02:36
Ballerburg9005 has quit [Ping timeout: 260 seconds]
03:05
<
chewitt >
f_[xmpp] any update?
03:25
JohnnyonFlame has quit [Ping timeout: 252 seconds]
04:43
hexdump0815 has quit [Ping timeout: 260 seconds]
04:45
hexdump0815 has joined #linux-amlogic
05:05
<
chewitt >
the previous (single) patch is missing the aiu binding so won't work
05:06
<
chewitt >
if all good, please pm me an email and real name to use with a "Tested-by:" tag .. and I'll send them upstream
05:40
buzzmarshall has quit [Quit: Konversation terminated!]
08:08
JohnnyonFlame has joined #linux-amlogic
08:10
f_ has joined #linux-amlogic
08:12
<
f_ >
No need to PM you for giving you a real name and email. They're all public.
08:12
<
f_ >
You can use: Ferass El Hafidi <vitali64pmemail@protonmail.com>
08:13
<
f_ >
(Not now. I didn't test yet)
08:16
<
f_ >
I'm `git clone`ing the repo. That'll take a while.
08:23
<
f_ >
Ok. I formatted the patches. Let's include them in the APKBUILD and see what happens!
08:33
<
f_ >
Time to reboot!
08:35
<
f_ >
chewitt: Well. ALSA definitely detected some sound cards!
08:35
<
f_ >
Wanna know what that means? Time to unplug the KII Pro and move it somewhere where I can actually test audio!
08:36
<
f_ >
Well. That's definitely a step in the right direction! :D
08:41
<
f_ >
chewitt: It partially works!
08:41
<
f_ >
When I say partially, I mean:
08:41
<
f_ >
Sound works, but it's a little glitchy at first, but then it works perfectly fine.
08:42
<
f_ >
chewitt: You can add the Tested-by tag with my name and email.
08:43
<
narmstrong >
great success !
08:44
<
f_ >
narmstrong: Yup! :D
08:45
<
f_ >
Now let's test bluetooth.
08:47
<
f_ >
Ok. `bluetoothctl` recognises a controller.
08:51
<
f_ >
One thing to note. I did have to fiddle around with alsamixer for sound to work.
08:51
<
f_ >
But that's normal.
09:04
<
f_ >
chewitt: However, bluetooth doesn't work.
09:05
<
f_ >
It detects a bluetooth controller but when discovering devices it doesn't show up anything.
09:05
<
f_ >
(I think that was the same on Armbian/Manjaro though)
09:06
<
f_ >
Still, progress!
09:28
<
chewitt >
f_ can you share dmesg log .. let's see what shows up now
09:28
<
chewitt >
btw, where are you in the world?
09:33
<
naoki >
is firmware needed to use vdec on g12(a)?
09:35
<
chewitt >
this also has slightly newer firmware than whats upstream
09:35
<
chewitt >
but whether that's a good or bad thing is unknown, as it's closed-source
09:36
* chewitt
rolls dice
09:37
<
chewitt >
and depending on what media you want to decode, software decoding might be better (or worse) .. or hardware decoding might be (currently) impossible
09:38
<
chewitt >
8-bit H264/HEVC up to 1080p are okay
09:38
<
chewitt >
10-bit or 4K media will wedge the board
09:39
<
chewitt >
VP9 same as H264/HEVC .. 8-bit
09:39
<
naoki >
I'm trying meson vdec first time, I don't know anything ;)
09:39
<
chewitt >
MEPG1/2/4 are broken and VC1 hasn't been written
09:39
<
naoki >
kernel: CONFIG_VIDEO_MESON_VDEC
09:39
<
chewitt >
no, driver
09:40
<
naoki >
ffmpeg: *_v4l2m2m
09:41
<
naoki >
is there any document I should read?
