Date: Wed, 29 Oct 2025 14:32:32 +0000
On Wed, Oct 29, 2025 at 2:29 PM Ville Voutilainen wrote:
>
> > Then memcpy/memmove don't have special behavior, but you
>> are also not to allowed to assume byte identity or copy/move manually.
>
> That's what we currently have. There are reasons why we have those
> relocation functions, including that vtbl re-signing.
But are we all in agreement that re-signing (or re-encrypting) a
pointer is __not__ trivial ? And therefore, that all polymorphic
object are __not__ trivially relocatable on arm64e?
>
> > Then memcpy/memmove don't have special behavior, but you
>> are also not to allowed to assume byte identity or copy/move manually.
>
> That's what we currently have. There are reasons why we have those
> relocation functions, including that vtbl re-signing.
But are we all in agreement that re-signing (or re-encrypting) a
pointer is __not__ trivial ? And therefore, that all polymorphic
object are __not__ trivially relocatable on arm64e?
Received on 2025-10-29 14:32:45
