12:03 <rperier> sjg1_: around ? did you notice the regression for back_to_brom on rk3288 with u-boot-rockchip.git and u-boot.git both master ? (just in case)

12:03 <rperier> I have started to investigate...

12:21 <mmind00> rperier: try https://www.mail-archive.com/u-boot@lists.denx.de/msg251268.html

12:23 <mmind00> rperier (and sjg1_): the recent refactorings did break some expectations of the bootrom (2k blocks and init_boot_size settings) and that series seem to fix these issues again

12:34 <rperier> oh interesting, I will take a look

12:34 <rperier> thanks :)

12:34 <rperier> tbh, I can debug and write code for uboot itself or driver , but I think that my brain is not ready yet to play with spl and back_to_brom stuffs :D

12:49 <mrjay> rperier: it's not that hard :-)

12:52 <mmind00> rperier: and also the spl is not that different from uboot-proper, just smaller [which is the overall goal ;-) ]

12:54 <mrjay> rperier: mmind00 yeap tpl and spl are separate builds of u-boot source, lets say spl is tiny and tpl even tinier :-) build

13:01 <rperier> mmind00: no that's more the hw initializing sequence phase that is complicated and for debugging also, compared to a normal uboot I mean

13:01 <mrjay> and back_to_bootrom() is just asm code to return to bootloader execution. For example on rk3066 bootloader first reads 1KB data(TPL) from nand and executes it. Than you need to do back_to_brom() to return to bootloader which reads another 32KB(SPL) of data to sram and executes it(sdram init code needs to be there). Than you need to do another back_to_bootrom() and bootloader reads ~200KB data to RAM and also executes it :-)

13:02 <rperier> I have tried this https://marc.info/?l=u-boot&m=149618000904178&w=2

13:02 <rperier> it does not work better :/

13:04 <rperier> ah, the 3/3 perhaps also...

13:10 <mmind00> rperier: yep, patch3 (init_boot_size) is _really_ important, because this tells the bootrom how much code of your real uboot to load into sdram :-)

13:13 <rperier> so basically the whole serie (patch 3 needs patch 1), not because I have to rebase it by hand onto master (does not apply here)

13:13 <rperier> :)

16:42 <ayaka> mmind00, I have tried many time with the upstream u-boot, it always stop at spl as the document said

17:24 <mmind00> ayaka: what document? If you're downloading and running via usb that is to be expected ... depending on which method (mmc vs. back_to_bootrom) you're using you need to write spl and uboot to specific positions on the emmc (in the back_to_bootrom case, mkimage the spl-part and then cat the uboot.bin after that)