[PATCH v3 3/7] mm: Add PG_arch_3 page flag

[Catalin Marinas]

On Tue, Sep 20, 2022 at 05:33:42PM +0100, Marc Zyngier wrote:
> On Tue, 20 Sep 2022 16:39:47 +0100,
> Catalin Marinas <catalin.marinas at arm.com> wrote:
> > On Mon, Sep 19, 2022 at 07:12:53PM +0100, Marc Zyngier wrote:
> > > On Mon, 05 Sep 2022 18:01:55 +0100,
> > > Catalin Marinas <catalin.marinas at arm.com> wrote:
> > > > Peter, please let me know if you want to pick this series up together
> > > > with your other KVM patches. Otherwise I can post it separately, it's
> > > > worth merging it on its own as it clarifies the page flag vs tag setting
> > > > ordering.
> > > 
> > > I'm looking at queuing this, but I'm confused by this comment. Do I
> > > need to pick this as part of the series? Or is this an independent
> > > thing (my hunch is that it is actually required not to break other
> > > architectures...).
> > 
> > This series series (at least the first patches) won't apply cleanly on
> > top of 6.0-rc1 and, of course, we shouldn't break other architectures. I
> > can repost the whole series but I don't have the setup to test the
> > MAP_SHARED KVM option (unless Peter plans to post it soon).
> 
> I don't feel brave enough to take a series affecting all architectures

It shouldn't affect the others, the only change is that PG_arch_2 is now
only defined for arm64 but no other architecture is using it. The
problem with loongarch is that it doesn't have enough spare bits in
page->flags and even without any patches I think it's broken with the
right value for NR_CPUS.

> so late in the game, and the whole thing had very little arm64
> exposure. The latest QEMU doesn't seem to work anymore, so I don't
> have any MTE-capable emulation (and using the FVP remotely is a pain
> in the proverbial neck).
> 
> I'll come back to this after the merge window, should Peter decide to
> respin the series.

It makes sense.

-- 
Catalin