On 01/04/2018 01:11 PM, Tim Harvey wrote:
On Thu, Jan 4, 2018 at 10:51 AM, Steve Longerbeam <slongerbeam@xxxxxxxxx> wrote:
On 01/04/2018 09:57 AM, Tim Harvey wrote:
On Tue, Jan 2, 2018 at 6:00 PM, Steve Longerbeam <slongerbeam@xxxxxxxxx>
wrote:
Hi Tim,
Happy New Year! And pardon the delay.
On 12/28/2017 01:56 PM, Tim Harvey wrote:
Steve/Hans,
I'm trying to get interlaced capture working with the TDA1997x driver
I've been working on which is connected to an IMX6 CSI.
The particular board I'm currently testing on is an IMX6Q which has
both a TDA19971 HDMI receiver as well as an ADV7180 analog video
decoder. The media-ctl topology for this board can be found at
http://dev.gateworks.com/docs/linux/media/imx6q-gw54xx-media.png.
For adv7180 everything appears to be working as expected:
- media-ctl --get-v4l2 '"adv7180 2-0020":0' shows:
[fmt:UYVY8_2X8/720x480 field:interlaced colorspace:smpte170m]
- he following captures/streams from the adv7180 using the vdic to
de-interlace:
media-ctl --link "adv7180 2-0020":0 -> "ipu2_csi1_mux":1[1]
media-ctl --link "ipu2_csi1_mux":2 -> "ipu2_csi1":0[1]
media-ctl --link "ipu2_csi1":1 -> "ipu2_vdic":0[1]
media-ctl --link "ipu2_vdic":2 -> "ipu2_ic_prp":0[1]
media-ctl --link "ipu2_ic_prp":2 -> "ipu2_ic_prpvf":0[1]
media-ctl --link "ipu2_ic_prpvf":1 -> "ipu2_ic_prpvf capture":0[1]
media-ctl --set-v4l2 'adv7180 2-0020':0 [fmt:UYVY2X8/720x480]
media-ctl --set-v4l2 'ipu2_csi1_mux':2 [fmt:UYVY2X8/720x480
field:interlaced]
media-ctl --set-v4l2 'ipu2_csi1':1 [fmt:UYVY2X8/720x480
field:interlaced]
media-ctl --set-v4l2 'ipu2_vdic':2 [fmt:UYVY2X8/720x480
field:interlaced]
media-ctl --set-v4l2 'ipu2_ic_prp':2 [fmt:UYVY2X8/720x480 field:none]
media-ctl --set-v4l2 'ipu2_ic_prpvf':1 [fmt:UYVY2X8/720x480 field:none]
v4l2-ctl -d /dev/video3
--set-fmt-video=width=720,height=480,pixelformat=UYVY
# capture 1 frame
v4l2-ctl -d /dev/video1 --stream-mmap --stream-skip=1
--stream-to=/tmp/x.raw --stream-count=1
# stream jpeg/rtp
gst-launch-1.0 v4l2src device=/dev/video3 !
"video/x-raw,width=720,height=480,format=UYVY" !
