sr (nobody.delete@this.nowhere.com) on October 2, 2021 11:16 am wrote:
> rwessel (rwessel.delete@this.yahoo.com) on October 1, 2021 6:19 am wrote:
> > The 386 had a 48(-ish*) bit virtual address space, but that was mapped into a 32 bit linear address
> > space. Increasing the size of the linear address space would be been fairly trivial, and essentially
> > invisible from the application layer (given the era, it's obvious why they wouldn't have initially
> > defined the linear address space bigger than 32 bits). Not really unlike how the 286 had a 32(-ish*)
> > bit virtual address space mapped into a 24-bit physical address space (and how a 16-bit OS could have
> > used the 386 enhancements to address more than 24-bits worth of memory, without using any "32-bit"
> > feature of the CPU - just extra, previously unused, bits in the descriptor tables).
> >
>
> By obvious you mean that by defining linear address space as big as segment offset it makes possible
> to map all memory with only offset? So it makes possible to not use segmentation at all - or use segmentation
> with possibility to access all needed segments within single segment where needed.


No. The linear address space was not made larger than that because there would be, for the time being, no plausible implementations where it could practically be bigger than that. To the extent that segmentation use would consider in the future, they left themselves with a trivial upgrade.


> Yeah it would be trivial to extend address space with segmentation - but benefits of being able to access
> all memory with single pointer far exceeds it. I believe Intel did pretty good job with their segmentation
> model with i386, it was just that programmers decided that segmentation wasn't worth the trouble.


You really can't have it both ways...