09:41
<
naoki >
to use vdec on meson
09:44
<
chewitt >
this contains a large number of improvements to v4l2_m2m support in ffmpeg and handles 'quirks' of the Amlogic driver
09:44
<
chewitt >
it's not perfect, but better than upstream ffmpeg
09:44
<
chewitt >
and there is an effort by RPi Foundation Devs (this is the version they use with RPi devices) to upstream those changes
09:45
<
chewitt >
a very slow effort, but still the only one being made
09:57
camus1 has joined #linux-amlogic
09:59
camus has quit [Ping timeout: 265 seconds]
09:59
camus1 is now known as camus
10:12
<
naoki >
I can see something...
10:26
naoki has quit [Quit: naoki]
10:42
rockosov has quit [Ping timeout: 260 seconds]
10:43
rockosov has joined #linux-amlogic
11:03
<
chewitt >
if oops was the board wedging.. you've played 10-bit or 4K media
11:12
camus has quit [Quit: camus]
11:29
<
f_ >
chewitt: I'm somewhere in the world.
11:30
<
chewitt >
the name sound sort of local/regional to where I am (GMT+4)
11:31
<
f_ >
I'll share dmesg logs shortly.
11:31
<
chewitt >
El Something normally means GMT+2?
11:32
<
f_ >
That's part of my last name.
11:32
<
f_ >
Nothing to do with where I live.
11:33
<
f_ >
I generally don't tell anyone where I live because......privacy :)
11:34
<
f_ >
Got some Bluetooth-related dmesg logs!
11:38
JohnnyonFlame has joined #linux-amlogic
11:39
<
f_ >
Yup. That's what I was thinking too.
11:39
<
chewitt >
put in /usr/lib/firmware/brcm/
11:39
<
f_ >
If that works, I'll make yet another package.
11:39
<
f_ >
firmware-videostrong-kii-pro-bluetooth
11:40
<
f_ >
Or something similar.
11:40
<
chewitt >
^ needed for WiFi
11:40
<
f_ >
chewitt: Wait Wi-Fi works?
11:41
<
chewitt >
no idea .. you only shared the BT bits from dmesg :)
11:41
<
f_ >
firmware-videostrong-kii-pro then.
11:41
<
f_ >
chewitt: It doesn't even appear in NetworkManager so I assumed it didn't work.
11:41
<
chewitt >
no firmware = no wifi
11:42
<
chewitt >
create brcmfmac4335-sdio.videostrong,kii-pro.bin and the .txt equivalent as symlinks and you get one less 'error' message in dmesg
11:43
<
f_ >
I did have a related error in dmesg. It said it failed to load it.
11:43
<
chewitt >
I tries to load the board specific file first, then falls back to the generic filename
11:44
<
f_ >
chewitt: Sure. Gotta go for a sec... See you later!
11:44
<
chewitt >
it also looks for a clm blob, but those are board specific and nobody has the originals for these boxes
11:44
<
f_ >
(My client will stay connected)
11:44
<
f_ >
chewitt: Sure.
11:57
ldevulder has quit [Remote host closed the connection]
11:57
ldevulder has joined #linux-amlogic
12:03
<
chewitt >
narmstrong: this is on 6.1 with the SDIO interrupt patch un-reverted and the latest 3x patches from Heiner applied
12:08
Ballerburg9005 has joined #linux-amlogic
12:09
<
chewitt >
and this is with the SDIO interrupt patch reverted
12:09
<
chewitt >
working wlan
12:10
<
chewitt >
over time I have the impression that the QCA9377 chip in this box is 'sensitive' .. but there's still a regression with the interrupts patch
12:10
<
chewitt >
i'm not seeing issues with devices with Broadcom chips
12:58
Ballerburg9005 has quit [Ping timeout: 264 seconds]
13:07
<
f_ >
chewitt: Let's see if it works.
13:09
<
f_ >
chewitt: Wi-Fi works!
13:09
<
f_ >
Well, it fetches available networks fine.
13:10
<
f_ >
Let's see if it can connect.
13:15
<
f_ >
Well. It can't, it seems. Unless it's that problem where after a few tries it works, like what's happening on my Pocophone F1?
13:25
<
f_ >
chewitt: Are you there?
13:26
<
f_ >
Wi-Fi doesn't seem to work right. NetworkManager doesn't recognise it anymore so I'm rebooting.
