08:17 <exkc> oh It seem i can get helf of core working by copying f's uboot spl.

08:17 <exkc> *half of the core

08:18 <exkc> i mean on belink gt1

08:18 <exkc> with uboot spl

08:19 <narmstrong> nice, now find out how to enable the second cluster and you're done!

08:21 <exkc> or i will get mad bc of the broken ram/ram contrller

08:22 <exkc> one of them is broken on mine

09:46 <f_> exkc: what?

09:46 <f_> show me logs!! :)

09:46 <f_> Also I'm not `f`, I'm `f_` lol

09:47 <f_> I want to see logs!

09:47 <f_> As for RAM it could be an SPL issue.

09:50 <f_> logs ~> https://bin.vitali64.duckdns.org or https://paste.debian.net

09:51 <f_> Just so that I can document how well it works

09:53 <f_> To be clear, I want SPL+BL30+BL31+U-Boot logs

09:54 <f_> Also did you try booting Linux on it with SPL? Try to see if it also doesn't panic/oops after a while (look closely at UART logs)

10:04 <f_> exkc I also assume you're running my upstream TF-A fork..so of course the 2nd cluster won't work AFAIK

10:04 <f_> Amlogic BL31 won't work

10:07 <f_> Have no doubt amlogic bl31.bin will freak out upon seeing how different SPL's params structure is

10:56 <f_> rockosov: BTW for flashing I plan to add DFU support in U-Boot SPL so that we can use it for flashing :)

10:57 <f_> (if you missed: U-Boot SPL boots on S905 and S905X, but my S905 box panics because of broken DRAM init (regression, I found the commit I made introducing the issue))

11:19 <lvrp16> narmstrong: in the end, it's these crappy SD cards that leak vddio to vcc. i added a hack to scp to drop vddio to 1.8v during poweroff and the voltage is low enough to reset the sd card so it doesn't hang.

11:20 <lvrp16> improves compatibility with a bunch of kingstons and that company that acquired lexar

11:21 <f_> lvrp16: Aha, amlogic SoCs now have a "low-quality" SD card detect function!

11:23 <f_> Fourtunately for SPL development I used a samsung SD card :)

11:26 <lvrp16> f_: yeah, after playing with so many SD cards, I kind of learn all their quirks. SanDisk doesn't care about IO voltage negotiation, you can straight drive that thing at 1.8V and it'll work

11:27 <lvrp16> Samsung adhere to spec pretty seriously

11:28 <f_> hmm

11:28 <f_> >The standard was introduced in August 1999 by SanDisk, Panasonic (Matsushita) and Toshiba

11:29 <f_> https://en.wikipedia.org/wiki/SD_card

12:47 <lvrp16> is there a way to hijack psci reboot? I can control suspend and power off from SCP but not reboot with the closed firmware

13:20 <rockosov> f_: What do you mean by DFU? Is it Amlogic custom flashing protocol?

13:22 <rockosov> f_: Because in the current moment we have finished ADNL protocol implementation (A1, C3, T7, S4, other new SoCs), and 50% finished old protocol, optimus (in the Amlogic terms)

13:42 <narmstrong> rockosov: for non-product boards, we plan to use DFU which is a standard USB prococol to fully chainload SPL, U-Boot and Linux via US

13:43 <narmstrong> rockosov: having ADNP protocol in U-boot implemented in a clean way is still a great thing!

13:43 <narmstrong> *ADNL

13:44 <rockosov> narmstrong: But DFU will not work in the case of BootROM-started flashing process

13:44 <rockosov> narmstrong: But I see, you mentioned "non-product" boards

13:45 <narmstrong> rockosov: no, the initial protocol would be the amlogic one, then SPL will switch to DFU to load scp/bl31/u-boot then u-boot would also enable SPL to load a fitImage

13:45 <narmstrong> rockosov: most board don't have a predefined storage media, so no partitions scheme

13:45 <rockosov> Ah, okay, I see

14:52 <f_> rockosov: basically, we would first load SPL using the stripped down amlogic protocol, then use DFU to load scp+BL31+u-boot, then U-Boot will also enter DFU mode where you'll be able to actually flash the device.

15:07 <f_> Or....do what narmstrong does and boot Linux from USB :)

15:08 <narmstrong> yes the second part is already done and is used to daily boot u-boot over USB to pass the pytest suite

15:09 <f_> Only the first part has to be done by someone....who still has to debug some other issues...

15:09 <f_> But it'll be done for sure!

15:47 <rockosov> f_: narmstrong: usually we use fastboot with the same workflow (but w/o SPL)

15:47 <f_> IMHO u-boot SPL is still not ready for production

15:48 <rockosov> Anyway, I hope custom Amlogic protocols will be helpful as well, so we continue to do hard work :)

17:19 <f_> I love standards :)

17:30 <exkc> i havent power on my gt1 for a while.i remeber it has memory error last time i powered it on. I was using u-boot-beelink-gt1 from pmOS.

18:55 <f_> exkc: You just said you were able to boot with u-boot SPL?

19:05 <exkc> sry Ah i dont what was thinking i should said i should able to.