- 19 Jun, 2013 1 commit
-
-
Sylwester Nawrocki authored
The ISP clock register content is not preserved over the ISP power domain off/on cycle. Instead of setting the clock frequencies once at probe time the clock rates set up is moved to the runtime_resume handler, which is invoked after the related power domain is already enabled, ensuring the clocks are properly configured when the device is actively used. This fixes the FIMC-IS malfunctions and STREAM ON timeout errors accuring on some boards: [ 59.860000] fimc_is_general_irq_handler:583 ISR_NDONE: 5: 0x800003e8, IS_ERROR_UNKNOWN [ 59.860000] fimc_is_general_irq_handler:586 IS_ERROR_TIME_OUT Signed-off-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com Signed-off-by:
Mauro Carvalho Chehab <mchehab@redhat.com>
-
- 17 Jun, 2013 1 commit
-
-
Gianluca Gennari authored
As suggested by Antti, this patch replaces: https://patchwork.kernel.org/patch/2649861/ The buffer overflow is fixed by reading only the r820t ID register. Signed-off-by:
Gianluca Gennari <gennarone@gmail.com> Acked-by:
Antti Palosaari <crope@iki.fi> Reviewed-by:
-
- 09 Jun, 2013 6 commits
-
-
Wenbing Wang authored
in soc_camera_close(), if ici->ops->remove() removes device firstly, and then call __soc_camera_power_off(), it has logic error. Since if remove device, it should disable subdev clk. but in __soc_camera_ power_off(), it will callback v4l2 s_power function which will read/write subdev registers to control power by i2c. and then i2c read/write will fail because of clk disable. So suggest to re-sequence two functions call. Change-Id: Iee7a6d4fc7c7c1addb5d342621eb8dcd00fa2745 Signed-off-by:
Wenbing Wang <wangwb@marvell.com> Signed-off-by:
Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by:
-
Katsuya Matsubara authored
The 'bytesperline' value only indicates the stride of the Y plane if the color format is planar, such as NV12. When calculating the total plane size, the size of CbCr plane must also be considered. Signed-off-by:
Katsuya Matsubara <matsu@igel.co.jp> Signed-off-by:
-
Katsuya Matsubara authored
In the sh_veu driver, only the interrupt handler 'sh_veu_bh' can invoke the v4l2_m2m_job_finish() function. So the hardware must be alive for handling interrupts until returning from v4l2_m2m_ctx_release(). Signed-off-by:
-
Katsuya Matsubara authored
v4l2_m2m_job_finish() should be invoked even if the current ongoing job has been aborted since v4l2_m2m_ctx_release() which has issued the job abort may wait until the finish function is invoked. Signed-off-by:
-
Hans Verkuil authored
The clips pointer is a userspace pointer, not a kernelspace pointer, so you can't dereference the clips pointer. Also add a few missing commas and newlines. Signed-off-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
-
Hans Verkuil authored
The radio filter function that filters controls that are valid for a radio device should also accept V4L2_CTRL_CLASS_FM_RX controls. Signed-off-by:
-
- 08 Jun, 2013 6 commits
-
-
Hans Verkuil authored
This fixes a NULL pointer deference when loading the cx88_dvb module for a Hauppauge HVR4000. The bugzilla bug report is here: https://bugzilla.kernel.org/show_bug.cgi?id=56271 The cause is that the wm8775 is optional, so even though the board info says there is one, it doesn't have to be there. Checking whether the module was actually loaded is much safer. Note that this driver is quite buggy when it comes to unloading and reloading modules. Unloading cx8800 and reloading it again will still cause a crash, most likely because either the i2c bus isn't unloaded at the right time and/or the v4l2_device_unregister isn't called at the right time. Signed-off-by:
