Re: [PATCH v1 2/3] ARM: GICv3 ITS: do not invalidate memory while sending a command

[Volodymyr Babchuk <Volodymyr_Babchuk@xxxxxxxx>]

Hi Mark,

I am writing to you, because you are GICv3 maintainer in Linux. We are
updating ITS driver in Xen and we have a question about cache
maintenance WRT memory shared with ITS. As I can see, the Linux ITS
driver uses gic_flush_dcache_to_poc() all over the code. This boils down
to "dc civac" instruction which does both clean and invalidate. But do
we really need to invalidate a cache when we are sending an ITS command?
In my understanding it is sufficient to clean a cache only and Linux
uses clean&invalidate just out of convenience. Is this correct?

Below you can find our discussion about this.

Julien Grall <julien@xxxxxxx> writes:

> On 19/09/2023 15:36, Volodymyr Babchuk wrote:
>> Julien Grall <julien@xxxxxxx> writes:
>>> On 19/09/2023 12:28, Volodymyr Babchuk wrote:
>>>> There is no need to invalidate cache entry because we just wrote into a
>>>> memory region. Writing itself guarantees that cache entry is valid.
>>>
>>> The goal of invalidate is to remove the line from the cache. So I
>>> don't quite understand the reasoning here.
>>>
>> Well, I may be wrong, but what is the goal in removing line from the
>> cache? As I see this, we want to be sure that ITS sees data written in
>> the memory, so we should flush a cache line. But why do we need to
>> remove it from CPU's cache?
>
> I don't exactly know. From a brief look I agree with you. However, our
> driver is based on Linux where the clean & invalidate is also used. So
> I am a little be cautious to remove it.
>
> The way forward would be to ask the Marc Zyngier (GICv3 maintainer)
> why it was added in Linux. Then we can decide whether this can be
> removed in Xen.
>
> Cheers,


-- 
WBR, Volodymyr