[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <3b19a2bb-20a0-c650-49b7-d966585b5ba9@nvidia.com>
Date: Thu, 6 Jun 2019 08:06:57 +0100
From: Jon Hunter <jonathanh@...dia.com>
To: Sameer Pujar <spujar@...dia.com>, <thierry.reding@...il.com>,
<robh+dt@...nel.org>, <mark.rutland@....com>
CC: <devicetree@...r.kernel.org>, <linux-tegra@...r.kernel.org>,
<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/2] arm64: tegra: add ACONNECT, ADMA and AGIC nodes
On 06/06/2019 06:58, Sameer Pujar wrote:
>
> On 6/4/2019 2:37 PM, Jon Hunter wrote:
>> On 06/05/2019 12:58, Sameer Pujar wrote:
>>> Add DT nodes for following devices on Tegra186 and Tegra194
>>> * ACONNECT
>>> * ADMA
>>> * AGIC
>>>
>>> Signed-off-by: Sameer Pujar <spujar@...dia.com>
>>> ---
>>> arch/arm64/boot/dts/nvidia/tegra186.dtsi | 67
>>> ++++++++++++++++++++++++++++++++
>>> arch/arm64/boot/dts/nvidia/tegra194.dtsi | 67
>>> ++++++++++++++++++++++++++++++++
>>> 2 files changed, 134 insertions(+)
>>>
>>> diff --git a/arch/arm64/boot/dts/nvidia/tegra186.dtsi
>>> b/arch/arm64/boot/dts/nvidia/tegra186.dtsi
>>> index 6e2b6ce..2c432c9 100644
>>> --- a/arch/arm64/boot/dts/nvidia/tegra186.dtsi
>>> +++ b/arch/arm64/boot/dts/nvidia/tegra186.dtsi
>>> @@ -1153,4 +1153,71 @@
>>> (GIC_CPU_MASK_SIMPLE(4) | IRQ_TYPE_LEVEL_LOW)>;
>>> interrupt-parent = <&gic>;
>>> };
>>> +
>>> + aconnect@...1000 {
>>> + compatible = "nvidia,tegra210-aconnect";
>>> + clocks = <&bpmp TEGRA186_CLK_APE>,
>>> + <&bpmp TEGRA186_CLK_APB2APE>;
>>> + clock-names = "ape", "apb2ape";
>>> + power-domains = <&bpmp TEGRA186_POWER_DOMAIN_AUD>;
>>> + #address-cells = <2>;
>>> + #size-cells = <2>;
>>> + ranges;
>> My recollection is that non-empty range is preferred from what Rob told
>> me in the past. See the Tegra210 binding.
> I see at various places empty ranges property is used. From DT spec, it
> means
> there is no translation in the address space between parent/child.
Yes that is correct, but this is what Rob told me in the past ...
"Use of non-empty ranges is preferred though there is not much benefit
if the parent and child sizes are the same. However, it does also
limit what is a valid address for those child nodes."
So I think it is best to be consistent.
> Also I looked at Tegra210 binding,
> ranges = <0x702c0000 0x0 0x702c0000 0x00040000>;
> Should it be encoded as a triplet(child addr, parent addr, length)?
Right but remember the parent address is 2 cells, hence the 4 values.
Jon
--
nvpublic
Powered by blists - more mailing lists