- 21 Oct, 2021 6 commits
-
-
Mansur Alisha Shaik authored
In existing implementation, driver is freeing and un-mapping all the decoded picture buffers(DPB) as part of dynamic resolution change(DRC) handling. As a result, when firmware try to access the DPB buffer, from previous sequence, SMMU context fault is seen due to the buffer being already unmapped. With this change, driver defines ownership of each DPB buffer. If a buffer is owned by firmware, driver would skip from un-mapping the same. Signed-off-by:
Mansur Alisha Shaik <mansur@codeaurora.org> Signed-off-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org> Signed-off-by:
Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
-
Stanimir Varbanov authored
According to stateful decoder docs a fatal failure of decoding (and encoding) could be recover it by closing the corresponding file handle and open new one or reinitialize decoding (and encoding) by stop streaming on both queues. In order to satisfy this requirement we add a mechanism ins sys_error_handler and corresponding decoder and encoder drivers to wait for sys_error_done waitqueue in reqbuf. Signed-off-by:
Vikash Garodia <vgarodia@codeaurora.org> Signed-off-by:
-
Stanimir Varbanov authored
Add a helper to mark source and destination vb2 queues fatal unrecoverable error. Signed-off-by:
-
Stanimir Varbanov authored
Check sys error flag for all hfi_session_xxx functions and exit with EIO in case of an error. Signed-off-by:
-
Stanimir Varbanov authored
Make the sys_error flag an atomic bitops in order to avoid locking in sys_error readers. Signed-off-by:
-
Stanimir Varbanov authored
Migrate encoder to pm-runtime autosuspend. Signed-off-by:
-
- 20 Oct, 2021 13 commits
-
-
Michael Tretter authored
The vui parameters are optional. However, the vui data allows to specify the color space of the encoded video. Write the vui parameters to make sure that decoders are able to pick up the correct color space. Also implement the necessary lookup functions to convert the values from the V4L2 controls to the values specified in the hevc standard. Signed-off-by:
Michael Tretter <m.tretter@pengutronix.de> Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
-
Michael Tretter authored
The NAL unit generator for HEVC does not support the generation of vui parameters. Implement it to allow drivers to set the vui parameters in the coded video stream. Signed-off-by:
-
Michael Tretter authored
Currently, the driver always writes PAL as video format into the SPS of the encoded stream. Set the video format to the default value 5 (unspecified) and use the color description that is already configured on the channel as color space. This fixes the color space definition in the coded data to reflect the configured color space of the video data that is encoded. Add lookup functions to convert the color primaries, transfer function and matrix coefficients from the V4L2 control values to the values specified in the h.264 standard. Signed-off-by:
-
Michael Tretter authored
The lookup of the h.264 or hevc values for the respective V4L2 controls is done by the driver that uses the sps/pps generator and not in the generator. Therefore, it is more intuitive to define these functions directly in the header and not in the module. Extract the functions to the headers as static inline functions. Also simplify the function name and add kernel-doc for the hevc functions. Signed-off-by:
-
Michael Tretter authored
The bit rate in the SPS can be scaled by an exponent, which allows to reduce the number of bits in the SPS in case of high bit rates. The driver did not scale the bitrate, but used a scaling exponent of 0. Fix this by properly calculating the scaling factor and writing the bit rate as value and scaling factor into the SPS. Signed-off-by:
-
Michael Tretter authored
The VCU allows to specify the QP per frame and coding unit. A buffer that specifies the QP is passed via the ep2 field in the ENCODE_FRAME message. The driver currently does not support the external QP table. Simplify the driver by not setting the ep2 fields at all. Signed-off-by:
-
Michael Tretter authored
The fields for the number of rows and columns in the encode frame response message are switched. This causes broken PPS, if the encoder uses tiles for encoding and the number of rows and columns differ. Write the fields of the response message into the correct fields of the the internal data structure when parsing the response message. Signed-off-by:
-
Michael Tretter authored
The encoder buffer can have a negative impact on the quality of the encoded video. Add a control to allow user space to disable the encoder buffer per channel if the VCU supports the encoder buffer but the quality is not sufficient. Signed-off-by:
