Thank you Alan!
<http://cr.openjdk.java.net/%7Eigerasim/7129312/2/webrev/>
This looks much better (although I'm still wondering about greedy
tests wanting to reserve 4GB when running tests concurrently). But
let's go with what you have for now. One small suggestion is to assume
"PARENT" when invoking without any args as that will make it easier to
run interactively. I'm happy to sponsor this, just send me a link to
the change-set.
I couldn't come up with any idea on how to test the overflow without
actually allocating that much memory.
I simplified the test a bit once again. Instead of invoking the same
class twice, I created a designated class for the child.
Here's the new webrev:
http://cr.openjdk.java.net/~igerasim/7129312/4/webrev/
Hope it looks fine.
And here's the exported change set:
http://cr.openjdk.java.net/~igerasim/2commit/7129312-jdk8-Negative-array-size.patch
Sincerely yours,
Ivan
