John Sheu [Wed, 6 Feb 2013 23:03:01 +0000 (20:03 -0300)]
[media] v4l2-mem2mem: drop rdy_queue on STREAMOFF
When a v4l2-mem2mem context gets a STREAMOFF call on either its CAPTURE
or OUTPUT queues, we should:
* Drop the corresponding rdy_queue, since a subsequent STREAMON expects
an empty queue.
* Deschedule the context, as it now has at least one empty queue and
cannot run.
Sakari Ailus [Tue, 22 Jan 2013 15:27:56 +0000 (12:27 -0300)]
[media] media: implement 32-on-64 bit compat IOCTL handling
Use the same handlers where the structs are the same. Implement a new
handler for link enumeration since struct media_links_enum is different on
32-bit and 64-bit systems.
I'm pasting the original code and my proposal on the commit message for
make it easy to compare the two versions.
Line 62 of cx25821-audio-upstream.h contains:
char *_defaultAudioName = "/root/audioGOOD.wav";
Original code after replace kmemdup for kstrdup, and after fix return error
code:
if (dev->input_audiofilename) {
dev->_audiofilename = kstrdup(dev->input_audiofilename,
GFP_KERNEL);
if (!dev->_audiofilename) {
err = -ENOMEM;
goto error;
}
/* Default if filename is empty string */
if (strcmp(dev->input_audiofilename, "") == 0)
dev->_audiofilename = "/root/audioGOOD.wav";
} else {
dev->_audiofilename = kstrdup(_defaultAudioName,
GFP_KERNEL);
if (!dev->_audiofilename) {
err = -ENOMEM;
goto error;
}
}
Code proposed in this patch:
if ((dev->input_audiofilename) &&
(strcmp(dev->input_audiofilename, "") != 0))
dev->_audiofilename = kstrdup(dev->input_audiofilename,
GFP_KERNEL);
else
dev->_audiofilename = kstrdup(_defaultAudioName,
GFP_KERNEL);
if (!dev->_audiofilename) {
err = -ENOMEM;
goto error;
}
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] dvb_frontend: Simplify the emulation logic
The current logic was broken and too complex; while it works
fine for DVB-S2/DVB-S, it is broken for ISDB-T.
Make the logic simpler, fixes it for ISDB-T and make it clearer.
This function is complex, and has different workflows, one for
DVBv3 calls, and another one for DVBv5 calls. Break it into 3
functions, in order to make easier to understand what each
block does.
No functional changes so far. A few comments got improved.
Frank Schaefer [Sun, 10 Mar 2013 10:25:25 +0000 (07:25 -0300)]
[media] em28xx-i2c: relax error check in em28xx_i2c_recv_bytes()
It turned out that some devices return less bytes then requested via i2c when
ALL of the following 3 conditions are met:
- i2c bus B is used
- there was no attempt to write to the specified slave address before
- no device present at the specified slave address
With the current code, this triggers an -EIO error and prints a message to the
system log.
Because it can happen very often during device probing, it is better to ignore
this error and bail out silently after the follwing i2c transaction success
check with -ENODEV.
[mchehab@redhat.com: a small CodingStyle fix] Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Frank Schaefer [Sat, 9 Mar 2013 09:53:01 +0000 (06:53 -0300)]
[media] em28xx: set the timestamp type for video and vbi vb2_queues
The em28xx driver obtains the timestamps using function v4l2_get_timestamp(),
which produces a montonic timestamp.
Fixes the warnings appearing in the system log since commit 6aa69f99
"[media] vb2: Add support for non monotonic timestamps"
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com> Acked-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] em28xx: add support for registering multiple i2c buses
Register both buses 0 and 1 via I2C API. For now, bus 0 is used
only by eeprom on all known devices. Later patches will be needed
if this changes in the future.
[media] em28xx: Add a separate config dir for secondary bus
Prepare to register a separate bus for the second bus.
For now, just add a new field. A latter patch will add the
bits to make it work.
This patch was generated by this script:
perl -e 'while (<>) { if (s/EM2874_I2C_SECONDARY_BUS_SELECT.*\n//) {
printf "\t\t.def_i2c_bus = 1,\n"; $found = 1; print $_ } else { if ($found) { s/^\s+// }; $found = 0; print $_; } }' \
drivers/media/usb/em28xx/em28xx-cards.c >a && mv a drivers/media/usb/em28xx/em28xx-cards.c
[media] em28xx: Prepare to support 2 different I2C buses
Newer em28xx devices have 2 buses. Change the logic to allow
using both buses.
This patch was generated by this small script:
for i in drivers/media/usb/em28xx/*.c; do
sed 's,->i2c_adap,->i2c_adap[dev->def_i2c_bus],g;s,->i2c_client,->i2c_client[dev->def_i2c_bus],'
done
Of course, em28xx.h needed manual edit.
