- 18 Dec, 2017 40 commits
-
-
Mauro Carvalho Chehab authored
The agreed format is to use /* */ comments inside header files. Unfortunately, I ended by using // on a few ones. Reported-by:
Andi Shyti <andi.shyti@samsung.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@s-opensource.com>
-
Flavio Ceolin authored
s5p_jpeg_runtime_resume() does not call clk_disable_unprepare() for jpeg->clocks[0] when one of the clk_prepare_enable() fails. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by:
Flavio Ceolin <flavio.ceolin@intel.com> Acked-by:
Andrzej Pietrasiewicz <andrzej.p@samsung.com> Signed-off-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by:
-
Simon Shields authored
If the subdev is not yet present (probably because the subdev module has not yet been loaded), the pipe will be NULL. Make sure that this is not the case before attempting to call the op. Signed-off-by:
Simon Shields <simon@lineageos.org> Signed-off-by:
-
Marek Szyprowski authored
Support for Exynos4212 SoCs has been removed by commit bca9085e ("ARM: dts: exynos: remove Exynos4212 support (dead code)"), so there is no need to keep remaining dead code related to this SoC version. Signed-off-by:
Marek Szyprowski <m.szyprowski@samsung.com> Acked-by:
Krzysztof Kozlowski <krzk@kernel.org> Signed-off-by:
-
Shuah Khan authored
Driver calls request_firmware() whenever the device is opened for the first time. As the device gets opened and closed, dev->num_inst == 1 is true several times. This is not necessary since the firmware is saved in the fw_buf. s5p_mfc_load_firmware() copies the buffer returned by the request_firmware() to dev->fw_buf. fw_buf sticks around until it gets released from s5p_mfc_remove(), hence there is no need to keep requesting firmware and copying it to fw_buf. This might have been overlooked when changes are made to free fw_buf from the device release interface s5p_mfc_release(). Fix s5p_mfc_load_firmware() to call request_firmware() once and keep state. Change _probe() to load firmware once fw_buf has been allocated. s5p_mfc_open() and it continues to call s5p_mfc_load_firmware() and init hardware which is the step where firmware is written to the device. This addresses the mfc_mutex contention due to repeated request_firmware() calls from open() in the following circular locking warning: [ 552.194115] qtdemux0:sink/2710 is trying to acquire lock: [ 552.199488] (&dev->mfc_mutex){+.+.}, at: [<bf145544>] s5p_mfc_mmap+0x28/0xd4 [s5p_mfc] [ 552.207459] but task is already holding lock: [ 552.213264] (&mm->mmap_sem){++++}, at: [<c01df2e4>] vm_mmap_pgoff+0x44/0xb8 [ 552.220284] which lock already depends on the new lock. [ 552.228429] the existing dependency chain (in reverse order) is: [ 552.235881] -> #2 (&mm->mmap_sem){++++}: [ 552.241259] __might_fault+0x80/0xb0 [ 552.245331] filldir64+0xc0/0x2f8 [ 552.249144] call_filldir+0xb0/0x14c [ 552.253214] ext4_readdir+0x768/0x90c [ 552.257374] iterate_dir+0x74/0x168 [ 552.261360] SyS_getdents64+0x7c/0x1a0 [ 552.265608] ret_fast_syscall+0x0/0x28 [ 552.269850] -> #1 (&type->i_mutex_dir_key#2){++++}: [ 552.276180] down_read+0x48/0x90 [ 552.279904] lookup_slow+0x74/0x178 [ 552.283889] walk_component+0x1a4/0x2e4 [ 552.288222] link_path_walk+0x174/0x4a0 [ 552.292555] path_openat+0x68/0x944 [ 552.296541] do_filp_open+0x60/0xc4 [ 552.300528] file_open_name+0xe4/0x114 [ 552.304772] filp_open+0x28/0x48 [ 552.308499] kernel_read_file_from_path+0x30/0x78 [ 552.313700] _request_firmware+0x3ec/0x78c [ 552.318291] request_firmware+0x3c/0x54 [ 552.322642] s5p_mfc_load_firmware+0x54/0x150 [s5p_mfc] [ 552.328358] s5p_mfc_open+0x4e4/0x550 [s5p_mfc] [ 552.333394] v4l2_open+0xa0/0x104 [videodev] [ 552.338137] chrdev_open+0xa4/0x18c [ 552.342121] do_dentry_open+0x208/0x310 [ 552.346454] path_openat+0x28c/0x944 [ 552.350526] do_filp_open+0x60/0xc4 [ 552.354512] do_sys_open+0x118/0x1c8 [ 552.358586] ret_fast_syscall+0x0/0x28 [ 552.362830] -> #0 (&dev->mfc_mutex){+.