#linux-sunxi on 2024-01-10 — irc logs at oftc.irclog.whitequark.org

2022-12-21 00:44 ChanServ changed the topic of #linux-sunxi to: Allwinner/sunxi development - Did you try looking at our wiki? https://linux-sunxi.org - Don't ask to ask. Just ask and wait for an answer! - This channel is logged at https://oftc.irclog.whitequark.org/linux-sunxi

00:06 machinehum has joined #linux-sunxi

00:14 machinehum has quit [Ping timeout: 480 seconds]

00:23 machinehum has joined #linux-sunxi

00:36 ftg has quit [Read error: Connection reset by peer]

01:03 Hypfer has quit [Ping timeout: 480 seconds]

01:06 machinehum has quit [Ping timeout: 480 seconds]

01:16 machinehum has joined #linux-sunxi

01:22 Hypfer has joined #linux-sunxi

01:24 machinehum has quit [Ping timeout: 480 seconds]

01:49 machinehum has joined #linux-sunxi

01:57 pg12_ has joined #linux-sunxi

01:57 pg12 has quit [Remote host closed the connection]

02:00 machinehum has quit [Ping timeout: 480 seconds]

02:02 pg12_ is now known as pg12

02:03 indy has quit [Read error: Connection reset by peer]

02:05 indy_ has joined #linux-sunxi

02:08 montjoie has joined #linux-sunxi

02:09 machinehum has joined #linux-sunxi

02:10 montjoie_ has quit [Ping timeout: 480 seconds]

02:17 machinehum has quit [Ping timeout: 480 seconds]

02:27 machinehum has joined #linux-sunxi

03:05 Hypfer has quit [Ping timeout: 480 seconds]

03:34 machinehum has quit [Ping timeout: 480 seconds]

03:40 Hypfer has joined #linux-sunxi

03:59 machinehum has joined #linux-sunxi

04:02 Hypfer has quit [Ping timeout: 480 seconds]

04:06 Hypfer has joined #linux-sunxi

04:09 machinehum has quit [Ping timeout: 480 seconds]

04:34 chewitt has quit [Ping timeout: 480 seconds]

04:35 machinehum has joined #linux-sunxi

04:41 hexdump0815 has joined #linux-sunxi

04:43 hexdump01 has quit [Ping timeout: 480 seconds]

04:45 machinehum has quit [Ping timeout: 480 seconds]

04:55 machinehum has joined #linux-sunxi

05:00 advancel has joined #linux-sunxi

05:04 machinehum has quit [Ping timeout: 480 seconds]

05:13 machinehum has joined #linux-sunxi

05:17 chewitt has joined #linux-sunxi

05:22 machinehum has quit [Ping timeout: 480 seconds]

05:24 JohnDoe_71Rus has joined #linux-sunxi

05:33 machinehum has joined #linux-sunxi

05:54 gsz has joined #linux-sunxi

06:29 macromorgan has quit [Read error: Connection reset by peer]

06:31 macromorgan has joined #linux-sunxi

06:47 DonkeyHotei has quit [Ping timeout: 480 seconds]

06:56 DonkeyHotei has joined #linux-sunxi

07:19 gsz has quit [Ping timeout: 480 seconds]

07:41 machinehum has quit [Ping timeout: 480 seconds]

07:49 machinehum has joined #linux-sunxi

07:58 machinehum has quit [Ping timeout: 480 seconds]

08:13 warpme has joined #linux-sunxi

08:13 * advancel Trying to play video on 6.1 Kernel with Allwinner A13. Using v4l2_request api and vlc.

08:13 <advancel> Hey i'm getting this error on VLC. Is there any advice for it? Error; "glconv_vaapi_x11 gl error: vaAcquireBufferHandle: an unsupported memory type was supplied"

08:21 gsz has joined #linux-sunxi

08:25 machinehum has joined #linux-sunxi

08:35 machinehum has quit [Ping timeout: 480 seconds]

08:44 <gamiee> Why it tries to use vaapi, when it should use v4l2 req API?