jpegenc ! rtpjpegpay ! udpsink host=$SERVER port=5000"
For the tda1997x I'm trying to do something similar:
- media-ctl --get-v4l2 '"tda19971 2-0048":0' shows:
[fmt:UYVY8_1X16/1920x1080 field:alternate colorspace:srgb]
^^^^ still not sure V4L2_FIELD_ALTERNATE/SRGB returned from tda1997x
get_fmt is correct
- I setup the pipeline with:
media-ctl --link "tda19971 2-0048":0 -> "ipu1_csi0_mux":1[1]
media-ctl --link "ipu1_csi0_mux":2 -> "ipu1_csi0":0[1]
media-ctl --link "ipu1_csi0":1 -> "ipu1_vdic":0[1]
media-ctl --link "ipu1_vdic":2 -> "ipu1_ic_prp":0[1]
media-ctl --link "ipu1_ic_prp":2 -> "ipu1_ic_prpvf":0[1]
media-ctl --link "ipu1_ic_prpvf":1 -> "ipu1_ic_prpvf capture":0[1]
media-ctl --set-v4l2 'tda19971 2-0048':0[fmt:UYVY8_1X16/1920x1080]
media-ctl --set-v4l2 'ipu1_csi0_mux':2[fmt:UYVY8_1X16/1920x1080
field:alternate]
media-ctl --set-v4l2 'ipu1_csi0':1[fmt:UYVY8_1X16/1920x1080
field:alternate]
media-ctl --set-v4l2 'ipu1_vdic':2[fmt:UYVY8_1X16/1920x1080
field:alternate]
media-ctl --set-v4l2 'ipu1_ic_prp':2[fmt:UYVY8_1X16/1920x1080
field:none]
media-ctl --set-v4l2 'ipu1_ic_prpvf':1[fmt:UYVY8_1X16/1920x1080
field:none]
v4l2-ctl -d /dev/video1
--set-fmt-video=width=1920,height=1080,pixelformat=UYVY
v4l2-ctl -d /dev/v4l-subdev1 --set-dv-bt-timings=query
v4l2-ctl -d /dev/video1 --stream-mmap --stream-skip=1
--stream-to=/tmp/x.raw --stream-count=1
ipu1_csi0: bayer/16-bit parallel buses must go to IDMAC pad
ipu1_ic_prpvf: pipeline start failed with -22
VIDIOC_STREAMON: failed: Invalid argument
Steve,
Thanks for the help.
Right, according to the i.MX6 reference manual, if the CSI muxes
are receiving from the parallel bus input with width >= 16 bits,
that data can't be passed to the IC. It never really made much sense
to me, and I can't remember if I ever tried it, maybe not, because
I don't have such hardware.
hmmm... that's not good. I may have to dig into what's being done in
my 3.14 kernel with the Freescale capture driver where I can capture
1080p60 fine with my tda1997x driver there.
Try this hack as an experiment: modify is_parallel_16bit_bus() in
imx-media-csi.c to simply return false, and see if the above pipeline
works.
I'm currently on 4.15-rc1 which doesn't have a
'is_parallel_16bit_bus()' but if I comment out the check we are
talking about in csi_link_validate as such:
--- a/drivers/staging/media/imx/imx-media-csi.c
+++ b/drivers/staging/media/imx/imx-media-csi.c
@@ -999,6 +999,7 @@ static int csi_link_validate(struct v4l2_subdev *sd,
is_csi2 = (sensor_ep->bus_type == V4L2_MBUS_CSI2);
incc = priv->cc[CSI_SINK_PAD];
+/*
if (priv->dest != IPU_CSI_DEST_IDMAC &&
(incc->bayer || (!is_csi2 &&
sensor_ep->bus.parallel.bus_width >= 16))) {
@@ -1007,6 +1008,7 @@ static int csi_link_validate(struct v4l2_subdev *sd,
ret = -EINVAL;
goto out;
}
+*/
if (is_csi2) {
int vc_num = 0;
I get a pipeline start failure for ipu1_ic_prpvf:
root@ventana:~# v4l2-ctl -d /dev/video1 --stream-mmap --stream-skip=1
--stream-to=/tmp/x.raw --stream-count=1
[ 909.993353] tda1997x 2-0048: tda1997x_get_pad_format
[ 909.998342] tda1997x 2-0048: tda1997x_fill_format
^^^^ my tda1997x driver debug messages
[ 910.004483] ipu1_ic_prpvf: pipeline start failed with -32
VIDIOC_STREAMON: failed: Broken pipe
The driver doesn't really support V4L2_FIELD_ALTERNATE, the CSI subdev
attempts to translate this to sequential-top-bottom or sequential-bottom-top
at the CSI output pads, since alternate holds no info about field order. I
