RFR (S): 7176220 'Full GC' events miss date stamp information occasionally

Bengt Rutisson bengt.rutisson at oracle.com
Tue May 14 06:23:01 UTC 2013


Hi Andreas,

Looks good.

One question regarding the change in GCTraceTime:

http://cr.openjdk.java.net/~aeriksso/7176220/webrev.00/src/share/vm/gc_implementation/shared/gcTraceTime.cpp.udiff.html

It looks to me like we are missing a ": " in the new version. Maybe it 
is ok to loose this character in the log format, but a safer change may be:

if (PrintGCTimeStamps || PrintGCDateStamps) {
   gclog_or_tty->date_stamp(PrintGCDateStamps);
   gclog_or_tty->stamp(PrintGCTimeStamps);
gclog_or_tty->print(": ");
}

This actually still changes the ": " behavior since since you didn't use 
to get a ": " if you only had PrintGCDateStamps enabled. But I think 
this is a more expected behavior.

Thanks,
Bengt

On 5/13/13 6:01 PM, Andreas Eriksson wrote:
> Hi all,
>
> Please take a look at my changes for bug 7176220.
> The problem is that CMS sometimes misses datestamps for time trace 
> printouts.
>
> The proposed fix is to move the datestamp print logic into the class 
> that handles time trace printouts, GCTraceTime.
> This would make it consistent with timestamps, which are already 
> printed from there.
>
> Bugs.sun:
> http://bugs.sun.com/view_bug.do?bug_id=7176220
>
> JIRA:
> https://jbs.oracle.com/bugs/browse/JDK-7176220
>
> Webrev:
> http://cr.openjdk.java.net/~aeriksso/7176220/webrev.00/
>
> Testing:
> jprt, jtreg
>
> Thanks,
> Andreas
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20130514/80f2525f/attachment.htm>


More information about the hotspot-gc-dev mailing list