This patchset upstreams some changes carried in the ChromeOS kernel tree
especially regarding the s5p-mfc video encoder and gsc-m2m color converter
hardware functionality.
Patch 4 in particular affects the V4L2 interface by allowing VIDIOC_REQBUFS(0)
to succeed when the queue is of type V4L2_MEMOR
k lifetime
appropriately.
Remove the specific check for type V4L2_MEMOMRY_MMAP when freeing all
buffers through VIDIOC_REQBUFS.
Signed-off-by: John Sheu
---
drivers/media/v4l2-core/videobuf2-core.c | 26 +-
1 file changed, 1 insertion(+), 25 deletions(-)
diff --git a/drivers/
(V4L2_CID_MPEG_VIDEO_BITRATE)
* (h264) frame type (V4L2_CID_MPEG_MFC51_VIDEO_FORCE_FRAME_TYPE)
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/regs-mfc-v6.h| 4 ++
drivers/media/platform/s5p-mfc/s5p_mfc_common.h | 31 ++--
drivers/media/platform/s5p-mfc/s5p_mfc_enc.c| 53
VIDIOC_STREAMOFF clears the encoder's destination queue -- routines run
from the interrupt handler cannot assume that the queue is non-empty.
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/s5p_mfc_enc.c | 20 +++-
1 file changed, 11 insertions(+), 9 dele
Explicitly specify sampling period for subsampled chroma formats, so
stride and image size are properly reported through VIDIOC_{S,G}_FMT.
Signed-off-by: John Sheu
---
drivers/media/platform/exynos-gsc/gsc-core.c | 154 +++
drivers/media/platform/exynos-gsc/gsc-core.h
On Thu, Oct 17, 2013 at 3:54 PM, Sylwester Nawrocki
wrote:
> On 10/18/2013 12:25 AM, John Sheu wrote:
>> On Thu, Oct 17, 2013 at 2:46 PM, John Sheu wrote:
>>> > Sweet. Thanks for spelling things out explicitly like this. The fact
>>> > that the CAPTURE and O
On Thu, Oct 17, 2013 at 2:46 PM, John Sheu wrote:
> Sweet. Thanks for spelling things out explicitly like this. The fact
> that the CAPTURE and OUTPUT queues "invert" their sense of "crop-ness"
> when used in a m2m device is definitely all sorts of confusing.
Jus
t the CAPTURE and OUTPUT queues "invert" their sense of "crop-ness"
when used in a m2m device is definitely all sorts of confusing.
I'll have an updated patch shortly.
-John Sheu
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of
AM, John Sheu wrote:
>> On Wed, Oct 9, 2013 at 11:49 PM, Hans Verkuil wrote:
>> In all cases, the crop boundary refers to the area in the source
>> image; for a CAPTURE device, this is the (presumably analog) sensor,
>
> Correct.
>
>> and for an OUTPUT device, this
same operation; cropping on the sink image, i.e.
whatever memory buffer I'm providing to the CAPTURE queue.
Not sure what you mean by "S_CROP on an output video node doesn't
crop, it composes", though.
Thanks,
-John Sheu
--
To unsubscribe from this list: send the line "
V4L2_CID_MPEG_MFC51_VIDEO_DECODER_H264_DISPLAY_DELAY_ENABLE is being
ignored, and the display delay is always being applied. Fix this.
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/s5p_mfc_opr_v6.c | 7 ++-
1 file changed, 2 insertions(+), 5 deletions(-)
diff --git a/drivers
These are various video processing driver fixes for the s5p-mfc and gsc-m2m
hardware blocks on the Samsung Exynos (ARM) platform, that have been carried
in the ChromeOS kernel tree for a while and should be pushed upstream.
John Sheu (6):
[media] s5p-mfc: fix DISPLAY_DELAY
[media] s5p-mfc
Allow userspace to set the crop rect of the input image buffer to
encode.
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/s5p_mfc_common.h | 6 ++-
drivers/media/platform/s5p-mfc/s5p_mfc_dec.c| 7 ++--
drivers/media/platform/s5p-mfc/s5p_mfc_enc.c| 54
Explicitly specify sampling period for subsampled chroma formats, so
stride and image size are properly reported through VIDIOC_{S,G}_FMT.
Signed-off-by: John Sheu
---
drivers/media/platform/exynos-gsc/gsc-core.c | 154 +++
drivers/media/platform/exynos-gsc/gsc-core.h
(V4L2_CID_MPEG_VIDEO_BITRATE)
* (h264) frame type (V4L2_CID_MPEG_MFC51_VIDEO_FORCE_FRAME_TYPE)
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/regs-mfc-v6.h| 4 ++
drivers/media/platform/s5p-mfc/s5p_mfc_common.h | 32 +++--
drivers/media/platform/s5p-mfc/s5p_mfc_enc.c| 53
k lifetime
appropriately.
Remove the specific check for type V4L2_MEMOMRY_MMAP when freeing all
buffers through VIDIOC_REQBUFS.
Signed-off-by: John Sheu
---
drivers/media/v4l2-core/videobuf2-core.c | 26 +-
1 file changed, 1 insertion(+), 25 deletions(-)
diff --git a/drivers/
VIDIOC_STREAMOFF clears the encoder's destination queue -- routines run
from the interrupt handler cannot assume that the queue is non-empty.
Signed-off-by: John Sheu
---
drivers/media/platform/s5p-mfc/s5p_mfc_enc.c | 20 +++-
1 file changed, 11 insertions(+), 9 dele
From: John Sheu
The video encode/decode paths have duplicated logic between
VIDIOC_TRY_FMT and VIDIOC_S_FMT that should be de-duped. Also, video
decode reports V4L2_PIX_FMT_NV12MT_16X16 output format, regardless of
what the actual output has been set at. Fix this.
Signed-off-by: John Sheu
Save flags correctly when taking spinlocks in v4l2_m2m_try_schedule.
Signed-off-by: John Sheu
---
drivers/media/v4l2-core/v4l2-mem2mem.c | 19 +++
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/drivers/media/v4l2-core/v4l2-mem2mem.c
b/drivers/media/v4l2-core
Arun Kumar K samsung.com> writes:
> MFC v6 needs minimum number of capture buffers to be queued
> for encoder depending on the stream type and profile.
> For achieving this the sequence for allocating buffers at
> the encoder is modified similar to that of decoder.
> The new sequence is as follow
From: John Sheu
Callers to dma_buf_mmap expect to fput() the vma struct's vm_file
themselves on failure. Not restoring the struct's data on failure
causes a double-decrement of the vm_file's refcount.
Signed-off-by: John Sheu
---
drivers/base/dma-buf.c | 21 +++
From: John Sheu
When a v4l2-mem2mem context gets a STREAMOFF call on either its CAPTURE
or OUTPUT queues, we should:
* Drop the corresponding rdy_queue, since a subsequent STREAMON expects
an empty queue.
* Deschedule the context, as it now has at least one empty queue and
cannot run
From: John Sheu
Callers to dma_buf_mmap expect to fput() the vma struct's vm_file
themselves on failure. Not restoring the struct's data on failure
causes a double-decrement of the vm_file's refcount.
Signed-off-by: John Sheu
---
drivers/base/dma-buf.c | 18 ++
From: John Sheu
In v4l2_m2m_try_schedule(), use the CAPTURE queue lock when accessing
the CAPTURE queue, instead of relying on just holding the OUTPUT queue
lock.
Signed-off-by: John Sheu
---
drivers/media/v4l2-core/v4l2-mem2mem.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers
24 matches
Mail list logo