On Tue, Jan 5, 2021 at 1:37 AM David Hildenbrand <david@xxxxxxxxxx> wrote:
>
> >> Yeah, obviously the first one. Being able to add+use PMEM is more
> >> important than using each and every last MB of main memory.
> >>
> >> I wonder if we can just stop adding any system RAM like
> >>
> >> [ Memory Section ]
> >> [ RAM ] [ Hole ]
> >>
> >> When there could be the possibility that the hole might actually be
> >> PMEM. (e.g., with CONFIG_ZONE_DEVICE and it being the last section in a
> >> sequence of sections, not just a tiny hole)
> >
> > I like the simplicity of it... I worry that the capacity loss
> > regression is easy to notice by looking at the output of free(1) from
> > one kernel to the next and someone screams.
>
> Well, you can always make it configurable and then simply fail to add
> PMEM later if impossible (trying to sub-section hot-add into early
> section). It's in the hands of the sysadmin then ("max out system ram"
> vs. "support any PMEM device that could eventually be there at
> runtime"). Distros would go for the second.
>
> I agree that it's not optimal, but sometimes simplicity has to win.
Here's where we left it last time, open to pfn_to_online_page hacks...
https://lore.kernel.org/linux-mm/CAPcyv4ivq=EPUePXiX2ErcVyF7+dV9Yv215Oue7X_Y2X_Jfw8Q@xxxxxxxxxxxxxx
I don't think a slow-path flag in the mem-section is too onerous, but
I'll withhold judgement until I have the patch I'm thinking of
in-hand. Let me give it a shot, you can always nack the final result.
