07:53 <ukleinek> mmind00: I have https://paste.debian.net/1200824/ now. I wonder about two things: Should I make the vcc5v0_usb regulator hook up to some device (then: how?) instead of using always on (which I copied from the vendor dts)?

07:54 <ukleinek> mmind00: and: IMHO it's a bit strange that I have to enable the phy twice. Wouldn't it be sensible to maybe drop the status="disabled" for &u2phy1 or &u2phy1_otg?

07:55 <ukleinek> ah, and a third one: vcc5v0-usb-en has no hw pull, should then the pinctrl better contain one?

07:56 <mmind00> ukleinek: sooo many questions :-D

07:56 <ukleinek> mmind00: I hope they are easy enough?!

07:57 <mmind00> ukleinek: for enabling ... not sure ... i.e. there are boards where only 1 of the 2 phys is connected, so having the disabled in the subnodes makes sense ... and somehow I think enabling a phy driver without actual phys connected is also somewhat strange ;-)

07:57 <mmind00> though I don't think anybody tried that yet

07:58 <ukleinek> yeah, I thought this was the reason

07:58 <mmind00> ukleinek: as for vcc5v0-usb-en ... that's probably an output gpio then ... so is there an actual pull needed when the 0/1 setting is explicitly set from the kernel?

07:59 <ukleinek> mmind00: a pull obviously only matters until the GPIO is switched to output and so drives the line. So maybe unimportant?

07:59 <mmind00> ukleinek: and finally for the regulator ... take a look at rk3399-nanopc-t4.dts ... and its vcc5v0_host0 -> phy-supply in the phy node

08:00 <mmind00> ukleinek: pull: yep and pinctrl also is set only directly before that ... so if its floating until the regulator probes, the last 1ms won't kill it ;-)

08:00 <ukleinek> ah, for the phy. My grep missed that because I looked into the usbdrd3_[01] nodes only.

08:01 <ukleinek> mmind00: 👍

08:03 <ukleinek> funny, rk3399-nanopc-t4.dts's vcc5v0_host0 is still regulator-always-on

13:51 <ukleinek> mmind00: I forgot to commit a ; in my helios64 dts commit. Under the assumption that the patch is otherwise fine for application, can you add this ; in line 482 when you apply? Or should I better send a v2?

13:52 <ukleinek> (context: https://lore.kernel.org/r/20210611081414.1448786-1-uwe@kleine-koenig.org)