, then that data_offset was not taken into account.
Note that read() or write() for formats with more than one plane is currently
not allowed, hence the use of 'planes[0]' since this is only relevant for a
single-plane format.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Modern kernels enable dynamic printk support, which is fine, except when it is
combined with a debug module option. Enabling debug in videobuf2-core now produces
no debugging unless it is also enabled through the dynamic printk support in debugfs.
Either use a debug module option + pr_info, or use pr_debug without a debug module
option. In this case the fact that you can set various debug levels is very useful,
so I believe that for videobuf2-core.c we should use pr_info.
The mix of the two is very confusing: I've spent too much time already trying to
figure out why I am not seeing any debug output in the kernel log when I do:
echo 1 >/sys/modules/videobuf2_core/parameters/debug
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The following lockdep warning has been there ever since commit a517cca6b2
one year ago:
[ 403.117947] ======================================================
[ 403.117949] [ INFO: possible circular locking dependency detected ]
[ 403.117953] 3.16.0-rc6-test-media #961 Not tainted
[ 403.117954] -------------------------------------------------------
[ 403.117956] v4l2-ctl/15377 is trying to acquire lock:
[ 403.117959] (&dev->mutex#3){+.+.+.}, at: [<ffffffffa005a6c3>] vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.117974]
[ 403.117974] but task is already holding lock:
[ 403.117976] (&mm->mmap_sem){++++++}, at: [<ffffffff8118291f>] vm_mmap_pgoff+0x6f/0xc0
[ 403.117987]
[ 403.117987] which lock already depends on the new lock.
[ 403.117987]
[ 403.117990]
[ 403.117990] the existing dependency chain (in reverse order) is:
[ 403.117992]
[ 403.117992] -> #1 (&mm->mmap_sem){++++++}:
[ 403.117997] [<ffffffff810d733c>] validate_chain.isra.39+0x5fc/0x9a0
[ 403.118006] [<ffffffff810d8bc3>] __lock_acquire+0x4d3/0xd30
[ 403.118010] [<ffffffff810d9da7>] lock_acquire+0xa7/0x160
[ 403.118014] [<ffffffff8118c9ec>] might_fault+0x7c/0xb0
[ 403.118018] [<ffffffffa0028a25>] video_usercopy+0x425/0x610 [videodev]
[ 403.118028] [<ffffffffa0028c25>] video_ioctl2+0x15/0x20 [videodev]
[ 403.118034] [<ffffffffa0022764>] v4l2_ioctl+0x184/0x1a0 [videodev]
[ 403.118040] [<ffffffff811d77d0>] do_vfs_ioctl+0x2f0/0x4f0
[ 403.118307] [<ffffffff811d7a51>] SyS_ioctl+0x81/0xa0
[ 403.118311] [<ffffffff8199dc69>] system_call_fastpath+0x16/0x1b
[ 403.118319]
[ 403.118319] -> #0 (&dev->mutex#3){+.+.+.}:
[ 403.118324] [<ffffffff810d6a96>] check_prevs_add+0x746/0x9f0
[ 403.118329] [<ffffffff810d733c>] validate_chain.isra.39+0x5fc/0x9a0
[ 403.118333] [<ffffffff810d8bc3>] __lock_acquire+0x4d3/0xd30
[ 403.118336] [<ffffffff810d9da7>] lock_acquire+0xa7/0x160
[ 403.118340] [<ffffffff81999664>] mutex_lock_interruptible_nested+0x64/0x640
[ 403.118344] [<ffffffffa005a6c3>] vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118349] [<ffffffffa0022122>] v4l2_mmap+0x62/0xa0 [videodev]
[ 403.118354] [<ffffffff81197270>] mmap_region+0x3d0/0x5d0
[ 403.118359] [<ffffffff8119778d>] do_mmap_pgoff+0x31d/0x400
[ 403.118363] [<ffffffff81182940>] vm_mmap_pgoff+0x90/0xc0
