Re: Out of memory kernel deat

Mike Jagdis (mike@roan.co.uk)
Tue, 13 May 1997 09:32:12 +0100 (GMT/BST)


On Mon, 12 May 1997, Raul Miller wrote:

> On May 12, Tim Hollebeek wrote
> > SIGDANGER is essentially useless because it is AIX specific; >99% of
> > software ignores it so it gains you nothing.
>
> Unless you have a daemon ready to implement some kind of process killing
> policy when it gets the signal. I think you'd need real-time extensions
> to write this daemon right (e.g. lock it into memory). And, of course,
> you'd want it to be very, very small.

We already have memory lock down. We already have real time
scheduling. We even have a spare bit in the standard signal
vector.

And, yes, I do believe that a broadcast SIGDANGER with a default
behaviour of ignore is a useful start. Ok, hardly anything uses
it at the moment - but then if it existed perhaps things might
change? :-)

Mike

-- 
.----------------------------------------------------------------------.
|  Mike Jagdis                  |  Internet:  mailto:mike@roan.co.uk   |
|  Roan Technology Ltd.         |                                      |
|  54A Peach Street, Wokingham  |  Telephone:  +44 118 989 0403        |
|  RG40 1XG, ENGLAND            |  Fax:        +44 118 989 1195        |
`----------------------------------------------------------------------'