+.}: -> #0 (&dev->mfc_mutex){+.+.}: [ 552.368379] lock_acquire+0x6c/0x88 [ 552.372364] __mutex_lock+0x68/0xa34 [ 552.376437] mutex_lock_interruptible_nested+0x1c/0x24 [ 552.382086] s5p_mfc_mmap+0x28/0xd4 [s5p_mfc] [ 552.386939] v4l2_mmap+0x54/0x88 [videodev] [ 552.391601] mmap_region+0x3a8/0x638 [ 552.395673] do_mmap+0x330/0x3a4 [ 552.399400] vm_mmap_pgoff+0x90/0xb8 [ 552.403472] SyS_mmap_pgoff+0x90/0xc0 [ 552.407632] ret_fast_syscall+0x0/0x28 [ 552.411876] other info that might help us debug this: [ 552.419848] Chain exists of: &dev->mfc_mutex --> &type->i_mutex_dir_key#2 --> &mm->mmap_sem [ 552.431200] Possible unsafe locking scenario: [ 552.437092] CPU0 CPU1 [ 552.441598] ---- ---- [ 552.446104] lock(&mm->mmap_sem); [ 552.449484] lock(&type->i_mutex_dir_key#2); [ 552.456329] lock(&mm->mmap_sem); [ 552.462222] lock(&dev->mfc_mutex); [ 552.465775] *** DEADLOCK *** Signed-off-by:Shuah Khan <shuahkh@osg.samsung.com> Signed-off-by:
-
Shuah Khan authored
s5p_mfc_load_firmware() will not get called if fw_buf.virt allocation fails. The allocation happens very early on in the probe routine and probe fails if allocation fails. There is no need to check if it is null in s5p_mfc_load_firmware(). Remove the check. Signed-off-by:
-
Marek Szyprowski authored
Setting both V4L2_CAP_VIDEO_CAPTURE_MPLANE and V4L2_CAP_VIDEO_OUTPUT_MPLANE for mem2mem video nodes is obsolete since commit f0476a83 ("[media] V4L: Add capability flags for memory-to-memory devices"). It was enough time to adapt all users to the new flags, so drop the legacy caps for now to match other mem2mem drivers. Signed-off-by:
-
Marek Szyprowski authored
Setting both V4L2_CAP_VIDEO_CAPTURE_MPLANE and V4L2_CAP_VIDEO_OUTPUT_MPLANE for mem2mem video nodes is obsolete since commit f0476a83 ("[media] V4L: Add capability flags for memory-to-memory devices"). It was enough time to adapt all users to the new flags, so drop the legacy caps for now to match other mem2mem drivers. Signed-off-by:
-
Arnd Bergmann authored
We copy the subdev frame format from a partially initialized structure, which is not entirely well-defined. Older compilers like gcc-4.4 can copy uninitialized stack data here and warn about it: drivers/media/platform/exynos4-is/fimc-isp.c: In function 'fimc_isp_subdev_open': drivers/media/platform/exynos4-is/fimc-isp.c:379: error: 'fmt.reserved[10u]' may be used uninitialized in this function drivers/media/platform/exynos4-is/fimc-isp.c:379: error: 'fmt.reserved[9u]' may be used uninitialized in this function ... drivers/media/platform/exynos4-is/fimc-isp.c:379: error: 'fmt.reserved[0u]' may be used uninitialized in this function drivers/media/platform/exynos4-is/fimc-isp.c:379: error: 'fmt.xfer_func' may be used uninitialized in this function On newer compilers, only the initialized fields get copied, but we should not rely on that, so this changes the code to zero-out the remaining fields first. Fixes: 9a761e43 ("[media] exynos4-is: Add Exynos4x12 FIMC-IS driver") Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Sylwester Nawrocki authored
This patch fixes the menu_skip_mask field initialization and addresses a following issue found by the SVACE static analysis: * NO_EFFECT.SELF: assignment to self in expression 'cfg.menu_skip_mask = cfg.menu_skip_mask' No effect at drivers/media/platform/s5p-mfc/s5p_mfc_enc.c:2083 Signed-off-by:
-
Jia-Ju Bai authored
The driver may sleep under a spinlock. The function call path is: bdisp_device_run (acquire the spinlock) bdisp_hw_update bdisp_hw_save_request devm_kzalloc(GFP_KERNEL) --> may sleep To fix it, GFP_KERNEL is replaced with GFP_ATOMIC. This bug is found by my static analysis tool(DSAC) and checked by my code review. Signed-off-by:Jia-Ju Bai <baijiaju1990@gmail.com> Reviewed-by:
