00:02 <apritzel> macromorgan: ok, that looks good: I can set exactly the bits described for the A80 NMI controller, at 0x07010320. There does not seem to be anything at 0x7021000, which would be the location for a sun6i-a31-r-intc compatible intc

00:03 <apritzel> so this is good news, as it means that no interrupts are routed through that r-intc, as in the H6, where it caused a major problem, when we introduced that later

00:05 <apritzel> macromorgan: tokyovigilante, acmeplus: what are the plans for the .dts submission? Now is a good time: if we can get this reviewed before the end of April, it has good chances to go into v6.10

00:06 <apritzel> drop everything we don't have a binding for (graphics, Mali, video, audio)

00:07 <apritzel> macromorgan: regarding USB: it is quite common for many boards to have separate GPIO controlled regulators (voltage switches, really) for USB VBUS

00:08 <tokyovigilante> apritzel: sure, have been thinking about getting a minimum DTS together, will try to do that in the next day or so

00:09 <apritzel> for peripheral mode you want that turned off, also by default it would be the sink of the power supply, not a source

00:09 <apritzel> tokyovigilante: great, thanks! Expect two or three rounds of submissions, a few days apart

00:10 <tokyovigilante> Should we have a separate one for the RG35xx-plain? (I don't have that specific one), but there are now a base model, the -plus that adds Wifi/BT, and the -H which is a different form-factor, and adds a second USB and analog sticks

00:10 <apritzel> so don't bother too much getting everything *perfect* on the first run, you will need to fix things anyway

00:10 <tokyovigilante> Will do, have seen a few go in recently so sort-of familiary with the process now

00:12 <apritzel> yes, it looks like there should be a rg35xx_2024.dts, which is included by the the other two

00:13 <apritzel> even if this new one is not a perfect subset: we can delete or overwrite nodes or properties in the other two as well

00:14 <tokyovigilante> makes sense, will just have to guess for the -2024 unless macromorgan has one, I don't think even if I order one now it will be here in time, have the -H and -Plus to hand though.

00:14 <apritzel> look at sun50i-h6-pine-h64-model-b.dts for an example

00:14 <tokyovigilante> cool

00:24 <macromorgan> let me post what I have right now... you want to take the lead tokyo? Otherwise I can (note I am working off the H, not the plus, but ADC aside it should be almost identical

00:25 <apritzel> macromorgan: which ADC is that? GPADC or LRADC? You can also leave that part of for now if that's not ready

00:43 <macromorgan> I don't know yet honestly

00:47 <macromorgan> here's what I am working with currently: https://paste.debian.net/1313726/

00:48 <macromorgan> I need to hook up the GPIOs to the USB/USB regulator, and remove the nmi interrupt stuff. I was also hoping to sort out all the regulator bits before I upstreamed it, so there's a lot of /* test */ that still needs to be defined or removed.

00:49 <macromorgan> CPU is the only regulator that I take for granted currently, all the other ones were figured out "the hard way"

00:50 <macromorgan> but if it doesn't say /* test */ it means I've already confirmed through testing that the value is correct

01:20 <nickAA> Jookia: I could ask him if he wants to submit a patch. You can find our chat here. https://forum.armbian.com/topic/29794-how-to-install-armbian-in-h618/page/4/

01:21 <Jookia> could be cool

05:19 <tokyovigilante> macromorgan: Yup I'm happy to, I'll pull a first draft together tonight or on the weekend

06:47 <Jookia> Has anyone managed to get u-boot usb gadget working reliably on the D1 or T113? It seems to randomly crash or time out

06:58 <Jookia> i sent about a years of backlogged patches to u-boot today. sorry reviewers :)

10:25 <apritzel> macromorgan: I think the boost regulator node name must be just "boost", not "reg-boost". The alias name can be anything, but the actual node name has to match the name in the driver

11:41 <apritzel> Jookia: which reviewers? There seem to be almost no people caring about reviewing U-Boot patches in the last months :-( If you know people that would care, please CC: and poke them, we can use any help we can get!

13:15 <macromorgan> Jookia: That's nothing, I've been meaning to resubmit a v2 version of a patch for over 2 years now dealing with UBIFS and slc-emulation. I'm probably the last madlad there is caring about mainline on my PocketCHIP

14:21 <macromorgan> so the nmi seems to do an ioremap to a different address... does that mean it should really be at the different address?

14:23 <macromorgan> sets "pad control" to 0x7000208: https://gist.github.com/macromorgan/7f7df7860e8d6d09af7fce1885523624#file-sun50i-h700-anbernic-rg35xxph-dts-L6093

14:23 <macromorgan> pad control = ioremap: https://github.com/CrealityTech/sonic_pad_os/blob/main/lichee/linux-4.9/drivers/irqchip/irq-sun8i-nmi.c#L129

14:59 <apritzel> macromorgan: that's interesting. I stumbled upon this pad conntrol in the manual as well, but I think it's unrelated, since it seems to be an *output*?

