RFR (L): 8067433: Keep waste at end of regions for further allocation during GC

Thomas Schatzl thomas.schatzl at oracle.com
Fri Oct 2 11:22:43 UTC 2015


Hi,

On Thu, 2015-10-01 at 18:34 +0300, Dmitry Fazunenko wrote:
> Hi Thomas,
> 
> I looked at the new regression test you provided. It looks good to me, a 
> few minor comments though:
> 
> 1) Please, use 4-space indent, this is java... :)
> 2) 8031686 --> 8067433
> 3) I think you need to clarify in the comments what the test does:
> The summary states: "Test to ensure that retaining regions during GC works".
> But what the test does is: it allocates memory hoping to cause 
> fragmentation and to enforce the region retaining mechanism, right?
> So, the test doesn't ensure that retaining works or even occurs, it 
> checks that retaining doesn't crash if happens.
> I'm fine with the test itself, I just want the comments were more precise.

  thanks for the (partial) review. I tried to fix all these concerns,
see

http://cr.openjdk.java.net/~tschatzl/8067433/webrev.1/ (full)
http://cr.openjdk.java.net/~tschatzl/8067433/webrev.0_to_1/ (diff)

Thanks,
  Thomas




More information about the hotspot-gc-dev mailing list