[00/11] x86/shadow: misc tidying

Message

Jan Beulich Jan. 5, 2023, 3:57 p.m. UTC

... or so I hope. The main observation was that we still have both
hash_vcpu_for_each() and hash_domain_for_each(), where the latter was
introduced in 2014/15 to replace the former. Only some eight years
later we can now complete this conversion. Everything else addresses
other things noticed along the road.

01: replace sh_reset_l3_up_pointers()
02: convert sh_audit_flags()'es 1st parameter to domain
03: drop hash_vcpu_foreach()
04: rename hash_domain_foreach()
05: move bogus HVM checks in sh_pagetable_dying()
06: drop a few uses of mfn_valid()
07: L2H shadow type is PV32-only
08: reduce effort of hash calculation
09: simplify conditionals in sh_{get,put}_ref()
10: correct shadow type bounds checks
11: sh_remove_all_mappings() is HVM-only

Jan

Comments

Andrew Cooper Jan. 6, 2023, 2:03 a.m. UTC | #1

On 05/01/2023 3:57 pm, Jan Beulich wrote:
> ... or so I hope. The main observation was that we still have both
> hash_vcpu_for_each() and hash_domain_for_each(), where the latter was

foreach

> introduced in 2014/15 to replace the former. Only some eight years
> later we can now complete this conversion. Everything else addresses
> other things noticed along the road.

Wow, it has been a long time.  That was the start of the "make Xen not
fall over NULL pointers if the toolstack issues some hypercalls out of
order", a task that is still ongoing...

>
> 01: replace sh_reset_l3_up_pointers()
> 02: convert sh_audit_flags()'es 1st parameter to domain
> 03: drop hash_vcpu_foreach()
> 04: rename hash_domain_foreach()
> 05: move bogus HVM checks in sh_pagetable_dying()
> 06: drop a few uses of mfn_valid()
> 07: L2H shadow type is PV32-only
> 08: reduce effort of hash calculation
> 09: simplify conditionals in sh_{get,put}_ref()
> 10: correct shadow type bounds checks
> 11: sh_remove_all_mappings() is HVM-only

Everything without code queries, Acked-by: Andrew Cooper
<[e-mail]>