RFR: JDK-8145180: Add back PrintGC, PrintGCDetails and -Xloggc
bengt.rutisson at oracle.com
Tue Jan 19 07:34:33 UTC 2016
Thanks for looking at this.
On 2016-01-19 08:15, kirk at kodewerk.com wrote:
> Hi Bengt,
> I think I commented on the loss of familiar flags earlier on so very happy to see them back.
Glad you like the change.
> Not sure what prompted you to go through all the effort to put back all this code. I think I would have taken the lazy way out and just provided some mapping in Arguments. Given the size of the change in logging, getting it exactly right isn’t that important. That said, what is done is done and I don’t see any obvious.
Not really sure what you mean. The change I propose is pretty much just
a mapping in Arguments. It does take the -Xloggc stuff into account, but
other than that it is pretty much just a simple mapping.
>> On Jan 18, 2016, at 4:52 PM, Bengt Rutisson <bengt.rutisson at oracle.com> wrote:
>> Hi everyone,
>> Could I have a couple of reviews for this follow-up change to the GC logging changes for the unified logging framework?
>> The change adds back the flags PrintGC and PrintGCDetails as well as -Xloggc. I've tried to keep the behavior similar to before, so if -Xloggc is used the PrintGC and PrintGCDetails settings should affect what is logged to the file not to stdout.
>> The change also adds warning messages for all three flags that will let the users know that the flags have been deprecated.
>> I've tried to add this code back in similar places where the code used to be. I've put PrintGC and PrintGCDetails back in globals.hpp at the same place where they used to be. I've added a method, handle_deprecated_print_gc_flags(), which is called from the place in Arguments::parse() where we used to enable PrintGC if PrintGCDetails was enabled (http://hg.openjdk.java.net/jdk9/hs-rt/hotspot/rev/f944761a3ce3#l134.259). I also re-introduce the Arguments::_gc_log_filename variable to hold the file name specified by -Xloggc.
More information about the hotspot-gc-dev