00:06 <smaeul> MoeIcenowy: in https://lore.kernel.org/linux-arm-kernel/20161213152252.53749-6-icenowy@aosc.xyz/ how did you determine if DCDC2 (vs DCDC3) was the CPU supply? is this reported somewhere in the FEX/dmesg/sysfs?

00:07 <smaeul> I have an A33 tablet (no schematic) with some evidence of DCDC3 being the CPU supply. I would be surprised if this tablet didn't follow the reference design, but I would also be surprised if DVFS worked with the wrong supply in the devicetree

00:09 <smaeul> evidence is this line in /sys/class/regulator/dump which matches the DVFS table for the current CPU frequency (1.2 GHz):

00:09 <smaeul> axp22_dcdc3 : enabled 0 1320000 supply_name: axp22_dcdc3 axp22_dcdc3

00:10 <smaeul> and that I can toggle DCDC3 in system suspend without hanging or breaking resume, but if I leave DCDC3 off, resume fails

00:24 <apritzel> smaeul: interestingly in this very file the regulator-name of reg_dcdc3 is "vdd-cpu" (even though indeed the cpu node points to reg_dcdc2)

00:43 <apritzel> also U-Boot's Kconfig says: "On A23 / A31 / A33 boards dcdc3 is VDD-CPU and should be 1.2V."

00:46 <apritzel> and with U-Boot setting DCDC3 to 1.2V, and Linux probably never changing this, the CPUs should always get a comfortable voltage, even for the highest frequency

00:52 <apritzel> so could it be that we indeed change the wrong rail, and the CPU is just somewhat resilient against VDD-SYS changing?

00:53 <apritzel> The A33 Olinuxino schematic says DCDC3 is CPU and DCDC2 is SYS

14:11 <junari> made pinctrl patch for h616 with extra gpio pins https://gist.github.com/iuncuim/0b1a871a52d164f42ac57a2fa44b08e3

14:12 <junari> not sure about pinmux names

14:31 <apritzel> junari: that looks alright, but we need that on the mailing list

14:58 <wens> apritzel: just looked at the a23 and a33 tablet reference designs, DCDC2 is SYS+GPU, DCDC3 is CPU

15:00 <wens> same for a31 reference design

15:01 <apritzel> wens: so this is then basically a typo in sun8i-reference-design-tablet.dtsi?

15:09 <wens> or misreading the schematic?

15:12 <apritzel> smaeul: can you do some experiments on your tablet to confirm this? Like running at the highest frequency, with DCDC2 at 1.2, but DCDC3 at 1.04V, and waiting for crashes?

15:13 <apritzel> then setting DCDC2 to 1.04V and DCDC3 to 1.2V, and not seeing any crashes?

22:12 <tuxd3v> hello guys,

22:12 <tuxd3v> for allwinner H2+(bananapi m2 zero), is there are development being made, that is not in the last lts kernel?

22:14 <tuxd3v> When you build your kernels with arm toolchain >8.3 doesn't a buindinfo file is created saying that your kernel is tainted?

22:14 <tuxd3v> buildinfo*

23:10 <apritzel> tuxd3v: what does "arm toolchain 8.3" refer to?