[v4,1/2] Documentation: Documentation for qcom, llcc

Commit Message

Rishabh Bhatnagar April 10, 2018, 8:08 p.m. UTC

Documentation for last level cache controller device tree bindings,
client bindings usage examples.

Signed-off-by: Channagoud Kadabi <[e-mail]>
Signed-off-by: Rishabh Bhatnagar <[e-mail]>
---
 .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58 ++++++++++++++++++++++
 1 file changed, 58 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt

--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Evan Green April 12, 2018, 10:07 p.m. UTC | #1

On Tue, Apr 10, 2018 at 1:09 PM Rishabh Bhatnagar <[e-mail]>
wrote:

> Documentation for last level cache controller device tree bindings,
> client bindings usage examples.

> Signed-off-by: Channagoud Kadabi <[e-mail]>
> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
> ---
>   .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58
++++++++++++++++++++++
>   1 file changed, 58 insertions(+)
>   create mode 100644
Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt

> diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
> new file mode 100644
> index 0000000..497cf0f
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
> @@ -0,0 +1,58 @@
> +== Introduction==
> +
> +LLCC (Last Level Cache Controller) provides last level of cache memory
in SOC,
> +that can be shared by multiple clients. Clients here are different cores
in the
> +SOC, the idea is to minimize the local caches at the clients and migrate
to
> +common pool of memory
> +
> +Properties:
> +- compatible:
> +        Usage: required
> +        Value type: <string>
> +        Definition: must be "qcom,sdm845-llcc"
> +
> +- reg:
> +        Usage: required
> +        Value Type: <prop-encoded-array>
> +        Definition: must be addresses and sizes of the LLCC registers
> +
> +- #cache-cells:
> +        Usage: required
> +        Value Type: <u32>
> +        Definition: Number of cache cells, must be 1
> +
> +- max-slices:
> +        usage: required
> +        Value Type: <u32>
> +        Definition: Number of cache slices supported by hardware
> +
> +Example:
> +
> +       llcc: qcom,llcc@1100000 {
> +               compatible = "qcom,sdm845-llcc";
> +               reg = <0x1100000 0x250000>;
> +               #cache-cells = <1>;
> +               max-slices = <32>;
> +       };
> +
> +== Client ==
> +
> +Properties:
> +- cache-slice-names:
> +        Usage: required
> +        Value type: <stringlist>
> +        Definition: A set of names that identify the usecase names of a
> +                       client that uses cache slice. These strings are
> +                       used to look up the cache slice entries by name.
> +
> +- cache-slices:
> +        Usage: required
> +        Value type: <prop-encoded-array>
> +        Definition: The tuple has phandle to llcc device as the first
> +                       argument and the second argument is the usecase
> +                       id of the client.
> +For Example:
> +       venus {
> +               cache-slice-names = "vidsc0", "vidsc1";
> +               cache-slices = <&llcc VIDSC0_ID>, <&llcc VIDSC1_ID>;

My git complains about some whitespace weirdness on the line above. Other
than that:

Reviewed-by: Evan Green <[e-mail]>

-Evan
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Rob Herring (Arm) April 16, 2018, 2:59 p.m. UTC | #2

On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
> Documentation for last level cache controller device tree bindings,
> client bindings usage examples.

"Documentation: Documentation ..."? That wastes a lot of the subject 
line... The preferred prefix is "dt-bindings: ..."

> 
> Signed-off-by: Channagoud Kadabi <[e-mail]>
> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
> ---
>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58 ++++++++++++++++++++++
>  1 file changed, 58 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
> 
> diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
> new file mode 100644
> index 0000000..497cf0f
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
> @@ -0,0 +1,58 @@
> +== Introduction==
> +
> +LLCC (Last Level Cache Controller) provides last level of cache memory in SOC,
> +that can be shared by multiple clients. Clients here are different cores in the
> +SOC, the idea is to minimize the local caches at the clients and migrate to
> +common pool of memory
> +
> +Properties:
> +- compatible:
> +	 Usage: required
> +	 Value type: <string>
> +	 Definition: must be "qcom,sdm845-llcc"
> +
> +- reg:
> +	 Usage: required
> +	 Value Type: <prop-encoded-array>
> +	 Definition: must be addresses and sizes of the LLCC registers

How many address ranges?

