Re: [PATCH v4 1/2] usb: xhci-pci: add AMD Promontory 21 PCI glue

[Mario Limonciello <mario.limonciello@amd.com>]

On 5/9/26 00:34, Jihong Min wrote:
> On 5/9/26, Mario Limonciello wrote:
>  > Promontory 21 is only on AMD platforms and AMD platforms are only x86.
>  > I think the Kconfig should be conditional on AMD CPU support being
>  > enabled and X86 architecture so that we don't bloat other architectures
>  > with dead code that will never run.
> 
> I agree with limiting this to x86, and I changed the current branch in that
> direction.
> 
> One detail I would like to double-check is whether CPU_SUP_AMD should 
> also be
> part of the dependency, or whether X86 alone would be more accurate for the
> PCI glue.
> 
> The PROM21 xHCI function is still a PCI device. I found one public 
> example of
> a B650/PROM21-based PCIe add-in card design which is reported to work in 
> both
> AMD and Intel systems:
> 
> https://www.tomshardware.com/pc-components/chipsets/pcie-card-unlocks- 
> amd-chipset-power-on-intel-motherboards-or-you-can-turn-any-b650- 
> motherboard-into-an-x670-one
> 
> That is clearly a niche/community hardware case, not a normal AMD platform,
> and I do not want to over-weight it. But it made me wonder if:
> 
>    depends on X86
> 
> would be the more accurate dependency than:
> 
>    depends on X86 && CPU_SUP_AMD
> 
> for a PCI-attached PROM21 xHCI controller. The option would still use:
> 
>    default USB_XHCI_PCI
> 
> and the hwmon sensor driver would still stay behind its own
> SENSORS_PROM21_XHCI option.
> 
> I am fine keeping CPU_SUP_AMD if you prefer; I just wanted to check before
> locking that part in for v5.
> 
> Sincerely,
> Jihong Min
> 

Fine by me either way.