"Stable" kernel 2.6.25.6

problem is, it doesn't. first, this feature is default off. second, even when it's enabled,
it's got exceptions so if any of those apps has a normal code execution bug, it's game over
again (and i doubt you can prove that, say, the Xorg server is free of such bugs). third, not
all 'NULL-deref' bugs actually access the first page or so only, it all depends on how the
NULL pointer is used in the code, a big enough offset added to it can easily lift it beyond
the mmap protected region (read LWN's explanation on the recent vmsplice bug and my comment on
how it's different on amd64 - not a NULL deref at all). fourth, not all invalid kernel pointer
dereference bugs are due to NULL pointers, maybe read Brad's post again and check out that
last URL ;).