On Wed, Apr 05, 2023 at 10:35:45AM -0400, Alan Stern wrote: > On Wed, Apr 05, 2023 at 10:20:45AM +0200, Johan Hovold wrote: > > Alan did point out though, that the ehci driver returns a string like > > "suspended" when trying to access registers for a suspended device. > > > > That behaviour dates back to before the git era though and long before > > we had runtime PM. In fact, ehci still does not seem to implement > > runtime PM so this check would essentially only kick when the HCD is > > dead IIUC. > > In fact there is runtime PM support for ehci-hcd. You probably missed > it because the routines are defined in core/hcd-pci.c. Ah, thanks. I grepped for callers of ehci_suspend, but missed the indirection via the pci_suspend pointer. Johan
