02:46 <anarsoul> mmind00: it didn't help

03:05 <anarsoul> it's stuck after this https://gist.github.com/anarsoul/3cb589e868ff99c6b9270d43231c805d

03:12 <anarsoul> dropping CONFIG_ROCKCHIP_SPL_RESERVE_IRAM helped

03:41 <anarsoul> OK, and vop probe fails because: clk_set_default_rates: failed to set rate on clock index 0 (219) for vop@ff8f0000

03:48 <anarsoul> removed assigned-clock-rates for vops and now it proceeds further

03:49 <anarsoul> fails at rk_display_init(vop@ff900000): no UCLASS_DISPLAY for remote-endpoint

04:00 <anarsoul> hm, it's not the issue actually, it just probes all the endpoints

04:21 <anarsoul> rk3399 clock driver doesn't support ACLK_VOP0 and ACLK_VOP1 that's why probe fails

04:21 <anarsoul> kevery: ^^

04:23 <anarsoul> also assertion at drivers/clk/rockchip/clk_rk3399.c:324 fails with DEBUG enabled

04:26 <anarsoul> HCLK_VOP0 is not supported either

04:27 <anarsoul> *sigh*

06:16 <levd> i managed to make TPL work on firefly-rk3288 using TF card, but found something quirky: 1. must turn on DEBUG in `drivers/mmc/dw_mmc.c` or mmc read of multiple sectors (aka loading the full u-boot image); 2. must adjust `u-boot,spl-boot-order` to "same-as-spl", &sdmmc, &emmc .

06:19 <levd> I also find that by turning DEBUG on spl-boot-order.c, the following error shows: board_spl_was_booted_from: brom_bootdevice_id 5 maps to 'dwmmc@ff0c0000'\n board_boot_order: could not find dwmmc@ff0c0000 in FDT .

06:23 <levd> that is the following call failed: node = fdt_path_offset(gd->fdt_blob, "dwmmc@ff0c0000"); but I can see that "dwmmc@ff0c0000" is defined in rk3288.dtsi. i know there must be some magic configs not turning on, any clue?

06:45 <levd> good, fdt_path_offset(gd->fdt_blob, "/dwmmc@ff0c0000") works!

06:46 <levd> kevery: shall entries of boot_devices array in arch/arm/mach-rockchip/rk3288/rk3288.c need to be prefixed with "/" ?

06:48 <kevery> Yes, seems the '/' is missing.

06:52 <levd> only rk3399.c has the '/' present :)

06:57 <levd> now i only got some quirky thing in making TPL work on firefly-rk3288 using TF card: must turn on DEBUG in `drivers/mmc/dw_mmc.c` to make read the whole u-boot image. note that with the DEBUG, reading header of the u-boot image actually works. any idea? kevery

06:58 <levd> fixme: note that *without* the DEBUG, reading *header* of the u-boot image actually works. any idea? kevery

07:08 <kevery> Rockchip U-Boot emmc dirver is very different with upstream, we have improve a lot features in local fork, so I have no idea about upstream version emmc driver. We use to try to make it better, but the emmc driver upstream is very very slow in the pass two years and we have to drop the upstream work for it.

07:09 <kevery> Instead of make the big improve of emmc driver, we could at least make the SD card work for boards, because it use to OK for firefly-rk3288.

07:48 <levd> what version of firefly-rk3288 board do u use?