On Tue, 7 Jul 2026 13:06:02 +1000, [email protected] wrote: > > > The whole bunch of template functions look like it could be useful for > > > initialization of the non-zone-device struct pages as well. > > > > > > As I mentioned previously, it's interesting to see if this approach speeds > > > up normal memory map initialization as well. If if does could have a > > > single > > > set of the template functions. > > > > Thanks for the suggestion. I agree this is worth exploring. > > > > For this series, I would prefer to keep the scope limited to the > > ZONE_DEVICE initialization path and get that part settled first. > > I think that makes sense, especially as my initial perf testing with other > types > of ZONE_DEVICE pages without altmap didn't show much of a perf impact. Meaning > it might not be so useful for normal page init. > > > After this series is settled, I plan to look into whether a similar > > template-based approach can also help normal page initialization. If it > > shows a real benefit there as well, I can follow up with a separate > > series. > > I agree it would be interesting though, please let us know what you find one > way > or the other.
Thanks. I will keep this series focused on the pmem-oriented ZONE_DEVICE path for now, and I will report back separately if the follow-up work on normal page initialization turns up anything useful, positive or negative. Thanks, Zhe
