> Date: Thu, 19 Jun 2025 16:10:16 +0200
> From: Stefan Sperling <stsp@stsp.name>
> 
> On Thu, Jun 19, 2025 at 03:55:31PM +0200, Mark Kettenis wrote:
> > Sure. 
> 
> Thanks, committed.
> 
> > I'm wondering at what point we decide that all AMD XHCI
> > controllers have this issue...
> 
> I also found an xhci-specific tweak you made to pci/pci.c in the
> function pci_set_powerstate(), quoted below.
> 
> It seems this code was added before the CSS workaround was added.
> Is it still needed?  The Z13 works fine without it.

Pretty sure that is still needed.

> 	/* Some AMD Ryzen xHCI controllers need a bit more time to wake up. */
> 	id = pci_conf_read(pc, tag, PCI_ID_REG);
> 	if (PCI_VENDOR(id) == PCI_VENDOR_AMD) {
> 		switch (PCI_PRODUCT(id)) {
> 		case PCI_PRODUCT_AMD_17_1X_XHCI_1:
> 		case PCI_PRODUCT_AMD_17_1X_XHCI_2:
> 		case PCI_PRODUCT_AMD_17_6X_XHCI:
> 			d3_delay = 20 * 1000;
> 		default:
> 			break;
> 		}
> 	}
> 
>