[ 403.118366] [<ffffffff81195cef>] SyS_mmap_pgoff+0x1df/0x2a0
[ 403.118369] [<ffffffff810085c2>] SyS_mmap+0x22/0x30
[ 403.118376] [<ffffffff8199dc69>] system_call_fastpath+0x16/0x1b
[ 403.118381]
[ 403.118381] other info that might help us debug this:
[ 403.118381]
[ 403.118383] Possible unsafe locking scenario:
[ 403.118383]
[ 403.118385] CPU0 CPU1
[ 403.118387] ---- ----
[ 403.118388] lock(&mm->mmap_sem);
[ 403.118391] lock(&dev->mutex#3);
[ 403.118394] lock(&mm->mmap_sem);
[ 403.118397] lock(&dev->mutex#3);
[ 403.118400]
[ 403.118400] *** DEADLOCK ***
[ 403.118400]
[ 403.118403] 1 lock held by v4l2-ctl/15377:
[ 403.118405] #0: (&mm->mmap_sem){++++++}, at: [<ffffffff8118291f>] vm_mmap_pgoff+0x6f/0xc0
[ 403.118411]
[ 403.118411] stack backtrace:
[ 403.118415] CPU: 0 PID: 15377 Comm: v4l2-ctl Not tainted 3.16.0-rc6-test-media #961
[ 403.118418] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[ 403.118420] ffffffff82a6c9d0 ffff8800af37fb00 ffffffff819916a2 ffffffff82a6c9d0
[ 403.118425] ffff8800af37fb40 ffffffff810d5715 ffff8802308e4200 0000000000000000
[ 403.118429] ffff8802308e4a48 ffff8802308e4a48 ffff8802308e4200 0000000000000001
[ 403.118433] Call Trace:
[ 403.118441] [<ffffffff819916a2>] dump_stack+0x4e/0x7a
[ 403.118445] [<ffffffff810d5715>] print_circular_bug+0x1d5/0x2a0
[ 403.118449] [<ffffffff810d6a96>] check_prevs_add+0x746/0x9f0
[ 403.118455] [<ffffffff8119c172>] ? find_vmap_area+0x42/0x70
[ 403.118459] [<ffffffff810d733c>] validate_chain.isra.39+0x5fc/0x9a0
[ 403.118463] [<ffffffff810d8bc3>] __lock_acquire+0x4d3/0xd30
[ 403.118468] [<ffffffff810d9da7>] lock_acquire+0xa7/0x160
[ 403.118472] [<ffffffffa005a6c3>] ? vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118476] [<ffffffffa005a6c3>] ? vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118480] [<ffffffff81999664>] mutex_lock_interruptible_nested+0x64/0x640
[ 403.118484] [<ffffffffa005a6c3>] ? vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118488] [<ffffffffa005a6c3>] ? vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118493] [<ffffffff810d8055>] ? mark_held_locks+0x75/0xa0
[ 403.118497] [<ffffffffa005a6c3>] vb2_fop_mmap+0x33/0x90 [videobuf2_core]
[ 403.118502] [<ffffffffa0022122>] v4l2_mmap+0x62/0xa0 [videodev]
[ 403.118506] [<ffffffff81197270>] mmap_region+0x3d0/0x5d0
[ 403.118510] [<ffffffff8119778d>] do_mmap_pgoff+0x31d/0x400
[ 403.118513] [<ffffffff81182940>] vm_mmap_pgoff+0x90/0xc0
[ 403.118517] [<ffffffff81195cef>] SyS_mmap_pgoff+0x1df/0x2a0
[ 403.118521] [<ffffffff810085c2>] SyS_mmap+0x22/0x30
[ 403.118525] [<ffffffff8199dc69>] system_call_fastpath+0x16/0x1b
The reason is that vb2_fop_mmap and vb2_fop_get_unmapped_area take the core lock
while they are called with the mmap_sem semaphore held. But elsewhere in the code
the core lock is taken first but calls to copy_to/from_user() can take the mmap_sem
semaphore as well, potentially causing a classical A-B/B-A deadlock.
However, the mmap/get_unmapped_area calls really shouldn't take the core lock
at all. So what would happen if they don't take the core lock anymore?
There are two situations that need to be taken into account: calling mmap while
new buffers are being added and calling mmap while buffers are being deleted.
