| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| ndufresne | pinchartl: an opinion on Jung Zhaho way to force non-parsing VP8 decoder into the kernel ?
since nothing is documented (not even a commit message) I'm only guessing, but I have the impression his way is to pass the parsed vp8 frame through a CID before doing qbuf ndufresne would move parsed structure into it's own header | [00:00] |
| shuah | pinchartl: hey! it works! | [00:11] |
| .... (idle for 18mn) | ||
| pinchartl | shuah: \o/
ndufresne: do you have a pointer to that ? | [00:29] |
| ndufresne | should be on the ML archive, let me find a link
pinchartl: https://lkml.org/lkml/2016/1/26/119 The patch that adds public API for "unparsed" VP8 support, https://lkml.org/lkml/2016/1/26/118 the patch pretends to add VP8 support, but that's already there if your driver can transform parse the bitstream I believe tomorrow we'll get more feedback from the others | [00:37] |
| pinchartl | ndufresne: V4L2_CTRL_TYPE_VP8_FRAME_HDR isn't going to fly...
you're probably more knowledgeable than me when it comes to VP8 does the v4l2_ctrl_vp8_frame_hdr structure map to the VP8 spec, or is it specific to their hardware ? posciak: ping | [00:47] |
| ...... (idle for 25mn) | ||
| posciak | pinchartl: pong | [01:15] |
| pinchartl | posciak: I was wondering whether you had any comment regarding the VP8 patches that ndufresne mentioned | [01:18] |
| ........................................................................................ (idle for 7h16mn) | ||
| *** | hverkuil1 has left | [08:34] |
| .................................. (idle for 2h48mn) | ||
| smartin has quit IRC (Ping timeout: 244 seconds)
ribalda has quit IRC (Ping timeout: 244 seconds) UukGoblin has quit IRC (Ping timeout: 244 seconds) | [11:22] | |
| mchehab | javier__, pinchartl: how VBI is handled on OMAP3?
tvp5051 provides both raw and sliced VBI outputs (and tvp5151 also supports both) I'm not seeing any /dev/vbi? interface at https://mchehab.fedorapeople.org/mc-next-gen/igepv2_omap3isp.png btw, on em28xx, we only implement support for raw VBI. not sure if it is worth to map sliced VBI as a different PAD | [11:36] |
| javier__ | mchehab: yeah, that is not supported by the driver AFAICT | [11:40] |
| mchehab | does the hardware support?
(omap3, I mean) | [11:41] |
| javier__ | mchehab: yes, I got the question but I don't kow the answer. I was looking at the OMAP3 Technical Reference Manual
mchehab: I didn't find anything in the Camera ISP section, so that's a question for pinchartl or sailus who are familiar with omap3isp | [11:43] |
| mchehab | hint: on em28xx, raw VBI is provided by setting a few registers that tell the harware on what lines VBI is found:
vinctrl |= EM28XX_VINCTRL_VBI_RAW; em28xx_write_reg(dev, EM28XX_R34_VBI_START_H, 0x00); em28xx_write_reg(dev, EM28XX_R36_VBI_WIDTH, v4l2->vbi_width/4); em28xx_write_reg(dev, EM28XX_R37_VBI_HEIGHT, v4l2->vbi_height); if (v4l2->norm & V4L2_STD_525_60) { /* NTSC */ em28xx_write_reg(dev, EM28XX_R35_VBI_START_V, 0x09); } else if (v4l2->norm & V4L2_STD_625_50) { /* PAL */ em28xx_write_reg(dev, EM28XX_R35_VBI_START_V, 0x07); } sliced VBI means that the tvp5151 will decode it directly, and provide the decoded data on a separate bus so, basically, raw vbi is actually decoded inside omap3, while sliced vbi is decoded inside the TV decoder the userspace interface is different for sliced VBI and raw VBI on raw VBI, the interface is a sampled IF data sliced VBI outputs text | [11:47] |
| ......................................... (idle for 3h21mn) | ||
| found one issue with the way subdev entities are handled when adding tuner support on em28xx:
