> Right now, NetBSD has separate caches for virtual memory (things
> referenced through a process's address space) and filesystem data.  In
> addition to being inefficient at times, this means that an mmap()'d
> region isn't always consistent with what's in the filesystem according
> to read() and write().

Not to start another one of those threads, but would not the *simplest*
solution to this ongoing problem be to do what SunOS does, and that is
to implement file system operations over mmap(). SunOS's read() and
write() etc all work through mmap(). I cannot remember at what level, but
high-kernel seems reasonable.

This of course assumes that all file system types support an mmap()
operation.

Regards,
-- 
Peter Galbavy                                           peter@wonderland.org
@ Home                                                 phone://44/973/499465
in Wonderland                              http://www.wonderland.org/~peter/
                               snail://UK/NW1_6LE/London/21_Harewood_Avenue/