Linux x86 SIMD: Add XSAVEC support to kfpu_begin()

[AttilaFueloep]

Motivation and Context

Since there is a register restore bug in XSAVES and XRSTORS add and use XSAVEC if available.

How Has This Been Tested?

Builds,, more tests needed.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Breaking change (fix or feature that would cause existing functionality to change)
• Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
• Documentation (a change to man pages or other documentation)

Checklist:

• My code follows the OpenZFS code style requirements.
• I have updated the documentation accordingly.
• I have read the contributing document.
• I have added tests to cover my changes.
• I have run the ZFS Test Suite with this change applied.
• All commit messages are properly formatted and contain Signed-off-by.

[AttilaFueloep]

Needs more testing and maybe some refinement.

[AttilaFueloep]

We need to decide if we prefer XSAVES or XSAVEC if both are available. Please see the discussion in #14557 (comment) for more details. Linux prefers XSAVES, this PR currently XSAVEC. I'd lean towards XSAVES for the sake of the modified optimization. Comments welcome!

[behlendorf]

Based on your nice analysis of the issue in #14557 my inclination would be to prefer XSAVES as well. Not only because of the extra optimization but also because that code path has a lot of real world miles on it and is known to work well (errata aside).

[AttilaFueloep]

also because that code path has a lot of real world miles on it and is known to work well (errata aside).

That's a good point. I'll update the PR, do the usual mprime test and remove the draft tag.

[behlendorf]

Looks good, pending the results of the mprime testing.

[AttilaFueloep]

I think I may have found some issues, need to do more thorough checks.

[behlendorf]

@AttilaFueloep any update on this?

[rincebrain]

#14989 may be germane.

[tonyhutter]

Related? #16452

[rincebrain]

Not unrelated, but as far as I can tell, the erratum this is for is either incomplete or not the problem there, since xsaves is forcibly disabled and this is still coming up.