In article <1992Oct26.083011.17192@fys.ruu.nl> hooft@fys.ruu.nl (Rob Hooft) writes:
quoted text
>In <1cfislINNeiu@tartarus.uwa.edu.au> oreillym@tartarus.uwa.edu.au (Michael O'Reilly) writes:
>
>>Try re-writing your program to malloc the memory, and then memset()
>>over it. Your machine will die. :) The malloc() is very unlikely to
>          ^^^^^^^^^^^^^^^^^^^^^
...
quoted text
>
>Aha! So it dies and doesn't die at the same time? Well: It doesn't die. It
>just starts making a core dump. But it has only a few buffers left, so it
>will take a loooooong time to finish it. I once had a 20MB core dump because
>I ran out of memory. It took linux about 40 minutes to finish dumping,
>but it came out unharmed.
>
>Don't be afraid. Don't be impatient. It won't die. 
>
>-- 
>Rob W.W. Hooft,  Department of crystal and structural chemistry

Bob, I know you are a big advocate of the "linux will not die" camp,
but we have to be reasonable.  If I told my boss I was going out for
lunch for 40 minutes because my workstation was busy creating a core
file, I don't think he would be too sympathetic.  

Does anybody know what would be involved in setting up some kind of
SysV like limits so that a few more buffers are left around when
linux runs out of virtual memory?  Linux uses GCC as its compiler ->
GCC uses virtual memory like it is going out of style == lots
of people waiting 40 minutes for core files to appear.

- Rob Chen