- 27 Nov, 2018 2 commits
-
-
Malathi Gottam authored
Max bitrate property is not supported for venus version 4xx. Return unsupported from packetization layer. Handle it in hfi_venus layer to exit gracefully to venc layer. Signed-off-by:
Malathi Gottam <mgottam@codeaurora.org> Tested-by:
Alexandre Courbot <acourbot@chromium.org> Acked-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org> Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
-
Malathi Gottam authored
Add USERPTR to queue access methods by adding this support to io_modes on both the planes. Signed-off-by:
-
- 23 Nov, 2018 38 commits
-
-
Mauro Carvalho Chehab authored
The dvb_frontend core already checks for the frequencies. No need for any additional check inside the driver. It is part of the fixes for the following bug: https://bugzilla.opensuse.org/show_bug.cgi?id=1116374 Fixes: a3f90c75 ("media: dvb: convert tuner_info frequencies to Hz") Reported-by:
Stakanov Schufter <stakanov@eclipso.eu> Reported-by:
Takashi Iwai <tiwai@suse.de> Cc: stable@vger.kernel.org # For 4.19 Signed-off-by:
-
Mauro Carvalho Chehab authored
Tuners should report frequencies in Hz. That works fine on most drivers, but, in the case of dvb-pll, some settings are for satellite tuners, while others are for terrestrial/cable ones. The code was trying to solve it at probing time, but that doesn't work, as, when _attach is called, the delivery system may be wrong. Fix it by ensuring that all frequencies are in Hz at the per-tuner max/min values. While here, add a debug message, as this would help to debug any issues there. It partially fixes the following bug: https://bugzilla.opensuse.org/show_bug.cgi?id=1116374 Fixes: a3f90c75 ("media: dvb: convert tuner_info frequencies to Hz") Reported-by:
-
Mauro Carvalho Chehab authored
As we did an internal change inside the subsystem to always represent min/max frequencies in Hz, add a debug message, as this would help to discover bugs on drivers, if any. Signed-off-by: -
Mauro Carvalho Chehab authored
The dvb_frontend dprintk() macro already prints __func__. So, we don't need to add it again at the printed message. Signed-off-by: -
Mauro Carvalho Chehab authored
As reported by smatch: drivers/media/common/videobuf2/videobuf2-core.c: drivers/media/common/videobuf2/videobuf2-core.c:2159 vb2_mmap() warn: inconsistent returns 'mutex:&q->mmap_lock'. Locked on: line 2148 Unlocked on: line 2100 line 2108 line 2113 line 2118 line 2156 line 2159 There is one error condition that doesn't unlock a mutex. Fixes: cd26d1c4 ("media: vb2: vb2_mmap: move lock up") Reviewed-by: -
Mauro Carvalho Chehab authored
As warned by smatch: drivers/media/platform/seco-cec/seco-cec.c:338:21: warning: symbol 'secocec_cec_adap_ops' was not declared. Should it be static? This struct should be static. Also, it is const, so declare it as such. Signed-off-by: -
Hans Verkuil authored
Fix the incorrect obj-y. Signed-off-by:
-
Hans Verkuil authored
Userspace shouldn't set bytesused to 0 for output buffers. vb2_warn_zero_bytesused() warns about this (only once!), but it also calls WARN_ON(1), which is confusing since it is not immediately clear that it warns about a 0 value for bytesused. Just drop the WARN_ON as it serves no purpose. Signed-off-by:
Ezequiel Garcia <ezequiel@collabora.com> Signed-off-by:
-
John Sheu authored
Videobuf2 presently does not allow VIDIOC_REQBUFS to destroy outstanding buffers if the queue is of type V4L2_MEMORY_MMAP, and if the buffers are considered "in use". This is different behavior than for other memory types and prevents us from deallocating buffers in following two cases: 1) There are outstanding mmap()ed views on the buffer. However even if we put the buffer in reqbufs(0), there will be remaining references, due to vma .open/close() adjusting vb2 buffer refcount appropriately. This means that the buffer will be in fact freed only when the last mmap()ed view is unmapped. 2) Buffer has been exported as a DMABUF. Refcount of the vb2 buffer is managed properly by VB2 DMABUF ops, i.e. incremented on DMABUF get and decremented on DMABUF release. This means that the buffer will be alive until all importers release it. Considering both cases above, there does not seem to be any need to prevent reqbufs(0) operation, because buffer lifetime is already properly managed by both mmap() and DMABUF code paths. Let's remove it and allow userspace freeing the queue (and potentially allocating a new one) even though old buffers might be still in processing. To let userspace know that the kernel now supports orphaning buffers that are still in use, add a new V4L2_BUF_CAP_SUPPORTS_ORPHANED_BUFS to be set by reqbufs and create_bufs. [p.zabel@pengutronix.de: added V4L2_BUF_CAP_SUPPORTS_ORPHANED_BUFS, updated documentation, and added back debug message] Signed-off-by:
John Sheu <sheu@chromium.org> Reviewed-by:
Pawel Osciak <posciak@chromium.org> Signed-off-by:
Tomasz Figa <tfiga@chromium.org> Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Acked-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
-
Ezequiel Garcia authored