13:26
<
f_ >
Honestly, it can just be the same issue as the one I have on my phone.
13:27
<
f_ >
Where it detects Wi-Fi access points just fine, but I need a few tries before it connects correctly.
13:27
<
f_ >
chewitt: Later, sure.
13:27
<
chewitt >
edit uEnv.ini to set the dtb name (meson-gxbb-kii-pro.dtb) .. add 'ssh' to boot params
13:28
<
f_ >
And where it can't seem to want to connect to some Wi-Fi APs.
13:28
<
f_ >
chewitt: Sure....Why add `ssh` to the cmdline though?
13:28
<
chewitt >
that's running 6.1.11 kernel with iwd (not wpa_supplicant) and connman (not NetworkManager)
13:28
<
chewitt >
so we get to see if a different stack influences the problem or not
13:29
<
f_ >
chewitt: I could also just get rid of wpa_supplicant and NetworkManager.
13:29
<
chewitt >
LE ships with ssh disabled by default, and if there's an issue we often have no GUI so it's a way to ensure easy access
13:29
<
f_ >
chewitt: Oh. Sure.
13:29
<
f_ >
Always nice when `ip a` hangs.
13:30
<
chewitt >
if you see similar issues with other devices it's probably something odd in your router config
13:30
<
chewitt >
broadcom SDIO support is quite robust otherwise
13:30
<
f_ >
chewitt: You mean the connection issues where I need to try a few times before it connects fine?
13:30
<
f_ >
Or the fact that `ip a` hangs?
13:31
<
chewitt >
router config won't cause local commands to hand
13:31
<
f_ >
The connection issue on my Pocophone F1 is a known issue. That has to do with postmarketOS IIRC.
13:31
<
f_ >
chewitt: I mean. Local commands could hang if they have problems accessing network interfaces.
13:32
<
f_ >
chewitt: I'll flash your LE image, but first let's test if BT works.
13:34
<
f_ >
chewitt: BT definitely works.
13:34
<
f_ >
Which is nice.
13:37
<
f_ >
chewitt: Ok it depends on wpa...
13:38
<
f_ >
chewitt: I'll flash now.
13:47
<
f_ >
It's resizing.
13:49
<
f_ >
chewitt: When trying to connect Kodi crashes.
13:49
<
f_ >
Also Kodi is run in X11 right?
13:50
<
f_ >
I think it was because of the resolution change, not Wi-Fi.
13:51
<
chewitt >
we run Kodi on the framebuffer using GBM/V4L2
13:51
<
f_ >
chewitt: I thought that because Kodi started with a 1080p resolution when my monitor supports only up to 1280x1024.
13:51
<
f_ >
chewitt: Well. It's working.
13:52
<
chewitt >
I'm not aware that we hardcode resolution anywhere.. it should be auto-detecting
13:52
<
f_ >
Sure. I guess that may also be because of the VGA-HDMI dongle I use.
13:52
<
chewitt >
We limit it to 1080p by default (as running GUI at 4K sucks balls)
13:52
<
chewitt >
but we don't force it
13:52
<
f_ >
Might be related to my dongle then.
13:53
<
f_ >
(Yes, this monitor only supports DVI and VGA (and I don't have any DVI cable rn))
13:53
<
chewitt >
WiFi connections are set in the LE settings add-on, or you can use connmanctl from SSH
13:53
<
chewitt >
connmanctl
13:53
<
f_ >
chewitt: Yeah I know. I already used LibreELEC before.
13:53
<
f_ >
I used it on a raspberrypi back then :)
13:53
<
chewitt >
most folks do
13:53
<
f_ >
And before LibreELEC, I used OpenELEC.
13:54
<
f_ >
(RIP OpenELEC)
13:54
<
f_ >
Anyway. Let's try installing an addon!
13:55
<
chewitt >
that's been a bit sucky recently.. we need to up specs on the add-ons server
13:55
<
chewitt >
or offload all the add-ons onto mirrors
13:55
<
f_ >
Downloading does work. Nice.