[media] media/v4l2: VIDEOBUF2_DMA_CONTIG should depend on HAS_DMA
m68k/sun3:
drivers/media/v4l2-core/videobuf2-dma-contig.c: In function ‘vb2_dc_mmap’:
drivers/media/v4l2-core/videobuf2-dma-contig.c:204: error: implicit declaration of function ‘dma_mmap_coherent’
drivers/media/v4l2-core/videobuf2-dma-contig.c: In function ‘vb2_dc_get_base_sgt’:
drivers/media/v4l2-core/videobuf2-dma-contig.c:387: error: implicit declaration of function ‘dma_get_sgtable’
Make VIDEOBUF2_DMA_CONTIG and VIDEO_SH_VEU (which selects the former and
doesn't have a platform dependency) depend on HAS_DMA to fix this.
Paul Bolle [Mon, 4 Mar 2013 12:43:20 +0000 (09:43 -0300)]
[media] m920x: let GCC see 'ret' is used initialized
Since commit 7543f344e9b06afe86b55a2620f5c11b38bd5642 ("[media] m920x:
factor out a m920x_write_seq() function") building m920x.o triggers this
GCC warning:
drivers/media/usb/dvb-usb/m920x.c: In function ‘m920x_probe’:
drivers/media/usb/dvb-usb/m920x.c:91:6: warning: ‘ret’ may be used uninitialized in this function [-Wuninitialized]
This warning is caused by m920x_write_seq(), which is apparently inlined
into m920x_probe(). It is clear why GCC thinks 'ret' may be used
uninitialized. But in practice the first seq->address will always be
non-zero when this function is called. That means we can change the
while()-do{} loop into a do{}-while() loop. And that suffices to make
GCC see that 'ret' will not be used uninitialized.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Apparently a copy-paste mistake; the similar sh_vou.h exists, and both
were added to MAINTAINERS by commit b618b69 ([media] MAINTAINERS: add
entries for sh_veu and sh_vou V4L2 drivers).
Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Matt Gomboc [Fri, 1 Mar 2013 22:53:30 +0000 (19:53 -0300)]
[media] cx231xx : Add support for OTG102 aka EZGrabber2
Thanks for the response, I have done as you suggested.
Below is an updated patch for the OTG102 device against http://git.linuxtv.org/hverkuil/media_tree.git/shortlog/refs/heads/cx231xx, kernel version 3.8.
With further testing it appears the extra clauses in cx231xx-cards.c were not necessary (in static in cx231xx_init_dev and static int cx231xx_usb_probe), so those have been also been removed.
Signed-off-by: Matt Gomboc <gomboc0@gmail.com> Acked-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Chen Gang [Thu, 28 Feb 2013 06:08:02 +0000 (03:08 -0300)]
[media] drivers/staging/media/as102: using ccflags-y instead of EXTRA_FLAGS in Makefile
need using ccflags-y instead of EXTRA_CFLAGS
can reference scripts/checkpatch.pl (1755..1766)
when make EXTRA_CFLAGS=-W, the compiling issue will be occured.
Signed-off-by: Chen Gang <gang.chen@asianux.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Even if bus is not hot-pluggable, the devices can be unbound from the
driver via sysfs, so we should not be using __exit annotations on
remove() methods. The only exception is drivers registered with
platform_driver_probe() which specifically disables sysfs bind/unbind
attributes.
Similarly probe() methods should not be marked __init unless
platform_driver_probe() is used.
Terratec Cinergy S2 USB HD Rev.2 support.
This commit is a corrected cherry-pick of 03228792 which got reverted in b7e38636 because it was rebased incorrectly and introduced compilation
errors.
Signed-off-by: Stephan Hilb <stephan@ecshi.net> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] stv090x: do not unlock unheld mutex in stv090x_sleep()
goto err and goto err_gateoff before mutex_lock(&state->internal->demod_lock)
lead to unlock of unheld mutex in stv090x_sleep().
Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru> Cc: Manu Abraham <abraham.manu@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] cx231xx: fix undefined function cx231xx_g_chip_ident()
This patch:
http://git.linuxtv.org/media_tree.git/commit/b86d15440b683f8634c0cb26fc0861a5bc4913ac
is missing a chunk when compared to an older version:
https://patchwork.kernel.org/patch/2063281/
probably because of an unresolved merging conflict.
This causes the following error:
WARNING: "cx231xx_g_chip_ident" [/home/jena/media_build/v4l/cx231xx.ko] undefined!
Hans Verkuil [Fri, 15 Feb 2013 09:07:13 +0000 (06:07 -0300)]
[media] s2255: don't zero struct v4l2_streamparm
All fields after 'type' are already zeroed by the core framework.