Now that the mem2mem framework guarantees that .device_run won't be called from interrupt context, it is safe to call v4l2_m2m_job_finish directly in the top-half. So this means the bottom-half is no longer needed and we can get rid of it. Signed-off-by:
Paul Kocialkowski <paul.kocialkowski@bootlin.com> Signed-off-by:
-
Ezequiel Garcia authored
v4l2_m2m_job_finish() is typically called when DMA operations complete, in interrupt handlers or DMA completion callbacks. Calling .device_run from v4l2_m2m_job_finish creates a nasty re-entrancy path into the driver. Moreover, some implementation of .device_run might need to sleep, as is the case for drivers supporting the Request API, where controls are applied via v4l2_ctrl_request_setup, which takes the ctrl handler mutex. This commit adds a deferred context that calls v4l2_m2m_try_run, and gets scheduled by v4l2_m2m_job_finish(). Before this change, device_run would be called from these paths: vb2_m2m_request_queue, or v4l2_m2m_streamon, or v4l2_m2m_qbuf v4l2_m2m_try_schedule v4l2_m2m_try_run .device_run v4l2_m2m_job_finish v4l2_m2m_try_run .device_run After this change, the latter is now gone and instead: v4l2_m2m_device_run_work v4l2_m2m_try_run .device_run Signed-off-by: -
Sakari Ailus authored
The __v4l2_m2m_try_schedule function acquires and releases multiple spinlocks. Simplify unlocking the job lock by adding labels to unlock the lock and exit the function. Signed-off-by:
-
Ezequiel Garcia authored
Now that the mutexes for output and capture vb2 queues match, it is possible to refer to the context q_lock as the m2m lock for a given m2m context. Remove the output/capture lock selection. Signed-off-by:
-
Ezequiel Garcia authored
Currently, all the mem2mem driver either use a single mutex to lock the capture and output videobuf2 queues, or don't set any mutex. This means the mutexes match, and so the mem2mem framework is able to set the m2m context lock. Enforce this by making it mandatory for drivers to set the same capture and output mutex, or not set any mutex at all. Signed-off-by:
-
Vikash Garodia authored
Add devicetree binding documentation for firmware loader for video hardware running on qualcomm chip. Signed-off-by:
Vikash Garodia <vgarodia@codeaurora.org> Reviewed-by:
Rob Herring <robh@kernel.org> Acked-by:
-
Vikash Garodia authored
Video hardware is mainly comprised of vcodec subsystem and video control subsystem. Video control has ARM9 which executes the video firmware instructions whereas vcodec does the video frame processing. This change adds support to load the video firmware and bring ARM9 out of reset for platforms which does not have trustzone. An iommu domain is associated and managed with the firmware device. Signed-off-by:
-
Stanimir Varbanov authored
This registers a firmware platform_device and associate it with video-firmware DT subnode. Then calls dma configure to initialize dma and iommu. Signed-off-by:
-
Vikash Garodia authored
Separate firmware loading part into a new function. Signed-off-by:
-
Vikash Garodia authored
Add routine to reset the ARM9 and brings it out of reset. Also abstract the Venus CPU state handling with a new function. This is in preparation to add PIL functionality in venus driver. Signed-off-by:
-
Hans Verkuil authored
When vivid_update_format_cap() is called it should free any overlay bitmap since the compose size will change. Signed-off-by:
-
Hans Verkuil authored
text[s] is a signed char, so using that as index into the font8x16 array can result in negative indices. Cast it to u8 to be safe. Signed-off-by:
-
Hans Verkuil authored
Update file paths in MAINTAINERS. Signed-off-by:
Joe Perches <joe@perches.com> Reviewed-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by:
-
Hans Verkuil authored
vb2_core_create_bufs did not check if the memory model for newly added buffers is the same as for already existing buffers. It should return an error if they aren't the same. Signed-off-by:
-
Hans Verkuil authored
The v4l2_dv_timings_cap struct is used to do sanity checks when setting and enumerating DV timings, ensuring that only valid timings as per the HW capabilities are allowed. However, many drivers just filled in 0 for the minimum width, height or pixelclock frequency. This can cause timings with e.g. 0 as width and height to be accepted, which will in turn lead to a potential division by zero. Fill in proper values are minimum boundaries. 640x350 was chosen since it is the smallest resolution in v4l2-dv-timings.h. Same for 13 MHz as the lowest pixelclock frequency (it's slightly below the minimum of 13.5 MHz in the v4l2-dv-timings.h header). Signed-off-by:
-
Hans Verkuil authored
The use of flush_schedule_work() made no sense and caused a syzkaller error. Replace with the correct cancel_delayed_work_sync(). Signed-off-by:
-
Hans Verkuil authored
If you create multiple vivid instances, each with their own media device, then there was no way to tell them apart. Fill in the bus_info so each instance has a unique bus_info string. Signed-off-by:
-
Julia Lawall authored
The video_device structure is only copied into another structure, so it can be const. Done with the help of Coccinelle. Signed-off-by:
Julia Lawall <Julia.Lawall@lip6.fr> Signed-off-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