Fabien Dessenne <fabien.dessenne@st.com> Signed-off-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
-
Stanimir Varbanov authored
Currently videobuf2-dma-sg checks for dma direction for every single page and videobuf2-dc lacks any dma direction checks and calls set_page_dirty_lock unconditionally. Thus unify and align the invocations of set_page_dirty_lock for videobuf2-dc, videobuf2-sg memory allocators with videobuf2-vmalloc, i.e. the pattern used in vmalloc has been copied to dc and dma-sg. Suggested-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org> Acked-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
-
Hans Verkuil authored
The pvrusb2 code appears to have a some old workaround code for xawtv that causes a WARN() due to an unrecognized pixelformat 0 in v4l2_ioctl.c. Since all other MPEG drivers fill this in correctly, it is a safe assumption that this particular problem no longer exists. While I'm at it, clean up the code a bit. Signed-off-by:
-
Philipp Zabel authored
V4L2_BUF_FLAG_LAST is a signal from the driver to userspace for buffers on the capture queue. When userspace queues back a capture buffer with the flag set, we should clear it. Otherwise, if userspace restarts streaming after EOS, without reallocating the buffers, mem2mem devices will erroneously signal EOS prematurely, as soon as the already flagged buffer is dequeued. Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Philipp Zabel authored
The mvcol buffer needs to be placed behind the chroma plane(s) when decoding MPEG-4, same as for the h.264 decoder. Use the real offset with the required rounding. Signed-off-by:
-
Philipp Zabel authored
The MPEG-4 decoder mvcol buffer was registered, but its size not added to a frame buffer allocation. This could cause the decoder to write past the end of the allocated buffer for large frame sizes. Signed-off-by:
-
Philipp Zabel authored
We need internal frames to be rounded up to full macroblocks for MPEG-4 decoding as well. Signed-off-by:
-
Philipp Zabel authored
Since bytesperline always fulfills VDOA width requirements, detile the whole buffer instead of limiting to visible width. This stops TRY_FMT from returning -EINVAL for YUYV capture buffers that are not a multiple of 16 wide. An alternative would be to always round up width to stride, as we report the valid image rectange via G_SELECTION (V4L2_SEL_TGT_COMPOSE_DEFAULT), but that would require all applications to handle the compose default rectangle properly. Signed-off-by:
-
Lucas Stach authored
The current driver implementation expects at least one buffer on all queues to start streaming. Properly signal this to the vb2 core, to avoid confusion when streamon is racing with qbuf. Signed-off-by:
Lucas Stach <l.stach@pengutronix.de> Signed-off-by:
-
Flavio Ceolin authored
pxa_camera_probe() was not calling pxa_camera_deactivate(), responsible to call clk_disable_unprepare(), on the failure path. This was leading to unbalancing source clock. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by:
Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by:
-
Jacopo Mondi authored
videobuf2 core reports an error when not all buffers have been returned to the framework: drivers/media/v4l2-core/videobuf2-core.c:1651 WARN_ON(atomic_read(&q->owned_by_drv_count)) Fix this returning all buffers currently in capture queue. Signed-off-by:
Jacopo Mondi <jacopo+renesas@jmondi.org> Signed-off-by:
-
Mauro Carvalho Chehab authored