Clearing the full struct also clears 'type', which causes a wrong
type value to be returned.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Fri, 15 Feb 2013 09:12:58 +0000 (06:12 -0300)]
[media] s2255: fixes in the way standards are handled
Instead of comparing against STD_NTSC and STD_PAL compare against 60 and
50 Hz formats. That's what you really want.
When the standard is changed, make sure the width and height of the format
are also updated to reflect the current standard.
Also replace the deprecated current_norm by the g_std ioctl.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Instead use v4l2_dbg and v4l2_err. Note that the PDEBUG macro is kept to
make this patch-set less invasive, but it is simply a wrapper around
v4l2_dbg now. Most of the other changes are there to make the dev parameter
for the v4l2_xxx macros available everywhere we do logging.
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>
Hans Verkuil [Mon, 11 Feb 2013 09:31:55 +0000 (06:31 -0300)]
[media] gspca_sonixb: Remove querymenu function (dead code)
We forgot to remove that when sonixb was converted to the control framework.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Mon, 11 Feb 2013 09:31:12 +0000 (06:31 -0300)]
[media] gspca_sonixj: Convert to the control framework
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[media] s3c-camif: Fail on insufficient number of allocated buffers
Ensure the driver gets always at least its minimum required
number of buffers allocated by checking actual number of
allocated buffers in vb2_reqbufs(). And free any partially
allocated buffer queue with signaling an error to user space.
Without this patch applications may wait forever to dequeue
a filled buffer, because the hardware didn't even start after
VIDIOC_STREAMON, VIDIOC_QBUF calls, due to insufficient number
of empty buffers.
Kamil Debski [Fri, 25 Jan 2013 09:29:57 +0000 (06:29 -0300)]
[media] vb2: Add support for non monotonic timestamps
Not all drivers use monotonic timestamps. This patch adds a way to set the
timestamp type per every queue.
In addition, set proper timestamp type in drivers that I am sure that use
either MONOTONIC or COPY timestamps. Other drivers will correctly report
UNKNOWN timestamp type instead of assuming that all drivers use monotonic
timestamps.
Hans Verkuil [Sun, 10 Feb 2013 17:37:58 +0000 (14:37 -0300)]
[media] stk-webcam: implement support for count == 0 when calling REQBUFS
The spec specifies that setting count to 0 in v4l2_requestbuffers
should result in releasing any streaming resources and the stream
ownership. Implement this.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Tested-by: Arvydas Sidorenko <asido4@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Mon, 4 Feb 2013 12:18:14 +0000 (09:18 -0300)]
[media] stk-webcam: zero the priv field of v4l2_pix_format
The priv field should be set to 0. In this case the driver abused the priv
field for internal housekeeping. Modify the code so priv is no longer used
for that purpose.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Tested-by: Arvydas Sidorenko <asido4@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Sun, 10 Feb 2013 17:32:52 +0000 (14:32 -0300)]
[media] stk-webcam: fix querycap and simplify s_input
Add device_caps support to querycap, fill in bus_info correctly and
do not set the version field (let the core handle that).
Also simplify the s_input ioctl.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Tested-by: Arvydas Sidorenko <asido4@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Tue, 12 Feb 2013 08:25:38 +0000 (05:25 -0300)]
[media] stk-webcam: add ASUS F3JC to upside-down list
And add an extensive comment relating the history of the upside-down
handling in this driver.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Tested-by: Arvydas Sidorenko <asido4@gmail.com> Thanks-to: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Thu, 7 Feb 2013 16:22:06 +0000 (13:22 -0300)]
[media] cx231xx: disable 417 support from the Conexant video grabber
The 417 support doesn't work. Until someone can dig into this driver to
figure out why it isn't working the 417 support is disabled.
Sometimes you can actually stream a bit, but very soon the whole machine
crashes, so something is seriously wrong.
For the record, this was not introduced by my recent changes to this driver,
it was broken before that.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Colorspace, field and priv were not set, and sizeimage was calculated
using the wrong values (dev->ts1.ts_packet_size and dev->ts1.ts_packet_count
can be 0 at module load).
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Tue, 29 Jan 2013 13:50:54 +0000 (10:50 -0300)]
[media] cx231xx: improve std handling
Set the initial standard of subdevices instead of leaving it undefined.
Also update the width and height when a new standard is chosen and
return -EBUSY when attempting to change the standard while videobuf is
busy.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Hans Verkuil [Mon, 17 Sep 2012 11:30:07 +0000 (08:30 -0300)]
[media] cx231xx: remove current_norm usage
The use of this field is deprecated since it will not work when multiple
device nodes reference the same video input (the video and vbi nodes in
this case). The norm field should be a device-global value.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
projects / linux-beck.git / log