-
Michael Tretter authored
The encoder buffer serves as a cache for reference frames during the encoding process. The encoder buffer significantly reduces the bandwidth requirement for read accesses on the AXI ports of the VCU, but slightly reduces the quality of the encoded video. The encoder buffer must be configured as a whole during the firmware initialization and later explicitly enabled for every channel that shall use the encoder buffer. Prior to firmware version 2019.2, it was necessary to explicitly set the size of the encoder buffer for every channel. Since 2019.2 it is sufficient to enable the encoder buffer and leave the rest to the firmware. Therefore, only support the encoder buffer for firmware 2019.2 and later. Signed-off-by:
-
Michael Tretter authored
The allegro driver must ensure that the mcu and core clocks are enabled and set to the expected clock rate before trying to load the firmware and to reset the MCU. Up until now, the driver assumed that the clocks are always enabled in the PL (programming logic), because the xlnx_vcu driver did not export the clocks to other drivers. This has changed and by explicitly enabling the clocks in the driver, this assumption can be dropped. It might even be possible to disable the clocks for the encoder if the encoder is not used. However, the behavior is not documented and it might be necessary to reinitialize the encoder after deactivating the clocks. Play it safe by sticking to the current behavior. Signed-off-by:
-
Michael Tretter authored
The VCU provides information about its configuration in a dedicated register space. These settings include, for example, the expected clock rates and the configuration of the encoder buffer. In the device tree, the settings are described by the "xlnx,vcu-settings" compatible. The settings are needed to correctly configure the clocks and the encoder buffer. Lookup the VCU settings in the device tree and make it accessible to the driver via a regmap. Signed-off-by:
-
Michael Tretter authored
If the module probe finished, but the firmware initialization failed, removing the module must not revert the firmware initialization. Add a field to track the status of the firmware initialization and only roll it back, if the firmware was successfully initialized. Signed-off-by:
-
Michael Tretter authored
The mailbox is initialized after the interrupt handler is installed. As the firmware is loaded and started even later, it should not happen that the interrupt occurs without the mailbox being initialized. As the Linux Driver Verification project (linuxtesting.org) keeps reporting this as an error, add a check to ignore interrupts before the mailbox is initialized to fix this potential null pointer dereference. Reported-by:
Yuri Savinykh <s02190703@gse.cs.msu.ru> Reported-by:
Nadezda Lutovinova <lutovinova@ispras.ru> Signed-off-by:
-
- 19 Oct, 2021 20 commits
-
-
Sean Young authored
On an Intel NUC6iSYK, no IR is reported after a receive overflow. When a receiver overflow occurs, this condition is only cleared by reading the fifo. Make sure we read anything in the fifo. Fixes: 28c7afb0 ("media: ite-cir: check for receive overflow") Suggested-by:
Bryan Pass <bryan.pass@gmail.com> Tested-by:
Sean Young <sean@mess.org> Signed-off-by:
-
Sean Young authored
This driver is a port of the lirc_sir driver to rc-core. However, for this driver I could not find any hardware to test, so it was done without testing. This is a mistake. There are clear bugs in the code. For example the two arguments to ktime_us_delta() are reversed, which means the result is garbage. The driver has been in the kernel for four years, and noone has ever reported an issue. So, remove this driver. Signed-off-by:
-
Sean Young authored
commit f0c15b36 ("media: ir_toy: prevent device from hanging during transmit") removed a cpu_to_be16() cast, which causes a sparse warning. Fixes: f0c15b36 ("media: ir_toy: prevent device from hanging during transmit") Reported-by:
Hans Verkuil <hverkuil@xs4all.nl> Signed-off-by:
-
Randy Dunlap authored