Sebastian Frei <sebastian@familie-frei.net> Signed-off-by:
-
Hans Verkuil authored
The version number was still 3.9: update to 3.10. Signed-off-by:
-
Sylwester Nawrocki authored
The "sysreg" clock is required by multiple subsystems and none of the other drivers handles this clock explicitly. It is currently assumed that this clock is always on, left in its default state after system reset. Remove handling of this clock from the FIMC-IS driver to avoid breaking other subsystems. Signed-off-by:
Kyungmin Park <kyungmin.park@samsung.com> Signed-off-by:
-
Sylwester Nawrocki authored
The FIMC-IS-ISP handles only Bayer formats thus V4L2_COLORSPACE_SRGB should be used. This change applies to the code first added in v3.10. Signed-off-by:
-
Sylwester Nawrocki authored
Use clk_prepare_enable/clk_unprepare_disable instead of preparing the clocks during the driver initalization and then using just clk_disable/ clk_enable. The clock framework doesn't guarantee a clock will not get enabled during e.g. clk_set_parent if clk_prepare has been called on it. So we ensure clk_prepare() is called only when it is safe to enable the clocks, i.e. the parent clocks and the clocks' frequencies are set. It must be ensured the FIMC-IS clocks have proper frequencies before they are enabled, otherwise the whole system will hang. Signed-off-by:
-
Sylwester Nawrocki authored
Ensure the firmware isn't accessed in the driver when the firmware loading routine has not completed. This fixes a potential kernel crash: [ 96.510000] Unable to handle kernel NULL pointer dereference at virtual address 00000000 [ 96.520000] pgd = ee604000 [ 96.520000] [00000000] *pgd=6e947831, *pte=00000000, *ppte=00000000 [ 96.530000] Internal error: Oops: 17 [#1] PREEMPT SMP ARM [ 96.530000] Modules linked in: [ 96.530000] CPU: 2 PID: 2787 Comm: camera_test Not tainted 3.10.0-rc1-00269-gcdbde37-dirty #2158 [ 96.545000] task: ee42e400 ti: edfcc000 task.ti: edfcc000 [ 96.545000] PC is at fimc_is_start_firmware+0x14/0x94 [ 96.545000] LR is at fimc_isp_subdev_s_power+0x13c/0x1f8 ... [ 96.745000] [<c03e0354>] (fimc_is_start_firmware+0x14/0x94) from [<c03e1cc4>] (fimc_isp_subdev_s_power+0x13c/0x1f8) [ 96.745000] [<c03e1cc4>] (fimc_isp_subdev_s_power+0x13c/0x1f8) from [<c03ed088>] (__subdev_set_power+0x70/0x84) [ 96.745000] [<c03ed088>] (__subdev_set_power+0x70/0x84) from [<c03ed164>] (fimc_pipeline_s_power+0xc8/0x164) [ 96.745000] [<c03ed164>] (fimc_pipeline_s_power+0xc8/0x164) from [<c03ee2b8>] (__fimc_pipeline_open+0x90/0x268) [ 96.745000] [<c03ee2b8>] (__fimc_pipeline_open+0x90/0x268) from [<c03ec5f0>] (fimc_capture_open+0xe4/0x1ec) [ 96.745000] [<c03ec5f0>] (fimc_capture_open+0xe4/0x1ec) from [<c03c5560>] (v4l2_open+0xa8/0xe4) [ 96.745000] [<c03c5560>] (v4l2_open+0xa8/0xe4) from [<c0112900>] (chrdev_open+0x9c/0x158) [ 96.745000] [<c0112900>] (chrdev_open+0x9c/0x158) from [<c010d3e0>] (do_dentry_open+0x1f4/0x27c) [ 96.745000] [<c010d3e0>] (do_dentry_open+0x1f4/0x27c) from [<c010d558>] (finish_open+0x34/0x50) [ 96.745000] [<c010d558>] (finish_open+0x34/0x50) from [<c011bea0>] (do_last+0x59c/0xbcc) [ 96.745000] [<c011bea0>] (do_last+0x59c/0xbcc) from [<c011c580>] (path_openat+0xb0/0x484) [ 96.745000] [<c011c580>] (path_openat+0xb0/0x484) from [<c011ca58>] (do_filp_open+0x30/0x84) [ 96.745000] [<c011ca58>] (do_filp_open+0x30/0x84) from [<c010d060>] (do_sys_open+0xe8/0x170) [ 96.745000] [<c010d060>] (do_sys_open+0xe8/0x170) from [<c000f040>] (ret_fast_syscall+0x0/0x30) Signed-off-by:
