On Mon, Nov 11, 2024 at 12:55:02PM -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]. Sorry to nag + add workload, but could you also add changes to the Documentation/mm/process_addrs.rst file to reflect this as part of the series? It'd be really cool to have the change as part of your series, so when it lands we update the documentation with it. This change fundamentally changes internals that are documented there so it makes sense to :) Thanks!
