[Cryptography] It's all K&R's fault

[Tom Mitchell]

On Mon, Apr 21, 2014 at 2:57 PM, Sandy Harris <sandyinchina at gmail.com> wrote:
> Jonathan Thornburg <jthorn at astro.indiana.edu> wrote:
>> On Sun, Apr 20, 2014 at 08:18:18PM -0700, Bill Frantz wrote:
>
>>> And encrypted swap is really the sweet spot for encryption because
>>> all the key management problems just go away:
>> [[...]]
....
> Do we need swap on current systems? Both server and desktop
> boxes now have gigabytes of RAM and I suspect that phones
> and such do not have enough or fast enough storage to make
> swap very useful. Why not just use an OS that does not swap?

Swap is still needed, the size is open to discussion.

The reason is the fork(); exec() pair.  Yes,  vfork() can help.
 /proc/sys/vm/overcommit_memory is an interesting hack...
It is kin to credit cards and getting paid before the payment is due.
Not paycheck to paycheck but paycheck+credit card float to
paycheck.... and never bounce a check.

Long running programs should not die because the system
permits over commitment of memory resources.  Thus backing
store aka swap is needed.

Lazy allocation by the VM for sbrk() brk() should not plant
a time bomb for moments or weeks ahead because there is
no backing store.

Eliminate the fork() exec() pair and the answer is different.

-- 
  T o m    M i t c h e l l