The following ioctl sequence causes fimc_dma_run() to start processing without
complete scaler and DMA initialization which causes missing interrupt and
blocking on DQBUF:
S_FMT, STREAMON, QBUF, DQBUF, STREAMOFF, STREAMON, QBUF, DQBUF.
Fix this regression caused by moving pm_runtime* calls to start/stop_streaming
callback by making sure the fimc_m2m_resume() is always invoked when expected.
Reported-by: Tomasz Stanislawski <t.stanislaws@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
It is not common for dvb_net_init to fail, but after the patch
"dvb_net_init: return -errno on error" it can fail due to running out
of memory. Handle this.
From an audit of dvb_net_init callers.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Reviewed-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
From an audit of dvb_net_init callers, now that that function
returns -errno on error.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Clean up and error out if dvb_net_init fails (for example due to
ENOMEM). This involves moving the dvb_net_init call to before
frontend_init to make cleaning up a little easier.
From an audit of dvb_net_init callers, now that dvb_net_init lets
callers know about errors.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Clean up and error out if dvb_net_init fails (for example when
running out of memory).
From an audit of dvb_net_init callers, now that dvb_net_init
has learned to return a nonzero value from time to time.
[mchehab.redhat.com: codingstyle fix: printk() should include KERN_ facility level]
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Bail out if dvb_net_init encounters an error (for example an
out-of-memory condition), now that it reports them.
[mchehab@redhat.com: CodingStyle fix: don't use "if ((ret = foo()) < 0)"]
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Avoid some repetition by adopting the usual "goto err" idiom for error
handling.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Repeating the same cleanup code in each error handling path makes life
unnecessarily difficult for reviewers, who much check each instance of
the same copy+pasted code separately. A "goto" to the end of the
function is more maintainable and conveys the intent more clearly.
While we're touching this code, also lift some assignments from "if"
conditionals for simplicity.
No functional change intended.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
videobuf_dvb_register_bus relies on dvb_net_init to set dvbnet->dvbdev
on success, but ever since commit fcc8e7d8c0 ("dvb_net: Simplify the
code if DVB NET is not defined"), ->dvbdev is left unset when
networking support is disabled. Therefore in such configurations
videobuf_dvb_register_bus always returns failure, tripping
little-tested error handling paths and preventing the device from
being initialized and used.
Now that dvb_net_init returns a nonzero value on error, we can use
that as a more reliable error indication. Do so.
Now your card be used with CONFIG_DVB_NET=n, and the kernel will pass
on a more useful error code describing what happened when
CONFIG_DVB_NET=y but dvb_net_init fails due to resource exhaustion.
Reported-by: David Fries <David@Fries.net>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
dvb_net_init unconditionally returns 0. Callers such as
videobuf_dvb_register_frontend examine dvbnet->dvbdev instead of the
return value to tell whether the operation succeeded. If it has been
set to a valid pointer, success; if it was left equal to NULL,
failure.
Alas, there is an edge case where that logic does not work as well:
when network support has been compiled out (CONFIG_DVB_NET=n), we want
dvb_net_init and related operations to behave as no-ops and always
succeed, but there is no appropriate value to which to set dvb->dvbdev
to indicate this.
Let dvb_net_init return a meaningful error code, as preparation for
adapting callers to look at that instead.
The only immediate impact of this patch should be to make the few
callers that already check for an error code from dvb_net_init behave
a little more sensibly when it fails.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Turn the pid filter off by caps option only.
This is so the full stream is passed to demuxer and not limited
by pid count.
Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The function v4l2_device_register() is called too late in vpif_probe().
This meant that vpif_obj.v4l2_dev is accessed before it is initialized
which caused a crash.
This used to work in the past, but video_register_device() is now actually
using the v4l2_dev pointer.
Note that vpif_display.c doesn't have this bug, there v4l2_device_register()
is called at the beginning of vpif_probe.
Signed-off-by: Georgios Plakaris <gplakari@cisco.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Manjunath Hadli <Manjunath.hadli@ti.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Somehow I didn't ever quite get around to implementing suspend/resume on
the MMP2 platform; this patch fixes that little oversight. A bit of core
work was necessary to do the right thing in the s/g DMA case.
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Turn the pid filter off by caps option only.
This is so the full stream is passed to demuxer and not limited
by pid count.
Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Reworked device probing to get rid of hacks to guess the maximum size of
dvb iso transfer packets. The new code also selects the first alternate
config which supports the largest possible iso transfers for dvb.
[mchehab@redhat.com: Fix a few checkpatch.pl CodingStyle compliants]
Signed-off-by: Holger Nelson <hnelson@hnelson.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Adds support for the Honestech Vidbox NW03 USB capture device.
Signed-off-by: Gareth Williams <gareth@garethwilliams.me.uk>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This adds support for the Terratec Cinergy HTC USB XS which is similar to
the Terratec H5 by adding the USB-ids to the table. According to
http://linux.terratec.de it uses the same ICs and DVB-C works for me
using the firmware of the H5.
