01:23 <Jookia> aperezdc: in anger? :O

08:19 <Jookia> yes

08:19 <Jookia> im sorry aperzdc :(

08:20 <Jookia> i tested and prototyped pwm clock setting code in python, i'm pretty pleased with the result. i think it's somewhat better than the linux code :)

08:22 <aperezdc> no woprries, it was a clear autocomplete thing, because they left so typing ap+tab probably hit me next in the list or something

08:23 <aperezdc> s/woprries/worries

08:34 <Jookia> :)

08:55 <tokyovigilante> Raqbit: sorry yeah my u-boot code is a bit behind. There is a TF-A branch which also includes the AXP driver which you may need, but I have previously had booting to linux working well without. I will refresh my branch either tonight or over the weekend. I do very occasionally see 2GB RAM reported, so I assume the DRAM settings are not entirely right unfortunately. However I see that maybe 1

08:55 <tokyovigilante> in 20 times, and the device boots ok otherwise.

08:58 <tokyovigilante> If you're using the device with a battery connected, maybe try unplugging it and just running 5v power from the USB cable, we had some issues during the AXP bringup where a rail was misconfigured and wouldn't set till both battery and USB power were pulled to fully reset it

08:59 <tokyovigilante> https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/27326

09:00 <tokyovigilante> for the AXP patches.

09:01 <tokyovigilante> apritzel: just confirming usb gadget ethernet should work from u-boot to allow booting a kernel over TFTP? I can't get u-boot to build with the gadget driver enabled currently, I think because I'm missing (or just not enabling the correct) peripheral driver

09:12 <apritzel> Jookia: the SPL does use DRAM later in its runtime, for the heap (used by the MMC code) and the BSS. But it's not until TF-A executing that code execution and caches come into play

09:13 <Jookia> tokyovigilante: you need a patch for sunxi musb

09:17 <apritzel> Jookia: really? what for?

09:17 <Jookia> so it compiles

09:17 <apritzel> I think you just need CONFIG_USB_MUSB_GADGET?

09:18 <Jookia> https://lore.kernel.org/all/20230608195631.55364-1-CFSworks@gmail.com/

09:18 <Jookia> i gave it a review/tested-by last month

09:19 <Jookia> someone else also made a duplicate patch and posted it to the mailing list earlier this year

09:20 <apritzel> but that's just for DM enablement, isn't it? gadget Ethernet should work already ...

09:21 <Jookia> i i dunno

09:22 <Jookia> maybe that's the error they got though

09:24 <Jookia> with the amount of compile errors you get if you enable a wrong setting it's hard to find what's broken or what's unsupported for your hardware

09:32 <tokyovigilante> Jookia: awesome, will give it a go. Was getting a few missing includes which were clearly defined in the various other peripheral drivers. will give it a roll

09:41 <apritzel> tokyovigilante: so the USB mass storage gadget in U-Boot definitely works, just tested that on the OPiZero3

09:42 <apritzel> just enable CONFIG_USB_MUSB_GADGET=y, for the generic MUSB part, this already automatically pulls in gadget Ethernet, and U-Boot prints some info about it

09:44 <apritzel> if you enable CONFIG_USB_FUNCTION_MASS_STORAGE=y and CONFIG_CMD_USB_MASS_STORAGE=y, you can export the SD card via "ums 0 mmc 0"

09:44 <apritzel> which would allow you to copy the kernel to the SD card, then load it from there. Not ideal, but better than pulling the SD card in and out

09:45 <tokyovigilante> cool, thanks. So you'd suggest that over trying TFTP if I get both working?

09:46 <apritzel> I will keep trying on the gadget Ethernet, it seems a bit impatient to me (timing out after one second already?)

09:46 <apritzel> so it depends on *your* impatience ;-)

09:48 <tokyovigilante> Extreme, but will see how I go :)

09:48 <apritzel> ah: setenv cdc_connect_timeout 20

09:49 <apritzel> maybe my host just doesn't have the USB Ethernet driver ready ...

09:53 <tokyovigilante> Raqbit: try https://git.sr.ht/~tokyovigilante/u-boot/tree/rg35xx-upstream

09:54 <tokyovigilante> that's u-boot master with the pending 6.10 version of the DT for the RGXXX devices, then with some extra nodes for DFVS, NMI etc on top

09:55 <tokyovigilante> ah, I didn't have CONFIG_USB_MUSB_GADGET enabled at all

09:55 <apritzel> ha, got it to work!

09:56 <apritzel> so it seems to work only on the first try, meaning you got to reset when something goes wrong

09:57 <apritzel> that's the sequence: setenv ipaddr 192.168.51.2; setenv serverip 192.168.51.1; setenv cdc_connect_timeout 100; tftpboot $kernel_addr_r Image-6.9-rc5.gz

09:58 <apritzel> then on your host: ifconfig usb0 192.168.51.1 (or whatever this is with ip ...)