the FM1236MK3 has internally a tda9887 chip entity_6 [label="{{<pad_7> 0} | entity_6\ntuner\ntda9887 | {<pad_8> 1}}", shape=Mrecord, style=filled, fillcolor=lightblue] entity_9 [label="{{<pad_10> 0} | entity_9\ntuner\nPhilips NTSC MK3 (FM1236MK3 or FM1236/F) | {<pad_11> 1}}", shape=Mrecord, style=filled, fillcolor=lightblue] IMHO, it should be shown as just one tuner, and not mapped as two tuners as it doesn't make any sense on controlling the tda9887 individually | [15:11] | |
| .... (idle for 18mn) | ||
| hverkuil1 | mchehab: they are controlled independently. The tda9887 may be on the same die, but it acts as two ICs. | [15:30] |
| mchehab | hverkuil1: yes, that's true
but it makes harder to build the media graph ;) the problem is that, on devices with tda9887, the tuner is actually: PLL -> tda9887 where "PLL" is the entity_9 above and it has only one source PAD while tda9887 has two source PADs http://www.mouser.com/catalog/specsheets/FQ1216ME.pdf page 5 (one pad with audio IF and one paudio with video) (one pad with audio IF and one pad with video) | [15:30] |
| ........... (idle for 52mn) | ||
| dc786 | after upgrade to ubuntu 14.04.3 LTS, mencoder v4l2 capture drops around 75% of frames
using Hauppauge wintv, old TV tuner card result is file with fast-forward video | [16:26] |
| jmleo | dc786, what is the kernel version of your distribution (uname -r) | [16:41] |
| *** | benjiG has left | [16:49] |
| pinchartl | javier__: there's no VBI support | [16:51] |
| dc786 | jmleo, 3.13.0-76-generic | [16:52] |
| jmleo | and which tv card is it exactly (lspci will tell you) | [16:55] |
| dc786 | 0e:05.0 Multimedia video controller: Brooktree Corporation Bt878 Video Capture (rev 02)
and similar with .1 for audio on analog channels, the audio goes from output jack on wintv to line or mic in on audio card for composite, I use stereo audio on the audio card lspci -v : Subsystem: Hauppauge computer works Inc. WinTV Series Flags: bus master, medium devsel, latency 132, IRQ 17 Memory at d8800000 (32-bit, prefetchable) [size=4K] Kernel driver in use: bttv same mencoder command worked OK with Ubuntu 12.04 well, that is, if I got pulseaudio and alsa to cooperate on getting audio so I updated for more consistent audio capability too bad, now video is broken | [16:58] |
| jmleo | are you sure it is a 3.13.0-76 in ubuntu 14.04.3 LTS ? according to the release note it should be a 3.19... | [17:10] |
| dc786 | I've seen that, too, but as of yesterday update manager says there are no updates needed
I read about a low-latency version, know anything about that? | [17:19] |
| ............... (idle for 1h11mn) | ||
| javier__ | pinchartl: thanks for the confirmation | [18:30] |
| ................... (idle for 1h30mn) | ||
| *** | awalls1 has left | [20:00] |
| .... (idle for 16mn) | ||
| dc786 has left "Leaving" | [20:16] | |
| .... (idle for 15mn) | ||
| mchehab | hverkuil: ok, I added the PLL + tda9887: https://mchehab.fedorapeople.org/mc-next-gen/wintv_usb2.png
(this simple device doesn't have DVB nor VBI) nor radio http://www.amazon.com/Hauppauge-1020-WinTV-USB2-External-Tuner/dp/B0006FS2HM | [20:31] |
| *** | awalls1 has left | [20:42] |
| pfallenop has quit IRC (Ping timeout: 250 seconds) | [20:47] | |
| ........ (idle for 37mn) | ||
| bparrot__ | hverkuil, ping
hverkuil, with "echo 1 >/sys/class/video4linux/video1/dev_debug" i can enable ioctl trace level. but is their a way to see the debug trace between a bridge and a subdevice driver with in a similar way? s/their/there/ | [21:24] |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
