Re: [PATCH v1] USB: dwc3: only call usb_phy_set_suspend in suspend/resume

[Johan Hovold]

On Sat, Nov 04, 2023 at 11:51:22AM +0000, Stefan Eichenberger wrote:
> > From: Johan Hovold <johan@xxxxxxxxxx>

> > When reviewing the driver I did find a bug in the xhci-plat driver which
> > is likely the cause for the imbalance you're seeing. I just sent a fix
> > here in case you want to give it a try:
> > 
> >         https://lore.kernel.org/lkml/20231103164323.14294-1-johan+linaro@xxxxxxxxxx/
> 
> I tested it and it solves the issue we have. Thanks a lot for the fix!
> Before the use count for our regulator always went up to 2 and now it is
> 1 as expected.
> root@verdin-imx8mp-14773241:~# cat /sys/kernel/debug/regulator/CTRL_SLEEP_MOCI#/use_count
> 1
> 
> Also when going to suspend the regulator is turned off now. With the
> suspend patch applied from us the use count will be one more but
> everything still works as expected.

Thanks for testing.

> > But, also, why are you using legacy PHYs? Which platform is this for?
> 
> We have an external hub that we want to turn off when the system goes
> into suspend. For the i.MX8MM we use the phy-generic driver to achieve
> this. When I saw that the dwc3 driver would support the phy-generic via
> usb-phy, I thought we could use the same approach for the i.MX8MP and,
> in the future, the AM62. Maybe I misunderstood, would the right solution
> be to add a suspend function to the fsl,imx8mp-usb-phy driver and use
> vbus instead? But what would we do for the AM62, as it doesn't have a
> phy driver if I'm not mistaken.

That's not how the phy driver is supposed be used, and for on-board hubs
we now have:

	drivers/usb/misc/onboard_usb_hub.c

Have you tried using that one instead?

Johan