Re: [PATCH v1] RISC-V: clarify what some RISCV_ISA* config options do
From: Andrew Jones
Date: Fri Apr 19 2024 - 10:09:09 EST
On Fri, Apr 19, 2024 at 12:06:59PM +0100, Conor Dooley wrote:
> On Fri, Apr 19, 2024 at 01:01:52PM +0200, Andrew Jones wrote:
> > > diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> > > index 6d64888134ba..c3a7793b0a7c 100644
> > > --- a/arch/riscv/Kconfig
> > > +++ b/arch/riscv/Kconfig
> > > @@ -503,8 +503,8 @@ config RISCV_ISA_SVNAPOT
> > > depends on RISCV_ALTERNATIVE
> > > default y
> > > help
> > > - Allow kernel to detect the Svnapot ISA-extension dynamically at boot
> > > - time and enable its usage.
> > > + Add support for the Svnapot ISA-extension when it is detected by
> > > + the kernel at boot.
> >
> > I'm not sure we need the 'by the kernel', since I guess that's implied by
> > being in a Kconfig help text, but either way is fine by me.
>
> I think we do, given some of the options are required for userspace to
> use it and others are not. Distinguishing between them doesn't cos us
> more than a few characters so I think it is worthwhile.
I agree we should ensure 'support in the kernel' type of text is present,
but here we're saying 'detected by the kernel' which I was thinking was
implied since this is kernel code. Maybe we should just add the 'the
kernel' text to where the support is rather than where the detection is?
I assumed it was left off of the 'Add support' because Svnapot is for
S-mode.
>
>
> > > @@ -686,7 +687,8 @@ config FPU
> > > default y
> > > help
> > > Say N here if you want to disable all floating-point related procedure
> > > - in the kernel.
> > > + in the kernel. Without this option enabled, neither the kernel nor
> > > + userspace may use floating-point procedures.
> > >
> > > If you don't know what to do here, say Y.
> > >
> >
> > Zicboz could also use some clarification, right? Or is the fact that
> > RISCV_ISA_ZICBOZ enables the use in both the kernel and userspace the
> > reason "Enable the use of the Zicboz extension (cbo.zero instruction)
> > when available." looks sufficient? Maybe Zicboz should follow the
> > "Say N here if..." pattern of V and FPU?
>
> Yeah, I think I just overlooked Zicboz. If the kernel option is needed
> for userspace to use it then yeah, it should follow the same wording as
> V/FPU.
Actually, never mind. I was thinking we only set the envcfg when this
config was selected, but that's not true. We'll set it whenever the
extension is present with or without this config. So I guess it can
follow Zicbom's pattern.
Thanks,
drew