Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:6 as permitted sender) client-ip=2620:137:e000::3:6;
From:   Emil Renner Berthing <emil.renner.berthing@canonical.com>
In-Reply-To: <20231203-mountable-snowsuit-24e4252c9eb5@spud>
References: <20231130-bobbing-valid-b97f26fe8edc@spud> <CAJM55Z9=smqcZ=su1oNdk1YZL_XdYAKrwtrZS7ScS=cAVmxZOA@mail.gmail.com>
 <20231203-mountable-snowsuit-24e4252c9eb5@spud>
Mime-Version: 1.0
Date:   Sun, 3 Dec 2023 08:09:00 -0800
Message-ID: <CAJM55Z-cb=0X2xToKnEwmqMkBSZkQ-kjAZyoGo1AY=edTLvxew@mail.gmail.com>
Subject: Re: [PATCH v1] riscv: dts: starfive: move timebase-frequency to .dtsi
To:     Conor Dooley <conor@kernel.org>,
        Emil Renner Berthing <emil.renner.berthing@canonical.com>
Cc:     linux-riscv@lists.infradead.org,
        Conor Dooley <conor.dooley@microchip.com>,
        Emil Renner Berthing <kernel@esmil.dk>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Paul Walmsley <paul.walmsley@sifive.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
        Walker Chen <walker.chen@starfivetech.com>,
        JeeHeng Sia <jeeheng.sia@starfivetech.com>,
        Leyfoon Tan <leyfoon.tan@starfivetech.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Conor Dooley wrote:
> On Fri, Dec 01, 2023 at 02:44:58PM +0100, Emil Renner Berthing wrote:
> > Conor Dooley wrote:
> > > From: Conor Dooley <conor.dooley@microchip.com>
> > >
> > > Properties fixed by the SoC should be defined in the $soc.dtsi, and the
> > > timebase-frequency is not sourced directly from an off-chip oscillator.
> >
> > Yes, according to the JH7100 docs[1] the mtime register is sourced from the
> > osc_sys external oscillator through u74rtc_toggle. However I haven't yet found
> > a place in the docs that describe where that clock is divided by 4 to get
> > 6.25MHz from the 25MHz.
> >
> > I expect the JH7110 mtime is set up in a similar way, but haven't yet dug into
> > the available documentation.
>
> Your other reply suggests that this is a fixed division for the jh7110,
> in which case it makes sense to leave it as-is. mpfs is different in
> that it is fixed to 1 MHz regardless of which of the permitted external
> oscillator frequencies you use.

This is what I've found for the JH7100:

osc_sys (25MHz) -> u74rtc_toggle (gate) -> ? (div 4) -> mtime

The divide by 4 is not in the regular clock tree, so if it is configurable it
must be some bits hidden in the syscon area or something. The only restriction
I've found in the docs is that it must be strictly less than half the rate of
the core clock.

For the JH7110 it goes:

osc (24MHz) -> rtc_toggle (div N) -> mtime

..where N defaults to 6 and this is also the maximum N.

/Emil