vmas are per-mapping: in general each call to mmap() establishes a new
mapping. So when different processes, map the same shared file, they
each have their own vmas. When a file is mapped several times in
different places or with different permissions, those use different vmas
too. (E.g. an ELF executable usually involves several mappings of the
same file into the same memory space, with different permissions).
If several cloned threads access the same memory-mapped area, they will
all hit the same vma: only one mmap() call is used to set up the
mapping, which is then visible to all the threads.
Exceptions: adjacent mappings in the same memory space may have their
vmas merged together, if the permissions, file offset and file
etc. match up at the ends. Mappings may also be split up by munmap()
and mmap(), and so on.
-- Jamie
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/