09:43 <wens> MoeIcenowy: I resent your "thermal/of: support thermal zones w/o trips subnode" patch

09:47 <MoeIcenowy> wens: I saw it

09:48 <MoeIcenowy> in fact I initially did it for kukui

09:48 <MoeIcenowy> although I know it's also a problem on a64

09:48 <MoeIcenowy> (maybe no longer a problem on a64)

09:48 <MoeIcenowy> okay gpu thermal zone on a64 still has no trips

09:50 <MoeIcenowy> (for kukui, I have a kodama at home

10:09 <wens> I see

13:58 <macromorgan> apritzel: I keep meaning to go back to that (the NTC Chip... I actually built a Debian Buster image for it with a mainline kernel). Problem is I get new devices all the time and use my spare time to try and support them too...

13:58 <macromorgan> I do have a full patch series to support slc-mode in U-boot that I never got upstreamed, that's really the last missing piece to full mainline support across the board for NAND booting on the NTC Chip

13:59 <macromorgan> https://github.com/macromorgan/chip-debroot/blob/main/u-boot_files/0001-sunxi-Add-support-for-slc-emulation-on-mlc-NAND.patch

14:01 <macromorgan> yeah the raw NAND situation is kind of crappy though. Luckily being SLC that should be easier to support

14:09 <macromorgan> Hypfer: So I'm dusting off a very old portion of my brain, but it looks like the datasheet for that NAND is here: https://www.skyhighmemory.com/download/ONFI_S34ML01G3_02_Page_2KB_002-19206.pdf

14:09 <macromorgan> suggests it's ONFI compatible which I think means the parameters can be auto detected. Is mainline Linux able to see the NAND when you light up the controller?

14:10 <macromorgan> here's me adding support for it in the NTC Chip: https://github.com/macromorgan/chip-debroot/blob/main/devicetree/sun5i-r8-chip.dts#L205

14:11 <macromorgan> for you it should work just fine since it's SLC, but if it was MLC you'd need to jump through a whole host of hoops

14:11 <macromorgan> you still can't "format" the NAND so much as you can create a UBIFS filesystem for it and apply it, it's way different than simply using a block device

16:54 <apritzel> macromorgan: ah, many thanks for the info! And yeah, it looks like Hypfer's particular chip is actually SLC (how rare these days, on cheap AW devices!)

16:58 <apritzel> one problem at hand is now that we don't support the NAND controller for anything >= H3 (I think). I have no idea how close the modern IP is to the old ones, but that probs needs some lines in the driver

16:58 <macromorgan> no problemo. One of these days I swear I'll finish the job on the CHIP

16:58 <macromorgan> I don't remember the old driver being all too hard, assuming the IP is still similar it shouldn't be too rough. If I recall DMA used to give me fits but that's about it

16:59 <apritzel> macromorgan: it seems like Anbernic keeps you busy by releasing tiny variations of their consoles ;-)

16:59 <macromorgan> this past week I've been trying to push out a charger driver for a rockchip board, but yeah working on Anbernic stuff keeps me swampped

17:00 <macromorgan> I'm working on one of the very few non-Anbernic devices now... unfortunately I think the guy may have "took the money and run" so to speak, but I still said I'd do mainline and there are at least a few boards in the wild.

17:00 <macromorgan> I'm also waiting for some additional comments to pour in on the U-Boot and AXP717 drivers, then I can resubmit them with additional fixes.

17:01 <macromorgan> I may need to modify the U-Boot axp_spl driver slightly... for my use case I need to make sure the cpusldo regulator is on and the dcdc4 regulator is off, and it's not enough to leave that to chance.

17:02 <macromorgan> this is for board auto detection, which will come later

17:13 <apritzel> macromorgan: I have your H700 DRAM series high on my review list, maybe this weekend ...

17:14 <macromorgan> there were a few comments I needed to address. I don't suppose you want to review assuming those comments make it into the next revision?

17:14 <macromorgan> otherwise I can try to resubmit early next week

17:15 <MoeIcenowy> apritzel: well technically I think small SLC is sometimes preferred by some HW vendor

17:16 <MoeIcenowy> e.g. some Japan one with name starts with N and ends with o

17:17 <apritzel> macromorgan: I saw those comments, and they made sense to me, but I can happily do the review of the rest, and would assume those get fixed, given you agree.

17:18 <macromorgan> I do

17:18 <macromorgan> assuming they test out okay... if they don't I'll respond back as such

17:18 <apritzel> macromorgan: so early next week for a re-submit, with or without my comments ;-)

17:18 <apritzel> I wanted to test them on a bunch of H616 boards as well

17:19 <apritzel> MoeIcenowy: I was under the impression that SLC these days has just a small niche in reliability, and where cost is not so much a problem?

17:19 <apritzel> MoeIcenowy: aka: outside of Allwinner :-D

17:20 <apritzel> but maybe it's a different situation for small chips

17:27 <MoeIcenowy> ah yes

17:27 <MoeIcenowy> but I think most vendors that uses SLC have switched to SPI NAND

17:29 <apritzel> indeed! Hence I am surprised to find an SLC NAND on some rather uninspired new-ish device (Hypfer: robot vacuum?)