04:22 <Ke> hanetzer: # CONFIG_USB_DWC2_HOST is not set

05:29 <teletype> hi all, i've tried several kernels, currently trying walle-kernel-rk3188, trying to get an rk3188 SOM of unknown provenance to boot from SD. this one is the furthest i've gotten, but it hangs early in the boot when the kernel tries to init DDR. anyone know what i might be doing wrong?

05:34 <Ke> does kernel init DDR?

05:35 <teletype> i don't know. been working with linux forever but pretty new to rockchip. i do know that with the same bootloader image, i tried a kernel i extracted from a rom for an olimex board and it did get past the ddr init, but crashed later in boot

05:36 <teletype> this is what i'm seeing w/ the kernel i built https://hastebin.com/xomubuyotu.cs

05:37 <Ke> right, it's booting kernel indeed, not sure that is DDR init though, just printing out some details

05:38 <teletype> well, the olimex kernel gets further and the next line it prints after that paste is: [ 0.000000] DDR DEBUG: init success!!! freq=300MHz

05:38 <teletype> so i assume it's hanging during ddr init

05:44 <teletype> hrm, don't have a proper debugger on this, but there's not a whole lot of code between the last line it's logging and the next line i expect to see. time for some printf debugging i guess

07:14 <raphari> hi all, trying to identify parts of the code in bsp (uboot/kernel) that need to be patched when you upgrade DDR3 to DDR4 for rk3399. RK3399MINIALL.ini

07:18 <raphari> identified ./linux/rkbin/RKBOOT/RK3399MINIALL.ini & ./linuxkernel/arch/arm64/boot/dts/rockchip/rk3399-dram-default-timing.dtsi

07:55 <raphari> The files seems to have the required patch, but the kernel fails to boot itself completely and these are the complete logs set https://pastebin.com/wgnDe54P

07:55 <raphari> last message from kernel is "rockchip-dmc dmc: could not find power_model node"

07:57 <teletype> printf debugging worked. for some reason the rk3188 dpll stuff doesn't agree with this board. not sure why yet but disabling it makes my kernel boot.

12:41 <hanetzer> Ke: yeah, but DUAL_ROLE is set, which is both _HOST and _PERIPHERAL

12:41 <Ke> ok, if you are sure

12:42 <hanetzer> I'm not sure :P

12:45 <Ke> I remember I once had ext3 journaling on and wondered, why I can't mount ext3 filesystems

12:50 <hanetzer> Ke: I'm not sure, but my grasp of how kbuild works leads me to believe that its correct

12:51 <Ke> if it would use the same code, I believe the convention would be to depend on CONFIG_USB_DWC2_HOST wouldn't it?

12:52 <hanetzer> apparently not. ifneq ($(filter y,$(CONFIG_USB_DWC2_HOST) $(CONFIG_USB_DWC2_DUAL_ROLE)),)

12:53 <hanetzer> this adds hcd.o, hdcd_intr.o, hcd_queue.o, hcd_ddma.o to the driver, which are the host dodads

13:02 <Ke> I kind of like that there is no way to know which compatibles are found, this is like one command in Microsoft Windows 10

13:03 <hanetzer> Ke: rephrase?

13:03 <Ke> compatible is the member in dtc that tells which drivers to expect to be needed for hw

13:04 <Ke> like pciid

13:04 <hanetzer> ah, you can check that in the sysfs

13:05 <Ke> I don't believe you can check drivers

13:10 <Ke> rockchip,rk3288-usb", "rockchip,rk3066-usb I believe is the device

13:11 <Ke> or snps,dwc2

13:12 <Ke> though seems to use the same device as the otg so I guess that would not even be informative

13:23 <hanetzer> usb@ff540000: rockchip,rk3288-usbrockchip,rk3066-usbsnps,dwc2

13:25 <Ke> yes that one

13:25 <Ke> you need to have a driver for it, but having a driver for it is not sufficient

13:33 <hanetzer> Ke: further investigation shows if I boot with a usb device plugged in, it shows up in lsusb, but replugging it to the same port looses it

13:33 <hanetzer> which explains why the webcam is always there

13:34 <Ke> webcam IIRC was different

13:34 <Ke> but anyway

13:34 <hanetzer> but it is usb at least

13:35 <Ke> yes USB, I remember having this sort of issue also caused by a missing driver

13:35 <Ke> thought I don't remember the details

13:47 <Ke> anyway, I don't think I have any more useful input

13:49 <Ke> https://users.aalto.fi/~jkarlson/files/config-4.9.60 well outside of this config that worked for me

22:04 <hanetzer> so, does anyone have working mali accelerated gfx with the libmali.so blob on rk3288, and would care to assist?

23:29 <amstan_> hanetzer: CONFIG_USB_DWC2_DUAL_ROLE is fine i think

23:30 <amstan_> it just means that it's capable of doing both, the dts is then consulted (the dwc2 entry should say : host;)

23:30 <hanetzer> amstan_: theoretically it brings in the same code, so yeah, but in practice I'm seeing usb devices only being recognized/enumerated if they are plugged in during boot

23:30 <amstan> have you tried older kernels? it might be a regression

23:31 <hanetzer> I went as far back as 4.17.0, same issue. haven't gone farther than that to be fair

23:33 <amstan> try something like 4.14, even 4.10 should work on that device

23:33 <amstan> if you ask nicely maybe mmind00 can give you a hand, he might have a couple of similar devices in his farm

23:34 <hanetzer> amstan: will do.