The changes for this XSA introduced an inefficiency and could do with some further hardening. Addressing of this wasn't sensible as part of the XSA, though (but you may take this as an explanation of why this starts out as v2).
1: x86: verify function type (and maybe attribute) in switch_stack_and_jump() 2: x86: clobber registers in switch_stack_and_jump() when !LIVEPATCH 3: x86/PV: avoid double stack reset during schedule tail handling 4: livepatch: adjust a stale comment Jan
