RFR(M): 8244660: Code cache sweeper heuristics is broken

Robbin Ehn robbin.ehn at oracle.com
Fri May 29 10:35:22 UTC 2020


Hi Nils,

On 2020-05-29 11:41, Tobias Hartmann wrote:
> Hi Nils,
> 
> On 27.05.20 22:42, Nils Eliasson wrote:
>> New webrev: http://cr.openjdk.java.net/~neliasso/8244660/webrev.03/
> 
> Very nice findings and cleanup! Looks good to me, below are some minor comments.

Looks good, thanks for fixing. (minus Tobias nits below)

Thanks, Robbin

> 
> globals.hpp
> - line 1723: "controling" -> "controlling"
> - line 1724: "percent" -> "percentage"
> 
> sweeper.cpp
> - line 258: ifs should be merged
> - line 484: "There can be data races on _bytes_changed. The data races are benign, since it does not
> matter if we loose a couple of bytes." This is not true anymore with atomic loads/stores, right?
> - Not sure if 'should_start_sweep' really needs it's own method. Inlining it into the single caller
> would have all the accesses to _bytes_changed in one place.
> 
> sweeper.hpp
> - line 73: empty comment
> 
> Best regards,
> Tobias
> 


More information about the hotspot-compiler-dev mailing list