Agner`s CPU blog

have you checked out the work of Darek Mihocka who some years ago wrote a multi-part series on fixing x86 (and, interestingly, proposed eliminating hardware paging)

Many of the security problems that Darek discusses can be solved with CRISC1. The memory map in my proposal is so small that it is no problem to give each thread in a process its own memory map and its own private data space that cannot be accessed from other threads. Do you know if any other systems have memory protection between threads? I can't find any (other than emulators). Of course we have thread-local storage, but that does not guarantee that one thread cannot corrupt data belonging to another thread in the same process in case of program bugs or malicious hacking.

I don't see any reason why different threads should have access to each other's private data. Synchronization structures (e. g. semaphores) and resource sharing between threads should go through the shared main memory of the application, not the private memory of one thread.