review request (XXS) - 7176220: missing date stamps

John Coomes John.Coomes at oracle.com
Wed Jun 13 06:53:43 UTC 2012


Jon Masamitsu (jon.masamitsu at oracle.com) wrote:
> John,
> 
> Why isn't the call to date_stamp() at 489 sufficient?  I can
> see from the logs that it is not but  don't get it?

The do_collection code is ... confusing.  When you have PrintGC &&
!PrintGCDetails, the TraceTime that's active is buried in the call to
collect() at 595; the TraceTimes in do_collection itself are not
active.  In that case, the date_stamp() at 489 covers only the first
iteration through the loop.

I've decided I don't like the huge distance between the call to
date_stamp() and the TraceTime that it's labeling (in both the
existing code and my change).  So ignore this webrev; I'll try for
something better.

-John

> On 6/12/2012 2:01 PM, John Coomes wrote:
> > Hi,
> >
> > Please review this tiny change to add date stamps for some Full GCs.
> > The bug has existed since -XX:+PrintGCDateStamps was added in
> > hs10/jdk6u4.
> >
> > http://cr.openjdk.java.net/~jcoomes/7176220-date-stamps/
> >
> > -John



More information about the hotspot-gc-dev mailing list