Re: [RFC PATCH] arm/gicv2: make GICv2 driver and vGICv2 optional

[Luca Fancellu <Luca.Fancellu@xxxxxxx>]

> On 3 Aug 2023, at 09:29, Julien Grall <julien@xxxxxxx> wrote:
> 
> 
> 
> On 03/08/2023 09:04, Luca Fancellu wrote:
>>>> There are so many things that can go wrong during early boot including the 
>>>> entire boofdt parsing
>>>> and having earlycon would be so desirable.
>>> It is not clear what to take from your reply. Earlier you were concerned 
>>> about the error not showing up in the log if the .config is not correct.
>>> 
>>> There is no really quick fix for that as a .config may work for platform A 
>>> but not platform B. The only viable solution is having an early console.
>>> 
>>> Anything else like forcing to always have one GICvX selected is just a hack 
>>> that would work for one set of people but not the others.
>> I agree with you in part, when you say that it would work for one set of 
>> people but not the others, isn’t it always the case? We are providing
>> a defconfig that would fit the majority of the people, but it’s always a set 
>> of them.
> 
> Correct.
> 
>> More specific to this patch, with the provided Kconfig “hack” as you say, it 
>> is not only providing the same user experience as the current state,
>> it is doing more, it is providing a way to exclude from the build the GICv2 
>> which is not possible currently, this work aims to provide a more
>> fine granular configuration so that experienced users can remove entire 
>> modules that they don’t need in their platform so that they don’t have
>> to take them into account when going to, for example, safety critical audits.
> 
> There seems to be some misunderstanding about the hack I am referring to. 
> This is not about the patch itself but the following line:
> 
> select GICV2 if !GICV3 && !NEW_VGIC
> 
> Now that GICv2 can be deselecting, I see no point to force select GICV2 if 
> GICV3 is not selected.

Ok I see

> 
> As I wrote earlier, there is so many way for someone to create .config that 
> is wrong or doesn't boot on their platform. So someone tweaking .config 
> should always be careful when selecting/deselecting options.

Yes indeed

> 
>> I agree with you that earlycon would be perfect, so that we could make the 
>> user remove every module and let him know quickly the issue,
>> but on the other hand it is a complete new feature that you are requesting 
>> (are you?) to make this patch go forward.
> 
> I didn't suggest it. I was making the point that if you have time, it is best 
> to spend it trying to enable the early console rather than trying to prevent 
> the user to disable both GIC.

It will be for sure something I will investigate, but in the mean time 
unfortunately I won’t have time at least until October
to even start an investigation.

I would like to ask if you see this one going forwards or not (as it is), 
because I have a set of patches to isolate and Kconfig-out
the dom0less code, that is depending on this one (for a small bit) and before 
sending them I need to understand if this one can
see the light or not.


> 
> Cheers,
> 
> -- 
> Julien Grall