On Wed, 2018-12-05 at 18:22 +0100, Michal Hocko wrote: > On Fri 30-11-18 13:53:18, Alexander Duyck wrote: > > Modify the set_page_links function to include the setting of the reserved > > flag via a simple AND and OR operation. The motivation for this is the fact > > that the existing __set_bit call still seems to have effects on performance > > as replacing the call with the AND and OR can reduce initialization time. > > > > Looking over the assembly code before and after the change the main > > difference between the two is that the reserved bit is stored in a value > > that is generated outside of the main initialization loop and is then > > written with the other flags field values in one write to the page->flags > > value. Previously the generated value was written and then then a btsq > > instruction was issued. > > > > On my x86_64 test system with 3TB of persistent memory per node I saw the > > persistent memory initialization time on average drop from 23.49s to > > 19.12s per node. > > I have tried to explain why the whole reserved bit doesn't make much > sense in this code several times already. You keep ignoring that and > that is highly annoying. Especially when you add a tricky code to > optimize something that is not really needed. > > Based on that I am not going to waste my time on other patches in this > series to review and give feedback which might be ignored again. I got your explanation. However Andrew had already applied the patches and I had some outstanding issues in them that needed to be addressed. So I thought it best to send out this set of patches with those fixes before the code in mm became too stale. I am still working on what to do about the Reserved bit, and plan to submit it as a follow-up set.
