In message <20010604191911.18110.qmail@mail.netbsd.org>eeh@netbsd.org writes

[...]

>I considered mapping the frame buffer with a single giant page, 
>but the current mmap() interface does not lend itself to that.)

Have we *still* not fixed that?  years ago someone suggested a new
devsw entry (solaris-inspired?)  to allow mmap'ing a `segment' in one
fell swoop. Seemed like the right approach to me, allowed incremental
(port-by-port) updates, via a simple, MI wrapper which just iterated
over all pages in the range.


Were there actual problems with that, or was it just lack of
time/motiviation?