Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date:   Thu, 28 Apr 2022 10:49:33 +0200
From:   Michael Walle <michael@walle.cc>
To:     Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzk+dt@kernel.org>,
        Nicolas Ferre <nicolas.ferre@microchip.com>,
        Kavyasree Kotagiri <kavyasree.kotagiri@microchip.com>
Cc:     devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] ARM: dts: lan966x: fix sys_clk frequency
In-Reply-To: <20220326194028.2945985-1-michael@walle.cc>
References: <20220326194028.2945985-1-michael@walle.cc>
User-Agent: Roundcube Webmail/1.4.13
Message-ID: <3e860b122533f488c053abe0f3ff03eb@walle.cc>
Precedence: bulk

Am 2022-03-26 20:40, schrieb Michael Walle:
> The sys_clk frequency is 165.625MHz. The register reference of the
> Generic Clock controller lists the CPU clock as 600MHz, the DDR clock 
> as
> 300MHz and the SYS clock as 162.5MHz. This is wrong. It was first
> noticed during the fan driver development and it was measured and
> verified via the CLK_MON output of the SoC which can be configured to
> output sys_clk/64.
> 
> The core PLL settings (which drives the SYS clock) seems to be as
> follows:
>   DIVF = 52
>   DIVQ = 3
>   DIVR = 1
> 
> With a refernce clock of 25MHz, this means we have a post divider clock
>   Fpfd = Fref / (DIVR + 1) = 25MHz / (1 + 1) = 12.5MHz
> 
> The resulting VCO frequency is then
>   Fvco = Fpfd * (DIVF + 1) * 2 = 12.5MHz * (52 + 1) * 2 = 1325MHz
> 
> And the output frequency is
>   Fout = Fvco / 2^DIVQ = 1325MHz / 2^3 = 165.625Mhz
> 
> This all adds up to the constrains of the PLL:
>     10MHz <= Fpfd <= 200MHz
>     20MHz <= Fout <= 1000MHz
>   1000MHz <= Fvco <= 2000MHz
> 
> Fixes: 290deaa10c50 ("ARM: dts: add DT for lan966 SoC and 2-port board 
> pcb8291")
> Signed-off-by: Michael Walle <michael@walle.cc>

Ping :)

Btw. this is also true for the new B0 silicon. I just verified it
with the CLK_MON output.

-michael