Signed-off-by: Holger Nelson <hnelson@hnelson.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
dw2102_properties et al refer to entries in the USB-id table using
hard-coded indices, as in "&dw2102_table[6]", which means adding new
entries before the end of the list has the potential to introduce bugs
in code elsewhere in the file.
Use C99-style initializers with symbolic names for each index to avoid
this. This way, other device tables wanting to reuse the USB ids can
use expressions like "&dw2102_table[TEVII_S630]" that do not change as
the entries in the table are reordered.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The conditional after the kzalloc says that the tested expression should
never be true, but if it were, the allocated data would have to be freed.
This change just moves the allocation below the test, to avoid any
possibility of the problem.
A simplified version of the semantic match that finds the problem is as
follows: (http://coccinelle.lip6.fr)
// <smpl>
@r exists@
local idexpression x;
statement S;
identifier f1;
position p1,p2;
expression *ptr != NULL;
@@
x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
<... when != x
when != if (...) { <+...x...+> }
x->f1
...>
(
return \(0\|<+...x...+>\|ptr\);
|
return@p2 ...;
)
@script:python@
p1 << r.p1;
p2 << r.p2;
@@
print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
vpbe_dev needs to be freed before leaving the function in an error case.
A simplified version of the semantic match that finds the problem is as
follows: (http://coccinelle.lip6.fr)
// <smpl>
@r exists@
local idexpression x;
statement S;
identifier f1;
position p1,p2;
expression *ptr != NULL;
@@
x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
<... when != x
when != if (...) { <+...x...+> }
x->f1
...>
(
return \(0\|<+...x...+>\|ptr\);
|
return@p2 ...;
)
@script:python@
p1 << r.p1;
p2 << r.p2;
@@
print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There were some curly braces missing so the probe() function always
failed.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
I changed the function definitions for dt3155_queue_setup() to match the
newer API. The dt3155_start_streaming() function didn't do anything so
I just removed it.
This silences the following gcc warnings:
drivers/staging/media/dt3155v4l/dt3155v4l.c:307:2: warning: initialization from incompatible pointer type [enabled by default]
drivers/staging/media/dt3155v4l/dt3155v4l.c:307:2: warning: (near initialization for ‘q_ops.queue_setup’) [enabled by default]
drivers/staging/media/dt3155v4l/dt3155v4l.c:311:2: warning: initialization from incompatible pointer type [enabled by default]
drivers/staging/media/dt3155v4l/dt3155v4l.c:311:2: warning: (near initialization for ‘q_ops.start_streaming’) [enabled by default]
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
To properly detect frame loss the driver must keep
track of a frame_count.
Furthermore, field_count use was erroneous because
in progressive format this must be incremented twice.
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch uses channel 2 of the eMMa-PrP to convert
format provided by the sensor to YUV420.
This format is very useful since it is used by the
internal H.264 encoder.
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
As our general practice, we use stream off before we close
the video node. So that the drivers its stream off function
would be called before its remove function.
But for the case for ctrl+c, the program would be force closed.
We have no chance to call that vb2 stream off from user space,
but directly call the remove function in soc_camera.
In that common code of soc_camera:
ici->ops->remove(icd);
if (ici->ops->init_videobuf2)
vb2_queue_release(&icd->vb2_vidq);
It would first call the device remove function, then release vb2,
in which stream off function is called. Thus it create different
order for the driver.
This patch change the order to make driver see the same sequence
to make it happy.
Signed-off-by: Lei Wen <leiwen@marvell.com>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
mt9m111 camera sensors support cropping and scaling. The current
implementation is broken. For example, .s_crop() sets output frame sizes
instead of the input cropping window. This patch adds a proper implementation
of these methods. Besides it adds a sensor-disable and -enable operations
on first open() and last close() respectively, to save power while closed and
to return the camera to the default power-on state.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Cleanly separating register contexts A and B will allow us to configure
the contexts independently.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The "pixfmt" parameter of the struct soc_camera_host_ops::set_bus_param()
method is redundant, because at the time, when this method is called,
pixfmt is guaranteed to be equal to icd->current_fmt->host_fmt->fourcc.
Remove this parameter and update all drivers accordingly.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Before this patch the resulting values from a try_fmt were different then
those from a s_fmt with the same parameters. try_fmt simply did not
touch / fill some values like bytesperline at all.