08:45 machinehum has joined #linux-sunxi

08:54 machinehum has quit [Ping timeout: 480 seconds]

09:00 machinehum has joined #linux-sunxi

09:09 machinehum has quit [Ping timeout: 480 seconds]

09:14 gsz is now known as Guest13578

09:14 gsz has joined #linux-sunxi

09:16 Guest13578 has quit [Ping timeout: 480 seconds]

09:20 machinehum has joined #linux-sunxi

09:27 tnovotny has joined #linux-sunxi

09:29 machinehum has quit [Ping timeout: 480 seconds]

09:45 gsz has quit [Ping timeout: 480 seconds]

09:54 machinehum has joined #linux-sunxi

10:00 <advancel> gamiee: Because V4L2_request is vaapi based ofc if i'm not wrong.

10:10 machinehum has quit [Ping timeout: 480 seconds]

10:15 <gamiee> advancel: no it's not?

10:15 <gamiee> (afaik)

10:19 machinehum has joined #linux-sunxi

10:24 <advancel> gamiee: "A media player that supports VAAPI (such as VLC) can then be used to decode a video in a supported format" quoted from v4l2_request github page.

10:31 <gamiee> advancel: aren't you trying to use that old, not-maintained V4L2 Req VAAPI bridge?

10:32 <gamiee> https://github.com/bootlin/libva-v4l2-request

10:35 <advancel> gamiee: Ofc i mean that one because i couldn't create a player that works with ffmpeg with the v4l2 patch

10:36 <gnarface> ffplay doesn't?

10:36 <advancel> gamiee: i tried mpv/vlc/QtAv/ffplay but no luck. None of them work.

10:38 <advancel> just green glitches on screen. I told this to Jernej and he told me please check your cedrus frequency maybe you should try lower freq(but i didn't try yet).

10:40 <advancel> i uploaded the screenshot of player. There it is ; https://snipboard.io/CfXTjW.jpg

10:46 machinehum2 has joined #linux-sunxi

10:51 <gamiee> advancel: note that without patched ffmpeg, V4L2 Req api will not work

10:51 <gamiee> gstreamer should have already v4L2 req support

10:52 machinehum has quit [Ping timeout: 480 seconds]

10:54 <advancel> gamiee: Yes i know the patch also i got from Jernej's repo(6.1.1). I couldn't make the gstreamer work.

11:09 warpme has quit []

11:20 exkc has joined #linux-sunxi

11:28 dsimic is now known as Guest13591

11:28 dsimic has joined #linux-sunxi

11:30 Guest13591 has quit [Ping timeout: 480 seconds]

11:33 warpme has joined #linux-sunxi

12:26 warpme has quit []

12:28 JohnDoe_71Rus has quit []

13:00 bauen1 has quit [Ping timeout: 480 seconds]

13:13 machinehum2 has quit [Ping timeout: 480 seconds]

13:17 machinehum has joined #linux-sunxi

13:18 vertex004 has joined #linux-sunxi

13:39 gsz has joined #linux-sunxi

13:39 JohnDoe_71Rus has joined #linux-sunxi

13:57 macromorgan has quit [Quit: Leaving]

13:59 macromorgan has joined #linux-sunxi

14:19 bauen1 has joined #linux-sunxi

14:27 <paulk> hey folks

14:28 <paulk> trying to get jpeg decode going with the v536 libcedarc on a33

14:28 <paulk> https://github.com/lindenis-org/lindenis-v853-external/blob/master/libcedarx/demo/jpegdemo/demojpeg.c#L683

14:28 <paulk> makes so much sense...

14:28 <paulk> anyway looks like the jpeg hw changed a bit between generations, I can't get it to decode

14:29 <paulk> does anyone know of known working libcedar + blobs for a33?

14:29 <linkmauve> Speaking of which, do you want me to send this series again, which didn’t work on either A10, A20 or A64? https://linkmauve.fr/dev/linux/cedrus-jpeg/

14:32 <paulk> linkmauve: I guess I'll get back to it when working on implemented jpeg dec in cedrus

14:32 <paulk> no need to send it if it doesn't work yet

14:32 <linkmauve> Alright.

