Re: [PATCH v3 2/2] EDAC/amd64: Set zn_regs_v2 for all Family 1Ah models

[Yazen Ghannam]

On Sun, Oct 04, 2026 at 09:04:12AM -0700, Borislav Petkov wrote:
> On Sun, Oct 04, 2026 at 11:14:47AM -0400, Yazen Ghannam wrote:
> > Because it becomes an empty case once this flag set is moved up.
> 
> No, models 0x40... are supported. The others which are not explicitly listed
> there are not.
> 
> And no, the code doesn't enforce it yet but probably it should. And there
> should be a default: label for the models too which returns -ENODEV, like it
> does for the unknown families.
> 

I see what you mean, but I think that's a bigger issue with the current
design. And I want to move away from that.

The current design is "opt-in" for each new model/group even if there's
no technical difference. So we keep having to write these minor model
check patches just to load the module.

I'd rather we load unconditionally for all models with the same base
behavior using sane defaults. Then we can have model-specific patches
for variations, if needed.

Essentially, we could avoid a whole class of patches for derivative
(client, embedded, etc.) products. The module would load with the sane
defaults. If the test folks find an issue, then we can have a
model-specific patch.

This is how we've been trending with AMD64 EDAC. The family init has
been shrinking over the years.

Thanks,
Yazen