> +
> +- #cache-cells:

This is all written as it is a common binding, but it is not one.

You already have most of the configuration data for each client in the 
driver, I think I'd just put the client connection there too. Is there 
any variation of this for a given SoC?

> +	 Usage: required
> +	 Value Type: <u32>
> +	 Definition: Number of cache cells, must be 1
> +
> +- max-slices:
> +	 usage: required
> +	 Value Type: <u32>
> +	 Definition: Number of cache slices supported by hardware

What's a slice?

> +
> +Example:
> +
> +	llcc: qcom,llcc@1100000 {
> +		compatible = "qcom,sdm845-llcc";
> +		reg = <0x1100000 0x250000>;
> +		#cache-cells = <1>;
> +		max-slices = <32>;
> +	};
> +
> +== Client ==
> +
> +Properties:
> +- cache-slice-names:
> +	 Usage: required
> +	 Value type: <stringlist>
> +	 Definition: A set of names that identify the usecase names of a
> +			client that uses cache slice. These strings are
> +			used to look up the cache slice entries by name.
> +
> +- cache-slices:
> +	 Usage: required
> +	 Value type: <prop-encoded-array>
> +	 Definition: The tuple has phandle to llcc device as the first
> +			argument and the second argument is the usecase
> +			id of the client.
> +For Example:
> +	venus {
> +		cache-slice-names = "vidsc0", "vidsc1";
> +	 	cache-slices = <&llcc VIDSC0_ID>, <&llcc VIDSC1_ID>;
> +	};
> --
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> a Linux Foundation Collaborative Project
> 
> --
> To unsubscribe from this list: send the line "unsubscribe devicetree" in
> the body of a message to [e-mail]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Rishabh Bhatnagar April 17, 2018, 5:43 p.m. UTC | #3

On 2018-04-16 07:59, Rob Herring wrote:
> On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
>> Documentation for last level cache controller device tree bindings,
>> client bindings usage examples.
> 
> "Documentation: Documentation ..."? That wastes a lot of the subject
> line... The preferred prefix is "dt-bindings: ..."
> 
>> 
>> Signed-off-by: Channagoud Kadabi <[e-mail]>
>> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
>> ---
>>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58 
>> ++++++++++++++++++++++
>>  1 file changed, 58 insertions(+)
>>  create mode 100644 
>> Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>> 
>> diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt 
>> b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>> new file mode 100644
>> index 0000000..497cf0f
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>> @@ -0,0 +1,58 @@
>> +== Introduction==
>> +
>> +LLCC (Last Level Cache Controller) provides last level of cache 
>> memory in SOC,
>> +that can be shared by multiple clients. Clients here are different 
>> cores in the
>> +SOC, the idea is to minimize the local caches at the clients and 
>> migrate to
>> +common pool of memory
>> +
>> +Properties:
>> +- compatible:
>> +	 Usage: required
>> +	 Value type: <string>
>> +	 Definition: must be "qcom,sdm845-llcc"
>> +
>> +- reg:
>> +	 Usage: required
>> +	 Value Type: <prop-encoded-array>
>> +	 Definition: must be addresses and sizes of the LLCC registers
> 
> How many address ranges?
> 
It consists of just one address range. I'll edit the definition to make
it more clear.
>> +
>> +- #cache-cells:
> 
> This is all written as it is a common binding, but it is not one.
> 
> You already have most of the configuration data for each client in the
> driver, I think I'd just put the client connection there too. Is there
> any variation of this for a given SoC?
> 
#cache-cells and max-slices won't change for a given SOC. So you want me
to hard-code in the driver itself?

