Code review request: 2 Native Memory Tracking bugs
zhengyu.gu at oracle.com
Wed Aug 1 13:20:14 PDT 2012
On 8/1/2012 4:01 PM, Karen Kinnear wrote:
> Thank you also for putting in the additional debug information to try
> to get more information on the duplicate pointer assertion which is so
> hard to catch.
> Couple of questions:
> 1. MemTrackWorker.cpp
> line 134 transites -> transits
> When you call shutdown, do you also want to break out of the loop?
> (which would change the assertion at the end)
> Or is there any potential hard in continuing to loop?
shutdown sets shutdown pending flag, which will break out the loop when
check loop condition (!shutdown_in_progress())
> 2. MemTracker.hpp
> Might it make sense to put the shutdown check in ceate_memory_record
> so you don't need to check all callers?
> But wait - it already is in create_memory_record ... so did you need
> to add it in those cases?
right, it is not needed.
> On Aug 1, 2012, at 1:39 PM, Zhengyu Gu wrote:
>> These are two simple fixes for NMT.
>> CR: 7185614 NMT ON: "check by caller" assertion failed on nsk
>> ThreadMXBean test
>> The assertion was incorrect, since there can be a race condition
>> between caller's check and assertion. The assertion is now removed,
>> shutdown in progress is checked instead.
>> CR: 7187429 NMT ON: Merge failure should cause NMT to shutdown
>> Out-of-memory is the only cause to merge failure, to be consistent,
>> NMT should shutdown itself.
>> Webrev: http://cr.openjdk.java.net/~zgu/7187429/webrev.00/
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the hotspot-runtime-dev