Code review request: 7198529 NPG: assert with NMT code in Thread destructor
David Holmes
david.holmes at oracle.com
Fri Sep 14 20:53:46 PDT 2012
On 15/09/2012 1:28 AM, Zhengyu Gu wrote:
> Thread's stack base address can be NULL, if the thread is never started
> or exited before calling Thread::record_stack_base_and_size().
>
> Webrev: http://cr.openjdk.java.net/~zgu/7198529/webrev.00/
> CR: http://monaco.sfbay.sun.com/detail.jsf?cr=7198529 (not available on
> bugs.sun.com)
Looks odd to check _stack_base and then call stack_base(). But otherwise ok.
Makes me wonder though whether there should be something explicit in the
thread startup that sets a "NMT is valid" flag, that then gets used by
all NMT related actions. Just a thought.
David
> Thanks,
>
> -Zhengyu
>
More information about the hotspot-dev
mailing list