>> +	 Usage: required
>> +	 Value Type: <u32>
>> +	 Definition: Number of cache cells, must be 1
>> +
>> +- max-slices:
>> +	 usage: required
>> +	 Value Type: <u32>
>> +	 Definition: Number of cache slices supported by hardware
> 
> What's a slice?
> 
System cache memory provided by LLCC is divided into smaller chunks
called slices. Each slice has its associated size and ID. Clients can
query slice details, activate and deactivate them.
>> +
>> +Example:
>> +
>> +	llcc: qcom,llcc@1100000 {
>> +		compatible = "qcom,sdm845-llcc";
>> +		reg = <0x1100000 0x250000>;
>> +		#cache-cells = <1>;
>> +		max-slices = <32>;
>> +	};
>> +
>> +== Client ==
>> +
>> +Properties:
>> +- cache-slice-names:
>> +	 Usage: required
>> +	 Value type: <stringlist>
>> +	 Definition: A set of names that identify the usecase names of a
>> +			client that uses cache slice. These strings are
>> +			used to look up the cache slice entries by name.
>> +
>> +- cache-slices:
>> +	 Usage: required
>> +	 Value type: <prop-encoded-array>
>> +	 Definition: The tuple has phandle to llcc device as the first
>> +			argument and the second argument is the usecase
>> +			id of the client.
>> +For Example:
>> +	venus {
>> +		cache-slice-names = "vidsc0", "vidsc1";
>> +	 	cache-slices = <&llcc VIDSC0_ID>, <&llcc VIDSC1_ID>;
>> +	};
>> --
>> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
>> Forum,
>> a Linux Foundation Collaborative Project
>> 
>> --
>> To unsubscribe from this list: send the line "unsubscribe devicetree" 
>> in
>> the body of a message to [e-mail]
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Rishabh Bhatnagar April 17, 2018, 10:12 p.m. UTC | #4

On 2018-04-17 10:43, [e-mail] wrote:
> On 2018-04-16 07:59, Rob Herring wrote:
>> On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
>>> Documentation for last level cache controller device tree bindings,
>>> client bindings usage examples.
>> 
>> "Documentation: Documentation ..."? That wastes a lot of the subject
>> line... The preferred prefix is "dt-bindings: ..."
>> 
>>> 
>>> Signed-off-by: Channagoud Kadabi <[e-mail]>
>>> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
>>> ---
>>>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58 
>>> ++++++++++++++++++++++
>>>  1 file changed, 58 insertions(+)
>>>  create mode 100644 
>>> Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>> 
>>> diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt 
>>> b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>> new file mode 100644
>>> index 0000000..497cf0f
>>> --- /dev/null
>>> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>> @@ -0,0 +1,58 @@
>>> +== Introduction==
>>> +
>>> +LLCC (Last Level Cache Controller) provides last level of cache 
>>> memory in SOC,
>>> +that can be shared by multiple clients. Clients here are different 
>>> cores in the
>>> +SOC, the idea is to minimize the local caches at the clients and 
>>> migrate to
>>> +common pool of memory
>>> +
>>> +Properties:
>>> +- compatible:
>>> +	 Usage: required
>>> +	 Value type: <string>
>>> +	 Definition: must be "qcom,sdm845-llcc"
>>> +
>>> +- reg:
>>> +	 Usage: required
>>> +	 Value Type: <prop-encoded-array>
>>> +	 Definition: must be addresses and sizes of the LLCC registers
>> 
>> How many address ranges?
>> 
> It consists of just one address range. I'll edit the definition to make
> it more clear.
>>> +
>>> +- #cache-cells:
>> 
>> This is all written as it is a common binding, but it is not one.
>> 
>> You already have most of the configuration data for each client in the
>> driver, I think I'd just put the client connection there too. Is there
>> any variation of this for a given SoC?
>> 
> #cache-cells and max-slices won't change for a given SOC. So you want 
> me
> to hard-code in the driver itself?
> 
I can use of_parse_phandle_with_fixed_args function and fix the number 
of
args as 1 instead of keeping #cache-cells here in DT. Does that look 
fine?
>>> +	 Usage: required
>>> +	 Value Type: <u32>
>>> +	 Definition: Number of cache cells, must be 1
>>> +
>>> +- max-slices:
>>> +	 usage: required
>>> +	 Value Type: <u32>
>>> +	 Definition: Number of cache slices supported by hardware
>> 
>> What's a slice?
>> 
> System cache memory provided by LLCC is divided into smaller chunks
> called slices. Each slice has its associated size and ID. Clients can
> query slice details, activate and deactivate them.
>>> +
>>> +Example:
>>> +
>>> +	llcc: qcom,llcc@1100000 {
>>> +		compatible = "qcom,sdm845-llcc";
>>> +		reg = <0x1100000 0x250000>;
>>> +		#cache-cells = <1>;
>>> +		max-slices = <32>;
>>> +	};
>>> +
>>> +== Client ==
>>> +
>>> +Properties:
>>> +- cache-slice-names:
>>> +	 Usage: required
>>> +	 Value type: <stringlist>
>>> +	 Definition: A set of names that identify the usecase names of a
>>> +			client that uses cache slice. These strings are
>>> +			used to look up the cache slice entries by name.
>>> +
>>> +- cache-slices:
>>> +	 Usage: required
>>> +	 Value type: <prop-encoded-array>
>>> +	 Definition: The tuple has phandle to llcc device as the first
>>> +			argument and the second argument is the usecase
>>> +			id of the client.
>>> +For Example:
>>> +	venus {
>>> +		cache-slice-names = "vidsc0", "vidsc1";
>>> +	 	cache-slices = <&llcc VIDSC0_ID>, <&llcc VIDSC1_ID>;
>>> +	};
>>> --
>>> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora 
>>> Forum,
>>> a Linux Foundation Collaborative Project
>>> 
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe devicetree" 
>>> in
>>> the body of a message to [e-mail]
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Rob Herring (Arm) April 18, 2018, 2:52 p.m. UTC | #5

