Re: [RFC 13/13] x86/mm: Try to preserve old TLB entries using PCID

[Dave Hansen]

On 01/08/2016 03:15 PM, Andy Lutomirski wrote:
> + * The guiding principle of this code is that TLB entries that have
> + * survived more than a small number of context switches are mostly
> + * useless, so we don't try very hard not to evict them.

Big ack on that.  The original approach tried to keep track of the full
4k worth of possible PCIDs, it also needed an additional cpumask (which
it dynamically allocated) for where the PCID was active in addition to
the normal "where has this mm been" mask.

That's a lot of extra data to mangle, and I can definitely see your
approach as being nicer, *IF* the hardware isn't doing something useful
with the other 9 bits of PCID that you're throwing away. ;)