14:37 warpme has joined #linux-sunxi

14:39 <gamiee> paulk: hi, I think best would be just to find some BSP and use it from there. btw. we found more latest cedar with more symbols.

14:40 <gamiee> https://github.com/Tina-Linux/d1s-melis/tree/master/elibrary/media/libcedarc/library

14:41 <paulk> nice

14:41 <paulk> there used to be the actual jpeg code for melis in an earlier release

14:41 apritzel has joined #linux-sunxi

14:42 <paulk> https://gitee.com/grantho/d1s-melis/tree/master/emodules/mod_jpeg_dec

14:44 <paulk> anyway I'm a bit surprised the mpeg decode side changed for v536

14:49 <gamiee> does it work on v3?

14:51 <paulk> mhh didn't try it

14:53 <gamiee> would be surprised if there is big diff between v3 and a33

14:54 <gamiee> btw paulk : is there big difference between v3/s3 and V536/V85X H.264 encoder cores?

15:11 warpme has quit []

15:17 warpme has joined #linux-sunxi

15:18 machinehum has quit [Ping timeout: 480 seconds]

15:42 jernej has quit [Remote host closed the connection]

15:43 jernej has joined #linux-sunxi

15:47 gsz has quit [Ping timeout: 480 seconds]

15:56 vertex004 has quit [Remote host closed the connection]

16:22 <paulk> gamiee: I haven't really looked closely at v536+ but it seems that the h264 enc has evolved quite a lot over time

16:22 <paulk> there is a significant difference between v3 and a20 at least

16:28 buZz has quit [Quit: Reconnecting]

16:28 buZz has joined #linux-sunxi

16:29 tnovotny has quit [Quit: Leaving]

16:46 warpme has quit []

16:59 vagrantc has joined #linux-sunxi

17:39 machinehum has joined #linux-sunxi

17:47 machinehum has quit [Ping timeout: 480 seconds]

17:49 machinehum has joined #linux-sunxi

17:51 <jernej> paulk: h264 encoding is the only process that radically changed between cedrus generations

17:51 <jernej> all decoding processes only had incremental changes, most of which are backward compatible

17:51 <paulk> jernej: yeah that's why I'm surprised the jpeg one doesn't work ootb

17:52 <paulk> I means it's the mpeg engine

17:52 <paulk> there's about no reason at all why it would change

17:52 <paulk> anyway I'm not yet done investigating what's happening with this A33 stuff

17:52 <paulk> I'll add I/O tracing soon

17:52 <paulk> that should help

17:53 <jernej> to CedarC blobs?

17:53 <paulk> and maybe it can work to return a fake VE version number too

17:53 <paulk> jernej: I'm using page fault trapping kernel-side

17:53 <paulk> so it just prints the I/O to the VE MMIO range

17:54 <jernej> ok, that's way more advanced that I used for RE

17:54 <paulk> and forwards it to another virt memory area where the range is actually mapped

17:54 <paulk> jernej: do you have some sort of I/O tracing going?

17:54 <paulk> I didn't really see any other way

17:54 <paulk> since it's not doing syscalls at all

17:54 <paulk> just mmap and direct access

17:59 <jernej> paulk: I just add some code to https://github.com/jernejsk/linux/blob/master/drivers/media/cedar-ve/cedar_ve.c#L627

17:59 <paulk> ohh like you dump current reg values?

17:59 <jernej> which opened a test file and dumped register values

18:00 <paulk> yeah that's a lot more reasonable

18:00 <jernej> It's way too much data for dmesg

18:01 <paulk> having the exact I/O is very readable though

18:01 <paulk> indeed dumping full ranges is a pain

18:01 <paulk> but you could also just cache and print what changed

18:01 <jernej> I almost figured out VC-1 a while back. It has still some issues though

18:02 <paulk> hehe

18:02 <jernej> eh, I added same full dump code to both sides. with some post-processing (removing volatile registers), it's easy to compare

18:03 <jernej> still, there is usually a lot of difference, since some bits are often don't care if feature isn't used

18:05 <jernej> and more I'm doing this, more I see that register values aren't all. It seems that parsing functionality *has* to be used to get properly decoded video.

