Re: [PATCH v2 2/5] mm: move per-vma lock into vm_area_struct

[Suren Baghdasaryan <surenb@google.com>]

On Wed, Nov 13, 2024 at 7:02 AM Lorenzo Stoakes
<lorenzo.stoakes@oracle.com> wrote:
>
> On Wed, Nov 13, 2024 at 02:28:16PM +0000, Lorenzo Stoakes wrote:
> > On Tue, Nov 12, 2024 at 11:46:32AM -0800, Suren Baghdasaryan wrote:
> > > Back when per-vma locks were introduces, vm_lock was moved out of
> > > vm_area_struct in [1] because of the performance regression caused by
> > > false cacheline sharing. Recent investigation [2] revealed that the
> > > regressions is limited to a rather old Broadwell microarchitecture and
> > > even there it can be mitigated by disabling adjacent cacheline
> > > prefetching, see [3].
> >
> > I don't see a motivating reason as to why we want to do this? We increase
> > memory usage here which is not good, but later lock optimisation mitigates
> > it, but why wouldn't we just do the lock optimisations and use less memory
> > overall?
>
> I worded this badly. To clarify:
>
> I don't see a motivating reason _in the commit message_ as to why we want
> to do this.
>
> I am certain there are, in fact Mateusz and Vlastimil have provided them.
>
> So my review is - let's just put these there :)

Yeah, I had trouble wording all the reasons because in my head it was
simply "the right thing to do". Now with all your input my job has
become much easier :) Thanks folks!