Re: [PATCH] arm64: mm: Defer read-only remap of data/bss linear alias

[Kevin Brodsky]

On 22/06/2026 18:42, Ard Biesheuvel wrote:
>> It would also be good to explain why we can't do this in
>> mark_linear_text_alias_ro() (presumably because this is too early). I do
>> agree that honouring rodata=off is a good thing, though.
>>
> I didn't consider that tbh - I'll check whether that is a more suitable
> location. But generally, I think doing it later rather than earlier is
> fine because it reduces the likelihood of hidden issues such as the one
> being addressed by this patch, and I don't think doing this very early
> makes a meaningful difference in terms of robustness.

That makes sense, probably not worth the hassle of checking if
mark_linear_text_alias_ro() would work.

>> BTW, if only kvm_arm_init() needs to read the data/bss via the linear
>> map, maybe we can still unmap it later?
>>
> Yes, it is still my intent to bring back the patches that unmap this alias
> entirely, including some additional tweaks for the fixmap PTE level table
> that I dropped after an earlier revision. But that will be 7.3 material,
> whereas this fixes a regression due to the changes that were pulled for 7.2.

Sounds good.

- Kevin