RFR (S): JDK-8077841: G1: Remove PrintReachable support
Bengt Rutisson
bengt.rutisson at oracle.com
Wed Apr 15 11:59:52 UTC 2015
Hi Stefan,
Thanks for looking at this.
On 2015-04-15 14:04, Stefan Karlsson wrote:
>
>
> On 2015-04-15 13:52, Bengt Rutisson wrote:
>>
>> Hi everyone,
>>
>> Could I have a couple of reviews for this change to remove the
>> PrintReachable support in G1?
>>
>> https://bugs.openjdk.java.net/browse/JDK-8077841
>> http://cr.openjdk.java.net/~brutisso/8077841/webrev.00/
>
> http://cr.openjdk.java.net/~brutisso/8077841/webrev.00/src/share/vm/gc_implementation/g1/g1CollectedHeap.cpp.udiff.html
>
>
> print_extended_on(gclog_or_tty);
> - gclog_or_tty->cr();
> -#ifndef PRODUCT
>
> The removal of cr() affects the product build and should probably not
> be done.
Good point. I'll add that back.
Thanks,
Bengt
>
> Thanks,
> StefanK
>
>>
>> There are three develop flags that allow G1 to dump all objects on
>> the heap to a text format:
>>
>> G1PrintReachableAtInitialMark
>> G1VerifyDuringGCPrintReachable
>> G1PrintReachableBaseFile
>>
>> Turning on this dumping code will dump all objects in all regions to
>> the file specified by G1PrintReachableBaseFile,
>>
>> Several things are disturbing with this code:
>> - The way the code opens a separate file and writes to it is pretty
>> uncommon for our logging code
>> - The flags are develop flags, so not usable in product code - even
>> druring verification
>> - The format the heap is dumped in is not a standard format so there
>> is no tool support for it
>> - There is already standard ways of dumping the heap, using hprof for
>> example.
>> - It is pretty unclear when the information logged by PrintReachable
>> is useful.
>>
>>
>> I've asked around a bit and as far as I can tell no one has been
>> using this feature for a long time. So, rather than trying to fix the
>> code I propose that we remove it.
>>
>> Thanks,
>> Bengt
>
More information about the hotspot-gc-dev
mailing list