first bits from a way to long flight

suggested by "Karsten W. Rohrbach" <karsten@rohrbach.de>

ok miod@, tedu@, pedro@

requires memory; try to make sure we have it. If all fails, leak
instead of crash. Test case originally found by cloder@, fix tested
by many.

on most architectures.  From Oliver Smith.  OK deraadt@ and henning@

spotted by ray some time ago

region succeeds, but allocation a required page dir failed. This
can happen if we're really close to ulimit after allocation the
region of the size requested.  See malloc_ulimit1 regress test.
Tested by many; thanks.

tested by quite a few developers. ok deraadt@

pave the way for future work; ok since this does not change the generated
MAKEDEV

doesn't trigger security(8)'s ownership checks. ok bob@

root node doesn't have a parent, and POSIX does not say what should be
done in that case. Warn developers that different implementations
may do different things.

pointer to the freed root node, but return a pointer to the new
root node. POSIX does not define, what should be returned in
that case.
Fixes Coverity CID 2528.
ok millert@ otto@

remove redundant check on size. ok millert@ deraadt@

actually help us find bugs later (with lint, not cc), if they get
introduced

- Fix semantics: seekdir(pos); telldir() shoud return pos. The code
that implements this will be made faster in a later commit.
- We loose documented behaviour (after closedir() the telldir()
positions are not valid anymore). This was never in Posix, and most
other systems have nothing like it.
Diff originally from Paul Thorn, rewritten by me using some FreeBSD
code.  "slap it in" deraadt@

updated but the function itself never was; ok millert

ok jmc@ otto@

driver

suggested and ok by kettenis@

we are making the move where ttya will become the zs devices, and tty00
will support com(4) devices. ok drahn

or if malloc must be used suggest check.
Get rid of "one".
OK deraadt@ and jmc@, OK kjell@ to earlier version with "one"s.

ok naddy@ todd@

grep should assume ASCII text, fixes umask detection
ok millert@ jaredy@

David Hill's diffs.
ok otto

to (for instance) interrupt dhclient with a ^C instead of a 'CAPS-LOCK C'
from e@molioner.dk checked by matthieu