- 19 Feb, 2016 40 commits
-
-
Laurent Pinchart authored
Implement internal control of the VSP pipeline to be used by the DU DRM/KMS driver when using the VSP as an internal composer handled through DRM/KMS only. Signed-off-by:
Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@osg.samsung.com>
-
Laurent Pinchart authored
As the pipeline is configured internally by the driver when the userspace API is disabled its configuration can be trusted and link validation isn't needed. Signed-off-by:
-
Laurent Pinchart authored
The RPF and WPF alpha values are set through V4L2 controls and applied when starting the video stream by a call to v4l2_ctrl_handler_setup(). As that function uses the control handler mutex it can't be called in interrupt context, where the VSP+DU pipeline handler might need to reconfigure the pipeline. Set the alpha value manually in the RPF and WPF s_stream handler to ensure that the hardware is properly configured even when controlled without the userspace API. If the userspace API is enabled protect that with the control lock to avoid race conditions with userspace. Signed-off-by:
-
Laurent Pinchart authored
The pipeline inputs array stores pointers to all RPFs contained in the pipeline. It's currently indexed contiguously by adding RPFs in the order they are found during graph walk. This can't easily support dynamic addition and removal of RPFs while streaming, which will be required for combined VSP+DU support. Make the array indexed by RPF index instead and skip NULL elements when iterating over RPFs. Signed-off-by:
-
Laurent Pinchart authored
The R-Car Gen3 SoCs include VSP instances dedicated to the DU that will be controlled entirely by the rcar-du-drm driver through the KMS API. To support that use case make the userspace V4L2 API optional. Signed-off-by:
-
Laurent Pinchart authored
Format information and the related helper function are not specific to the V4L2 API, move them from vsp1_video.c to vsp1_pipe.c. Signed-off-by:
-
Laurent Pinchart authored
Not all VSP instances have a BRU on R-Car Gen3, make it optional. Set the feature unconditionally for now, this will be fixed when adding Gen3 support. Signed-off-by:
-
Laurent Pinchart authored
The R-Car Gen3 family has 5-inputs BRUs, support them by making the number of BRU inputs configurable. As the driver assumes that the number of BRU inputs is equal to the number of RPFs, replace the BRU_MAX_INPUTS macro with VSP1_MAX_RPF to make the assumption apparent. Signed-off-by:
-
Laurent Pinchart authored
The function will be used by the DU code, move it out of vsp1_video.c. Signed-off-by:
-
Laurent Pinchart authored
Several module read functions are not used, remove them. Signed-off-by:
-
Laurent Pinchart authored
Commit 58f896d8 ("[media] v4l: vsp1: sru: Make the intensity controllable during streaming") refactored the stream start code and removed the SRU CTRL0 register write by mistake. Add it back. Fixes: 58f896d8 ("[media] v4l: vsp1: sru: Make the intensity controllable during streaming") Signed-off-by:
-
Laurent Pinchart authored
Rename the VI6_DISP_IRQ_STA_DSE register bit to VI6_DISP_IRQ_STA_DST to fix a typo and match the datasheet. Signed-off-by:
-
Laurent Pinchart authored
Signed-off-by:
-
Laurent Pinchart authored
Link creation will be handled differently for the DU pipeline. Signed-off-by:
-
Laurent Pinchart authored
No need to waste CPU cycles when the value we need is already available. Signed-off-by:
-
Laurent Pinchart authored
The code will be reused outside of vsp1_video.c. Signed-off-by:
-
Laurent Pinchart authored
Those functions are specific to video nodes, rename them for consistency. Signed-off-by:
-
Laurent Pinchart authored
The code will be used to control the vsp1 driver from the DU driver without using video nodes. Signed-off-by:
-
Laurent Pinchart authored
To make the pipeline structure and operations usable without video devices the frame end processing must be decoupled from struct vsp1_video. Implement this by calling the video frame end function indirectly through a function pointer in struct vsp1_pipeline. Signed-off-by:
-
Laurent Pinchart authored
The field is always equal to the num_inputs field plus one, remove the duplicate. Signed-off-by:
-
Laurent Pinchart authored
Only RPFs and WPFs can be associated with video nodes, don't waste memory by storing the video pointer in all entities. Signed-off-by:
-
Laurent Pinchart authored
Not all VSP1 instances include a UDS. Make the renesas,#uds DT property optional and accept a number of UDS equal to 0 as valid. Signed-off-by:
-
Laurent Pinchart authored
The rwpf queue operation doesn't queue a buffer but sets the memory address for the next run. Rename it to set_memory and pass it a new structure independent of the video buffer than only contains memory information. Signed-off-by:
-
Laurent Pinchart authored
To make the video device nodes optional we need to decouple the [rw]pf instances from the video devices. Move video devices out of struct vsp1_rwpf and instantiate them dynamically in the core driver code. Signed-off-by:
-
Laurent Pinchart authored
The structure represent a vsp1 videobuf2 buffer, name it accordingly. Signed-off-by:
-
Laurent Pinchart authored