doubt
alternate is correct for the TDA19971 anyway, so that should be fixed in the
tda19971 driver. Until then, set to "seq-bt" downstream from the CSI, as in:
media-ctl --set-v4l2 'tda19971 2-0048':0[fmt:UYVY8_1X16/1920x1080]
media-ctl --set-v4l2 'ipu1_csi0_mux':2[fmt:UYVY8_1X16/1920x1080
field:alternate]
media-ctl --set-v4l2 'ipu1_csi0':1[fmt:UYVY8_1X16/1920x1080 field:alternate]
media-ctl --set-v4l2 'ipu1_vdic':2[fmt:UYVY8_1X16/1920x1080 field:seq-bt]
media-ctl --set-v4l2 'ipu1_ic_prp':2[fmt:UYVY8_1X16/1920x1080 field:none]
media-ctl --set-v4l2 'ipu1_ic_prpvf':1[fmt:UYVY8_1X16/1920x1080 field:none]
Steve,
This makes sense... I had a feeling that ALTERNATE was incorrect. HDMI
should be top/bot (with the exception of 720x480i (NTSC) which sends
bot then top). If I change my TDA1997x driver to set format->field =
V4L2_FIELD_SEQ_TB:
root@ventana:~# ./media-ctl --get-v4l2 '"tda19971 2-0048":0'
[fmt:UYVY8_1X16/1920x1080 field:seq-tb colorspace:srgb]
and
media-ctl --link "tda19971 2-0048":0 -> "ipu1_csi0_mux":1[1]
media-ctl --link "ipu1_csi0_mux":2 -> "ipu1_csi0":0[1]
media-ctl --link "ipu1_csi0":2 -> "ipu1_csi0 capture":0[1]
media-ctl --set-v4l2 'tda19971 2-0048':0[fmt:UYVY8_1X16/1920x1080]
media-ctl --set-v4l2 'ipu1_csi0_mux':2[fmt:UYVY8_1X16/1920x1080 field:seq-tb]
media-ctl --set-v4l2 'ipu1_csi0':2[fmt:UYVY8_1X16/1920x1080 field:none]
v4l2-ctl -d /dev/video4 --set-fmt-video=width=1920,height=1080,pixelformat=UYVY
v4l2-ctl -d /dev/v4l-subdev1 --set-dv-bt-timings=query
v4l2-ctl -d /dev/video4 --stream-mmap --stream-skip=1
--stream-to=/tmp/x.raw --stream-count=1
^^^ captures 4147200 byte frame yet not quite sure how to convert it (below)
convert -size 1920x1080 -depth 16 uyuv:x.raw frame.png
convert: delegate failed `"ufraw-batch" --silent --create-id=also
--out-type=png --out-depth=16 "--output=%u.png" "%i"' @
error/delegate.c/InvokeDelegate/1310.
convert: unable to open image `/tmp/magick-118907Itnr-kDtzA7.ppm': No
such file or directory @ error/blob.c/OpenBlob/2712.
convert: no images defined `frame.png' @
error/convert.c/ConvertImageCommand/3210.
^^^ not sure why I can't convert it using imagemagick but I can use
gstreamer instead:
gst-launch-1.0 v4l2src device=/dev/video4 num-buffers=1 !
video/x-raw,width=1920,height=1080,format=UYVY ! jpegenc ! filesink
location=/var/www/html/files/frame.jpg
^^^ captures 1 frame and jpeg's it
It's now capturing both fields to one frame but they are not lined up correctly:
see http://dev.gateworks.com/docs/linux/media/frame_1080p.jpg (1080p)
vs http://dev.gateworks.com/docs/linux/media/frame_1080i.jpg
So frame_1080p.jpgis what the image _should_ look like, and frame_1080i.jpg
is what was actually captured using the above gstreamer pipeline?
One thing you might try is set the field type at ipu1_csi0:2 to seq-tb,
which will
turn off the idmac de-interlacing, so you will get the top field and
bottom field
separated in the frame, maybe that will tell you something.
Also, fyi, I had forgotten that the IPU IC is only capable of outputting
scaled frames
up to 1024x1024, so you're earlier attempts to de-interlace 1920x1080i
with the
VDIC and IC paths failed with broken pipe error because the
ipu1_ic_prpvf renegotiated
output to 1024x1024p.
Steve
I'm wondering if I've got some issue with my sync signal
polarity/timing such that for interlaced sources one/both fields are
mis-aligned (but correct for progressive).
Tim