15:00 <apritzel> so the RTC can signal its interrupt condition on this pad/pin, and this actually wakes up the PMIC, for instance?

15:01 <apritzel> but you said it works with the mainline driver?

15:04 <apritzel> I played with the NMI registers in U-Boot, and managed to trigger an IRQ (probably due to changing the trigger type), and I could clear it through the status register, so that seems fine to me

15:04 <apritzel> I haven't checked whether it's forwarded to the GIC, though

15:52 <macromorgan> I'm just trying to figure out what the heck that even does. Turns out I know a lot less about RAM than I thought I did...

16:40 <acmeplus> tokyovigilante: I received a 2024 the other day, so will check the DTS and post the stock one for comparison. I assume it's 99% the plus without wifi/bt, but most likely battery is different too

16:41 <acmeplus> And apparently the RG28XX is going to be very similar too

17:05 <apritzel> if we are not sure about the RG35xx-2024, we could leave that out, for the initial post. How each DTB is compiled from what source files can change internally anytime, without breaking something. We can easily add support later.

18:07 <macromorgan> while it will require working on all 3 (which I eventually planned to do, just chose the H model as my primary target for bringup) I'd be shocked if the difference wasn't missing wifi for the 2024 model and missing adc joysticks for the H model

18:07 <macromorgan> tiny chance there could also be a difference with the audio output path as well as some button layout

18:58 <macromorgan> hmm... trying to figure out the right way to plumb the GPIOs for the USB host port. So I've got 1 GPIO that turns on a 5v regulator for the host port, 1 GPIO that seems to power the USB controller on the SoC (wired into the power domain?), and 1 GPIO that detects whether or not a device is plugged in.

20:54 <Jookia> aperezdc: you reviewers :) though i am dropping some reviewed-by on the list

21:04 <Jookia> oops

21:04 <Jookia> i tabbed the wrong person

21:04 <Jookia> apritzel: ^

21:06 <Jookia> I'm not too sure how much reviewed-by/tested-by is going to help get patches merged but I feel bad not tagging patches with those

21:20 <tokyovigilante> oh nice, thanks acmeplus. Good to see they are hopefully basing a few devices around the H700 for consistency

21:25 <Jookia> i have to admit, using b4 for patches has turned my intense hatred of mailing lists in to a little bit of enjoyment

21:34 <Jookia> btw the 'post patches even if they don't get merged, other people can use them' approach is why i'm doing it at this point

21:34 <Jookia> no pressure on merging my amazing patches :)

22:14 <dok> Jookia: you're working with sunxi devices?

22:14 <Jookia> dok: yes at the moment :(

22:14 <Jookia> novena stuff is still on my backlog

22:14 <Jookia> what about you?

22:15 <dok> trying to dust off my patchset to support sunxi a64 in barebox

22:16 <dok> currently in the rabbit hole of also adding support for H6

22:17 <dok> respectively pine64 and orange pi board

22:19 <Jookia> oh cool! i'd like more sunxi in barebox

22:19 <dok> Jookia: was novena something related to a stm SoC ?

22:19 <Jookia> oh! i'm confusing you with someone else

22:19 <Jookia> ok i remember you now

22:19 <dok> its alright

22:21 <Jookia> the t113 might go well with barebox, not sure

22:21 <Jookia> it's a nice chip

22:22 <dok> is novena the laptop thing on crowdsupply ?

22:26 <Jookia> yeah, it's an old piece of hardware i want to get working again and mainlined

22:31 <apritzel> Jookia: just wanted to avoid disappointment by stating that U-Boot patches are severely review-limited

22:31 <Jookia> apritzel: yeah i understand :)

22:52 <Jookia> without DMA usb gadget is taking 2 minutes to upload a 7 megabyte file to RAM. is this normal?

22:54 <Jookia> that's 60KiB/s :(

23:22 <Jookia> ok, fastboot actually works really fast. but bootm breaks. hmm

23:23 <Jookia> (is that why they call it fastboot?)

23:43 <apritzel> Jookia: there is surely something fishy with the 60 KiB/s. With PIO you would expect something in the ballpark of the FEL upload speed, which is typically about 500 KiB/s (differs per SoC)

23:44 <Jookia> fastboot manages to upload 7MB in 0.7 seconds

23:44 <apritzel> what's the other method, exactly?

23:44 <Jookia> dfu

23:45 <apritzel> ah, have you tried ums or the Ethernet gadget?

23:45 <Jookia> ums won't work for this, Ethernet gadget crashes