18:05 apritzel has quit [Ping timeout: 480 seconds]

18:06 <jernej> That's the case for H264, H265 (rare cases), but in VP8 and VC-1 for most if not all

18:14 <jernej> paulk: I believe this source is from A33 era: https://github.com/allwinner-zh/media-codec/tree/master/sunxi-cedarx/SOURCE/plugin/vdecoder/mjpeg

18:17 <gamiee> Well, I think it makes sense that H.264 / HEVC encoder core is heavily improved

18:18 <gamiee> Since the improvements to encoder efficiency, together with ISP etc... Where on decoder side, you just decode it, you can only improve support for HDR, etc.

18:19 <paulk> jernej: yeah I'm guessing there is some internal state stuff we're not seeing which can matter in subtle cases

18:19 <paulk> jernej: thanks, I didn't remember jpeg dec was part of that code drop

18:28 <paulk> the code is dated from 2014/2015

18:28 <paulk> and A33 came out in Q1 2014

18:28 <paulk> could work out :)

18:30 <jernej> I believe code from linkmauve should be close to working, just some kinks to work out

18:47 advancel has quit [Remote host closed the connection]

19:03 JohnDoe_71Rus has quit []

19:52 bauen1 has quit [Ping timeout: 480 seconds]

20:00 warpme has joined #linux-sunxi

20:13 gsz has joined #linux-sunxi

20:31 gsz has quit [Ping timeout: 480 seconds]

20:47 KREYREN_oftc has quit [Remote host closed the connection]

20:47 KREYREN_ has joined #linux-sunxi

20:48 bauen1 has joined #linux-sunxi

20:53 ftg has joined #linux-sunxi

21:16 machinehum has quit [Ping timeout: 480 seconds]

21:19 warpme has quit []

21:22 machinehum has joined #linux-sunxi

21:30 KREYREN_ has quit [Remote host closed the connection]

21:31 KREYREN_ has joined #linux-sunxi

21:32 machinehum has quit [Ping timeout: 480 seconds]

21:36 KREYREN_ has quit [Remote host closed the connection]

21:37 KREYREN_ has joined #linux-sunxi

21:41 KREYREN__ has joined #linux-sunxi

21:42 KREYREN_ has quit [Remote host closed the connection]

21:49 KREYREN__ has quit [Remote host closed the connection]

21:49 KREYREN__ has joined #linux-sunxi

22:01 vagrantc has quit [Quit: leaving]

22:03 KREYREN__ has quit [Remote host closed the connection]

22:04 KREYREN__ has joined #linux-sunxi

22:06 KREYREN_ has joined #linux-sunxi

22:07 KREYREN__ has quit [Remote host closed the connection]

22:11 KREYREN_ has quit [Remote host closed the connection]

22:12 KREYREN_ has joined #linux-sunxi

22:12 machinehum has joined #linux-sunxi

22:20 KREYREN__ has joined #linux-sunxi

22:20 machinehum has quit [Ping timeout: 480 seconds]

22:20 KREYREN_ has quit [Remote host closed the connection]

22:27 KREYREN__ has quit [Remote host closed the connection]

22:28 KREYREN__ has joined #linux-sunxi

22:33 apritzel has joined #linux-sunxi

22:41 KREYREN__ has quit [Remote host closed the connection]

22:41 KREYREN__ has joined #linux-sunxi

22:53 KREYREN__ has quit [Remote host closed the connection]

22:54 KREYREN__ has joined #linux-sunxi

23:04 machinehum has joined #linux-sunxi

23:12 machinehum has quit [Ping timeout: 480 seconds]

23:39 apritzel has quit [Ping timeout: 480 seconds]

23:40 machinehum has joined #linux-sunxi

23:48 ftg has quit [Read error: Connection reset by peer]

23:49 machinehum has quit [Ping timeout: 480 seconds]