Request for Review (s) - 8149343: assert(rp->num_q() == no_of_gc_workers) failed: sanity

Thomas Schatzl thomas.schatzl at oracle.com
Mon Mar 21 11:07:24 UTC 2016


Hi Jon,

On Fri, 2016-03-18 at 18:16 -0400, Kim Barrett wrote:
> > On Mar 18, 2016, at 2:39 PM, Jon Masamitsu <
> > jon.masamitsu at oracle.com> wrote:
> > 
> > New versions of the webrev's
> > 
> > Changes from 01
> > http://cr.openjdk.java.net/~jmasa/8149343/webrev_delta.01_02/
> > 
> > Complete
> > http://cr.openjdk.java.net/~jmasa/8149343/webrev.02/
> > 
> > This wrong indentation does not show as changed lines
> > in the webrev but if you look for them, they are fixed in 02.
> 
> ---------------------------------------------------------------------
> --------- 
> src/share/vm/gc/shared/referenceProcessor.cpp
>  696   for (uint i = active_length; i < _max_num_q; i++)
>  697     assert(ref_lists[i].length() == 0, "%u unexpected References
> in %u",
>  698            ref_lists[i].length(), i); 
> 
> Missing braces around for-loop body.
> 
> Also, though I expect the compiler will eliminate the empty loop
> entirely in release mode, for readability I'd prefer the whole
> for-loop be surrounded by #ifdef ASSERT ... #endif.
> 
> ---------------------------------------------------------------------
> ---------
> 
> I don't need a new webrev for those.
> 
> Otherwise looks good.
> 

  apart from the issue Kim showed, looks good. I do not need a new
webrev either.

Thanks,
  Thomas



More information about the hotspot-gc-dev mailing list