-
- 03 Jun, 2013 12 commits
-
-
Sachin Kamat authored
In certain cases, dma_alloc_coherent returns NULL. Add check for NULL pointer. Signed-off-by:
Sachin Kamat <sachin.kamat@linaro.org> Signed-off-by:
Kamil Debski <k.debski@samsung.com> Signed-off-by:
-
Andrzej Hajda authored
Many debug messages missed end-of-line. Signed-off-by:
Andrzej Hajda <a.hajda@samsung.com> Signed-off-by:
-
Andrzej Hajda authored
Callback .start_streaming is called once for every queue, so v4l2_ctrl_handler_setup was called twice during stream start. Moving v4l2_ctrl_handler_setup to context initialization reduces numbers of calls and seems to be more consistent with API. Signed-off-by:
-
Andrzej Hajda authored
This patch fixes a bug which caused overwriting h264 codec parameters by mpeg4 parameters during V4L2 control setting. Signed-off-by:
-
Arun Kumar K authored
MFC uses two clocks - MFC gate clock and special clock which is named as "sclk_mfc" in exynos4 and "aclk_333" in exynos5 SoC. The driver was doing just a clk_prepare on this special clock without a clk_enable call. As this sclk is the parent of gate clock, it gets prepared and enabled along with the gate clock. So there is no need for the driver to use this sclk. This patch removes the sclk usage from driver. Signed-off-by:
Arun Kumar K <arun.kk@samsung.com> Signed-off-by:
-
Sylwester Nawrocki authored
This patch fixes following compilation warning: CC [M] drivers/media/platform/s5p-mfc/s5p_mfc_opr_v6.o drivers/media/platform/s5p-mfc/s5p_mfc_opr_v6.c:1733:12: warning: ‘s5p_mfc_get_decoded_status_v6’ defined but not used It assigns existing but not used s5p_mfc_get_dec_status_v6() function to the get_dec_status callback. It seems the get_dec_status callback is not used anyway, as there is no corresponding s5p_mfc_hw_call(). Signed-off-by:
Sylwester Nawrocki <sylvester.nawrocki@gmail.com> Signed-off-by:
-
John Sheu authored
Save flags correctly when taking spinlocks in v4l2_m2m_try_schedule. Signed-off-by:
John Sheu <sheu@google.com> Signed-off-by:
-
Philipp Zabel authored
Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Philipp Zabel authored
Signed-off-by:
-
Seung-Woo Kim authored
The v4l2_m2m_poll() does not need to wait if there is already a buffer in done_list of source and destination queues, but current v4l2_m2m_poll() always waits. So done_list of each queue is checked before calling poll_wait(). Signed-off-by:
Seung-Woo Kim <sw0312.kim@samsung.com> Acked-by:
Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by:
-
Seung-Woo Kim authored
The vb2_poll() does not need to wait next vb_buffer_done() if there is already a buffer in done_list of queue, but current vb2_poll() always waits. So done_list is checked before calling poll_wait(). Signed-off-by:
-
Arun Kumar K authored
MFC v6 needs minimum number of output buffers to be queued for encoder depending on the stream type and profile. The patch modifies the driver so that encoding cannot be started with lesser number of OUTPUT buffers than required. This also fixes the crash happeninig during multi instance encoder-decoder simultaneous run due to memory allocation happening from interrupt context. Signed-off-by:
-
- 27 May, 2013 2 commits
-
-
Xiong Zhou authored
