02:05 <chris__> succesfully snooped i2c transfers by patching the binary-only kernel... but didn't find traffic to peripheral

02:05 <chris__> consulting the decompiled device tree reveals i2c address in linux is right shifted one bit from the peripheral data sheet!

02:05 <chris__> apparently the direction bit. now I go the traffic in my /proc/kmsg :-)

09:32 <karlp> chris2: yeah, i2c is annoying like that.

09:33 <karlp> some docs use shifted, some unshifted.

12:42 <rperier> hi all, correct me if I am wrong, but there are no ways to give the address of an ll_debug uart directly into the devicetree ? (stdout-path... is a kind of earlycon right ?)

12:44 <rperier> it's too bad, because if this value is hardcoded in the kernel image... you cannot use this image with different platforms (with the same SoC) when each of these platform use a different uart interface. Even, if yes... ll_debug is for debugging normally... I agree...

13:17 <mmind00> rperier: there is now also earlycon support for arm32 :-)

13:18 <mmind00> rperier: and yes debug_ll phys is hard-coded, but it is also usable in the decompressor ... so just switch to earlycon if the decompressor works :-)

13:37 <rperier> mmind00: mhhhh... in fact I thought that earlycon was only to reconfigure the uart (I mean completly), apparently it strongly depends on the drivers which is used... (we use "cadence" for uart... it should do the trick)

13:37 <rperier> hehe yeah... it seems to work :D

13:38 <mmind00> rperier: yes, earlycon is connected with the uart driver, but that makes it possible not having to rely on the bootloader to set up stuff

13:39 <rperier> indeed

16:36 <mmind00> rperier: as you're the one with speedy, could you take a look at https://github.com/mmind/linux-rockchip/commits/tmp/analogixdp-veyron ... especially if the edp hotplug works there or we need the override too?

17:09 <rperier> mmind00: I will add this to my todolist for this week end , I already need to merge the branch for veyron into the master branch for meta-rockchip , I will test what you asked this week end. I started to play a bit with edp and backlight on speedy but it was with the chromeos driver not with analogix. However I can rebase my work on top of it :)

17:10 <rperier> mmind00: btw, do you still have this "white flickering" issue when you boot mainline ?

17:13 <mmind00> rperier: not sure what flickering issue you mean ... my jerry comes up with a somewhat gray screen and then goes to the console

17:14 <rperier> on my veryon, the driver pwm-backlight is started and probed before the edp is bounded to the vop and before we get something to display on the screen. In this case, the pwm-backlight driver defaults to the default brightness

17:15 <mmind00> rperier: but dianders pointed me to https://chromium-review.googlesource.com/#/c/287490/ today ... maybe your speedy has a similar issue?

17:16 <rperier> mhhh... it seems to be another way to solve the issue... (at least... as I understand, but I need to investigate to have more context...)

17:18 <rperier> in the chromeos kernel, the "backlight-boot-off" property initializes the backlight to be power down by default until it is used by the edp

17:33 <amstan> rperier: you mean you get a white screen for a few seconds?

17:33 <amstan> i think i saw that a few times

17:39 <rperier> amstan: exactly

17:39 <amstan> one way to surely do that though, is to load the wrong dts file

17:39 <rperier> when I investigated, I found that a way to solve this issue is to turn the backlight to "powerdown" until something is bounded to it

17:39 <amstan> i booted a few times forgetting to add it in, and it tought it was a firefly

17:39 <amstan> so white screen and crash halfway through booting

17:40 <amstan> but i'm sure i've seen it with the proper dts too

17:40 <rperier> in my case, I used the right dts ;)

18:07 <mmind00> norris: if you have time at some point, could you please also try https://github.com/mmind/linux-rockchip/commits/tmp/analogixdp-veyron on your jaq and see if the edp-hotplug works or we need some other special handling

18:09 <norris> mmind00: will put it on my TODO; I'm trying to recall at what point (and with what extra patches) I had eDP working...

18:10 <mmind00> norris: thanks ... that branch with Yakir's new and shiny analogix-edp should work nearly directly ... only the board-specifc magic is missing

22:11 <bashintosh> Does someone happen to know if u-boot can perform a thorough test of RAM ICs? A memtest86-like thing...

22:52 <sjoerd> bashintosh: there is some memtest cpability, don't know how thorogh it is though