Message-ID: <ff1fdb31-d739-ee85-9ce0-8386065e297f@canonical.com>
Date:   Wed, 20 Oct 2021 13:42:53 +0200
From:   Heinrich Schuchardt <heinrich.schuchardt@...onical.com>
To:     Anup Patel <anup@...infault.org>
Cc:     Daniel Lezcano <daniel.lezcano@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Guo Ren <guoren@...ux.alibaba.com>,
        Bin Meng <bmeng.cn@...il.com>, Xiang W <wxjstz@....com>,
        Samuel Holland <samuel@...lland.org>,
        Atish Patra <atish.patra@....com>,
        Rob Herring <robh+dt@...nel.org>,
        Palmer Dabbelt <palmer@...belt.com>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        Anup Patel <anup.patel@....com>,
        "linux-kernel@...r.kernel.org List" <linux-kernel@...r.kernel.org>,
        DTML <devicetree@...r.kernel.org>,
        linux-riscv <linux-riscv@...ts.infradead.org>,
        OpenSBI <opensbi@...ts.infradead.org>
Subject: Re: [PATCH 1/1] dt-bindings: T-HEAD CLINT

On 10/20/21 13:27, Anup Patel wrote:
> On Wed, Oct 20, 2021 at 3:06 PM Heinrich Schuchardt
> <heinrich.schuchardt@...onical.com> wrote:
>>
>> The CLINT in the T-HEAD 9xx CPUs is similar to the SiFive CLINT but does
>> not support 64bit mmio access to the MTIMER device.
>>
>> OpenSBI currently uses a property 'clint,has-no-64bit-mmio' to indicate the
>> restriction and the "sifive,cling0" compatible string. An OpenSBI
>> patch suggested to use "reg-io-width = <4>;" as the reg-io-width property
>> is generally used in the devicetree schema for such a condition.
>>
>> As the design is not SiFive based it is preferable to apply a compatible
>> string identifying T-HEAD instead.
>>
>> Add a new yaml file describing the T-HEAD CLINT.
>>
>> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@...onical.com>
>> ---
>> @Palmer, @Anup
>> I copied you as maintainers from sifive,clint.yaml. Please, indicate if
>> this should be changed.
>>
>> For the prior discussion see:
>> https://lore.kernel.org/all/20211015100941.17621-1-heinrich.schuchardt@canonical.com/
>> https://lore.kernel.org/all/20211015120735.27972-1-heinrich.schuchardt@canonical.com/
>>
>> A release candidate of the ACLINT specification is available at
>> https://github.com/riscv/riscv-aclint/releases
> 
> T-HEAD supporting only 32bit accesses to MTIME and MTIMECMP
> registers are totally allowed. The RISC-V privileged specification does
> not enforce RV64 platforms to support 64bit accesses to MTIME and
> MTIMECMP registers. Also, the ACLINT specification only states
> that MTIME and MTIMECMP registers are 64-bit wide but it does
> not enforce platforms to support 64-bit accesses.
> 
> Here are some discussions from tech-aia mailing list:
> https://lists.riscv.org/g/tech-aia/message/115
> https://lists.riscv.org/g/tech-aia/message/119
> https://lists.riscv.org/g/tech-aia/message/120
> 
> In other words, the T-HEAD CLINT (MTIMER+MSWI) is compliant
> with the RISC-V ACLINT specification.
> 
> I think we should add implementation specific compatible strings
> for Allwinner D1 in the ACLINT MTIMER and ACLINT MSWI
> DT bindings.
> 
> How about including the following two compatible strings in
> ACLINT DT bindings ?
> allwinner,sun20i-d1-aclint-mtimer
> allwinner,sun20i-d1-aclint-mswi

If the Allwinner CLINT is sufficiently compliant, this makes sense to me.

Will there be a new round of
[RFC PATCH v4 08/10] dt-bindings: timer: Add ACLINT MTIMER bindings
https://lore.kernel.org/all/20211007123632.697666-9-anup.patel@wdc.com/
were you could add the Allwinner device? Or is that series already merged?

Should the riscv,aclint-mtimer.yaml file mention that there are 
different access sizes and either state per compatibility string what 
that size is or provide a parameter for that purpose?

Best regards

Heinrich

> 
> Regards,
> Anup
> 
>> ---
>>   .../bindings/timer/thead,clint.yaml           | 62 +++++++++++++++++++
>>   1 file changed, 62 insertions(+)
>>   create mode 100644 Documentation/devicetree/bindings/timer/thead,clint.yaml
>>
>> diff --git a/Documentation/devicetree/bindings/timer/thead,clint.yaml b/Documentation/devicetree/bindings/timer/thead,clint.yaml
>> new file mode 100644
>> index 000000000000..02463fb2043a
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/timer/thead,clint.yaml
>> @@ -0,0 +1,62 @@
>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/timer/thead,clint.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: SiFive Core Local Interruptor
>> +
>> +maintainers:
>> +  - Palmer Dabbelt <palmer@...belt.com>
>> +  - Anup Patel <anup.patel@....com>
>> +
>> +description:
>> +  T-HEAD (and other RISC-V) SOCs include an implementation of the T-HEAD
>> +  Core Local Interruptor (CLINT) for M-mode timer and M-mode inter-processor
>> +  interrupts. It directly connects to the timer and inter-processor interrupt
>> +  lines of various HARTs (or CPUs) so RISC-V per-HART (or per-CPU) local
>> +  interrupt controller is the parent interrupt controller for CLINT device.
>> +  The clock frequency of the CLINT is specified via "timebase-frequency" DT
>> +  property of "/cpus" DT node. The "timebase-frequency" DT property is
>> +  described in Documentation/devicetree/bindings/riscv/cpus.yaml
>> +
>> +properties:
>> +  compatible:
>> +    items:
>> +      - const:
>> +          - allwinner,sun20i-d1-clint
>> +      - const:
>> +          - thead,clint0
>> +
>> +    description:
>> +      Should be "<vendor>,<chip>-clint" and "thead,clint<version>" for
>> +      the T-HEAD derived CLINTs.
>> +      Supported compatible strings are -
>> +      "allwinner,sun20i-d1-clint" for the CLINT in the Allwinner D1 SoC
>> +      and "thead,clint0" for the T-HEAD IP block with no chip
>> +      integration tweaks.
>> +
>> +  reg:
>> +    maxItems: 1
>> +
>> +  interrupts-extended:
>> +    minItems: 1
>> +
>> +additionalProperties: false
>> +
>> +required:
>> +  - compatible
>> +  - reg
>> +  - interrupts-extended
>> +
>> +examples:
>> +  - |
>> +    timer@...0000 {
>> +      compatible = "allwinner,sun20i-d1-clint", "thead,clint0";
>> +      interrupts-extended = <&cpu1intc 3 &cpu1intc 7
>> +                             &cpu2intc 3 &cpu2intc 7
>> +                             &cpu3intc 3 &cpu3intc 7
>> +                             &cpu4intc 3 &cpu4intc 7>;
>> +       reg = <0x2000000 0x10000>;
>> +    };
>> +...
>> --
>> 2.32.0
>>

Powered by blists - more mailing lists