Re: [PATCH] riscv: RISCV_ISA_SV48 and RISCV_ISA_SV57 should depend on MMU
From: Conor Dooley
Date: Fri Oct 16 2026 - 15:22:41 EST
On Thu, Oct 15, 2026 at 02:24:42PM +0200, Geert Uytterhoeven wrote:
> Hi Jisheng,
>
> On Thu, 15 Oct 2026 at 14:08, Jisheng Zhang <jszhang@xxxxxxxxxx> wrote:
> > On Thu, Oct 15, 2026 at 09:43:40AM +0200, Geert Uytterhoeven wrote:
> > > On Thu, 15 Oct 2026 at 01:48, Jisheng Zhang <jszhang@xxxxxxxxxx> wrote:
> > > > On Wed, Oct 14, 2026 at 12:20:39PM +0200, Geert Uytterhoeven wrote:
> > > > > These options are irreleant for nommu systems.
> > > > >
> > > > > Fixes: 36de74049ec8cb1a ("riscv: introduce RISCV_ISA_SV48 and RISCV_ISA_SV57")
> > > > > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
> > >
> > > > > --- a/arch/riscv/Kconfig
> > > > > +++ b/arch/riscv/Kconfig
> > > > > @@ -605,7 +605,7 @@ config RISCV_ISA_SSQOSID
> > > > >
> > > > > config RISCV_ISA_SV48
> > > > > bool "Page-Based 48-bit Virtual-Memory System"
> > > > > - depends on 64BIT
> > > > > + depends on 64BIT && MMU
> > > > > default y
> > > > > help
> > > > > Add support for the Page-Based 48-bit Virtual-Memory System.
> > > > > @@ -617,7 +617,7 @@ config RISCV_ISA_SV48
> > > > >
> > > > > config RISCV_ISA_SV57
> > > > > bool "Page-Based 57-bit Virtual-Memory System"
> > > > > - depends on 64BIT
> > > > > + depends on 64BIT && MMU
> > > >
> > > > hmm, I think we also need to make RISCV_ISA_SV57 depend on
> > > > RISCV_ISA_SV48. Otherwise, there's a case: RISCV_ISA_SV57 &&
> > > > !RISCV_ISA_SV48. Human won't make this wrong combination, but
> > > > the random config can.
> > >
> > > Is that really an issue?
> > > If I build a kernel for an SV57-system, do I really need to enable
> > > support for SV48, too? It builds fine without...
> >
> > This is to align with the spec:
> > "implementations that support Sv57 must also support Sv48"
>
> IMO that is something different.
I think you're right Geert, "implementation" here surely refers to the
device. If someone wants to make their kernel for sv57 capable hardware
only, I think that's fine.
Attachment:
signature.asc
Description: PGP signature