- 02 Feb, 2015 40 commits
-
-
Lars-Peter Clausen authored
Add media controller support to the adv7180 driver by registering a media entity instance for it as well as implementing pad ops for configuring the format. As there currently don't seem to be any users of the video ops format operations those are removed as well in this patch. Also set the V4L2_SUBDEV_FL_HAS_DEVNODE flag for the subdevice so it is possible to create a subdevice device node. Since the driver now depends on VIDEO_V4L2_SUBDEV_API all drivers which select the driver need to depend on that symbol as well. Signed-off-by:
Lars-Peter Clausen <lars@metafoo.de> Acked-by:
Federico Vaga <federico.vaga@gmail.com> Acked-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@osg.samsung.com>
-
Lars-Peter Clausen authored
Reset the device when initializing it so it is in a good known state and the assumed register settings matches the actual register settings. Signed-off-by:
-
Lars-Peter Clausen authored
The ad7180 has multiple register pages which can be switched between by writing to a register. Currently the driver manually switches between pages whenever a register outside of the default register map is accessed and switches back after it has been accessed. This is a bit tedious and also potential source for bugs. This patch adds two helper functions that take care of switching between pages and reading/writing the register. The register numbers for registers are updated to encode both the page (in the upper 8-bits) and the register (in the lower 8-bits) numbers. Having multiple pages means that a register access is not a single atomic i2c_smbus_write_byte_data() or i2c_smbus_read_byte_data() call and we need to make sure that concurrent register access does not race against each other. Signed-off-by:
-
Lars-Peter Clausen authored
Consistently prefix register defines with ADV7180_REG. Also remove the "ADI" from register names, the ADV7180 prefix should provide enough of a namespace separation. Signed-off-by:
-
Lars-Peter Clausen authored
Most IRQ controllers support different types of interrupts. The adv7180 generates falling edge interrupts, so make sure to pass IRQF_TRIGGER_FALLING to request_threaded_irq() so the IRQ controller is configured for the correct mode. Signed-off-by:
-
Lars-Peter Clausen authored
Currently the IRQ is requested from within the init_device() function. This function is not only called during device probe, but also during resume causing the driver to try to request the IRQ again. Move requesting the IRQ from init_device() to the probe function to make sure that it is only requested once. Signed-off-by:
-
Arnd Bergmann authored
The davinci media drivers use videobuf2, which they enable through a 'select' statement. If one of these drivers is built-in, but the v4l2 core is a loadable modules, we end up with a link error: drivers/built-in.o: In function `vb2_fop_mmap': :(.text+0x113e84): undefined reference to `video_devdata' drivers/built-in.o: In function `vb2_ioctl_create_bufs': :(.text+0x114710): undefined reference to `video_devdata' drivers/built-in.o: In function `vb2_ioctl_reqbufs': :(.text+0x114ed8): undefined reference to `video_devdata' drivers/built-in.o: In function `vb2_ioctl_querybuf': :(.text+0x115530): undefined reference to `video_devdata' To solve this, we need to add a dependency on VIDEO_V4L2, which enforces that the davinci drivers themselves can only be loadable modules if V4L2 is not built-in, and they do not cause the videobuf2 code to be built-in. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Acked-by:
Lad, Prabhakar <prabhakar.csengg@gmail.com> Signed-off-by:
-
Hans Verkuil authored
Keep the colorspace and encoding names consistent with what is used elsewhere (primarily the utilities in v4l-utils.git). Signed-off-by:
-
Arnd Bergmann authored
The USB and MMC front-ends to the siano driver both only make sense when combined with the SMS_SIANO_MDTV driver. That driver already requires RC_CORE to not be a module, so we also need to add that dependency here. drivers/built-in.o: In function `smssdio_remove': :(.text+0x155bd8): undefined reference to `smscore_putbuffer' :(.text+0x155bdc): undefined reference to `smscore_unregister_device' drivers/built-in.o: In function `smssdio_interrupt': :(.text+0x155e4c): undefined reference to `smsendian_handle_rx_message' :(.text+0x155e50): undefined reference to `smscore_onresponse' :(.text+0x155e54): undefined reference to `smscore_getbuffer' :(.text+0x155e58): undefined reference to `smscore_putbuffer' drivers/built-in.o: In function `smssdio_sendrequest': :(.text+0x155f20): undefined reference to `smsendian_handle_tx_message' drivers/built-in.o: In function `smssdio_probe': :(.text+0x15610c): undefined reference to `sms_get_board' :(.text+0x156114): undefined reference to `smscore_register_device' :(.text+0x156118): undefined reference to `smscore_set_board_id' :(.text+0x156128): undefined reference to `smscore_unregister_device' :(.text+0x156140): undefined reference to `smscore_start_device' Signed-off-by:
-
Arnd Bergmann authored
Large udelay values are not allowed on the ARM architecture and result in a build error like ERROR: "__bad_udelay" [drivers/media/radio/radio-aimslab.ko] undefined! This changes the aimslab radio driver to use an equivalent mdelay statement. Signed-off-by:
-
Arnd Bergmann authored
The timberdale media driver requires the use of the respective dma engine driver, but that may not be enabled, causing a Kconfig warning: warning: (VIDEO_TIMBERDALE) selects TIMB_DMA which has unmet direct dependencies (DMADEVICES && MFD_TIMBERDALE) This fixes the dependency by removing the inappropriate 'select' statement and replacing it with a direct dependency on the drivers that provide the services this needs. Fixes: 7155043c ("[media] enable COMPILE_TEST for media drivers") Signed-off-by:
-
Prabhakar Lad authored
This patch fixes following spare warnings: drivers/media/platform/am437x/am437x-vpfe.c:66:28: warning: symbol 'vpfe_standards' was not declared. Should it be static? drivers/media/platform/am437x/am437x-vpfe.c:2202:57: warning: incorrect type in argument 2 (different address spaces) drivers/media/platform/am437x/am437x-vpfe.c:2202:57: expected void [noderef] <asn:1>*params drivers/media/platform/am437x/am437x-vpfe.c:2202:57: got void *param include/linux/spinlock.h:364:9: warning: context imbalance in 'vpfe_start_streaming' - unexpected unlock Reported-by:
Fengguang Wu <fengguang.wu@intel.com> Signed-off-by:
-
Nicholas Mc Guire authored
This is only an API consolidation and should make things more readable Signed-off-by:
Nicholas Mc Guire <der.herr@hofr.at> Signed-off-by:
-
Ismael Luceno authored
Re-add Ismael Luceno as co-maintainer (with personal email address). Signed-off-by:
Ismael Luceno <ismael@iodev.co.uk> Signed-off-by:
-
Mauro Carvalho Chehab authored
changeset 5eeb3014 added a fixup at the error check code. However, it introduced a new error: drivers/media/usb/cx231xx/cx231xx-cards.c:1586 cx231xx_usb_probe() error: we previously assumed 'dev' could be null (see line 1430) This happens when dev = kmalloc() fails. So, instead of relying on it to succeed, just change the parameter of clear_bit() from 'dev->devno' to 'nr'. Cc: Alexey Khoroshilov <khoroshilov@ispras.ru> Cc: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
-
Alexey Khoroshilov authored
Commit b7085c08 ("cx231xx: convert from pr_foo to dev_foo") moves usb_get_dev(interface_to_usbdev(interface)) to the beginning of cx231xx_usb_probe() to use udev->dev in dev_err(), but it does not make sure usbdev is put on all failure paths. Later dev_err(udev->dev) was replaced by dev_err(d). So the patch moves usb_get_dev() below (before the first use) and fixes another failure path. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by:
Alexey Khoroshilov <khoroshilov@ispras.ru> Signed-off-by:
-
William Towle authored
This commit moves the "buffer in use" logic from the .buf_cleanup handler into .stop_streaming, based on advice that this is its proper logical home. By ensuring the list of pointers in priv->queue_buf[] is managed as soon as possible, we avoid warnings concerning buffers in ACTIVE state when the system cleans up after streaming stops. This fixes a problem with modification of buffers after their content has been cleared for passing to userspace. After the refactoring, the buf_init and buf_cleanup functions were found to contain only initialisation/release steps as are carried out elsewhere if omitted; these functions and references were removed. Signed-off-by:
William Towle <william.towle@codethink.co.uk> Signed-off-by:
Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by:
-
Ian Molton authored
The code that tests that capture from a stream has stopped is presently insufficient and the potential for a race condition exists where frame capture may generate an interrupt between requesting the capture process halt and freeing buffers. This patch refactors code out of rcar_vin_videobuf_release() and into rcar_vin_wait_stop_streaming(), and ensures there are calls in places where we need to know that capturing has finished. Signed-off-by:
Ian Molton <ian.molton@codethink.co.uk> Signed-off-by:
-
Josh Wu authored
Reuse the v4l2 size definitions from v4l2-image-sizes.h. So we can remove the rudundent definitions from ov2640.c. Signed-off-by:
Josh Wu <josh.wu@atmel.com> Signed-off-by:
-
Guennadi Liakhovetski authored
A hunk, removing this code has been lost between versions 6 and 7 of patch "soc-camera: add V4L2-async support." The code is harmless, but redundant. Signed-off-by:
-
Lad, Prabhakar authored
This patch drops driver specific wait_prepare() and wait_finish() callbacks from vb2_ops and instead uses the the helpers vb2_ops_wait_prepare/finish() provided by the vb2 core, the lock member of the queue needs to be initalized to a mutex so that vb2 helpers vb2_ops_wait_prepare/finish() can make use of it. Signed-off-by:
-
Philipp Zabel authored
Now that the bitstream buffer is only allocated for the BIT decoder case, we can use bitstream.size to check for bitstream ringbuffer operation. Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
Kamil Debski <k.debski@samsung.com> Signed-off-by:
-
Philipp Zabel authored
The BIT encoder does not use a per-context bitstream ringbuffer as it encodes directly into the videobuf2 capture queue's buffers. Avoid allocation of the bitstream ringbuffer for encoder contexts. Signed-off-by:
-
Philipp Zabel authored
In preparation for CODA9 JPEG support, allow contexts that control hardware units directly, without the BIT processor. Signed-off-by:
-
Philipp Zabel authored
Make sure the buffer_mutex lock is taken in coda_bit_release while coda_free_framebuffers and coda_free_context_buffers are called. Signed-off-by:
-
Philipp Zabel authored
In preparation for CODA9 JPEG support, which doesn't have to call SEQ_END on the BIT processor. Signed-off-by:
-
Philipp Zabel authored
Since we have to copy from input buffers into the bitstream ringbuffer with the CPU, there is no need for contiguous DMA buffers on the decoder input side. Signed-off-by:
-
Philipp Zabel authored
Call coda_estimate_sizeimage from both try_fmt and set_defaults to avoid this v4l2-compliance warning on the h.264 decoder video device: G_FMT: 1920x1088, 32315559, 1, 1920, 1048576, 3, 0, 0, feedcafe TRY/S_FMT: 1920x1088, 32315559, 1, 1920, 3133440, 3, 0, 0, feedcafe fail: v4l2-test-formats.cpp(948): Video Capture: S_FMT(G_FMT) != G_FMT Signed-off-by: -
Philipp Zabel authored
On i.MX6, two subsampling ping-pong buffers are used for motion estimation and deblocking They should not be counted as framebuffers, or they will be also used to store reconstructed frames, causing visible artifacts in P-frames. Signed-off-by:
-
Philipp Zabel authored
This patch queues seq_end_work and flushes the queue during stop_streaming and clears the ctx->initialized flag. This allows to start streaming again after stopping streaming without releasing the context. Signed-off-by:
-
Philipp Zabel authored
The isequence counter is never used, qsequence counts the buffers queued into the bit decoder bitstream ringbuffer. It needs to be reset in stop_streaming. Signed-off-by:
-
Markus Pargmann authored
Compare rounded up width to fit into bytesperline. Signed-off-by:
Markus Pargmann <mpa@pengutronix.de> Signed-off-by:
-
Philipp Zabel authored
To make sure a set RET_DEC_PIC_SUCCESS flag is not a leftover from a previous successful run, clear it in prepare_decode. Signed-off-by:
-
Philipp Zabel authored
Zeroing the SRAM on probe helps with debugging SRAM contents. Signed-off-by:
-
Philipp Zabel authored
Properly zero the structure on the stack before using it. Signed-off-by:
-
Philipp Zabel authored
v4l2-compliance complains about invalid colorspace settings being accepted on the output side. This patch only allows REC709 and JPEG. Signed-off-by:
-
Philipp Zabel authored
Clarify whether job_ready returns false because the context is on hold, waiting for new input buffers, whether there are not enough input buffers to fill two into the bitstream, or whether there is not enough data in the bitstream buffer for the bitstream reader hardware to read a whole frame. Signed-off-by:
-
Lucas Stach authored
If userspace doesn't properly separate the bitstream input into individual frames (which may happen for example on slightly corrupted streams) the CODA hardware may decode more frames than we expect. We already log an error in this case, but it's also necessary to adjust the sequence offset. Otherwise we spam the log with a sequence number mismatch on every frame frame after the unexpected one. Signed-off-by:
Lucas Stach <l.stach@pengutronix.de> Signed-off-by:
-
Philipp Zabel authored
Handle an empty buffer metadata list without crashing. This can happen if the decoder is fed a broken stream, or multiple compressed frames in a single queued buffer. Signed-off-by:
-
Philipp Zabel authored
Do not remove the per-context debugfs directory before the per-buffer debugfs entries contained therein. Signed-off-by:
-
