09:38 <loki666> apritzel u-boot ram detection on H700 is still failing about 1 time over 4, any idea or patch I could try. I've already all the patch recently applied to u-boot custodian

19:55 <apritzel> loki666: that's sad, also not the first time I hear this

19:56 <apritzel> This is on an Anbernic device? And when does this happen, exactly? On a reset (watchdog reset? reset button?) or on a cold boot? So can start the SPL, then reset multiple times and it eventually happens?

19:57 <apritzel> also: do you have an idea or can dig in to find out what exactly happens? So at what point the DRAM init code takes a wrong turn to calculate the size wrongly?

20:12 <loki666> Cold boot, a couple of reset usually does the trick

20:15 <loki666> I don't have the slightest clues at how to debug this issue... I'm more of a high level developer (Kotlin, python, go). I can read and write c, but the HW programming is... Difficult

20:18 <loki666> I think tokyovigilante uboot fork was working better (could still fails sometimes, but more like 1 over 10) but it's not working anymore since mainline DTS moved PMIC to I2C bus

20:26 <loki666> That being said, if you can guide me to debug stuff (print regs or I don't know) that could help you, I'm all good to try (and learn something in the process)

20:46 <apritzel> loki666: was it on those Anbernic devices that we saw the DRAM voltage increased to 1.16V (on the BSP)? Or do I mix this up?

20:47 <loki666> It's on an Anbernic device the RG35XX family, but I don't know about this voltage increase.

20:47 <loki666> Uboot detects 2Gb instead of just 1

20:48 <loki666> And then crash

20:58 <apritzel> can you try to change that line in the defconfig to CONFIG_AXP_DCDC3_VOLT=1160, and see if that changes anything?

21:45 <loki666> ok