- 24 Jun, 2017 17 commits
-
-
Dan Carpenter authored
Mostly this adds some unlocks to error paths. But, if you see where there were "break;" statements before, I changed those paths to return error codes instead of returning success. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@s-opensource.com>
-
Dan Carpenter authored
This patch mostly adds unlocks to error paths. But one additional small change is that I made the first "break;" a "goto unlock;" which means that now we return failure instead of success on that path. Signed-off-by:
-
Gustavo A. R. Silva authored
Values assigned to variables Fmax and Fmin at lines 2740 and 2741 are overwritten at lines 2754 and 2755 before they can be used. This makes such variable assignments useless. Addresses-Coverity-ID: 1226952 Addresses-Coverity-ID: 1226953 Signed-off-by:
Gustavo A. R. Silva <garsilva@embeddedor.com> Signed-off-by:
-
Mauro Carvalho Chehab authored
Avoid the following warning when building documentation: checking consistency... /devel/v4l/patchwork/Documentation/media/v4l-drivers/imx.rst:: WARNING: document isn't included in any toctree While here, avoid placing all driver authors at just one line at the html/pdf output. Acked-by:
Steve Longerbeam <steve_longerbeam@mentor.com> Signed-off-by:
-
Mauro Carvalho Chehab authored
It is not trivial to implement the logic that collects DVBv5 statistics. As we're seein lately too many implementations that are not quite right when reviewing patchsets, add a detailed explanation, adding a few examples about the right thing to be done. Signed-off-by: -
Daniel Scheller authored
Signed-off-by:
Daniel Scheller <d.scheller@gmx.net> Signed-off-by:
-
Daniel Scheller authored
Side effect: KERN_DEBUG messages aren't written to the kernel log anymore. This also improves the tda18212_ping reporting a bit so users know that if pinging wasn't successful, bad things will happen. Since in module_init_ddbridge() there's no dev yet, pr_info is used instead. Signed-off-by:
-
Daniel Scheller authored
Reported by smatch: drivers/media/pci/ddbridge/ddbridge-core.c:1246 input_tasklet() warn: this loop depends on readl() succeeding drivers/media/pci/ddbridge/ddbridge-core.c:1768 flashio() warn: this loop depends on readl() succeeding drivers/media/pci/ddbridge/ddbridge-core.c:1788 flashio() warn: this loop depends on readl() succeeding Fix this by introducing safe_ddbreadl() which will wrap ddbreadl and checks for all bits set in the return which indicates failure, and return 0 in that case. Usable as drop-in-replacement in all affected while loops w/o having to change the logic. Signed-off-by:
-
Philipp Zabel authored
I would have liked the the picture run timeout error handler to be renamed to something a bit more descriptive in the original commit fb2be08f ("[media] coda: first step at error recovery"). Somehow v1 [1] was merged instead of v2 [2]. [1] https://patchwork.kernel.org/patch/9663965/ [2] https://patchwork.kernel.org/patch/9774239/Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
Hans Verkuil <hansverk@cisco.com> Signed-off-by:
-
Colin Ian King authored
The current pre-decrement is incorrect and should be replaced with a post-decrement. Consider the case where the very first clk_prepare_enable fails when i is 0; in this case the error clean up will decrement the unsigned int which wraps to the largest unsigned int value causing an array out of bounds read on core->clks[i]. Detected by CoverityScan, CID#1446590 ("Out-of-bounds read") Signed-off-by:Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
Gustavo A. R. Silva authored
Check return value from call to core->write(), so in case of error print error message, jump to goto label fail and eventually return. Addresses-Coverity-ID: 1226943 Signed-off-by:
-
Arnd Bergmann authored
Back in April I created a patch to address a false-positive warning: drivers/media/usb/rainshadow-cec/rainshadow-cec.c: In function 'rain_irq_work_handler': drivers/media/usb/rainshadow-cec/rainshadow-cec.c:171:31: error: 'data' may be used uninitialized in this function [-Werror=maybe-uninitialized] My patch was totally wrong and introduced a real bug, and Colin Ian King thankfully noticed it now and fixed my mistake. Unfortunately, fixing the actual uninitialized data in this case brought back the original bogus warning. This is a new version of the patch, which simplifies the code to the point where gcc notices the behavior is correct. Fixes: ca33784b ("[media] rainshadow-cec: ensure exit_loop is intialized") Fixes: ea6a69de ("[media] rainshadow-cec: avoid -Wmaybe-uninitialized warning") Cc: Colin Ian King <colin.king@canonical.com> Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Joe Perches authored
