RFR(XXS): 8007084: EnableTracing asserts in non-product builds

Markus Grönlund markus.gronlund at oracle.com
Wed Jan 30 07:08:22 PST 2013


Greetings,

 

Asking for review and putback sponsorship for the following very simple fix:

 

Bugid:  http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8007084

Webrev: http://cr.openjdk.java.net/~mgronlun/8007084/webrev01/

 

Comment: The effect of this fix is that stringStream will now always assign a valid timestamp upon constructing its outputStream. The impact of this is considered to be minimal. I elaborated with adding additional constructors to stringStream with options for enabling timestamping, but this proved cumbersome, there is already a default constructor taking a size_t for initial size. In addition, adding new constructors would mean chasing down the oop::print_string, oop::print_on, oop::print_value_on chains and updating all stringStream alloc sites. Since the timestamp field is already in outputStream, I made this tradeoff for this bugfix instead - which I believe will have little impact in practice.

 

Cheers

Markus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/attachments/20130130/aeb74de8/attachment.html 


More information about the hotspot-runtime-dev mailing list