[PATCH v1] arm64: dts: rockchip: Add cache information to the Rockchip RK3566 and RK3568 SoC

Dragan Simic dsimic at manjaro.org
Sun Mar 3 11:10:55 PST 2024


Hello Anand,

On 2024-02-28 18:50, Dragan Simic wrote:
> On 2024-02-28 11:42, Dragan Simic wrote:
>> On 2024-02-27 15:58, Dragan Simic wrote:
>>> On 2024-02-27 13:49, Anand Moon wrote:
>>>> On Tue, 27 Feb 2024 at 00:39, Dragan Simic <dsimic at manjaro.org> 
>>>> wrote:
>>>>> On 2024-02-26 19:23, Anand Moon wrote:
>>>>> > As per RK3568 Datasheet and TRM add missing cache information to
>>>>> > the Rockchip RK3566 and RK3568 SoC.
>>>>> >
>>>>> > - Each Cortex-A55 core has 32KB of L1 instruction cache available and
>>>>> >       32KB of L1 data cache available with ECC.
>>>>> > - Along with 512KB Unified L3 cache with ECC.
>>>>> >
>>>>> > With adding instruction cache and data cache and a write buffer to
>>>>> > reduce the effect of main memory bandwidth and latency on data
>>>>> > access performance.
>>>>> 
>>>>> I was about to send my own patch that adds the same missing cache
>>>>> information, so please allow me to describe the proposed way to 
>>>>> move
>>>>> forward.
>>>>> 
>>>>> The way I see it, your commit summary and description need a rather
>>>>> complete rewrite, to be more readable, more accurate, and to avoid
>>>>> including an irrelevant (and slightly misleading) description of 
>>>>> the
>>>>> general role of caches.
>>>>> 
>>>>> Also, the changes to the dtsi file would benefit from small 
>>>>> touch-ups
>>>>> here and there, for improved consistency, etc.
>>>>> 
>>>>> With all that in mind, I propose that you withdraw your patch and 
>>>>> let
>>>>> me send my patch that will addresses all these issues, of course 
>>>>> with
>>>>> a proper tag that lists you as a co-developer.  I think that would
>>>>> save us a fair amount of time going back and forth.
>>>>> 
>>>>> I hope you agree.
>>>> 
>>>> I have no issue with this,.If you have a better version plz share 
>>>> this.
>>> 
>>> Thank you, I'll send my patch within the next couple of days.
>> 
>> Here's a brief update...  Basically, not all of the cache-size values
>> found in your patch were correct, but I've got all of them calculated
>> again, double-checked, and cross-compared with the way values in my
>> earlier patch for the RK3399 SoC dtsi were calculated. [2]
>> 
>> It all checked out just fine.  It's all based on the RK3566 and RK3568
>> SoC datasheets and a couple of ARM specifications, which I'll describe
>> in detail in my patch description.  I'll send the patch after I test
>> it a bit, to make sure it all works as expected.
>> 
>> [1] 
>> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=b72633ba5cfa932405832de25d0f0a11716903b4
> 
> Pretty much the same patch for the RK3328 is also ready for testing.

Just sent the patches to the mailing list, please have a look. [2][3]

I've "downgraded" the previously proposed "Co-developed-by" tag in the
RK356x patch [3] to a "Helped-by" tag, just because the cache-size 
values
in your patch mostly weren't correct and, as a result, differed from the
cache-size values in my patch, making the "Co-developed-by" tag 
technically
not applicable.  For that tag to be applicable, the most important parts
of the patches need to be pretty much identical.

I hope you agree.

[2] 
https://lore.kernel.org/linux-rockchip/e61173d87f5f41af80e6f87f8820ce8d06f7c20c.1709491127.git.dsimic@manjaro.org/
[3] 
https://lore.kernel.org/linux-rockchip/2285ee41e165813011220f9469e28697923aa6e0.1709491108.git.dsimic@manjaro.org/



More information about the Linux-rockchip mailing list