winscape 2269

It would be wonderful if it did do a clean crash. MS-DOS does not; again, I will repeat myself. MS-DOS is an example of what not to do.

Again, it depends on the stuff you have on hand, both hardware and software.

What do mean "if it cannot"? Sheesh! It's only hardware.

You are being obdurate. I'm beginning to suspect that you are doing this on purpose.

That is an option but takes a longer time to implement. There are all kinds of ways. Those ways are determined by what you have both, in software and in hardware.

buttume you have paging hardware. The hardware will have some flavor of page fault flag. So you arrange your page mapping in such a way that any out of bounds will cause a page fault. Your page fault handler can then determine whether the reference was legitimate or not. If you don't have paging hardware but a flat core, your handler will have to take another approach based on the usage of the system. Is it guaranteed to be only one task at a time? Does it have to allow two sections of code in core (this could be two users, two tasks, two segments, two thingies)?

There is no One True Way to deal with memory management. Threatening precious body parts of the CPU designers to include a hardware buttist is quite effective. I think the reason the Intels of this world don't provide this service is precisely because they were not using their gear as part of their production. It's amazing what CPU designers will provide when they are the ones who have to wrestle with absence of these features.

BAH