13:55
<
chewitt >
make it someone-elses-problem
13:56
<
f_ >
Be lazy and make it someone else's problem :S
13:56
<
chewitt >
so looks like you can point fingers at postmarketOS for wifi issues
13:56
<
f_ >
Or at NetworkManager and wpa_supplicant.
13:56
<
chewitt >
dmesg | paste .. share the URL
13:57
<
f_ >
Sure but wait.
13:57
<
chewitt >
let's just check for any other odd things in dts to fix
13:57
<
f_ >
`paste` is a LibreELEC command?
13:57
* f_
just `curl`s their own pastebin XD
13:57
<
chewitt >
throws things at ix.io
13:58
<
f_ >
Sure, but I prefer bin.vitali64.duckdns.org. Mainly because I'm the one who hosts it :D
13:58
<
chewitt >
sure, whatever's easiest
14:00
<
f_ >
New paste has been created: 63e64ddc
14:05
<
chewitt >
all looks good, thanks
14:06
<
f_ >
Now let's get Wi-Fi to work on postmarketOS.....after I build my pmOS image.
16:51
<
f_ >
chewitt: I'll package the firmware needed for Wi-Fi and BT shortly.
16:52
<
f_ >
chewitt: Also, can I use your patches for the kernel package?
16:52
<
chewitt >
Sure, it's all in a repo so people can raid the patches
16:53
<
chewitt >
If you track my stuff, beware that I rebase the branches frequently
16:53
<
chewitt >
I try to leave everything in a working state, but if I've given the impression of skilled kernel developer, you've been fooled :)
16:54
<
f_ >
chewitt: Good thing.
16:54
<
f_ >
(The fact that you frequently rebase your branches)
16:54
<
chewitt >
it means all the patches are at the top of the branch, not lost out of sight behind merges
16:55
<
f_ >
chewitt: Also. If I've given the impression of a skilled kernel developer, you've also been fooled :)
16:55
<
f_ >
I do know C though (and like it), BTW
16:58
<
f_ >
chewitt: Still impressive, the fact that the tv box is pretty usable without any firmware though.
16:59
<
chewitt >
the other firmware that you'll maybe need to bundle (depending on use-case) is the vdec firmware
16:59
<
f_ >
>depending on use-case
16:59
<
chewitt >
and if wanting to use ffmpeg .. pick the sources I linked earlier
17:00
<
f_ >
I suppose the vdec firmware is used for hardware decoding?
17:03
* f_
thought they were dumb.
17:04
* f_
can't find the firmware
17:09
<
chewitt >
github.com/LibreELEC/meson-firmware/
17:09
<
chewitt >
or in the upstream linux-firmware repo
17:10
<
chewitt >
if you want things to work .. if not, use upstream ffmpeg
17:10
<
chewitt >
(changes in that branch will eventually go upstream, insh'allah)
17:11
<
chewitt >
if you like poking C code, there are lots of things that need love :)
17:11
* f_
checks if alpine linux packages that.
17:13
<
chewitt >
best to grab all the files
17:13
<
chewitt >
the h264 is only for gxbb, but it will use the gxl hevc firmware
17:14
<
chewitt >
and hopefully someone investigates why mpeg1/2 are currently broken
17:15
<
chewitt >
In my branch I've disabled them in the hardware decode driver to force software decode which is better than a blank screen
17:15
<
chewitt >
I have a hunch some of the upstream UAPI 'matured' and we're trying to use things that no longer exist
17:15
<
chewitt >
so probably an easy fix .. if you can read the code
17:16
<
chewitt >
I can half-read it .. but not enough to really understand it
17:20
<
f_ >
chewitt: Where's the code?
17:23
<
chewitt >
the (staging) driver is in drivers/staging/media/meson/vdec
17:24
<
chewitt >
I started to poke around with DVB cards
17:24
<
f_ >
.max_width = 1920,
17:24
<
f_ >
.max_height = 1080,
17:24
<
chewitt >
I've imported some (but not all) of the demod/tuner drivers used with WeTek Play2 (a box I have)
17:25
<
f_ >
Ok ignore what I said earlier.
