RFR: JDK-8143255: Remove debug logging from SymbolTable::unlink() and SymbolTable::possibly_parallel_unlink() (original) (raw)

David Holmes david.holmes at oracle.com
Wed Nov 18 21:42:03 UTC 2015

Hi Bengt,

On 19/11/2015 5:47 AM, Bengt Rutisson wrote:

Hi everyone, Could I have a couple of reviews for this patch to remove some debugging code? http://cr.openjdk.java.net/~brutisso/8143255/webrev.00/ https://bugs.openjdk.java.net/browse/JDK-8143255 From the bug report: The logging inside SymbolTable::unlink() and SymbolTable::possiblyparallelunlink() is guarded by Verbose and WizardMode, which are both develop flags. This means that the log output is not available in product builds. So, this logging is most likely just leftover debugging code.

As this is primarily guarded by PrintGCDetails the GC team owns this as far as I am concerned and so they can do as they desire. However ...

Most, if not all, non-product logging code could be characterized as "leftover debugging code" - but that doesn't mean it should all be ripped out. The use of Verbose && WizardMode does seem a little excessive in this case - one should suffice. As documented the aim is to ensure that this logging, while logically wanted as part of PrintGCDetails, is not included in the product mode because people parse that logging information.

In the GC team we don't use this logging (even though it is for some reason guarded by PrintGCDetails too) and I discussed it with Coleen and Rachel in the Runtime team. The Runtime team also don't use the logging.

This kind of logging is used to expose additional details while testing/debugging issues. It isn't expected to be used on a regular basis, and may well have not been used for years, or may have last been used extensively by people no longer working on OpenJDK.

It is pretty straight forward to add this logging back if it is ever needed for a debugging session. Until then we should remove this from the code base.

Mostly likely someone would not even know it ever existed and would simply re-write a new piece of logging code - which might even get checked in. And the cycle repeats. ;-)

Given the trend elsewhere I'm a little surprised this isn't simply being included in the product mode logging. :)

Cheers, David

Thanks, Bengt

More information about the hotspot-runtime-dev mailing list