review (XS) for 6970566: runThese fails with SIGSEGV

Tom Rodriguez tom.rodriguez at oracle.com
Thu Jul 22 13:41:43 PDT 2010


Duh.  I started the change elsewhere and moved to a different workspace but left that out.  Thanks for catching this.  I was able to reproduce the original crash so I couldn't easily test it.  I've updated the webrev.

tom

On Jul 21, 2010, at 7:04 PM, Vladimir Kozlov wrote:

> Tom,
> 
> You did not remove CodeCache::commit() from AdapterBlob::create().
> 
> Vladimir
> 
> Tom Rodriguez wrote:
>> http://cr.openjdk.java.net/~never/6970566
>> 6970566: runThese fails with SIGSEGV
>> Reviewed-by:
>> In the fix for 6965184 I added some code to track the number of
>> adapters in the code cache by calling CodeCache::commit after the
>> AdapterBlob was created.  The code was missing a null check in case
>> the allocation failed and that would lead to crashes if the code cache
>> filled up.  Instead of adding a null check I moved it into the
>> AdapterBlob constructor to mirror other calls to commit.  I also
>> deleted a dead constructor.



More information about the hotspot-compiler-dev mailing list