OmniVision sensors are used as well in Empiatech based cameras such as the
"SpeedLink Vicious And Devine Laplace" webcam (EM2765 + Omnivision OV2640).
With this patch applied, OminiVision sensors with 8 bit address and register
width are detected (recent models have a 16 bit address width and use different
client addresses).
The most commonly used sensors (including the ones listed by Empiatech) are
detected properly, although there is no support for them yet.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Other sensors like the ones from OmniVision need a different probing procedure,
so it makes sense have separate functions for each manufacturer/sensor type.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add further Micron chip IDs to be able to identify all Micron sensors listed
by Empiatech.
Also probe the two alternate i2c addresses used by Micron sensors with 8 bit
address and 16 bit register width.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
em28xx-cards.c is very large and the sensor/camera related code is growing,
so move this code to a separate source code file em28xx-camera.c.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Now that the board hints and the sensor initialization/configuration have been
separated, em28xx_detect_sensor() is the better name for this function.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Sensor detection and initialization/configuration are currently mixed together.
This works as long as all devices with a particular sensor are working with the
same board configuration. In the long run, this will be not sufficient, so
separate these both steps to make the code more flexible and future proof.
This also makes the code more consistent, because the initialization of the
MT9V011 sensor subdevice is already separated.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The current board hint code is mixed together with the sensor detection and
initialization code. It actually selects a board depending on the detected
sensor type only, with the result that 3 of the 6 webcam boards are currently
dead.
Separate it and move it to em28xx_hint_board() which already contains the board
hints for analog capturing+TV and DVB devices.
This way, we have all board hints at a common place which makes it easier
to extend the code and reduces the risk of regressions.
It also makes it possible again to use the boards EM2750_BOARD_DLCW_130,
EM2820_BOARD_VIDEOLOGY_20K14XUSB and EM2860_BOARD_NETGMBH_CAM (using the module
parameter "card").
NOTE: the current board hint logic for webcams is preserved. Not more not less.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Some devices without DVB support (such as the "Terratec Grabby" and
"Easycap DC-60") provide isochronous DVB USB endpoints with wMaxPacketSize set
to 0 bytes for all alt settings.
Ignore these endpoints and avoid registering a DVB device node and loading the
DVB driver extension.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Tested-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
-ERANGE is the right error code when the value is outside of the menu
range, but -EINVAL must be reported for invalid values inside the range.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The Windows driver writes the output resolution to registers 0x34 (width / 16)
and 0x35 (height / 16) always.
We don't know yet what these registers are used for.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Streaming doesn't work with the EM2765 if bit 5 of the output format register
0x27 is set.
It's actually not clear if really has to be set for the other chips, but for
now let's keep it to avoid regressions and add a comment to the code.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The em25xx/em276x/em277x/em278x frame data format is different to the one used
by the em2710/em2750/em28xx chips.
With the recent cleanups and reorganization of the frame data processing code it
can be easily extended to support these devices.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This chip can be found in the SpeedLink VAD Laplace webcam (1ae7:9003 and 1ae7:9004).
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The webcam "SpeedLink VAD Laplace" (em2765 + ov2640) uses a special algorithm
for i2c communication with the sensor, which is connected to a second i2c bus.
We don't know yet how to find out which devices support/use it.
It's very likely used by all em25xx and em276x+ bridges.
Tests with other em28xx chips (em2820, em2882/em2883) show, that this
algorithm always succeeds there although no slave device is connected.
The algorithm likely also works for real i2c client devices (OV2640 uses SCCB),
because the Windows driver seems to use it for probing Samsung and Kodak
sensors.
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add the missing unlock before return from function af9035_ctrl_msg()
in the error handling case.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Antti Palosaari <crope@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
- open/close/read and poll need to take the core lock as well.
- when the tuner goes to sleep we should set std_set_in_tuner_core
to 0 since the tuner loses the firmware at that time.
- initialize the tuner if std_set_in_tuner_core == 0 whenever:
1) g/s_tuner, s_std or s_frequency is called
2) read or poll is called
3) streamon is called
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
After loading the module the initial video routing is not setup.
Explicitly call s_input to get this right.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Just opening a device shouldn't have any side-effects.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Turn it into a simple function.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The driver crashed when the device was disconnected while an application
still had a device node open. Fixed by using the release() callback of struct
v4l2_device.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Also get rid of unnecessary format type check.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
- V4L2_CAP_AUDIO was set, but enumaudio was not implemented.
