RFR (M): 8035406: Improve data structure for Code Cache remembered sets

Mikael Gerdin mikael.gerdin at oracle.com
Tue Feb 25 04:40:31 PST 2014


On Tuesday 25 February 2014 13.04.02 Thomas Schatzl wrote:
> Hi,
> 
> On Tue, 2014-02-25 at 11:20 +0100, Mikael Gerdin wrote:
> > Hi,
> > 
> > > > Since the mutex is no longer only covering the OtherRegionsTable,
> > > > should
> > > > it be renamed to "HeapRegionRemSet lock"?
> > > > Also, it appears that the Mutex constructor actually copies the name
> > > > string
> > > > into the Mutex object, so we can format the string with whatever
> > > > contents
> > > > we want, such as "HeapRegionRemSet lock for region #4711" without
> > > > increasing footprint.
> > > 
> > > Fixed. That required some minor changes to other code though that
> > > manages the mutex.
> > 
> > Ok, I was thinking something more along the lines of:
> > 
> > _m(Mutex::leaf, FormatBuffer("HeapRegionRemSet lock #"UINT32_FORMAT, hr-
> > 
> > >hrs_index()), true),
> > 
> > FormatBuffer is the same class being used by err_msg used in asserts and
> > has a char* conversion operator, it encapsulates the char[] and snprintf
> > call so you don't need to worry about it.
> 
> I did not know about FormatBuffer. Undid the recent changes, using
> FormatBuffer now. Thanks.

Great! Code looks good, ship it!

/Mikael

> 
> New webrev at
> http://cr.openjdk.java.net/~tschatzl/8035406/webrev.2/
> 
> Testing:
> jprt
> 
> Thanks a lot,
>   Thomas



More information about the hotspot-gc-dev mailing list