Take 2: RFR: 8197429: Increased stack guard causes segfaults on x86-32

[Andrew Haley]

Webrev amended.

http://cr.openjdk.java.net/~aph/8197429-2/

Copyrights fixed.

Test case changed to be built with the rest of the native code.

We now only bang down the stack to its maximum size if we're going to
run Java code on the primordial stack.  This only happens with the
JNI invocation interface.

The new logic is:

  if we're on the primordial stack
    bang down the stack to its maximum size

  try to map the codebuf just below the primordial stack

  if that didn't work
    try to map again, but 1 megabyte lower

OK?

-- 
Andrew Haley
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671