Re: [PATCH] x86/cpu/hygon: Work around XSAVES Erratum 1386
From: Dave Hansen
Date: Fri Nov 20 2026 - 09:57:56 EST
On 11/20/26 06:44, David Laight wrote:
>> Did you mean that the kernel could prefer XSAVEC when no supervisor
>> XSAVE state needs to be saved, rather than clearing X86_FEATURE_XSAVES
>> on the affected processors?
> Or when there can never be any supervisor XSAVE state to be saved.
> I rather lost track of these instructions after XSAVEOPT.
Could you explain how that would help, exactly?
First, there needs to be a reference to the erratum in the changelog.
For me, searching for this string git the PDF to pop out of Google:
"Revision Guide for AMD Family 19h Models 00h-0Fh (Publication #56683)".
Second, I don't see anything in the erratum that says it is triggered by
supervisor state. Do you know something about this erratum that's not in
the public document?
I'm not even sure AMD _has_ supervisor state components on Zen 2 that
Linux uses.
Third, even if we could move to XSAVEC, I'm not sure it buys us much. I
certainly don't want to add the complexity to our already horrid XSAVE
mess. The right fix is in the microcode. Let's not give our dear
hardware vendors to mitigate even fewer issues than they do already.