RFR (XS): 8033443: Test8000311 fails after latest changes to parallelize string and symbol table unlink

Bengt Rutisson bengt.rutisson at oracle.com
Tue Feb 4 08:48:47 PST 2014


Hi Thomas,

This looks good to me.

One minor thing that I leave up to you to fix if you want to.

It seems more natural to me to pass an "is parallel" value to the 
constructor of G1StringSymbolTableUnlinkTask rather than have it look at 
G1CollectedHeap::use_parallel_gc_threads() itself to set the 
_do_in_parallel value.

Thanks,
Bengt

On 2/3/14 5:04 PM, Thomas Schatzl wrote:
> Hi all,
>
>    can I have reviews for the following change? It fixes the check that
> verify the correct amount of processed entries during string/symbol
> table unlink.
>
> The change adds an exception for the non-parallel case (i.e. the checks
> will always succeed in that case).
>
> The alternative would be to make the non-parallel case to also "claim"
> work, and use the parallel code. That would require somewhat larger
> refactoring of the string/symbol table code. In that case it would also
> be nice if the string/symbol table iteration did the same. I.e. it seems
> a good enough fix for now.
>
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8033443/webrev/
>
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8033443
>
> Testing:
> jprt, jtreg (all gc tests)
>
> Thomas
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20140204/c4b707ce/attachment-0001.html 


More information about the hotspot-gc-dev mailing list