Make sure FONT_8x16 can be found by find_font(). Signed-off-by:
Xiong Zhou <jencce.kernel@gmail.com> Signed-off-by:
-
Lad, Prabhakar authored
from commit 3778d050 [media: davinci: kconfig: fix incorrect selects] VIDEO_VPFE_CAPTURE was removed but there was a negative dependancy for building the DM365 VPFE MC based capture driver (VIDEO_DM365_VPFE), This patch fixes this dependency by replacing the VIDEO_VPFE_CAPTURE with VIDEO_DM365_ISIF, so as when older DM365 ISIF v4l driver is selected the newer media controller driver for DM365 isnt visible. Reported-by:
Paul Bolle <pebolle@tiscali.nl> Signed-off-by:
Lad, Prabhakar <prabhakar.csengg@gmail.com> Signed-off-by:
-
- 21 May, 2013 12 commits
-
-
Sylwester Nawrocki authored
Ensure the compatible property for FIMC-LITE IP blocks is properly documented, a cut&paste error fix. Signed-off-by:
-
Axel Lin authored
Current code uses is->config_index as array subscript, thus the valid value range is 0 ... ARRAY_SIZE(cmd) - 1. Signed-off-by:
Axel Lin <axel.lin@ingics.com> Signed-off-by:
-
Axel Lin authored
Current code uses fie->index as array subscript, thus the valid value range is 0 ... ARRAY_SIZE(s5c73m3_intervals) - 1. Signed-off-by:
-
Sachin Kamat authored
'rotation' was an 8 bit variable and hence could not have values greater than 255. Since we need higher values, change it to 16 bit type. Signed-off-by:
-
Sachin Kamat authored
When 'node' is NULL, the print statement tries to dereference it. Hence replace the variable with the one that is accessible. Signed-off-by:
-
Hans Verkuil authored
Signed-off-by:
-
Hans Verkuil authored
The Codec section in the V4L2 specification was marked as 'suspended', even though codec support has been around for quite some time. Update this section, explaining a bit about memory-to-memory devices and pointing to the MPEG controls section. Signed-off-by:
-
Arnd Bergmann authored
It is not possible to select SND_SOC_SI476X if we have not also enabled SND_SOC. warning: (RADIO_SI476X) selects SND_SOC_SI476X which has unmet direct dependencies (SOUND && !M68K && !UML && SND && SND_SOC) Signed-off-by:
Arnd Bergmann <arnd@arndb.de> [hans.verkuil@cisco.com: fixed wrong driver name in subject] Acked-by:
-
Geert Uytterhoeven authored
Gcc < 4.3 doesn't understand binary constanrs (0b*): drivers/media/radio/radio-si476x.c:862:20: error: invalid suffix "b10000000" on integer constant Hence use a hexadecimal constant (0x*) instead. Signed-off-by:
Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by:
-
Wei Yongjun authored
Fix to return a negative error code from the error handling case instead of 0, as done elsewhere in this function. Signed-off-by:
Wei Yongjun <yongjun_wei@trendmicro.com.cn> Acked-by:
-
Lad, Prabhakar authored
The error path on failure was calling mutex_unlock(), but there was no actuall call before for mutex_lock(). This patch fixes this issue by pointing it to proper go label. Reported-by:
Jose Pablo Carballo <jose.carballo@ridgerun.com> Signed-off-by:
-
Lad, Prabhakar authored
For NV12 format, even if display data is single image, both VIDWIN0 and VIDWIN1 parameters must be used. The start address of Y data plane and C data plane is configured in VIDEOWIN0ADH/L and VIDEOWIN1ADH/L respectively. cuurently only one layer was requested, which is suffice for yuv422, but for yuv420(NV12) two layers are required and fix the same by requesting for other layer if pix fmt is NV12 during set_fmt. Signed-off-by:
-
