RFR: Remove NMethodSizeLimit adjustment for Shenandoah

[Roman Kennke]

Yes, thank you! This does indeed fix my C1 woes on aarch64.

Roman

> Turns out, this is not a good idea. AArch64 asserts later because some instructions become
> non-encodable on C1 paths.
> 
> diff -r 551c72da087a src/hotspot/share/gc/shenandoah/shenandoahArguments.cpp
> --- a/src/hotspot/share/gc/shenandoah/shenandoahArguments.cpp   Mon Sep 10 10:27:14 2018 +0200
> +++ b/src/hotspot/share/gc/shenandoah/shenandoahArguments.cpp   Mon Sep 10 17:55:24 2018 +0200
> @@ -179,10 +179,2 @@
> 
> -  // Shenandoah needs more space in generated code to put barriers in.
> -  // TODO: NMethodSizeLimit should not be develop.
> -#ifdef ASSERT
> -  if (FLAG_IS_DEFAULT(NMethodSizeLimit)) {
> -    FLAG_SET_DEFAULT(NMethodSizeLimit, NMethodSizeLimit * 3);
> -  }
> -#endif
> -
>    // Shenandoah needs more C2 nodes to compile some methods with lots of barriers.
> 
> 
> Testing: tier3_gc_shenandoah (running)
> 
> Thanks,
> -Aleksey
>