-
Hans Verkuil authored
The capture DV timings capabilities allowed for a minimum width and height of 0. So passing a timings struct with 0 values is allowed and will later cause a division by zero. Ensure that the width and height must be >= 16 to avoid this. Signed-off-by:
-
Hans Verkuil authored
kthread_run returns an error pointer, but elsewhere in the code dev->kthread_vid_cap/out is checked against NULL. If kthread_run returns an error, then set the pointer to NULL. I chose this method over changing all kthread_vid_cap/out tests elsewhere since this is more robust. Signed-off-by:
-
Julia Lawall authored
The fields of a v4l2_subdev_ops structure are all const, so the structures that are stored there and are not used elsewhere can be const as well. Done with the help of Coccinelle. Signed-off-by:
Helen Koike <helen.koike@collabora.com> Signed-off-by:
-
Julia Lawall authored
The v4l2_m2m_ops structure can be const as it is only passed to v4l2_m2m_init whose parameter is const. Done with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
The v4l2_ctrl_ops structure is only stored in the ops field of a v4l2_ctrl_config structure, and this field is const, or passed as the second argument of v4l2_ctrl_new_std, and the corresponding parameter is declared as const. Accordingly, the structure can also be const. Done with the help of Coccinelle. Signed-off-by:
-
Sean Young authored
Do a series of minor corrections at the V4L2 uAPI documentation. Signed-off-by:
Sean Young <sean@mess.org> Signed-off-by:
-
Hans Verkuil authored
I noticed that repeatedly running 'cec-ctl --playback' would occasionally select 'Playback Device 2' instead of 'Playback Device 1', even though there were no other Playback devices in the HDMI topology. This happened both with 'real' hardware and with the vivid CEC emulation, suggesting that this was an issue in the core code that claims a logical address. What 'cec-ctl --playback' does is to first clear all existing logical addresses, and immediately after that configure the new desired device type. The core code will poll the logical addresses trying to find a free address. When found it will issue a few standard messages as per the CEC spec and return. Those messages are queued up and will be transmitted asynchronously. What happens is that if you run two 'cec-ctl --playback' commands in quick succession, there is still a message of the first cec-ctl command being transmitted when you reconfigure the adapter again in the second cec-ctl command. When the logical addresses are cleared, then all information about outstanding transmits inside the CEC core is also cleared, and the core is no longer aware that there is still a transmit in flight. When the hardware finishes the transmit it calls transmit_done and the CEC core thinks it is actually in response of a POLL messages that is trying to find a free logical address. The result of all this is that the core thinks that the logical address for Playback Device 1 is in use, when it is really an earlier transmit that ended. The main transmit thread looks at adap->transmitting to check if a transmit is in progress, but that is set to NULL when the adapter is unconfigured. adap->transmitting represents the view of userspace, not that of the hardware. So when unconfiguring the adapter the message is marked aborted from the point of view of userspace, but seen from the PoV of the hardware it is still ongoing. So introduce a new bool transmit_in_progress that represents the hardware state and use that instead of adap->transmitting. Now the CEC core waits until the hardware finishes the transmit before starting a new transmit. Signed-off-by:
-
Hans Verkuil authored
If debug_phys_addr is set, then CEC_CAP_PHYS_ADDR is added to the CEC adapter capabilities. This allows for testing CEC even if the physical address isn't set. This makes it possible to connect two HDMI outputs together and still use CEC. Very useful for testing CEC if you don't have access to an HDMI receiver under linux. Signed-off-by:
-
Hans Verkuil authored
The CEC specification requires that the Vendor ID (if any) is reported after a logical address was claimed. This was never done, so add support for this. Signed-off-by:
-
Hans Verkuil authored
The CEC IP is very similar between the three HDMI receivers, but not identical. Add support for all three variants. Tested with an adv7604 and an adv7612. Signed-off-by:
Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se> Signed-off-by:
-
Hans Verkuil authored
If a filehandle is dup()ped, then it is possible to close it from one fd and call mmap from the other. This creates a race condition in vb2_mmap where it is using queue data that __vb2_queue_free (called from close()) is in the process of releasing. By moving up the mutex_lock(mmap_lock) in vb2_mmap this race is avoided since __vb2_queue_free is called with the same mutex locked. So vb2_mmap now reads consistent buffer data. Signed-off-by:
Hans Verkuil <hverkuil@xs4all.nl> Reported-by: syzbot+be93025dd45dccd8923c@syzkaller.appspotmail.com Signed-off-by:
Hans Verkuil <hansverk@cisco.com> Signed-off-by:
-