09:58 <apritzel> 3.7MB/s, quite decent

10:02 <tokyovigilante> nice!

10:09 <apritzel> tokyovigilante: I ask people to add CONFIG_USB_MUSB_GADGET=y in the defconfig for every board that describes USB0 as being peripheral in the DT

10:09 <tokyovigilante> cool, will do

10:09 <tokyovigilante> hmm...Trying to boot from MMC1

10:09 <tokyovigilante> mmc_load_image_raw_sector: mmc block read error

10:09 <tokyovigilante> This is with u-boot at the 256 sector offset

10:10 <apritzel> oh yeah, you need a U-Boot patch to support that high offset on the H616

10:11 <apritzel> I have one, just didn't manage to send it out yet ...

10:11 <tokyovigilante> First time I've actually tried booting from the card

10:12 <tokyovigilante> although using FEL - eth0: usb_ether

10:12 <tokyovigilante> so that looks like its working

10:15 <tokyovigilante> actually, I could still use this with FEL-boot, because as soon as u-boot loads it will switch out of FEL and set up as a USB peripheral port presumably

10:15 <tokyovigilante> Obviously good to get the SD support sorted for an eventual downstream image

10:17 <apritzel> try this: https://gist.github.com/apritzel/423f96d84599d107383b557ca79bbf96

10:17 <apritzel> and yeah, FEL and gadget Ethernet pair up nicely

10:29 <tokyovigilante> Bus 001 Device 025: ID 1f3a:1010 Allwinner Technology Android device in fastboot mode

10:30 <tokyovigilante> Do I want this? Just getting a bunch of fastboot messages in the console currently

10:30 <tokyovigilante> oh no I just didn't interrupt the boot in time

10:32 <tokyovigilante> high speed config #1: 2 mA, Ethernet Gadget, using CDC Ethernet

10:32 <tokyovigilante> USB network up!

10:32 <tokyovigilante> now I just need networkmanager out of the way ;)

10:36 <tokyovigilante> thanks for the SD patch too

10:37 <apritzel> maybe you can actually talk networkmanager into automatically bringing up the usb0 interface when it appears, with a static IP, then the default three second timeout should be enough already

10:38 <tokyovigilante> oh yup

11:08 <tokyovigilante> awww

11:08 <tokyovigilante> => booti $kernel_addr - $fdt_addr

11:08 <tokyovigilante> Bad Linux ARM64 Image magic!

11:09 <tokyovigilante> tftp seemed to work though

11:09 <tokyovigilante> Bytes transferred = 24764928 (179e200 hex)

11:12 <apritzel> you should use $kernel_addr_r and $fdt_addr_r, I guess?

11:15 <tokyovigilante> sorry, just figured that out

11:15 <tokyovigilante> Starting kernel ...

11:15 <tokyovigilante> "Synchronous Abort" handler, esr 0x02000000

11:15 <tokyovigilante> elr: 000000000b5940e0 lr : 000000004a002fd4 (reloc)

11:15 <tokyovigilante> still nope, but awesome to get this up and running

11:28 <tokyovigilante> hmm, do I mean $fdtcontroladdr?

11:28 <tokyovigilante> Haven't built a FIT image, just a regular one so was going to rely on the u-boot DTS

12:23 <aperezdc> paru

12:23 <aperezdc> lol, that was supposed to go to a shell

12:24 <apritzel> tokyovigilante: yes, of course $fdtcontroladdr would be best

15:36 <Raqbit> tokyovigilante: Thanks, I'll see if I can get your updated u-boot working with the AXP TF-A branch. I initially tried without battery, but stopped doing so as the stock firmware was shutting down after a minute of running without (I'd assume that is coming from the PMIC)

21:59 <macromorgan> which is the latest U-Boot branch we're working from? Sorry but I am just now taking stuff back up as I had some unrelated U-boot bugs to quash

22:00 <macromorgan> this is the A-TF stuff I'm using at the moment: https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/27326/3

22:23 <apritzel> macromorgan: tokyovigilante pushed this earlier today: https://git.sr.ht/~tokyovigilante/u-boot/log/rg35xx-upstream

22:24 <macromorgan> sweet, I'll give that a whirl

22:24 <apritzel> I guess plus this patch if you want to load U-Boot from an SD card's 256K offset: https://gist.github.com/apritzel/423f96d84599d107383b557ca79bbf96

23:08 <tokyovigilante> apritzel: macromorgan: yup that is the 6.10 kernel DTS plus the minimum changes for DRAM and the AXP driver needed to boot, then some random GPU and video nodes on top which shoud probably be ignored ;) will add in or reference the 256k offset patch, and then start thinking about u-boot patches?

23:15 <apritzel> tokyovigilante: thanks, and yeah, it would be good to get them into shape. I will post my AXP717 patches in the next days, and I think we need some cleanup of the DRAM patches