When we split VB2 into an independent streaming module and a V4L2 one, some vb2-core functions started to have a wrong description: they're meant to be used only by the API-specific parts of VB2, like vb2-v4l2, as the functions that V4L2 drivers should use are all under videobuf2-v4l2.h. Correct their descriptions. Acked-by:
-
Mauro Carvalho Chehab authored
As kernel-doc doesn't support documenting #define values, and using enum makes easier to identify where the values are used, convert V4L2_MBUS_FRAME_DESC_FL_* to enum, and use BIT() macro. While here, fix the description at v4l2_mbus_frame_desc_entry, in order to match what's described for V4L2_MBUS_FRAME_DESC_FL_LEN_MAX. Signed-off-by: -
Mauro Carvalho Chehab authored
Convert V4L2_SUBDEV_IO_PIN_* to enums, use BIT() and document via kernel-doc. Signed-off-by: -
Mauro Carvalho Chehab authored
ownner -> owner Acked-by:
-
Mauro Carvalho Chehab authored
Despite the struct says "color16", it was actually using 32 bits for each color. Fix it. Suggested-by:
Hans Verkuil <hverkuil@xs4all.nl> Acked-by:
-
Mauro Carvalho Chehab authored
The color structs right now are just "color" and "color16". That may lead into conflicts, and don't define precisely what they meant. As those are used by two drivers (vivid and vimc), this is even on a somewhat public header! So rename them to: color -> tpg_rbg_color8 color16 -> tpg_rbg_color16 Acked-by:
-
Mauro Carvalho Chehab authored
The v4l2-tpg*.h headers are meant to be used only internally by vivid and vimc. There's no sense keeping them together with the V4L2 kAPI headers. Also, one header includes the other as they're meant to be used together. So, merge them. Acked-by:
-
Mauro Carvalho Chehab authored
Add cross-references where needed and add periods at the end of each kernel-doc paragraph, in order to make it coherent with other VB2 descriptions. Acked-by:
-
Mauro Carvalho Chehab authored
There are several VB2 core functions that aren't documented. Signed-off-by: -
Mauro Carvalho Chehab authored
There are several issues on the current markups: - lack of cross-references; - wrong cross-references; - lack of a period of the end of several phrases; - Some descriptions can be enhanced. Signed-off-by: -
Mauro Carvalho Chehab authored
Use the existing macros to identify vb2_io_modes bits. Acked-by:
-
Mauro Carvalho Chehab authored
The description there is completely broken and it mentions an ioctl that doesn't exist. Fix it. Signed-off-by: -
Mauro Carvalho Chehab authored
When generating Sphinx output, create cross-references for the callbacks for each ioctl. While here, fix a few wrong names for ioctls. Signed-off-by: -
Mauro Carvalho Chehab authored
Convert #defines to enums and add kernel-doc markups for V4L2 video_device flags. Acked-by:
-
Mauro Carvalho Chehab authored
The V4L_DIR_* direction flags document the direction for a V4L2 device node. Convert them to enum and document. Acked-by:
-
Mauro Carvalho Chehab authored
In the past, the same I2C address were used on multiple places. After I2C rebinding changes, this is no longer needed. So, we can just get rid of this header, placing the I2C address where they belong, e. g. either at bttv driver or at tvtuner. Acked-by:
-
Mauro Carvalho Chehab authored
Some of the previously used I2C addresses there aren't used anymore. So, get rid of them. Acked-by:
-
Mauro Carvalho Chehab authored
Using enums makes easier to document, as it can use kernel-doc markups. It also allows cross-referencing, with increases the kAPI readability. Please notice that now cx88_querycap() has to have a default for the VFL type, as there are more types than supported by the driver. Acked-By:
Mike Isely <isely@pobox.com> Acked-by:
-
Mauro Carvalho Chehab authored
Instead of using (1 << n) for bits, use the BIT() macro, as it makes a difference from documentation point of view. Acked-by:
-
