05:11 <zamon> I was tring to enter FEL mode yesterday. I have tried a lot of methods one of which is https://linux-sunxi.org/FEL#Through_a_special_SD_card_image. Now I can get into FEL mode, But I found my SD card that burned the image was broken. :(

05:12 <zamon> I was tring to enter FEL mode yesterday. I have tried a lot of methods one of which is https://linux-sunxi.org/FEL#Through_a_special_SD_card_image. Now I can get into FEL mode, But I found my SD card that burned the image was broken. :(

05:12 <zamon> This is my log, https://gist.github.com/sycamoremoon/435509fc80416e18f4d0b8a892599cd6

05:21 <jernej> Lightsword: That is just for lower consumption when in standby. It's not something that's important for normal operation, so it's not implemented (yet)

06:07 <Lightsword> jernej, it's not used for say resetting a stuck bus or anything like that?

06:09 <jernej> not that I would know

06:10 <jernej> I don't think it would help with stuck buses. It only puts pins in hi-Z state via pin controller.

06:11 <jernej> if, let's say, I2C device would be stuck, you would need to bit bang special sequence to unstuck it or power cycle it

07:44 <Lightsword> jernej, hmm, I saw some stuff about the kernel having some functionality that can be used to bit bang gpio's or something, possible that's what's needed to enable the i2c3 bus normally? I've been trying a bunch of different ways to try and unlock that bus but I haven't been able to find anything so far that has any real effect. Have you ever seen sunxi i2c buss locked errors before?

12:37 <dlan> montjoie: FYI, i've made ethernet(emac0) work, check https://github.com/dlan17/u-boot/tree/allwinner/A523

14:48 <loki666> jernej: I found this https://lore.kernel.org/dri-devel/cbf20469-02ab-403a-8db7-2b66e9936b4f@samsung.com/T/#m72ec9cac75b5d705b42fdbded9c6b114fc96cddd which looks very similar to what we need for h616 g31

18:06 <jernej> loki666: not really, no. besides, we have it solved already

18:08 <jernej> Lightsword: I still believe you have issue with pwm and that's why i2c doesn't work. Maybe I'll find some time next week to test this on one of my H616 boards

18:09 <Lightsword> jernej, that's the "allwinner,sun50i-h616-pwm" device tree node?

18:11 <Lightsword> I think I did modify that driver a bit to try and force it into bypass mode

18:11 <jernej> correct

18:11 <Lightsword> maybe I didn't do that right though

18:11 <jernej> in any case, I'm out of ideas to help you, so I need to test it myself

18:12 <loki666> ok, I rocknix team member did test my patches on various panfrost platforms without issues, but I didn't saw his mail lands in dri-devel (but did on sunxi)

18:12 <Lightsword> jernej, this is the diff of the change I made to try and debug the pwm and force it to bypass: https://gist.github.com/jameshilliard/55b0c300e5de708226e63ed2940cba6c

18:35 <jernej> override looks ok, but it's always good to check with direct reg read via devmem2 or similar tool

18:38 <Lightsword> jernej, hmm, yeah, I had tried poking some i2c3 config registers via devmem2 and was getting results that looked off

18:49 <Lightsword> jernej, should "devmem 0x0300A040 w" show the pwm enable register? looks like it's not being enabled properly

19:33 <montjoie> dlan: thanks, I will try it soon

23:37 <Lightsword> jernej, hmm, so I'm not sure how to obtain accurate devmem values since the pwm driver seems to always try and revert the configuration due to the i2c failure, is there a way to override it so it retains the configuration even on i2c3 failure?