This patch also corrects bytesperline to the proper value for a planar
format such as the YUV420P format the pwc driver produces, which is
the bytesperline value for the biggest plane, rather then those
of all planes added together.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Instead of making this a module parameter, automatically fallback to
higher compression settings if there is not enough bandwidth.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The pwc driver claims to support any resolution between 160x120
and 640x480, but emulates this by simply drawing a black border
around the image. Userspace can draw its own black border if it
really wants one.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This stems from the v4l1 era, with v4l2 everything can be done with
standardized v4l2 API calls.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The in kernel version of the pwc driver has never supported snapshot
mode, and now that we no longer support the pixfmt.priv abuse there also
no longer is a way for userspace to request it, rendering all the code in
question dead (never called), so remove it.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The .priv field never was intended for this, setting a framerate is
support using the standardized S_PARM ioctl.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Setting pan/tilt should be done with v4l2 controls, like with other
cams. The button is available as a standard input device
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
So that events get generated for the new red + blue bal values when switching
to a preset. This allows apps to solely rely on events instead of needing
to do a query + g_ctrl on all controls when a control with the update flag
set is gets set or gets an event.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
While testing gtk-v4l's new ctrl event code, I hit the following deadlock
in the pwc driver:
Thread 1:
-Does a VIDIOC_G_CTRL
-video2_ioctl takes the modlock
-video2_ioctl calls v4l2_g_ctrl
-v4l2_g_ctrl takes the ctrl_handler lock
-v4l2_g_ctrl calls pwc_g_volatile_ctrl
-pwc_g_volatile_ctrl releases the modlock as the usb transfer can take a
significant amount of time and we don't want to block DQBUF / QBUF too long
Thread 2:
-Does a VIDIOC_FOO_CTRL
-video2_ioctl takes the modlock
-video2_ioctl calls v4l2_foo_ctrl
-v4l2_foo_ctrl blocks while trying to take the ctrl_handler lock
Thread 1:
-Blocks while trying to re-take the modlock, as its caller will eventually
unlock that
Now we have thread 1 waiting for the modlock while holding the ctrl_handler
lock and thread 2 waiting for the ctrl_handler lock while holding the
modlock -> deadlock.
Conclusion:
1) We cannot unlock modlock from pwc_s_ctrl / pwc_g_volatile_ctrl,
but this can cause QBUF / DQBUF to block for up to a full second
2) After evaluating various option I came to the conclusion that pwc should
stop using the v4l2 core locking, and instead do its own locking
Thus this patch stops pwc using the v4l2 core locking, and replaces that with
it doing its own locking where necessary.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Currently auto white balance speed and delay are only available through custom
ioctls, which are deprecated and will be going away in 3.3 .
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This is a preperation patch for adding support for control events. Actually
enabling support for control events will be done in a separate patch, as that
depends on the necessary poll changes going upstream
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Somehow the code has ended up assuming 1400 packets/sec which of course
is wrong for usb1 devices like the ov511 cameras. usb1 only does 1000
(isoc) packets / sec.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This got broken by some gscpa core fixes, this patch restores the proper
min/max values for these controls.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Written by Theodore Kilgore
With minor changes by Hans de Goede:
-Code style fixes
-Correct the verbose level on various PDEBUG messages
-Make error messages use pr_err instead of PDEBUG
-Document the jl20 pixel format
Signed-off-by: Theodore Kilgore <kilgota@auburn.edu>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
* tag 'v3.2': (83 commits)
Linux 3.2
minixfs: misplaced checks lead to dentry leak
ptrace: ensure JOBCTL_STOP_SIGMASK is not zero after detach
ptrace: partially fix the do_wait(WEXITED) vs EXIT_DEAD->EXIT_ZOMBIE race
Revert "rtc: Expire alarms after the time is set."
[CIFS] default ntlmv2 for cifs mount delayed to 3.3
cifs: fix bad buffer length check in coalesce_t2
Revert "rtc: Disable the alarm in the hardware"
hung_task: fix false positive during vfork
security: Fix security_old_inode_init_security() when CONFIG_SECURITY is not set
fix CAN MAINTAINERS SCM tree type
mwifiex: fix crash during simultaneous scan and connect
b43: fix regression in PIO case
ath9k: Fix kernel panic in AR2427 in AP mode
CAN MAINTAINERS update
net: fsl: fec: fix build for mx23-only kernel
sch_qfq: fix overflow in qfq_update_start()
drm/radeon/kms/atom: fix possible segfault in pm setup
gspca: Fix falling back to lower isoc alt settings
futex: Fix uninterruptible loop due to gate_area
...
This patch implements necessary changes for enabling dm365 and
dm355 hardware for vpbe. The patch contains additional HD mode
support for dm365 (720p60, 1080i30) and appropriate register
modifications based on version numbers.
VPBE_VERSION_2 = dm365 specific
VPBE_VERSION_3 = dm355 specific
Signed-off-by: Manjunath Hadli <manjunath.hadli@ti.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add OSD block changes to enable dm365 and dm355 for vpbe driver.
Changes are based on version number of OSD, which have incremental
changes over 644x OSD hardware interms of few registers.
VPBE_VERSION_2 = dm365 specific
VPBE_VERSION_3 = dm355 specific
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Manjunath Hadli <manjunath.hadli@ti.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