Prevent the use of page table macros and types from 2 conflicting places. This fixes multiple build errors and warnings, e.g.: ../arch/x86/include/asm/pgtable_64_types.h:21:34: error: conflicting types for ‘pte_t’ typedef struct { pteval_t pte; } pte_t; ^~~~~ In file included from ../include/linux/mm_types_task.h:16:0, from ../include/linux/mm_types.h:5, from ../include/linux/buildid.h:5, from ../include/linux/module.h:14, from ../drivers/media/pci/ivtv/ivtv-driver.h:40, from ../drivers/media/pci/ivtv/ivtvfb.c:29: ../arch/um/include/asm/page.h:57:39: note: previous declaration of ‘pte_t’ was here typedef struct { unsigned long pte; } pte_t; ../arch/x86/include/asm/pgtable_types.h:284:43: error: expected ‘)’ before ‘prot’ static inline pgprot_t pgprot_nx(pgprot_t prot) ^ ../include/linux/pgtable.h:914:26: note: in definition of macro ‘pgprot_nx’ #define pgprot_nx(prot) (prot) ^~~~ In file included from ../arch/x86/include/asm/memtype.h:6:0, from ../drivers/media/pci/ivtv/ivtvfb.c:40: ../arch/x86/include/asm/pgtable_types.h:288:0: warning: "pgprot_nx" redefined #define pgprot_nx pgprot_nx ../arch/x86/include/asm/page_types.h:11:0: warning: "PAGE_SIZE" redefined #define PAGE_SIZE (_AC(1,UL) << PAGE_SHIFT) In file included from ../include/linux/mm_types_task.h:16:0, from ../include/linux/mm_types.h:5, from ../include/linux/buildid.h:5, from ../include/linux/module.h:14, from ../drivers/media/pci/ivtv/ivtv-driver.h:40, from ../drivers/media/pci/ivtv/ivtvfb.c:29: ../arch/um/include/asm/page.h:14:0: note: this is the location of the previous definition #define PAGE_SIZE (_AC(1, UL) << PAGE_SHIFT) Fixes: 68f5d3f3 ("um: add PCI over virtio emulation driver") Signed-off-by:Randy Dunlap <rdunlap@infradead.org> Cc: Johannes Berg <johannes@sipsolutions.net> Cc: Andy Walls <awalls@md.metrocast.net> Cc: linux-um@lists.infradead.org Cc: Richard Weinberger <richard@nod.at> Signed-off-by:
-
Jernej Skrabec authored
Except for VP8 probability coefficients buffer, all other buffers are never accessed by CPU. That allows us to mark them with DMA_ATTR_NO_KERNEL_MAPPING flag. This helps with decoding big (like 4k) videos on 32-bit ARM platforms where default vmalloc size is relatively small - 240 MiB. Since auxiliary buffer are not yet efficiently allocated, this can be easily exceeded. Even if allocation is optimized, 4k videos will still often exceed this limit. Signed-off-by:
Jernej Skrabec <jernej.skrabec@gmail.com> Signed-off-by:
-
Chen-Yu Tsai authored
The mem-to-mem stateless decoder API specifies support for dynamic resolution changes. In particular, the decoder should accept format changes on the OUTPUT queue even when buffers have been allocated, as long as it is not streaming. Relax restrictions for S_FMT as described in the previous paragraph, and as long as the codec format remains the same. This aligns it with the Hantro and Cedrus decoders. This change was mostly based on commit ae02d494 ("media: hantro: Fix s_fmt for dynamic resolution changes"). Since rkvdec_s_fmt() is now just a wrapper around the output/capture variants without any additional shared functionality, drop the wrapper and call the respective functions directly. Fixes: cd33c830 ("media: rkvdec: Add the rkvdec driver") Cc: <stable@vger.kernel.org> Signed-off-by:
Chen-Yu Tsai <wenst@chromium.org> Reviewed-by:
Nicolas Dufresne <nicolas.dufresne@collabora.com> Signed-off-by:
-
Chen-Yu Tsai authored
The rkvdec H.264 decoder currently overrides sizeimage for the output format. This causes issues when userspace requires and requests a larger buffer, but ends up with one of insufficient size. Instead, only provide a default size if none was requested. This fixes the video_decode_accelerator_tests from Chromium failing on the first frame due to insufficient buffer space. It also aligns the behavior of the rkvdec driver with the Hantro and Cedrus drivers. Fixes: cd33c830 ("media: rkvdec: Add the rkvdec driver") Cc: <stable@vger.kernel.org> Signed-off-by:
-
Mirela Rabulea authored
Save some power by disabling/enabling the jpeg clocks with every stream stop/start. Do not use DL_FLAG_RPM_ACTIVE in mxc_jpeg_attach_pm_domains, to ensure power domains are off after probe. Signed-off-by:
Mirela Rabulea <mirela.rabulea@oss.nxp.com> Signed-off-by:
-
Jeff Chase authored
Jeff is leaving Google so Joe will take over. Signed-off-by:
Jeff Chase <jnchase@google.com> Signed-off-by:
-
Krzysztof Hałasa authored
Confirmed to work with ADV7610 HDMI receiver. Signed-off-by:
Krzysztof Hałasa <khalasa@piap.pl> Reviewed-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Krzysztof Hałasa authored
ADV7610 is another HDMI receiver chip, very similar to the ADV7611. Also: print chip names in upper case. Fix an error message claiming that no ADV761x has been found, while in reality a chip different than requested (though still supported) may have been found. Tested on TinyRex BaseBoard Lite. Signed-off-by:
-
Krzysztof Hałasa authored
ADV7610 is another HDMI receiver chip, very similar to the ADV7611. Tested on TinyRex BaseBoard Lite. Signed-off-by:
-
Krzysztof Hałasa authored
