Re: [PATCH 3/4] RISC-V: Support case insensitive ISA string parsing.

From: Anup Patel <anup@brainfault.org>
To: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Anup Patel <Anup.Patel@wdc.com>,
	Albert Ou <aou@eecs.berkeley.edu>,
	Alan Kao <alankao@andestech.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Daniel Lezcano <daniel.lezcano@linaro.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Johan Hovold <johan@kernel.org>,
	Atish Patra <Atish.Patra@wdc.com>,
	Palmer Dabbelt <palmer@sifive.com>,
	"linux-riscv@lists.infradead.org"
	<linux-riscv@lists.infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Allison Randal <allison@lohutok.net>
Subject: Re: [PATCH 3/4] RISC-V: Support case insensitive ISA string parsing.
Date: Sat, 27 Jul 2019 13:35:36 +0530	[thread overview]
Message-ID: <CAAhSdy0Eycc0ORSnh6LJeC_D_x9yLOkoc7OkPNuN6qOcZEGVWg@mail.gmail.com> (raw)
In-Reply-To: <alpine.DEB.2.21.9999.1907270043190.26998@viisi.sifive.com>

On Sat, Jul 27, 2019 at 1:23 PM Paul Walmsley <paul.walmsley@sifive.com> wrote:
>
> On Sat, 27 Jul 2019, Anup Patel wrote:
>
> > > -----Original Message-----
> > > From: Paul Walmsley <paul.walmsley@sifive.com>
> > > Sent: Saturday, July 27, 2019 5:00 AM
> > >
> > > On Fri, 26 Jul 2019, Atish Patra wrote:
> > >
> > > > On 7/26/19 1:47 PM, Paul Walmsley wrote:
> > > > > On Fri, 26 Jul 2019, Atish Patra wrote:
> > > > >
> > > > > > As per riscv specification, ISA naming strings are case
> > > > > > insensitive. However, currently only lower case strings are parsed
> > > > > > during cpu procfs.
> > > > > >
> > > > > > Support parsing of upper case letters as well.
> > > > > >
> > > > > > Signed-off-by: Atish Patra <atish.patra@wdc.com>
> > > > >
> > > > > Is there a use case that's driving this, or
> > > >
> > > > Currently, we use all lower case isa string in kvmtool. But somebody
> > > > can have uppercase letters in future as spec allows it.
> > > >
> > > >
> > > > can we just say, "use
> > > > > lowercase letters" and leave it at that?
> > > > >
> > > >
> > > > In that case, it will not comply with RISC-V spec. Is that okay ?
> > >
> > > I think that section of the specification is mostly concerned with someone
> > > trying to define "f" as a different extension than "F", or something like that.
> > > I'm not sure that it imposes any constraint that software must accept both
> > > upper and lower case ISA strings.
> > >
> > > What gives me pause here is that this winds up impacting DT schema
> > > validation:
> > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Docu
> > > mentation/devicetree/bindings/riscv/cpus.yaml#n41
> >
> > If 'f' and 'F' mean same extension as-per RISC-V spec then software should also
> > interpret it that way hence this patch.
>
> The list of valid RISC-V ISA strings is already constrained by the DT
> schema to be all lowercase.  Anything else would violate the schema.
>
> I'd take a patch that would pr_warn() or a BUG() if any uppercase letters
> were found in the riscv,isa DT property, though, in case the developer
> skipped the DT schema validator.

If your only objection is uppercase letter not agreeing with YMAL schema
then why not fix the YMAL schema to have regex for RISC-V ISA string?

The YMAL schema should not enforce any artificial restriction which is
theoretically allowed in the RISC-V spec.

Regards,
Anup