The first case works almost fine without a lock: in all cases mmap relies on
correctly filled-in q->num_buffers/q->num_planes values and those are only
updated by reqbufs and create_buffers *after* any new buffers have been
initialized completely. Except in one case: if an error occurred while allocating
the buffers it will increase num_buffers and rely on __vb2_queue_free to
decrease it again. So there is a short period where the buffer information
may be wrong.
The second case definitely does pose a problem: buffers may be in the process
of being deleted, without the internal structure being updated.
In order to fix this a new mutex is added to vb2_queue that is taken when
buffers are allocated or deleted, and in vb2_mmap. That way vb2_mmap won't
get stale buffer data. Note that this is a problem only for MEMORY_MMAP, so
even though __qbuf_userptr and __qbuf_dmabuf also mess around with buffers
(mem_priv in particular), this doesn't clash with vb2_mmap or
vb2_get_unmapped_area since those are MMAP specific.
As an additional bonus the hack in __buf_prepare, the USERPTR case, can be
removed as well since mmap() no longer takes the core lock.
All in all a much cleaner solution.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
vb2_poll should always return POLLOUT | POLLWRNORM as long as there
are fewer buffers queued than there are buffers available. Poll for
an output stream should only wait if all buffers are queued and nobody
is dequeuing them.
Signed-off-by: Hans Verkuil <hansverk@cisco.com>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The original report from Nikhil was that if data_offset > 0 and bytesused == 0,
then the check in __verify_length() would fail, even though the spec says that
if bytes_used == 0, then it will be replaced by the actual length of the
buffer.
After digging into it a bit more I realized that there were several other
things wrong:
- in __verify_length() it would use the application-provided length value
for USERPTR and the vb2 core length for other memory models, but it
should have used the application-provided length as well for DMABUF.
- in __fill_vb2_buffer() on the other hand it would replace bytesused == 0
by the application-provided length, even for MMAP buffers where the
length is determined by the vb2 core.
- in __fill_vb2_buffer() it tries to figure out if all the planes have
bytesused == 0 before it will decide to replace bytesused by length.
However, the spec makes no such provision, and it makes for convoluted
code. So just replace any bytesused == 0 by the proper length.
The idea behind this was that you could use bytesused to signal empty
planes, something that is currently not supported. But that is better
done in the future by using one of the reserved fields in strucy v4l2_plane.
This patch fixes all these issues.
Regards,
Hans
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Cc: Nikhil Devshatwar <nikhil.nd@ti.com>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
When a fatal error occurs that render the device unusable, the only
options for a driver to signal the error condition to userspace is to
set the V4L2_BUF_FLAG_ERROR flag when dequeuing buffers and to return an
error from the buffer prepare handler when queuing buffers.
The buffer error flag indicates a transient error and can't be used by
applications to detect fatal errors. Returning an error from vb2_qbuf()
is thus the only real indication that a fatal error occurred. However,
this is difficult to handle for multithreaded applications that requeue
buffers from a thread other than the control thread. In particular the
poll() call in the control thread will not notify userspace of the
error.
This patch adds an explicit mechanism to report fatal errors to
userspace. Drivers can call the vb2_queue_error() function to signal a
fatal error. From this moment on, buffer preparation will return -EIO to
userspace, and vb2_poll() will set the POLLERR flag and return
immediately. The error flag is cleared when cancelling the queue, either
at stream off time (through vb2_streamoff) or when releasing the queue
with vb2_queue_release().
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
The V4L2 specification states that
"When the application did not call VIDIOC_QBUF or VIDIOC_STREAMON yet
the poll() function succeeds, but sets the POLLERR flag in the revents
field."
The vb2_poll() function sets POLLERR when the queued buffers list is
empty, regardless of whether this is caused by the stream not being
active yet, or by a transient buffer underrun.
Bring the implementation in line with the specification by returning
POLLERR if no buffer has been queued only when the queue is not
streaming. Buffer underruns during streaming are not treated specially
anymore and just result in poll() blocking until the next event.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Pawel Osciak <pawel@osciak.com>
Signed-off-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
