> mdio_bus_phy_resume() does trigger the state machine via > phy_start_machine(), so link state changes *are* detected after wakeup. > > But you're saying that's not sufficient and you really want the > PHY driver's IRQ handler to be called, do I understand that correctly? It is an interrupt, so i would expect the handler to be called. I've never looked deeply how the kernel handles this, but maybe there is some core support for this. The kernel does know about wake up interrupts. The interesting bit is how do you defer the interrupt until you have enough of the system running again you can actually service the interrupt. PHY interrupts mostly are level, not edge, because there are multiple sources of interrupts within the PHY. So you do need to clear the interrupt source, or you are going to get a storm, as you pointed out. But being a level might actually help you. It fires once to get you out of sleep, and then fires again when the interrupt controller is resumed and is enabled. Andrew
