[crac] RFR: 8373027: [CRaC] [CRIU] x86: FPU xsave area present, but host cpu doesn't support it

[Radim Vansa]

On Wed, 3 Dec 2025 15:08:08 GMT, Jan Kratochvil <jkratochvil at openjdk.org> wrote:

> In some cases (such as AWS `t3.nano` -> `m1.small`) CRIU refuses to restore:
> 
> x86: FPU xsave area present, but host cpu doesn't support it
> 
> 
> I think (based on information of @rvansa `xsave` is not needed for the safepoint of JDK checkpoint/restore.

OK. So there is a chance that native thread will use invalid instruction and fail later after restore - while this would be unfortunate, I think that possibility of such crashes is still outweighed by ability to restore in the CPU combos you mention. In other worse, we allow obviously detectable crashes to ignore some false positives.

What I am worried about is miscalculation; if (a part of) the xsave area is not restored correctly, can this produce a calculation error without triggering SIGILL?

-------------

PR Comment: https://git.openjdk.org/crac/pull/279#issuecomment-3665530137