On Tue, Apr 17, 2018 at 5:12 PM,  <[e-mail]> wrote:
> On 2018-04-17 10:43, [e-mail] wrote:
>>
>> On 2018-04-16 07:59, Rob Herring wrote:
>>>
>>> On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
>>>>
>>>> Documentation for last level cache controller device tree bindings,
>>>> client bindings usage examples.
>>>
>>>
>>> "Documentation: Documentation ..."? That wastes a lot of the subject
>>> line... The preferred prefix is "dt-bindings: ..."
>>>
>>>>
>>>> Signed-off-by: Channagoud Kadabi <[e-mail]>
>>>> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
>>>> ---
>>>>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58
>>>> ++++++++++++++++++++++
>>>>  1 file changed, 58 insertions(+)
>>>>  create mode 100644
>>>> Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>> b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>> new file mode 100644
>>>> index 0000000..497cf0f
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>> @@ -0,0 +1,58 @@
>>>> +== Introduction==
>>>> +
>>>> +LLCC (Last Level Cache Controller) provides last level of cache memory
>>>> in SOC,
>>>> +that can be shared by multiple clients. Clients here are different
>>>> cores in the
>>>> +SOC, the idea is to minimize the local caches at the clients and
>>>> migrate to
>>>> +common pool of memory
>>>> +
>>>> +Properties:
>>>> +- compatible:
>>>> +        Usage: required
>>>> +        Value type: <string>
>>>> +        Definition: must be "qcom,sdm845-llcc"
>>>> +
>>>> +- reg:
>>>> +        Usage: required
>>>> +        Value Type: <prop-encoded-array>
>>>> +        Definition: must be addresses and sizes of the LLCC registers
>>>
>>>
>>> How many address ranges?
>>>
>> It consists of just one address range. I'll edit the definition to make
>> it more clear.
>>>>
>>>> +
>>>> +- #cache-cells:
>>>
>>>
>>> This is all written as it is a common binding, but it is not one.
>>>
>>> You already have most of the configuration data for each client in the
>>> driver, I think I'd just put the client connection there too. Is there
>>> any variation of this for a given SoC?
>>>
>> #cache-cells and max-slices won't change for a given SOC. So you want me
>> to hard-code in the driver itself?
>>
> I can use of_parse_phandle_with_fixed_args function and fix the number of
> args as 1 instead of keeping #cache-cells here in DT. Does that look fine?

No, I'm saying why even put cache-slices properties in DT to begin
with? You could just define client id's within the kernel and clients
can use those instead of getting the id from the DT.

