00:23 <iscle> I've pushed all changes in a single clean commit here: https://github.com/iscle/linux-v853/commit/801bcc4e376eb3b8537c92f019dff608c48875bf

00:23 <iscle> Host mode has NOT been tested, so I expect it to not work or just work partially

00:37 <apritzel> you would at least need to split off the V853 enablement, since they are independent from the DMA rest

00:38 <apritzel> and those #ifdef's look odd, though I think they are an existing pattern for the various MUSB glue DMA bits?

00:39 <apritzel> try at least to use IF_ENABLED(), where possible, for instance for the writeb(SUNXI_MUSB_VEND0_DMA_MODE, ...) call

00:41 <apritzel> and you can move the stubs for _create() and _destroy() into some header file, as static inline, #ifdef'ed with CONFIG_USB_SUNXI_DMA, that keeps that contained there and doesn't disturb the actual C code

00:48 <iscle> apritzel: "you would at least need to split off the V853 enablement" > yep, I know, for now I'm keeping it all in one commit because I'm sure I'll have to modify something more when testing host mode, but I'm keeping it in mind

00:49 <iscle> the #ifdef

00:50 <iscle> the #ifdef's are indeed a common pattern in other MUSB drivers, but I do agree that they are a bit odd. In fact, the driver won't work if you compile without DMA if you don't specify CONFIG_MUSB_PIO_ONLY

00:55 <iscle> Not really sure why they decided to do it like this, but for me it does not make much sense: https://github.com/iscle/linux-v853/blob/a460436e68e2044b2a85d678b332e75ea22a344b/drivers/usb/musb/musb_core.c#L2420

05:44 <gamiee> iscle : well done!

05:44 <gamiee> by PIO you mean Polling mode?

09:27 <loki666> apritzel tokyovigilante macromorgan, any idea how to enable mmc2 on H700 ?

09:34 <apritzel> loki666: I guess you mean the second SD card on the Anbernic devices? There was something odd about a (GPIO?) regulator powering that port, don't remember the details, but they would be on the mailing list

09:38 <loki666> so I need a powerseq ?

09:43 <apritzel> that was an unanswered question, IIRC it didn't work reliably?

09:44 <loki666> looking at anbernic dt, https://gist.github.com/acmeplus/c109b4ada59151588c5a1ba5e71ac097#file-gistfile1-txt-L3727

09:44 <loki666> it seems there is a pwr-gpios

10:40 <apritzel> ... which is a property made up by AW, also the format doesn't seem to make much sense. I guess we need to wait for macromorgan to shed some light on this ...

18:45 <macromorgan> yes, it's a GPIO... give me a bit and I can remember what I did previously. Honestly I thought I already upstreamed the work. TL;DR it's just a GPIO controlled fixed regulator that's missing.

18:45 <macromorgan> also, anyone know where the PWM ended up? This is the last message I can see (from July): https://lore.kernel.org/linux-sunxi/dkoqgfvl2s4ly7jjzvln6x2nt4ztti4vxuiazx45vhwm5nb6e3@5ia7l3h2cnge/

19:40 <apritzel> macromorgan: PWM> cannot remember any update since then either, so it's up for grabs, I guess

19:41 <macromorgan> I might send an email first... since it got all the way to v9 and then just died

19:41 <apritzel> re: 2nd SD: wasn't there more than just a GPIO? I dimly remember this didn't work easily, or there was another problem?

19:42 <apritzel> well, it looks like it needs some actual work, according to Uwe's comments, or where they addressed already?