01:01 <KREYREN_oftc> Does anyone know what's the idle and load power consumption of AllWinner A20?

01:03 <Jookia> i have a lime2 emmc, not sure about power use

01:03 <Jookia> i don't really have the tools to measure

01:03 <Jookia> i also have it modded to run from usb

01:24 <KREYREN_oftc> Jookia, it reports voltage_now and amperage_now from the AXP803 if you have it connected to the battery

01:24 <Jookia> i don't

01:24 <KREYREN_oftc> welp

01:24 <KREYREN_oftc> i have A20-SOM i guess i will end up jumper-wiring things to it to test u.u

01:26 <Jookia> oh wait

01:26 <Jookia> i'm an actual not smart person

01:26 <Jookia> i can just measure it using a usb power tester

01:53 <Jookia> idles at 4.75V 0.4A, cpu use at 100%, 0.4.63V 0.6A

01:53 <Jookia> so 2W and 3W

01:55 <Jookia> idk if mainline linux has power saving for sunxi

09:30 <diego71> mainline linux have cpufreq for sunxi, but the last time i've measured it saves less than 0.5W

09:57 <tokyovigilante> https://lore.kernel.org/linux-sunxi/20240426095141.13602-2-ryan@testtoast.com/T/#t

09:57 <tokyovigilante> v4 of the RG35XX DTS set, fingers crossed this time!

10:05 <Jookia> tokyovigilante: nice!

11:09 <Jookia> apritzel: i spent a while today working on the clock stuff. i found out i have to somehow tell clocks that the parents are named by firmware nodes in the dt. i started to set clk->dev but now that hangs as the driver is binding to it aaaa why is life so complicated :(

11:27 <apritzel> Jookia: U-Boot or the kernel?

11:31 <Jookia> oh, u-boot

11:32 <apritzel> ah, I see, because for the kernel we can deal with that quite nicely, I think

11:34 <Jookia> yes the kernel has fw_node in the clocks but in u-boot we only have strings

11:35 <Jookia> so my thought was to have parent clocks allowed to be named fw:hosc or something like that

11:36 <Jookia> i'll see how it goes a bit more tomorrow :)

13:40 <apritzel> macromorgan_: I guess tokyovigilante is sound asleep now, so can you test his v4 series, but with the vqmmc of mmc2 set to the new vcc_3v3_sd2 regulator?

13:52 <macromorgan> I've been using that vqmmc as vcc_3v3_sd2 for quite some time already.

13:54 <apritzel> with vcc_3v3_sd2 being a pure 3.3V fixed regulator?

14:10 <apritzel> macromorgan: also, can you please reply to the list, stating that it works? Seems like it's the only blocker left in this series

14:28 <macromorgan> okay, give me a bit to test the specific tree on my 35XXH, mine still diverges a little bit

14:44 <jernej> apritzel: sadly, nmi can't be merged yet. bindings go through irqchip branch

14:45 <jernej> this also delays Anbernic series

14:46 <jernej> eh, it's applied

14:47 <jernej> sorry

14:49 <apritzel> that was close!

20:41 <tokyovigilante> apritzel: was kipped out but just making those last changes now for a v5

20:47 <tokyovigilante> macromorgan: you've got your mmc2 vmmc-supply as vcc_3v3_sd2 rather than your vqmmc-supply in the DTS you posted previously, and aldo1 (at 1.8v) as vqmmc

20:54 <tokyovigilante> I can't get that to work here, will do a bit of experimenting

20:56 <tokyovigilante> using vcc_3v3_sd2 as either vmmc or vqmmc doesn't work, but just having it defined while using cldo4/aldo1 as vmmc/vqmmc does, which doesn't make sense but presumably reflects the missing gpio-switched regulator

21:04 <tokyovigilante> so I would propose leaving it as is for now, but trying to figure out a proper switched regulator for a followup

21:20 <tokyovigilante> Even having just vmmc defined as cldo4, with no vqmmc, and removing the vcc_3v3_sd2 regulator doesn't work, and I see lots of bluetooth out-of-order packet messages in the log with it gone as well.

21:20 <tokyovigilante> I wonder how many hours spent on this before it would be worth flying to China, breaking into Anberic's offices, and stealing the schematics? ;)

21:45 <tokyovigilante> ok, have made the other changes and sent a v5, keeping the regulators the same for now given works for me here. I wonder if the different vmmc working for macromorgan is because all the regulators in his DT were -always-on? If this gets on I'm going to work on the audio codec next, and will look at a GPIO regulator for the SD2 slot

22:59 <apritzel> tokyovigilante: I dimly remember we tried that before, but what happens if you set ALDO1 to 3.3V (both min and max)?

23:00 <apritzel> looks like aldo1 is used exclusively for MMC2, but you say without it being referenced it doesn't work, so it looks to be the required regulator

23:01 <apritzel> does the kernel complain about it being 1.8V only? I'd assume the MMC core wants it 3.3V initially?