Re: [PATCH] riscv: Add RISC-V svpbmt extension

[Anup Patel]

On Thu, Sep 23, 2021 at 5:21 PM Alexandre Ghiti
<alexandre.ghiti@xxxxxxxxxxxxx> wrote:
>
> On Thu, Sep 23, 2021 at 12:11 PM Nick Kossifidis <mick@xxxxxxxxxxxx> wrote:
> >
> > Στις 2021-09-23 13:04, Philipp Tomsich έγραψε:
> > >
> > > How if we expand this to a mmu subnode in cpu@x and add a booleans for
> > > adornments like svnapot and svpbmt?
> > > The older mmu-type could then treated to indicate a mmu w/o any
> > > adornments specified.  I am aware that this generates an additional
> > > parsing-path that will be maintained, but it will allow future
> > > properties to be grouped.
> > >
> > > cpu@0 {
> > > ...
> > > mmu {
> > > type = "riscv,sv39";
> > > supports-svpbmt;
> > > }
> > > ...
> > > }
> >
> > I was about to propose the same thing, we can do this now without
> > breaking backwards compatibility, we don't really use mmu-type property
> > at this point, we are either sv39 or nommu.
>
> Indeed, this property is only informative and not useful since we can
> directly "ask" the hw what it supports (cf sv48 patchset). And it
> cannot actually be used to force a certain svXX since reading the
> device tree happens way too late in the boot process (I have this
> issue with my sv48 patchset where I used to read the device tree to
> set the size of the address space, but it actually breaks KASAN).
>
> Isn't there a way to know if it supports svPBMT at runtime?

Unfortunately, we can't detect Svpmbt through CSR read/write
or traps.

A DT property seems to be the only way for Svpmbt.

Regards,
Anup

>
> Alex
>
> >
> > _______________________________________________
> > linux-riscv mailing list
> > linux-riscv@xxxxxxxxxxxxxxxxxxx
> > http://lists.infradead.org/mailman/listinfo/linux-riscv