17:25
<
f_ >
chewitt: Nice!
17:26
<
f_ >
My DVB tuner is connected to pin headers. It's pretty easy to remove it.
17:27
<
chewitt >
I've done a little cleanup since the drivers historically supported Meson8 (32-bit) SoCs which have really (really) old GPIO
17:28
<
chewitt >
but there are parts where Amlogic GPU headers were used, which prob require some changes
17:28
<
f_ >
I didn't expect less from a vendor downstream kernel.
17:29
<
chewitt >
and some of the drivers are just horrifically ifdef'd with fugly code and a bazillion //commented lines
17:29
<
f_ >
'and a bazillion //commented lines' Reminds me of some dumb person called `f_`....Oh wait, it's me.
17:30
<
f_ >
I still think my coding style is cleaner than theirs though.
17:30
<
chewitt >
I have most of them in a state where the drivers are in the kernel, with Kconfig/Makefile entries, and they individually compile
17:31
<
chewitt >
The tricky bit will be glueing them together, and getting things to probe from device-tree
17:31
<
f_ >
Basically, that's what you all do - Understanding ugly code and writing actually good code from that.
17:32
<
chewitt >
that likely requires a level of knowledge that I don't have and aren't likely to acquire
17:32
<
chewitt >
but, hopefully sees some of the donkey work has been done and takes the bait :)
17:32
<
f_ >
chewitt: But hardware acceleration works, right?
17:33
<
chewitt >
it is likely that we are missing a 'demux' driver to take the output of the tuner/demod combo and feed something to a hardware decoder
17:34
<
chewitt >
and for that the vendor code probably isn't helpful since Amlogic wrote their own DRM framework
17:34
<
chewitt >
to their credit, they have 4K/HDR working aeons before anything in the upstream kernel (3.10 era)
17:35
<
chewitt >
to their discredit, they've been forward-porting that crap ever since :(
17:35
* chewitt
waves at various Amlogic staff in this channel .. we love you really :)
17:36
<
chewitt >
anyway.. I will continue to poke stuff in my branch(es) to see what I can learn and make happen
17:37
<
chewitt >
next driver on my cleanup list is the AVL6211 driver which is probably the demod in your box
17:38
<
f_ >
I would love helping you!
17:38
<
chewitt >
if yes, MxL603 is probably the tuner
17:38
<
chewitt >
(seems to be a regular combo)
17:39
* f_
wants to be sure it is the tuner/demod used.
17:39
<
chewitt >
if/when we reach the point where the DVB stuff mostly works, we will likely need the mpeg1/2/4 drivers working again
17:40
<
chewitt >
and then someone needs to tackle the deinterlace ip block
17:40
<
chewitt >
right now ffmpeg will software deinterlace, but the good methods are compute intensive and not possible on lowly ARM boards
17:41
<
f_ >
The TV box I use looks like it's quite powerful though.
17:41
<
f_ >
At least as powerful as my laptop, probably.
18:00
vagrantc has joined #linux-amlogic
18:20
naoki has joined #linux-amlogic
18:25
<
f_ >
chewitt: I packaged the firmware files.
18:37
hexdump0815 has quit [Quit: WeeChat 1.9.1]
18:39
<
f_ >
chewitt: Got Wi-Fi to work!
18:39
<
f_ >
Oh well. It connects but networking doesn't actually work...
19:32
naoki has quit [Quit: naoki]
19:32
naoki has joined #linux-amlogic
19:36
naoki has quit [Client Quit]
19:37
naoki has joined #linux-amlogic
19:57
naoki has quit [Quit: naoki]
20:04
Ballerburg9005 has joined #linux-amlogic
20:08
rockosov has quit [Quit: WeeChat 3.4-dev]
20:51
Ballerburg9005 has quit [Ping timeout: 265 seconds]
20:53
Ballerburg9005 has joined #linux-amlogic
21:36
f_ has quit [Ping timeout: 264 seconds]
22:34
naoki has joined #linux-amlogic
23:50
vagrantc has quit [Quit: leaving]