The TDA1997x (HDMI receiver) driver currently uses a specific video format detection scheme. The frame (or field in interlaced mode), line and HSync pulse durations are compared to those of known, standard video modes. If a match is found, the mode is assumed to be detected, otherwise -ERANGE is returned (then possibly ignored). This means that: - another mode with similar timings will be detected incorrectly (e.g. 2x faster clock and lines twice as long) - non-standard modes will not work. This patch replaces this scheme with a direct read of geometry registers. This way all modes recognized by the chip are supported. In interlaced modes, the code assumes the V sync signal has the same duration for both fields. While this may be not necessarily true, I can't see any way to get the "other" V sync width. This is most probably harmless. All tests have been performed on Gateworks' Ventana GW54xx board, with a TDA19971 chip. Signed-off-by:
Tim Harvey <tharvey@gateworks.com> Signed-off-by:
-
Hans Verkuil authored
Newer clang versions are suspicious of definitions that mix concatenated strings with comma-separated arrays of strings, this has found real bugs elsewhere, but this seems to be a false positive: drivers/media/usb/gspca/gl860/gl860-mi1320.c:62:37: error: suspicious concatenation of string literals in an array initialization; did you mean to separate the elements with a comma? [-Werror,-Wstring-concatenation] "\xd3\x02\xd4\x28\xd5\x01\xd0\x02" "\xd1\x18\xd2\xc1" ^ , drivers/media/usb/gspca/gl860/gl860-mi1320.c:62:2: note: place parentheses around the string literal to silence warning "\xd3\x02\xd4\x28\xd5\x01\xd0\x02" "\xd1\x18\xd2\xc1" Replace the string literals by compound initializers, using normal hex numbers. Signed-off-by:Arnd Bergmann <arnd@arndb.de> Reviewed-by:
-
Scott K Logan authored
Similar configuration to Kworld PC150-U. Tested: Composite, S-Video, NTSC, ATSC Unsupported: IR remote Signed-off-by:
Scott K Logan <logans@cottsay.net> Signed-off-by:
-
Jammy Huang authored
To show video real-time information as below: Capture: Signal : Unlock Width : 1920 Height : 1080 FRC : 30 Performance: Frame# : 0 Frame Duration(ms) : Now : 0 Min : 0 Max : 0 FPS : 0 [hverkuil: make aspeed_video_proc_open() static, fixes sparse warning] Signed-off-by:Jammy Huang <jammy_huang@aspeedtech.com> Signed-off-by:
-
Sergey Senozhatsky authored
We need to always link allocated vb2_dc_buf back to vb2_buffer because we dereference vb2 in prepare() and finish() callbacks. Signed-off-by:
Sergey Senozhatsky <senozhatsky@chromium.org> Tested-by:
Tomasz Figa <tfiga@chromium.org> Signed-off-by:
-
Vladimir Barinov authored
This adds g/s_parm ioctls for parallel interface. Signed-off-by:
Vladimir Barinov <vladimir.barinov@cogentembedded.com> Signed-off-by:
Nikita Yushchenko <nikita.yoush@cogentembedded.com> Reviewed-by:
Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se> Signed-off-by:
-
Arnd Bergmann authored
I came across a link failure from randconfig builds: x86_64-linux-ld: drivers/media/i2c/ths8200.o: in function `ths8200_remove': ths8200.c:(.text+0x491): undefined reference to `v4l2_async_unregister_subdev' x86_64-linux-ld: drivers/media/i2c/ths8200.o: in function `ths8200_probe': ths8200.c:(.text+0xe49): undefined reference to `v4l2_async_register_subdev' x86_64-linux-ld: drivers/media/i2c/tw9910.o: in function `tw9910_remove': tw9910.c:(.text+0x467): undefined reference to `v4l2_async_unregister_subdev' x86_64-linux-ld: drivers/media/i2c/tw9910.o: in function `tw9910_probe': tw9910.c:(.text+0x1123): undefined reference to `v4l2_async_register_subdev' These clearly lack a 'select' statement, but I don't know why this started happening only now. I had a bit of a look around to find other configs that have the same problem, but could not come up with a reliable way and found nothing else through experimentation. It is likely that other symbols like these exist that need an extra select. Signed-off-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
-
Hans Verkuil authored
Adding kfree(dvb) to vidtv_bridge_remove() will remove the memory too soon: if an application still has an open filehandle to the device when the driver is unloaded, then when that filehandle is closed, a use-after-free access takes place to the freed memory. Move the kfree(dvb) to vidtv_bridge_dev_release() instead. Signed-off-by:
-
- 18 Oct, 2021 1 commit
-
-
Daniel Almeida authored
If v4l2_ctrl_new_custom fails in cedrus_init_ctrls the error path will free ctx->ctrls, which is also freed in cedrus release. Fix this by setting ctx->ctrls to NULL instead of inadvertently removing kfree calls. Signed-off-by:
Daniel Almeida <daniel.almeida@collabora.com> Signed-off-by:
-