I have a couple of hesitations with putting this into the DT. First, I
think a cache is just one aspect of describing the interconnect
between masters and memory (and there's been discussions on
interconnect bindings too) and any binding needs to consider all of
the aspects of the interconnect. Second, I'd expect this cache
architecture will change SoC to SoC and the binding here is pretty
closely tied to the current cache implementation (e.g. slices). If
there were a bunch of SoCs with the same design and just different
client IDs (like interrupt IDs), then I'd feel differently.

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Channagoud Kadabi April 18, 2018, 6:11 p.m. UTC | #6

On 2018-04-18 07:52, Rob Herring wrote:
> On Tue, Apr 17, 2018 at 5:12 PM,  <[e-mail]> wrote:
>> On 2018-04-17 10:43, [e-mail] wrote:
>>> 
>>> On 2018-04-16 07:59, Rob Herring wrote:
>>>> 
>>>> On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
>>>>> 
>>>>> Documentation for last level cache controller device tree bindings,
>>>>> client bindings usage examples.
>>>> 
>>>> 
>>>> "Documentation: Documentation ..."? That wastes a lot of the subject
>>>> line... The preferred prefix is "dt-bindings: ..."
>>>> 
>>>>> 
>>>>> Signed-off-by: Channagoud Kadabi <[e-mail]>
>>>>> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
>>>>> ---
>>>>>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58
>>>>> ++++++++++++++++++++++
>>>>>  1 file changed, 58 insertions(+)
>>>>>  create mode 100644
>>>>> Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>> 
>>>>> diff --git 
>>>>> a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>> b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>> new file mode 100644
>>>>> index 0000000..497cf0f
>>>>> --- /dev/null
>>>>> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>> @@ -0,0 +1,58 @@
>>>>> +== Introduction==
>>>>> +
>>>>> +LLCC (Last Level Cache Controller) provides last level of cache 
>>>>> memory
>>>>> in SOC,
>>>>> +that can be shared by multiple clients. Clients here are different
>>>>> cores in the
>>>>> +SOC, the idea is to minimize the local caches at the clients and
>>>>> migrate to
>>>>> +common pool of memory
>>>>> +
>>>>> +Properties:
>>>>> +- compatible:
>>>>> +        Usage: required
>>>>> +        Value type: <string>
>>>>> +        Definition: must be "qcom,sdm845-llcc"
>>>>> +
>>>>> +- reg:
>>>>> +        Usage: required
>>>>> +        Value Type: <prop-encoded-array>
>>>>> +        Definition: must be addresses and sizes of the LLCC 
>>>>> registers
>>>> 
>>>> 
>>>> How many address ranges?
>>>> 
>>> It consists of just one address range. I'll edit the definition to 
>>> make
>>> it more clear.
>>>>> 
>>>>> +
>>>>> +- #cache-cells:
>>>> 
>>>> 
>>>> This is all written as it is a common binding, but it is not one.
>>>> 
>>>> You already have most of the configuration data for each client in 
>>>> the
>>>> driver, I think I'd just put the client connection there too. Is 
>>>> there
>>>> any variation of this for a given SoC?
>>>> 
>>> #cache-cells and max-slices won't change for a given SOC. So you want 
>>> me
>>> to hard-code in the driver itself?
>>> 
>> I can use of_parse_phandle_with_fixed_args function and fix the number 
>> of
>> args as 1 instead of keeping #cache-cells here in DT. Does that look 
>> fine?
> 
> No, I'm saying why even put cache-slices properties in DT to begin
> with? You could just define client id's within the kernel and clients
> can use those instead of getting the id from the DT.

The reason to add cache-slices here is to establish a connection between
client and system cache. For example if we have multiple instances of
system cache blocks and client wants to choose a system cache instance
based on the usecase then its easier to establish this connection using
device tree than hard coding in the driver.

> 
> I have a couple of hesitations with putting this into the DT. First, I
> think a cache is just one aspect of describing the interconnect
> between masters and memory (and there's been discussions on
> interconnect bindings too) and any binding needs to consider all of
> the aspects of the interconnect. Second, I'd expect this cache
> architecture will change SoC to SoC and the binding here is pretty
> closely tied to the current cache implementation (e.g. slices). If
> there were a bunch of SoCs with the same design and just different
> client IDs (like interrupt IDs), then I'd feel differently.

