00:06 <apritzel> Jookia: not sure what you are asking about, exactly? I am all for display support for modern SoCs in U-Boot, if that's what you meant, but we have to do it properly, so using DT and DM

01:04 <apritzel> jernej: Excellent, saving/restoring SP_IRQ worked! Many thanks for digging into this! Made a proper (fix) patch: https://github.com/apritzel/u-boot/commits/a523/

01:39 <apritzel> jernej: How is JTAG handled, exactly, as a bit-bang serial protocol? Could you choose other pins than ADBUS[0-3]?

01:39 <apritzel> because then you could connect the serial pins as well, to ADBUS[0-1], and switch between serial and JTAG, via USB?

01:43 <apritzel> (with serial pins I mean the UART0 muxes on PF2 and PF4)

02:48 <Jookia> apritzel: i have other u-boot patches for an SPI panel, PWM, SPI NAND booting, but there are no in tree users of those either

04:43 <jernej> apritzel: nice! FTDI datasheet specify which pins can be used for JTAG, so I think there is special logic for it inside. FT2232 chip can have UART on second channel. But I find this not that useful due to special code needed to initialize UART on port F.

07:38 <gamiee> apritzel / jernej : switching between JTAG and UART would be probably painful on this FTDI chip, mainly due driver issues.

07:38 <gamiee> Also, good job folks on solving that issue. It was fascinating to watch the progress.

10:53 <apritzel> Jookia: many new boards use SPI NAND flash, so that's definitely interesting

10:53 <apritzel> Jookia: also, as I said before: by submitting your DT, you make it an upstream board ...

10:56 <Jookia> Submit a DT for prototype dev boards?

10:59 <Jookia> There's no PCB yet, just perf boards at this point at best

11:03 <apritzel> Well then try to make it work on existing boards. Your changes seem to be around generic peripherals? Like panels and SPI flash, that's not board specific.

11:04 <Jookia> Yes, they do work on the mangopi

11:05 <apritzel> for instance the Lctech Pi F1C200s has SPI NAND, as do some other D1/T113 boards I believe. And many of the A523/T527 boards have SPI NAND as well

11:09 <Jookia> But the mango pi supports spi nand?

11:10 <Jookia> all of this has been done using the mango pi dual

11:11 <apritzel> so what is the question, exactly?

11:12 <Jookia> I don't even know any more, sorry

11:14 <apritzel> what I am saying: if you choose UART2 on PortD on *your* particular board, and that causes major pain in U-Boot, or your support patches break something, then I am not really happy about that

11:15 <apritzel> but: if you want SPI NAND support (in general), then sure: there are plenty of users out there who would benefit, so we want this anyway, regardless of your particular use case

11:16 <apritzel> same with display engine support or even panels

11:39 <Jookia> ok

19:58 <jernej> apritzel: should U-Boot proper by default use PSCI handlers for reset and such? Because it doesn't currently on A523.

22:55 <apritzel> jernej yes, should be easy, it's just one Kconfig option. H616 as well, we are missing that mostly for historical reasons

22:57 <jernej> I enabled psci reset but by default it enables also spl version and breaks build due to missing DM symbols in spl

22:57 <jernej> SYSRESET_PSCI enables also SPL_ARM_PSCI_FW, which requires DM

22:58 <apritzel> mmh, wasn't there a patch for that? i thought smaeul fixed that back then?

22:59 <jernej> and I narrowed down TF-A issue by enabling and disabling cores - reading PPU_PWSR() hangs cpu

22:59 <jernej> apritzel: no idea

23:00 <apritzel> maybe it was a patch of mine, but i never merged it? somehow rings a bell

23:12 <apritzel> jernej i think this one https://lore.kernel.org/u-boot/20240509230600.11306-1-andre.przywara@arm.com/

23:13 <jernej> looks about right