let's think of the drive as being accessible. It has a faceplate with
a button on it. There is the boot partition of the disk which is
read-only except for a window of time after that button is pushed.
That boot section contains the system bootup and validator code. The
bootup and validator is carefully tested software which only needs to
be patched every few years. Then there is another section of the disk
that is write-protected unless unlocked by a crypto key the validator
software must produce. This section is used for most of the OS. It is
only made writable when the OS is installing an update to itself. Then
the rest of the disk is an ordinary partition for general use.
-
Hypothetical OS, read-only @ an integral for writing the OS to accept
input as self-modifying, as a state-validator/permission contigent to
the HD 'button';- 3-tier (bootstrap, ancillary OS extensions, & flat
sectoring following), same as MS uses at a core level resource OS
protection, roughly in puported principle, from NT ported to XP as
uncrashable -- hardwiring that attempted MS assurance furthermore,
which oughtn't be different from any valid in-house programming staff
in need of no-nonsense levels of security. Sounds applicable to how a
bank might think.