history is hard

Jon Elson elson at pico-systems.com
Sun May 31 10:04:53 CDT 2020


On 05/31/2020 02:06 AM, jim stephens via cctalk wrote:
>
>
> On 5/30/2020 11:15 PM, Eric Smith via cctalk wrote:
>> On 05/29/2020 02:38 PM, Noel Chiappa via cctalk wrote:
>>>      <snip>
>>> Low-level machines did not even have storage protection
>>> keys, and on the /40 and /50 (I think) it was an option,
>>> although I'd guess almost any /50 had it installed.
> Our /50 had it and I have never seen any indication in the 
> documentation for the hardware that
> indicated that it was an option.
>
> I don't think that either MVT or MFT would have been very 
> stable without it.  I certainly spent a lot of time
> studying how to get around it, and am responsible for a 
> couple of SPIE patches in the MVT product
> from exploits trying to get into supervisor mode to muck 
> with such.
>
Yes, the SPIE call as supplied from IBM was surely the 
security hole big enough for 5 ocean liners abreast to steam 
right through! Everybody had to patch that, and the patch 
was fairly simple.  But, it was a clear indication of how 
LITTLE IBM thought about security. Of course, they were 
thinking about banks where 3 teams reviewed code before it 
ever ran on the machine, not universities where kids would 
try all sorts of mischief.
> Yes, I know supervisor state isn't tied to the storage 
> keys, but that was the way I went to
> try to circumvent the storage keys.
Oh, once you have the P bit set to zero, you can do 
anything, such as changing the storage protection key of 
your own program.

Jon


More information about the cctalk mailing list