Commit 680d87c0 ("[media] tuner-core: use pr_foo, instead of internal printk macros") removed the use of PREFIX, remove the #define Signed-off-by:
Joe Perches <joe@perches.com> Signed-off-by:
-
Joe Perches authored
Convert STK_<LEVEL> to pr_<level> to use the typical kernel logging. Add a define for pr_fmt. No change in logging output. Miscellanea: o Remove now unused PREFIX and STK_<LEVEL> macros o Realign arguments o Use pr_<level>_ratelimited o Add a few missing newlines to formats Signed-off-by:
-
Philipp Zabel authored
This fixes a smatch warning: drivers/media/platform/coda/coda-bit.c:415 coda_alloc_framebuffers() error: we previously assumed 'ctx->codec' could be null (see line 396) coda_alloc_framebuffers() is called from coda_start_encoding() and __coda_start_decoding(). Both dereference ctx->codec before calling coda_alloc_framebuffers() in lines 935 and 1649, so ctx->codec can not be NULL. Signed-off-by: -
Arnd Bergmann authored
This one got applied twice, causing a build error with clang: drivers/media/i2c/et8ek8/et8ek8_driver.c:1499:1: error: redefinition of '__mod_of__et8ek8_of_table_device_table' Fixes: 9ae05fd1 ("[media] et8ek8: Export OF device ID as module aliases") Signed-off-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Acked-by:
Pavel Machek <pavel@ucw.cz> Signed-off-by:
-
Markus Elfring authored
Seven single characters (line breaks) should be put into a sequence. Thus use the corresponding function "seq_putc". This issue was detected by using the Coccinelle software. Signed-off-by:
Markus Elfring <elfring@users.sourceforge.net> Signed-off-by:
-
- 23 Jun, 2017 17 commits
-
-
Kieran Bingham authored
Drivers must not perform unbalanced calls to stop the entity pipeline, however if they do they will fault in the core media code, as the entity->pipe will be set as NULL. We handle this gracefully in the core with a WARN for the developer. Replace the erroneous check on zero streaming counts, with a check on NULL pipe elements instead, as this is the symptom of unbalanced media_pipeline_stop calls. Signed-off-by:
Kieran Bingham <kieran.bingham+renesas@ideasonboard.com> Signed-off-by:
Laurent Pinchart <laurent.pincharts@ideasonboard.com> Signed-off-by:
-
Hans Verkuil authored
Unfortunately the use of 'type' was inconsistent for multiplanar buffer types. Starting with 4.13 both the normal and _MPLANE variants are allowed, thus making it possible to write sensible code. Yes, we messed up :-( Signed-off-by:
Hans Verkuil <hans.verkuil@cisco.com> Reviewed-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Acked-by:
-
Hans Verkuil authored
The type field in struct v4l2_selection is supposed to never use the _MPLANE variants. E.g. if the driver supports V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE, then userspace should still pass V4L2_BUF_TYPE_VIDEO_CAPTURE. The reasons for this are lost in the mists of time, but it is really annoying. In addition, the exynos drivers didn't follow this rule and instead expected the _MPLANE type. To fix that code is added to the v4l2 core that maps the _MPLANE buffer types to their regular equivalents before calling the driver. Effectively this allows for userspace to use either _MPLANE or the regular buffer type. This keeps backwards compatibility while making things easier for userspace. Since drivers now never see the _MPLANE buffer types the exynos drivers had to be adapted as well. Signed-off-by:
-
Helen Fornazier authored
Declare vimc_sen_video_ops as static, remove warning from sparse tool Signed-off-by:
Helen Koike <helen.koike@collabora.com> Signed-off-by:
-
Helen Fornazier authored
Implement scaler and integrated with the core Signed-off-by:
-
Helen Fornazier authored
Implement the debayer filter and integrate it with the core Signed-off-by:
-
Helen Fornazier authored
Change the core structure for adding subdevices in the topology. Instead of calling the specific create function for each subdevice, inject a child platform_device with the driver's name. Each type of node in the topology (sensor, capture, debayer, scaler) will register a platform_driver with the corresponding name through the component subsystem. Implementing a new subdevice type doesn't require vimc-core to be altered. This facilitates future implementation of dynamic entities, where hotpluging an entity in the topology is just a matter of registering/unregistering a platform_device in the system. It also facilitates other implementations of different nodes without touching the core code and remove the need of a header file for each type of node. Signed-off-by:
-
Helen Fornazier authored
