02:39 <smaeul> IIRC megi, you were working on DMA for MUSB at some point... how did that end up?

05:43 <megi> smaeul: like this https://megous.com/git/linux/log/?h=musb-sunxi-dma-wip

05:49 <megi> I didn't find any documentation for the controller, and just trying to match BSP driver's was fruitless (maybe I missed something), so I gave up. core musb driver is quite packed with a lot of special cases for various variants/musb integrations/glues so it was not particularly easy to follow or guess what's actually needed on sunxi

05:53 <megi> drivers/usb/musb/sunxi_dma.c is probably reasonably complete, for trying the tx path

05:57 <megi> looking at the code, I kinda have an itch to try again, now that I've worked with USB a bunch more in the meantime on various projects :)

06:16 <disctanger> megi: that would be great :)

06:20 <disctanger> megi: Based on your comment RX part is left of DMA? I am getting started with Embeded/kernel development (some details: https://oftc.irclog.whitequark.org/linux-sunxi/2022-06-29#31053802) and slowly getting comfortable with C and tiny kernel modifications/rebuilding stuffs.

06:21 <disctanger> For my USB Webcam project probably only TX side would suffise i think. Because data stream is mostly one way: USB Gadget (Video Stream) -> USB Host

06:24 <disctanger> i could help with testing the fixes on PineCube, PinePhone, PineBook Pro

07:14 <gamiee> disctanger: this effort will not help Pinebook Pro, as it's using Rockchip SoC

07:30 <disctanger> gamiee: Sure, then i could help with testing DMA support on PineCube and PinePhone

07:40 <megi> disctanger: yes, RX is left out, because I started with TX and didn't manage to get that working

07:41 <megi> usually not a great idea to change too many things at once if the code is full of unknown unknowns :)

07:43 <megi> so I left RX path in PIO mode

07:47 <disctanger> megi: noted well. What would be prerequisites for follow up that work? I have no experience in Kernel development but getting started with it. Learning low level stuffs bit by bit. PineCube USB Webcam happen to be my first project.

07:51 <megi> :)

07:52 <megi> given the lack of docs, only help can be comparison against Allwinner BSP code and experimentation

08:08 <disctanger> megi: Can you refer BSP source code? Internet search is yielding different results. (Is it this one? https://github.com/HandsomeMod/linux-allwinner-4.9)

08:08 <megi> I don't know where BSP code for S3 is

08:10 <disctanger> Okay then i will start my journey from finding BSP code for S3 :D. I will try to tinker around with it once i find it. I hope i wont destroy my only PineCube

08:11 <disctanger> I will be summarizing my results(findings) on official wiki page of PineCube (https://wiki.pine64.org/wiki/PineCube#PineCube_as_a_webcam)

08:11 <megi> I tried to make it work on A64, and used this for comparison https://megous.com/git/linux/tree/drivers/usb/sunxi_usb/udc?h=h6-4.9-bsp

08:12 <disctanger> megi: thanks for directions!

14:33 <MoeIcenowy> https://github.com/Tina-Linux/uboot-2018/blob/master/arch/arm/mach-sunxi/clock_sun55iw3.c intereseting

14:33 <MoeIcenowy> it's sun55i

14:53 <apritzel> so they used the opportunity to update! ... to 2018.05 :-(

14:55 <apritzel> so does that mean that the SoC with Cortex-A55s is still alive?

14:55 <apritzel> or is this just some leftover in the code?

14:56 <apritzel> And they seem to have indeed an "Allwinner A53" SoC, probably to complete the naming confusion

14:58 <jakllsch> heh

15:14 <jernej> apritzel: that seems pretty nice code for AC200. Too bad it doesn't work properly :)

15:15 <jernej> what is the issue again?

15:15 <apritzel> I have the feeling it's something stupid ;-)

15:16 <apritzel> it complains when the PHY instantiates the regmap gate clock, something about the refcount being 0, which points to some use-after-free

15:16 <apritzel> (don't have the exact dump here)

15:17 <apritzel> it could be something not cleaned up properly after some EPROBE_DEFER, so that it still finds the old freed clock, or something

15:21 <jernej> ah, that is fun to debug