buzzmarshall has quit [Quit: Konversation terminated!]
<wens>
setting up hosting with proper access controls and all the bells and whistles, making sure it runs smoothly for everyone using it, hosting your own network and machines, is a whole different skill set that normal developers don't really have
jacobk has joined #linux-amlogic
luka177 has quit [Ping timeout: 240 seconds]
jacobk has quit [Ping timeout: 240 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 246 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 258 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
tortoise has quit [Server closed connection]
tortoise has joined #linux-amlogic
Daanct12 has quit [Ping timeout: 246 seconds]
Daanct12 has joined #linux-amlogic
luka177 has quit [Ping timeout: 258 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 255 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 258 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 246 seconds]
jacobk has joined #linux-amlogic
jacobk has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
Terry13732293409 has quit [Quit: Bye Bye]
Terry13732293409 has joined #linux-amlogic
Daanct12 has quit [Quit: WeeChat 4.0.4]
f_ has joined #linux-amlogic
f_[xmpp] has joined #linux-amlogic
<f_>
wens: I didn't have to deal with access controls?
<f_>
Also you don't have to host your own network and machines
<f_>
One single VPS is more than enough for a git server
<f_>
Only two patches we apply are: 'ARM64: dts: meson-gxl: add support for the Xiaomi Mi Box 3' and 'add x96-mini support', both from exkc...
<f_>
I wonder if they got applied now..
<f_>
(all other amlogic devices we support run with no patches, e.g. lepotato and kii-pro)
<f_>
chewitt: Doubt we support anything AXG though
<f_>
(patches welcome though)
<chewitt>
the AXG audio driver is used with G12A/B and SM1 boards .. so you probably do
<f_>
Hm..
* f_
looks at pmOS wiki.
<f_>
I don't think exkc got much working on any of the G12/SM1 set-top boxes we support
<f_>
ah.
<f_>
We don't support any G12/SM1 board whatsoever, with the exception of the X96 Air and X96 Max Plus, but support hasn't been merged yet.
<f_>
Even then, audio seems to be broken on both
<f_>
(MR is stale)
<f_>
chewitt: so no, we don't (officially) support anything using the AXG audio driver
luka177 has quit [Ping timeout: 240 seconds]
luka177 has joined #linux-amlogic
<f_>
anyway, upgrading the kernel has to go in my TODO list
<f_>
(also includes upgrading BL21/BL301 for lepotato/lafrite and building them from source, on all boards as well as merging all U-Boot packages targeting Amlogic)
luka177 has quit [Ping timeout: 246 seconds]
<narmstrong>
f_: there's not a lot AXG boards in the wild, and no sbc for sure
<f_>
Perhaps one more reason why nobody bothered to support one of these AXG boards, in postmarketOS.
<narmstrong>
yep
<narmstrong>
they are used in small audio and iot stuff
<narmstrong>
like smart speakers
<f_>
Yeah I know.
<f_>
The A in AXG means Audio right?
<lvrp16>
I wonder how the acronyms are created.
<f_>
GXBB stands for generation x baby right?
luka177 has joined #linux-amlogic
<f_>
As for GXL, the L remains a mystery.
luka177 has quit [Ping timeout: 244 seconds]
<f_>
Aha.
<f_>
RK3036 DRAM register names look very similar
<f_>
>PCTL_DFIODTCFG
<f_>
>PCTL_SCFG
<f_>
That looks similar to me!
<f_>
Additionally, sunxi boot0/boot1 DRAM init code looks similar!
<f_>
Apart from, perhaps, different function names, that looks similar to me
<f_>
Not quite the same, but similar.
Danct12 has joined #linux-amlogic
<f_>
Interesting.
<f_>
Of course, in Allwinner language, PCTL is called MCTL.
Danct12 has quit [Client Quit]
<f_>
That must be documented somewhere, but that also means we could borrow some of the docs the sunxi people have written to try and understand how the S905(X) DRAM works.
<f_>
*try to understand
luka177 has joined #linux-amlogic
<narmstrong>
GXL is probably light, GXM is medium
<narmstrong>
G12 is the 12th generation
<narmstrong>
SM1 is the real mystery
<f_>
SoMehow1
<narmstrong>
GXL = Generation 10 Light
<chewitt>
S4? .. T7?
<f_>
narmstrong: By the way nice picture in the linux-meson website
<narmstrong>
they made it interesting to have pin2pin compatible packages across different families
luka177 has quit [Ping timeout: 246 seconds]
<f_>
means you can replace one SoC with another pin2pin compatible?
<f_>
*another one that's...
<narmstrong>
yes
<f_>
lvrp16: BTW what happened to tartiflette?
<lvrp16>
Lol @ generation x baby
<f_>
G means generation
<f_>
X means 10
<f_>
GXBB is sometimes written as 'GXBaby'
<f_>
so....generation x baby
<narmstrong>
yep it was their first arm64 soc
<f_>
I knew it!!
