Re: [PATCH 2/4] x86/amd_nb: add support for newer PCI topologies

[Borislav Petkov]

On Tue, Nov 06, 2018 at 03:42:56PM -0600, Bjorn Helgaas wrote:
> This isn't some complicated new device where the programming model
> changed on the new CPU.  This is a thermometer that was already
> supported.  ACPI provides plenty of functionality that could be used
> to support this generically, e.g., see drivers/acpi/thermal.c,
> drivers/thermal/int340x_thermal/processor_thermal_device.c, etc.

Ok, you say ACPI but how do you envision practically doing that? I mean,
this is used by old boxes too - ever since K8. So how do we go and add
ACPI functionality to old boxes?

Or do you mean it should simply be converted to do pci_register_driver()
with a struct pci_driver pointer which has all those PCI device IDs in a
table? I'm looking at the last example
drivers/thermal/int340x_thermal/processor_thermal_device.c you gave above.

> But maybe there's some real value in the nitty-gritty device-specific
> code in amd_nb.c.  If so, I guess you're stuck with updates like this
> and negotiating with the distros to do backports and new releases.

Well, even if it is converted to a different registration scheme, you
still need to add new PCI device IDs to the table, no? So *some* sort of
enablement still needs to happen.

And then the argument about needing enablement for distros is moot
because it still needs enablement/backporting - regardless of the
registration scheme.

Or do you mean something else?

I mean, don't get me wrong, I don't mind if it gets converted to
pci_register_driver() if you think it fits better this way with the
drivers registering with PCI devices - I'm just trying to understand the
reasoning for it.

Thx.

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.