This is partially true, a bunch of SoCs would support this design but
clients IDs are not expected to change. So Ideally client drivers could
hard code these IDs.

However I have other concerns of moving the client Ids in the driver.
The way the APIs implemented today are as follows:
#1. Client calls into system cache driver to get cache slice handle
with the usecase Id as input.
#2. System cache driver gets the phandle of system cache instance from
the client device to obtain the private data.
#3. Based on the usecase Id perform look up in the private data to get
cache slice handle.
#4. Return the cache slice handle to client

If we don't have the connection between client & system cache then the
private data needs to declared as static global in the system cache 
driver,
that limits us to have just once instance of system cache block.


> 
> Rob

Channagoud Kadabi April 20, 2018, 6:51 p.m. UTC | #7

On 2018-04-18 11:11, Channa wrote:
> On 2018-04-18 07:52, Rob Herring wrote:
>> On Tue, Apr 17, 2018 at 5:12 PM,  <[e-mail]> wrote:
>>> On 2018-04-17 10:43, [e-mail] wrote:
>>>> 
>>>> On 2018-04-16 07:59, Rob Herring wrote:
>>>>> 
>>>>> On Tue, Apr 10, 2018 at 01:08:12PM -0700, Rishabh Bhatnagar wrote:
>>>>>> 
>>>>>> Documentation for last level cache controller device tree 
>>>>>> bindings,
>>>>>> client bindings usage examples.
>>>>> 
>>>>> 
>>>>> "Documentation: Documentation ..."? That wastes a lot of the 
>>>>> subject
>>>>> line... The preferred prefix is "dt-bindings: ..."
>>>>> 
>>>>>> 
>>>>>> Signed-off-by: Channagoud Kadabi <[e-mail]>
>>>>>> Signed-off-by: Rishabh Bhatnagar <[e-mail]>
>>>>>> ---
>>>>>>  .../devicetree/bindings/arm/msm/qcom,llcc.txt      | 58
>>>>>> ++++++++++++++++++++++
>>>>>>  1 file changed, 58 insertions(+)
>>>>>>  create mode 100644
>>>>>> Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>>> 
>>>>>> diff --git 
>>>>>> a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>>> b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>>> new file mode 100644
>>>>>> index 0000000..497cf0f
>>>>>> --- /dev/null
>>>>>> +++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
>>>>>> @@ -0,0 +1,58 @@
>>>>>> +== Introduction==
>>>>>> +
>>>>>> +LLCC (Last Level Cache Controller) provides last level of cache 
>>>>>> memory
>>>>>> in SOC,
>>>>>> +that can be shared by multiple clients. Clients here are 
>>>>>> different
>>>>>> cores in the
>>>>>> +SOC, the idea is to minimize the local caches at the clients and
>>>>>> migrate to
>>>>>> +common pool of memory
>>>>>> +
>>>>>> +Properties:
>>>>>> +- compatible:
>>>>>> +        Usage: required
>>>>>> +        Value type: <string>
>>>>>> +        Definition: must be "qcom,sdm845-llcc"
>>>>>> +
>>>>>> +- reg:
>>>>>> +        Usage: required
>>>>>> +        Value Type: <prop-encoded-array>
>>>>>> +        Definition: must be addresses and sizes of the LLCC 
>>>>>> registers
>>>>> 
>>>>> 
>>>>> How many address ranges?
>>>>> 
>>>> It consists of just one address range. I'll edit the definition to 
>>>> make
>>>> it more clear.
>>>>>> 
>>>>>> +
>>>>>> +- #cache-cells:
>>>>> 
>>>>> 
>>>>> This is all written as it is a common binding, but it is not one.
>>>>> 
>>>>> You already have most of the configuration data for each client in 
>>>>> the
>>>>> driver, I think I'd just put the client connection there too. Is 
>>>>> there
>>>>> any variation of this for a given SoC?
>>>>> 
>>>> #cache-cells and max-slices won't change for a given SOC. So you 
>>>> want me
>>>> to hard-code in the driver itself?
>>>> 
>>> I can use of_parse_phandle_with_fixed_args function and fix the 
>>> number of
>>> args as 1 instead of keeping #cache-cells here in DT. Does that look 
>>> fine?
>> 
>> No, I'm saying why even put cache-slices properties in DT to begin
>> with? You could just define client id's within the kernel and clients
>> can use those instead of getting the id from the DT.
> 
> The reason to add cache-slices here is to establish a connection 
> between
> client and system cache. For example if we have multiple instances of
> system cache blocks and client wants to choose a system cache instance
> based on the usecase then its easier to establish this connection using
> device tree than hard coding in the driver.
> 
>> 
>> I have a couple of hesitations with putting this into the DT. First, I
>> think a cache is just one aspect of describing the interconnect
>> between masters and memory (and there's been discussions on
>> interconnect bindings too) and any binding needs to consider all of
>> the aspects of the interconnect. Second, I'd expect this cache
>> architecture will change SoC to SoC and the binding here is pretty
>> closely tied to the current cache implementation (e.g. slices). If
>> there were a bunch of SoCs with the same design and just different
>> client IDs (like interrupt IDs), then I'd feel differently.
> 
> This is partially true, a bunch of SoCs would support this design but
> clients IDs are not expected to change. So Ideally client drivers could
> hard code these IDs.
> 
> However I have other concerns of moving the client Ids in the driver.
> The way the APIs implemented today are as follows:
> #1. Client calls into system cache driver to get cache slice handle
> with the usecase Id as input.
> #2. System cache driver gets the phandle of system cache instance from
> the client device to obtain the private data.
> #3. Based on the usecase Id perform look up in the private data to get
> cache slice handle.
> #4. Return the cache slice handle to client
> 
> If we don't have the connection between client & system cache then the
> private data needs to declared as static global in the system cache 
> driver,
> that limits us to have just once instance of system cache block.
> 
> 
>> 
>> Rob

