Fixes for Zero ppc[32/64] issues:

[Andrew Haley]

On 01/10/14 04:50, Chris Phillips wrote:
> - )
> -ZERO_ONLY (
> - os::Linux::min_stack_allowed = MAX2(os::Linux::min_stack_allowed,           
> -            (size_t)(StackYellowPages+StackRedPages+StackShadowPages+
> -                     2*BytesPerWord COMPILER2_PRESENT(+1)) * Linux::page_size());
> - )
> +
> +#ifdef ZERO
> +  // If this is Zero, allow at the very minimum one page each for the
> +  // Zero stack and the native stack.  This won't make any difference
> +  // for 4k pages, but is significant for large pages.
> +  os::Linux::min_stack_allowed = MAX2(os::Linux::min_stack_allowed,
> +             (size_t)(StackYellowPages+StackRedPages+StackShadowPages+2) * Linux::page_size());
> +#endif
> +

Thanks for posting those.

We should look at upstreaming them, but I can't really justify this
hunk.  It gives us some extra wiggle room for the native Zero stack,
but I don't really know how much that should be.

On AArch64, StackShadowPages == 2 when 64k pages are is use.  That
really should be enough for our use: it should result in 5 * 64k ==
320k.  128k of that is the red and yellow zone, leaving 192k.

Andrew.