RFR (S/M): 8155094: Add logging for long lasting methods found in JDK-8152948

Kim Barrett kim.barrett at oracle.com
Sun Feb 26 08:03:30 UTC 2017


> On Feb 23, 2017, at 6:06 AM, Thomas Schatzl <thomas.schatzl at oracle.com> wrote:
> 
> Hi all,
> 
>   can I have reviews for this change that adds straightforward logging
> for some phases that were found to sometimes take a significant amount
> of time to allow us to identify them more easily in the future?
> 
> It's basic adding of the necessary code - I agree that we should make
> the logging more convenient to use - but I would like to do this kind
> of refactoring in another change.
> 
> CR:
> https://bugs.openjdk.java.net/browse/JDK-8155094
> Webrev:
> http://cr.openjdk.java.net/~tschatzl/8155094/webrev/
> Testing:
> jprt (updated test case), tier 2+3 gc tests
> 
> Thanks,
>   Thomas

------------------------------------------------------------------------------ 
src/share/vm/gc/g1/g1GCPhaseTimes.hpp
 103   double _cur_dpt_update_time_ms;
 190   void record_dpt_update_time(double ms) {

What is dpt?  Oh, DerivedPointerTable.  The abbreviation is not
helpful; please expand.

------------------------------------------------------------------------------ 
src/share/vm/gc/g1/g1CollectedHeap.cpp
3004 void G1CollectedHeap::start_new_collection_set() {
...
3009   guarantee(_eden.length() == 0, "eden should have been cleared");
3010   g1_policy()->transfer_survivors_to_cset(survivor());

The call site for this new function at 3215 had those two lines, but
the call site at 1382 did not.  That change seem suspicious.

------------------------------------------------------------------------------




More information about the hotspot-gc-dev mailing list