On Tue, Feb 11, 2025 at 01:45:06PM +0000, Conor Dooley wrote:
> On Tue, Feb 11, 2025 at 08:42:39AM +0800, Inochi Amaoto wrote:
> > On Mon, Feb 10, 2025 at 03:38:58PM +0100, Clément Léger wrote:
> > >
> > >
> > > On 06/12/2024 06:58, Inochi Amaoto wrote:
> > > > Add parsing for Zfbmin, Zvfbfmin, Zvfbfwma ISA extension which
> > > > were ratified in 4dc23d62 ("Added Chapter title to BF16") of
> > > > the riscv-isa-manual.
> > > >
> > > > Signed-off-by: Inochi Amaoto <inochiama@xxxxxxxxx>
> > > > ---
> > > > arch/riscv/include/asm/hwcap.h | 3 +++
> > > > arch/riscv/kernel/cpufeature.c | 3 +++
> > > > 2 files changed, 6 insertions(+)
> > > >
> > > > diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
> > > > index 869da082252a..14cc29f2a723 100644
> > > > --- a/arch/riscv/include/asm/hwcap.h
> > > > +++ b/arch/riscv/include/asm/hwcap.h
> > > > @@ -100,6 +100,9 @@
> > > > #define RISCV_ISA_EXT_ZICCRSE 91
> > > > #define RISCV_ISA_EXT_SVADE 92
> > > > #define RISCV_ISA_EXT_SVADU 93
> > > > +#define RISCV_ISA_EXT_ZFBFMIN 94
> > > > +#define RISCV_ISA_EXT_ZVFBFMIN 95
> > > > +#define RISCV_ISA_EXT_ZVFBFWMA 96
> > > >
> > > > #define RISCV_ISA_EXT_XLINUXENVCFG 127
> > > >
> > > > diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> > > > index c0916ed318c2..5cfcab139568 100644
> > > > --- a/arch/riscv/kernel/cpufeature.c
> > > > +++ b/arch/riscv/kernel/cpufeature.c
> > > > @@ -341,6 +341,7 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> > > > __RISCV_ISA_EXT_DATA(zacas, RISCV_ISA_EXT_ZACAS),
> > > > __RISCV_ISA_EXT_DATA(zawrs, RISCV_ISA_EXT_ZAWRS),
> > > > __RISCV_ISA_EXT_DATA(zfa, RISCV_ISA_EXT_ZFA),
> > > > + __RISCV_ISA_EXT_DATA(zfbfmin, RISCV_ISA_EXT_ZFBFMIN),
> > >
> > > Hi Inochi,
> > >
> > > You could add a validation callback to that extension:
> > >
> > > static int riscv_ext_f_depends(const struct riscv_isa_ext_data *data,
> > > const unsigned long *isa_bitmap)
> > > {
> > > if (__riscv_isa_extension_available(isa_bitmap, RISCV_ISA_EXT_f))
> > > return 0;
> > >
> > > return -EPROBE_DEFER;
> > > }
> > >
> > > ...
> > > __RISCV_ISA_EXT_DATA_VALIDATE(zfbfmin, RISCV_ISA_EXT_ZFBFMIN,
> > > riscv_ext_f_depends),
> > >
> > >
> > > But I'm ok with the current state of that patch since I have the same
> > > thing coming for other extensions as well.
> >
> >
> > I think it is good for me to add the check, and I wonder it is possible
> > to add the extra check for zvfbfmin and zvfbfwma like this:
> >
> > static int riscv_ext_zvfbfmin_validate(const struct riscv_isa_ext_data *data,
> > const unsigned long *isa_bitmap)
> > {
> > if (__riscv_isa_extension_available(isa_bitmap, RISCV_ISA_EXT_v))
> > return 0;
>
> This is not needed I think, V "turns on" Zve32f. If anything, you should
> be checking for CONFIG_RISCV_ISA_V here ^^
>
Thanks for pointing it. I will change the check.
> You /could/ call the resulting riscv_vector_f_validate(), since this is
> nothing specific to Zvfvfmin, and could be used for another extension
> that requires a Zve32f or Zve64 minimum base.
>
It is OK for me, I will change its name.
Regards,
Inochi