- audioset was never filled by enum_input
- ctrl_ainput was never updated when switching the video input
- g_audio was broken due to faulty logic: g_audio should set the
index, it doesn't receive it from the user.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
- define an initial frequency
- return an error if g_frequency is called for an invalid tuner index
- get the clamped frequency value after setting it: i.e. the tuner driver
may clamp the given frequency to a valid frequency range and ctrl_freq
should get that actual clamped frequency.
- remove obsolete tuner type checks (done by the core).
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add support for vidioc_g_chip_name, allowing AC97 to be implemented as a
second chip on the bridge with the name "ac97". v4l2-dbg can just match the
name with that string in order to detect a ac97-compliant set of registers.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
It was just a placeholder and we want to get rid of the AC97 matching
define.
Also replace MATCH_HOST with MATCH_BRIDGE since we are here anyway.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This ioctl is defined as IOW, so pass the argument as const.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This ioctl is defined as IOW, so pass the argument by value instead of by
reference. I could have chosen to add const instead, but this is 1) easier
to handle in drivers and 2) consistent with the s_std subdev operation.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Jonathan Corbet <corbet@lwn.net>
Acked-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This ioctl is defined as IOW, so pass the argument as const.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Alexey Klimov <klimov.linux@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This ioctl is defined as IOW, so pass the argument as const.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch enable dual tuner for Avermedia Twinstar.
Signed-off-by: Jose Alberto Reguero <jareguero@telefonica.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
usb_free_coherent() is safe with NULL addr and this check is
not required.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
I just made commit: "dvb_usb_v2: rework USB streaming logic" that
breaks that driver PID filter.
it913x driver checks use of PID filter directly from DVB USB v2 core
internal variable "adap->pid_filtering" and stores it to own state.
Calling order of .pid_filter_ctrl() and .pid_filter() was changed
and due to that state was updated too late. Update state earlier.
TODO: checking PID filter usage from DVB USB v2 is not very good idea
as PID filter callbacks are called only when PID filter is enabled.
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Control flow order changed a little bit. HW PID filter is now
disabled also when streaming is stopped - earlier it was just
set only when streaming was started.
Control flow is now:
* set streaming status bit
* submit USB streaming packets
* enable HW PID filter
* ask device to start streaming
* N x add PID to device HW PID filter
... streaming video ...
* N x remove PID from device HW PID filter
* ask device to stop streaming
* disable HW PID filter
* kill USB streaming packets
* clear streaming status bit
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
New checkpatch version likes to see strings not to split multiple
lines even those are exceeding 80 line length.
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
New checkpatch version likes to see strings not to split multiple
lines even those are exceeding 80 line length.
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There was sync mutex which was held over userspace. That is very
wrong and could cause deadlock if different userspace process is
used to "unlock". Wait queue seems to be correct solution for
that kind of synchronizing issue so use it instead.
lock debug gives following bug report:
================================================
[ BUG: lock held when returning to user space! ]
3.9.0-rc1+ #38 Tainted: G O
------------------------------------------------
tzap/4614 is leaving the kernel with locks still held!
1 lock held by tzap/4614:
Signed-off-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
As reported by Geert Uytterhoeven <geert@linux-m68k.org>:
drivers/media/usb/hdpvr/hdpvr-video.c: warning: "CONFIG_I2C_MODULE" is not defined [-Wundef]
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Thanks-to: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
dvb_usb_v2_generic_io() was not declared. It should be static.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Commit aab3125c43 broke em28xx. I traced
this eventually to the change in what algo_data points to. This pointer
is also passed to em28xx_tuner_callback() through several hidden tuner
layers (yuck!) and that callback was not updated.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There's currently a bug on em28xx-i2c that makes it write the
wrong values to register 06, that controlls the I2C bus speed
and bus.
Fix it to change only the I2C bus flag.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
It used to be a code that returns arror at ov7660_init.
However, this was removed by changeset c84e412f:
@@ -231,33 +116,40 @@ int ov7660_init(struct sd *sd)
if (dump_sensor)
ov7660_dump_registers(sd);
- err = ov7660_set_gain(&sd->gspca_dev, sensor_settings[GAIN_IDX]);
- if (err < 0)
- return err;
+ return 0;
+}
- err = ov7660_set_auto_white_balance(&sd->gspca_dev,
- sensor_settings[AUTO_WHITE_BALANCE_IDX]);
- if (err < 0)
- return err;
As complained by gcc:
drivers/media/usb/gspca/m5602/m5602_ov7660.c: In function 'ov7660_init':
drivers/media/usb/gspca/m5602/m5602_ov7660.c:99:9: warning: variable 'err' set but not used [-Wunused-but-set-variable]
It should be noticed that the original error code was crappy, as it wasn't
returning any error if sensor init fails.
Fix it by returning an error if the sensor can't be initialized.
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