Hi Rob:

Can you please provide your opinion on the approach here?

Thanks,
Channa

Patch

diff --git a/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
new file mode 100644
index 0000000..497cf0f
--- /dev/null
+++ b/Documentation/devicetree/bindings/arm/msm/qcom,llcc.txt
@@ -0,0 +1,58 @@ 
+== Introduction==
+
+LLCC (Last Level Cache Controller) provides last level of cache memory in SOC,
+that can be shared by multiple clients. Clients here are different cores in the
+SOC, the idea is to minimize the local caches at the clients and migrate to
+common pool of memory
+
+Properties:
+- compatible:
+	 Usage: required
+	 Value type: <string>
+	 Definition: must be "qcom,sdm845-llcc"
+
+- reg:
+	 Usage: required
+	 Value Type: <prop-encoded-array>
+	 Definition: must be addresses and sizes of the LLCC registers
+
+- #cache-cells:
+	 Usage: required
+	 Value Type: <u32>
+	 Definition: Number of cache cells, must be 1
+
+- max-slices:
+	 usage: required
+	 Value Type: <u32>
+	 Definition: Number of cache slices supported by hardware
+
+Example:
+
+	llcc: qcom,llcc@1100000 {
+		compatible = "qcom,sdm845-llcc";
+		reg = <0x1100000 0x250000>;
+		#cache-cells = <1>;
+		max-slices = <32>;
+	};
+
+== Client ==
+
+Properties:
+- cache-slice-names:
+	 Usage: required
+	 Value type: <stringlist>
+	 Definition: A set of names that identify the usecase names of a
+			client that uses cache slice. These strings are
+			used to look up the cache slice entries by name.
+
+- cache-slices:
+	 Usage: required
+	 Value type: <prop-encoded-array>
+	 Definition: The tuple has phandle to llcc device as the first
+			argument and the second argument is the usecase
+			id of the client.
+For Example:
+	venus {
+		cache-slice-names = "vidsc0", "vidsc1";
+	 	cache-slices = <&llcc VIDSC0_ID>, <&llcc VIDSC1_ID>;
+	};