This removes the dependency of vsp1_rpf and vsp1_wpf on vsp1_video, making it possible to reuse the operations without a V4L2 video device node. Signed-off-by:
-
Laurent Pinchart authored
Move the format from struct vsp1_video to struct vsp1_rwpf to prepare for VSPD KMS support that will not instantiate V4L2 video device nodes. Signed-off-by:
-
Laurent Pinchart authored
The rwpf field contains a pointer to the rpf or wpf associated with the video node. Instead of storing it as a vsp1_entity, store the corresponding vsp1_rwpf pointer to allow accessing the vsp1_rwpf fields directly. Signed-off-by:
-
Laurent Pinchart authored
There's no need to spread the code across multiple source files. Signed-off-by:
-
Laurent Pinchart authored
Tri-planar memory formats store the Y, U and V components in separate planes. The VSP hardware supports them, the driver now does too. Signed-off-by:
-
Laurent Pinchart authored
The formats use three planes through the multiplanar API, allowing for non-contiguous planes in memory. Signed-off-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
-
Benoit Parrot authored
Fix the following sparse warnings: ti-vpe/cal.c:387:26: warning: incorrect type in return expression (different address spaces) ti-vpe/cal.c:459:26: warning: incorrect type in return expression (different address spaces) ti-vpe/cal.c:503:27: warning: incorrect type in argument 6 (different address spaces) ti-vpe/cal.c:509:47: warning: incorrect type in argument 6 (different address spaces) ti-vpe/cal.c:518:47: warning: incorrect type in argument 6 (different address spaces) ti-vpe/cal.c:526:31: warning: incorrect type in argument 6 (different address spaces) ti-vpe/cal.c:1807:24: warning: Using plain integer as NULL pointer ti-vpe/cal.c:1844:16: warning: Using plain integer as NULL pointer Signed-off-by:
Benoit Parrot <bparrot@ti.com> Signed-off-by:
-
Wu-Cheng Li authored
There is a new control V4L2_CID_MPEG_VIDEO_FORCE_KEY_FRAME to force an encoder key frame. It is the same as requesting V4L2_MPEG_MFC51_VIDEO_FORCE_FRAME_TYPE_I_FRAME. Signed-off-by:
Wu-Cheng Li <wuchengli@chromium.org> Acked-by:
Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by:
-
Wu-Cheng Li authored
Some drivers also need a control like V4L2_CID_MPEG_MFC51_VIDEO_FORCE_FRAME_TYPE to force an encoder key frame. Add a general V4L2_CID_MPEG_VIDEO_FORCE_KEY_FRAME so the new drivers and applications can use it. Signed-off-by:
-
Hans de Goede authored
bytesperline should be the bytesperline for the first plane for planar formats, not that of all planes combined. This fixes a crash in xawtv caused by the wrong bpl. BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1305389Reported-and-tested-by:
Stas Sergeev <stsp@list.ru> Cc: stable@vger.kernel.org Signed-off-by:
Hans de Goede <hdegoede@redhat.com> Signed-off-by:
-
Sudip Mukherjee authored
The build of m32r allmodconfig fails with the error: drivers/media/v4l2-core/videobuf2-dma-contig.c:492:28: error: implicit declaration of function 'dma_get_cache_alignment' The build of videobuf2-dma-contig.c depends on HAS_DMA and it is correctly mentioned in the Kconfig but the symbol VIDEO_TI_CAL also selects VIDEOBUF2_DMA_CONTIG, so it is trying to compile videobuf2-dma-contig.c even though HAS_DMA is not defined. Fixes: 343e89a7 ("[media] media: ti-vpe: Add CAL v4l2 camera capture driver") Cc: Benoit Parrot <bparrot@ti.com> Signed-off-by:
Sudip Mukherjee <sudip@vectorindia.org> Signed-off-by:
-
Hans Verkuil authored
The Timberdale FPGA video driver has not seen any real development since 2011 (and very little before that). One of the problems with the timblogiw driver is that it uses videobuf instead of the newer vb2 framework. The long term goal is to either convert or remove any driver still using videobuf. Since none of the core v4l developers has the hardware, we cannot convert it ourselves. As far as I can tell it was only used in an Intel demo board in 2009 using Meego: http://www.chinait.com/intelcontent/intelprc/admin/PDFFile/20106411545.pdf which has since been superseded. Moving this driver to staging is the first step towards removal. After 2 or 3 kernel cycles it will be removed altogether unless someone steps up to clean up this driver. Signed-off-by:
Richard Röjfors <richard@puffinpack.se> Signed-off-by:
-
Hans Verkuil authored
The V4L2_CID_TX_EDID_PRESENT control reports if an EDID is present. The adv7511 however still reported the EDID present after disconnecting the HDMI cable. Fix the logic regarding this control. And when the EDID is disconnected also call ADV7511_EDID_DETECT to notify the bridge driver. This was also missing. Signed-off-by:
-
Dan Carpenter authored
It makes sense to make the min_t() cast unsigned here since we don't really want negative sizes. Making it signed causes a static checker warning in Smatch. Smatch knows "fw->size - i" is positive but it doesn't know that fw->size is less than INT_MAX so in theory casting it to int might lead to a negative. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Insu Yun authored
Since kthread_run returns -ENOMEM if failed, it needs to be checked whether it is error, not whether it is null. Signed-off-by:
Insu Yun <wuninsu@gmail.com> Signed-off-by:
-
