Branch/Steps required to build linux kernel for Cortex R52 FVP

[Arnd Bergmann]

On Tue, Apr 12, 2022 at 12:12 AM Ayan Kumar Halder
<ayan.kumar.halder at xilinx.com> wrote:
> On 31/03/2022 16:54, Vladimir Murzin wrote:
> > On 3/31/22 12:04 PM, Ayan Kumar Halder wrote:
> > Unfortunately (or fortunately) I'm at beginning of my sabbatical
> > month, so cannot really
> > help, yet you can try ARM Support - they might have some instructions
> > (or they come to me).
> >
> > In case you have strong interest I can find bandwidth (once I'm back)
> > to submit support for
> > AN536 ("Example Cortex™-R52x2 subsystem FPGA") for MPS3, so we get
> > R-class representative mainline ;)
>
> It will be great to see it mainline. Let's touch base when you are back.
>
> To give a bit more context, I am in the process of porting Xen on Cortex
> R52 (using FVP).
>
> I am using
> https://github.com/zephyrproject-rtos/zephyr/pull/43581/commits (as a
> reference for running an app on Cortex R52 FVP) and
> https://gitlab.com/xen-project/fusa/xen-integration/-/merge_requests/15
> to see how MPU is supported.
>
> I will be needing some kernel port to run it as a Dom0 kernel for Xen.
> So having the uCLinux will be awesome.

I am rather skeptical about this, can you give some more background
about the purpose of running a full-featured Xen on a Cortex-R52
microcontroller?

I would argue that you likely have either the wrong software stack or
the wrong hardware here. If you are interested in real-time or reliability
aspects of a Cortex-R52, then adding in an experimental Linux Dom0
into the critical path will destroy both, and you should be looking at
a different hypervisor or a Dom0-less Xen with static partitioning.

On the other hand, if you want a full-featured virtualized machine
with Linux as the Dom0, it's probably better to look into using a SoC with
Cortex-A cores or similar.

       Arnd