Allow user space to change the image format as the frame size, the pixel format, colorspace, quantization, field YCbCr encoding and the transfer function Signed-off-by:
-
Helen Fornazier authored
Allow user space to change the image format as the frame size, the media bus pixel format, colorspace, quantization, field YCbCr encoding and the transfer function Signed-off-by:
-
Helen Fornazier authored
Colorimetry value will always be checked in the same way. Adding a helper macro for that Signed-off-by:
-
Helen Fornazier authored
All links will be checked in the same way. Adding a helper function for that Signed-off-by:
-
Helen Fornazier authored
Move the vimc_cap_pipeline_s_stream from the vimc-cap.c to vimc-common.c as this core will be reused by other subdevices to activate the stream in their directly connected nodes Signed-off-by:
-
Helen Fornazier authored
As all the subdevices in the topology will be initialized in the same way, to avoid code repetition the vimc_ent_sd_{register, unregister} helper functions were created Signed-off-by: -
Helen Fornazier authored
Remove helper functions from vimc-core and add it in vimc-common to clean up the core. Signed-off-by:
-
Helen Fornazier authored
Initialize the test pattern generator on the sensor Generate a colored bar image instead of a grey one Signed-off-by:
-
Jacopo Mondi authored
Commit e7890297 ("i2c: sh_mobile: don't send a stop condition by default inside transfers") makes the i2c_sh_mobile I2C-adapter emit a stop/start sequence between messages in a single transfer only when explicitly requested with I2C_M_STOP. This breaks the ov772x driver in the SH4 Migo-R board as the Omnivision sensor uses the I2C-like SCCB protocol that doesn't support repeated starts: i2c-sh_mobile i2c-sh_mobile.0: Transfer request timed out ov772x 0-0021: Product ID error 92:92 Fix it by marking the client as SCCB, forcing the emission of a stop/start sequence between all messages. As I2C_M_STOP requires the I2C adapter to support protocol mangling, ensure that the I2C_FUNC_PROTOCOL_MANGLING functionality is available. Tested on SH4 Migo-R board, with OV772x now successfully probing soc-camera-pdrv soc-camera-pdrv.0: Probing soc-camera-pdrv.0 ov772x 0-0021: ov7725 Product ID 77:21 Manufacturer ID 7f:a2 Signed-off-by:
Jacopo Mondi <jacopo@jmondi.org> Suggested-by:
Laurent Pinchart <laurent.pinchart@ideasonboard.com> Acked-by:
Wolfram Sang <wsa+renesas@sang-engineering.com> Reviewed-by:
-
Mauro Carvalho Chehab authored
Since 2015, the documentation for FE_DISEQC_SEND_BURST, FE_SET_TONE and FE_SET_VOLTAGE are incorrectly saying that the enums are passed by reference. They aren't: they're passed by value. Fix the documentation to reflect reality. Fixes: 81959d99 ("[media] DocBook: better document FE_DISEQC_SEND_BURST ioctl") Fixes: d6b6d346 ("[media] DocBook: better document FE_SET_VOLTAGE ioctl") Fixes: 6dc59e7a ("[media] DocBook: better document FE_SET_TONE ioctl") Reported-by:
Thierry Lelegard <thierry.lelegard@free.fr> Signed-off-by:
-
- 20 Jun, 2017 6 commits
-
-
Arnd Bergmann authored
'struct video_event' is used for the VIDEO_GET_EVENT ioctl, implemented by drivers/media/pci/ivtv/ivtv-ioctl.c and drivers/media/pci/ttpci/av7110_av.c. The structure contains a 'time_t', which will be redefined in the future to be 64-bit wide, causing an incompatible ABI change for this ioctl. As it turns out, neither of the drivers currently sets the timestamp field, and it is presumably useless anyway because of the limited resolutions (no sub-second times). This means we can simply change the structure definition to use a 'long' instead of 'time_t' and remain compatible with all existing user space binaries when time_t gets changed. If anybody ever starts using this field, they have to make sure not to use 1970 based seconds in there, as those overflow in 2038. Signed-off-by:
-
Stanimir Varbanov authored
We want all media drivers to build with COMPILE_TEST, as the Coverity instance we use on Kernel works only for x86. Also, our test workflow relies on it, in order to identify git bisect breakages. Signed-off-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org>
-
Antti Palosaari authored
Move power-up and power-down functionality to init/sleep ops and get rid of old function. Fixes and simplifies power-up functionality slightly. Signed-off-by:
Antti Palosaari <crope@iki.fi> Signed-off-by:
-
Antti Palosaari authored
Refactor firmware download routine. Signed-off-by:
-
Antti Palosaari authored
We need to wait 2nd demod power-up before download firmware. Move that wait to more correct location. Signed-off-by:
-
Antti Palosaari authored
Removed register writes are already chip defaults, are not required, are set later or belong to AF9015 USB interface. Signed-off-by:
-