<f_>
Also first SoC to use a fork of Arm TF-A
<narmstrong>
they moved from an ARC SCP core and switched to TF-A so it was a big move
<f_>
They were too late
<f_>
S905 came out in 2015
<narmstrong>
a little yes but at the same time plenty of chinese socs switches to armv8
<lvrp16>
Tartiflette components have been sitting in our WH for 4 years.
<narmstrong>
so I suspect ARM pushed them with a really cheap cortex-a53 license
<f_>
Perhaps
<narmstrong>
and mali
<lvrp16>
Waiting for u-boot 2023.10 release before we do anything. Bandwidth limited on me.
<f_>
Cottonwood will be the big one.
<lvrp16>
After semi-launching La Frite, I needed the software to bake a while.
<lvrp16>
Every time I rebased LOST on La Frite, it took a week and basically had to rewrite the whole thing.
<f_>
BTW where's lost's source code?
<lvrp16>
Not scalable across multiple boards so it was a waste to launch.
<lvrp16>
Closed source 🤣
<f_>
Why?
<lvrp16>
Until I finished the pluggable architecture.
<f_>
Then it will be libre?
<f_>
Understandable
<narmstrong>
BTW where's the upstream DT for cottonwood boards ???
<narmstrong>
because so far I merged nothing :-p
<f_>
BTW I saw many of your scripts being licensed under CC BY-NC (non-libre) and the portability script being licensed under CC-NC-ND (proprietary)
<lvrp16>
Basically OS installation would be open source. But the driving components for LOST is SAAS based so not open source.
<f_>
seems like you don't enforce the license on the portability script (you merged some PRs)
luka177 has joined #linux-amlogic
<lvrp16>
I am not terribly great with organizing licensing. I just shove them there willy nilly sometimes.
<f_>
I'd recommend using a permissive license at the very least, for scripts
<f_>
(copyleft doesn't make sense on scripts anyway)
<lvrp16>
For all intents and purposes, public repos should all be open source.
<f_>
My favourite permissive license is BSD-3-Clause
<lvrp16>
I think that is part of GitHub's licensing requirement as well.
<f_>
I don't care about GitHub :P
<f_>
Never used it to host my stuff
<lvrp16>
narmstrong: It seems ARM is reducing cost of A76 so we should see those pop up more.
<f_>
I think more people should move to something other than github or gitlab..host their own git server, or join a public forgejo instance...
<lvrp16>
I am an realist, not an idealist. 🤣
<f_>
Know codeberg?
<lvrp16>
Incremental impactful changes are better writing books.
<f_>
(forgejo/gitea looks pretty similar to github tbh)
<lvrp16>
IMHO. Everyone differs.
<f_>
sure
buzzmarshall has joined #linux-amlogic
<f_>
As far as github-like forges go, I highly recommend codeberg.org/forgejo/gitea
<f_>
(if you ever wanted to move away from github)
<f_>
A little off-topic here though :)
<f_>
so I'll stop talking about git
<f_>
lvrp16: but you could pretty much use 3BSD or MIT for your scripts. Both are pretty nice
<f_>
GPL-3.0+ and MPL-2.0 are also fine.
<f_>
(but who wants copyleft on scripts?)
luka177 has quit [Ping timeout: 245 seconds]
<f_>
But the CC licenses you currently use on your scripts are not libre/open-source
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 255 seconds]
<lvrp16>
I think the original policy was that everything to create the board bootloader, linux, and hardware integration was libre/open source to make a "libre" computer. All other value adds did not have to be depending on their complexity.
<f_>
but still, code publicly available should be libre
<f_>
Having a board that uses libre firmware to boot and a libre OS **by default** is really nice though, even better if it's mainline linux
<f_>
(and mainline u-boot)
<lvrp16>
That is already the case. We use upstream to the maximum extent but some tweaks just cannot go upstream. Neil, Jerome, and others did their level best.
<lvrp16>
Again, theoretical vs real.
Danct12 has joined #linux-amlogic
<narmstrong>
Intle hides all the tweaks in ACPI, easy :-p;
<lvrp16>
Haha
<narmstrong>
*Intel
<lvrp16>
ACPIOS
<f_>
MINIX/ME
<lvrp16>
Lets call it for what it is.
<narmstrong>
but vendors like Qcom are now moving critical stuff into firmwares
<narmstrong>
like usb/pd & battery management
<f_>
narmstrong: saw you were working on qcom too
<f_>
nice
<f_>
(I own qcom hardware so it's nice to see mainline running on those)
<lvrp16>
narmstrong: I guess the thinking is people will break their hardware if they have source.
<f_>
lvrp16: there's a reason as to why there that big all caps disclaimer on each and every libre license :P
<narmstrong>
lvrp16: not necessarely, I think they want to be sure the devices won't burn (very complex modern batteries) and still passes usb4/usb-pd/thunderbolt certifications even with running an open source os
<lvrp16>
Yeah it gets hairy when people have regulator control in their own hands with overclocks.
<lvrp16>
At one point Armbian was running some high voltage and clocks on Allwinner and Rockchip boards copied from SDK trees.
<lvrp16>
Amlogic was saved by the grace of their closed ATF implementation.
<narmstrong>
until now :-p
<f_>
lvrp16: "saved"
<lvrp16>
Haha
<f_>
It wasn't even saved lol
<f_>
*they weren't
<f_>
there was that old tree rotting around :P
<f_>
and most of their code remained unchanged :P
<lvrp16>
The hardware was. There were patches running 1.4V on the SoCs.
luka177 has joined #linux-amlogic
<f_>
would be nice to implement the rest of BL31 upstream
<f_>
(and easy for newer SoCs)
<f_>
lvrp16: IIRC you told me amlogic rebased on Carlo's work, right?
<f_>
Ironically that makes it much easier to reverse-engineer
<lvrp16>
I have no idea TBH.
<f_>
you only have to reverse-engineer their mods ¯\_(ツ)_/¯
<f_>
Even if it wasn't it's clearly based on TF-A
<lvrp16>
They updated PSCI for me.
<narmstrong>
lvrp16: on gxl ?
<lvrp16>
narmstrong: yeah 1.0
<narmstrong>
lvrp16: nice for system suspend/resume!
<lvrp16>
Suspend, poweroff, resume all works within our tree.
<narmstrong>
noice
<lvrp16>
Noiceeeee
<f_>
which tree?
<narmstrong>
now runtime pm :-p
<lvrp16>
Our linux/u-boot/blx.
<f_>
Did BL2 get modified too?
<narmstrong>
nop it's only the PSCI interface
<narmstrong>
in bl31
<lvrp16>
Yeah, although GXL boards are still missing a PMIC so suspend is not 0W.
<narmstrong>
you don't want to deal with pmics :-p
<lvrp16>
I know it needs a lot of firmware work but 0W power off is nice.
<f_>
yeah sure
<f_>
but have they done any mods on the bl2 binaries you're using?
<lvrp16>
f_: no same bl2
<f_>
Ok
<f_>
I wonder if I can use the sun6i dram init codebase...
<f_>
(with some mods, of course)
<f_>
I could use it as documentation
<lvrp16>
I know sunxi had someone do the timing window testing. It would be cool to have infrastructure to do DDR tuning based on that.
<f_>
What I'm specifically looking for, is what the PGSR0 errors mean
<f_>
I almost want to mod BL2 for debugging purposes..
luka177 has joined #linux-amlogic
<f_>
...
<f_>
0xc0000fff might indicate success?
luka177 has quit [Ping timeout: 250 seconds]
vagrantc has joined #linux-amlogic
<aeroraptor>
Hello everyone! I'm hoping someone can help me update a device tree source file to change a pin definition. This (should) be an easy change but I'm having trouble finding some of the necessary information.
<f_>
And of course it throws an error 403 on me :/
<chewitt>
there's an excel sheet there which might be useful
<aeroraptor>
What I want to do is change it from using GPIOAO_7 to using GPIOCLK_0 (which is GPIO 98). From reading the s905x datasheet I believe that GPIOCLK_0 can be used as an interrupt, so I just need to swap out the pinctrl-o, gpios, and interrupts parameters. Right now I'm digging through in the example dts files to see if I can find anything else using GPIOCLK_0
<f_>
I hate when people block Tor users for no reason..
<f_>
(or when they use Cloudflare, which blocks Tor users)
<chewitt>
DropBox or Ti ?
<f_>
TI.
<f_>
No idea if Dropbox blocks Tor
<chewitt>
click the link and you'll find out..
<f_>
That's what I did
<f_>
It's slow.
<f_>
but seems to work
<f_>
(does it?)
<f_>
It doesn't, let's once again fallback
<lvrp16>
aeroraptor: look at the meson-gxl-gpio.h file.
<lvrp16>
#define GPIOCLK_0 98
<lvrp16>
But the gpio controller is not gpio_ao
<f_>
I just noticed something funny
<f_>
Some PCTL registers are documented in bl301
<aeroraptor>
lvrp16: thanks, I actually just found that in the 1wire example dts they use this pin, and they have it in just &gpio "gpios = <&gpio GPIOCLK_0 GPIO_ACTIVE_HIGH>;"
<narmstrong>
You get the number in the s905x datasheet
<f_>
Glad sc-im supports excel files :)
<aeroraptor>
narmstrong: okay, I can look there next. I also have to figure out two more things - pinctrl-o = <&spdif_out_ao_6_pins>; --- where are they getting this name? I can't find a table of these names anywhere
<lvrp16>
aeroraptor: amlogic platforms before sm1 do not support double edged interrupts so do look for warnings in your kernel if you are attempting to use those.