This is the mail archive of the
java-discuss@sourceware.cygnus.com
mailing list for the Java project.
RE: GC statistics (was Re: big project ported)
- To: "Boehm, Hans" <hboehm at exch dot hpl dot hp dot com>
- Subject: RE: GC statistics (was Re: big project ported)
- From: Tom Tromey <tromey at cygnus dot com>
- Date: Thu, 28 Oct 1999 15:40:38 -0700
- Cc: "'Jonathan P. Olson'" <olson at mmsi dot com>, Godmar Back <gback at cs dot utah dot edu>, krab at daimi dot au dot dk, bryce at albatross dot co dot nz, jsturm at sigma6 dot com, java-discuss at sourceware dot cygnus dot com
- References: <34E36C05935CD311AE5000A0C9B6B0BF011A150E@hplex3.hpl.hp.com>
>>>>> "Hans" == Boehm, Hans <hboehm@exch.hpl.hp.com> writes:
Hans> Our collector does currently clear objects before allocation if
Hans> they are not known to be completely pointer-free to the
Hans> collector, and I believe they never are for gcj. My guess is
Hans> that's a mistake for gcj, and the collector should be fixed to
Hans> accomodate gcj better.
We currently rely on the zeroing behavior. We do this to properly
initialize some objects, eg arrays of references. We used to have an
explicit memzero, but we removed it (as it was redundant) once we
discovered that the collector could do it for us. (See _Jv_InitGC,
where we set "ok_init = TRUE" on each object kind we allocate.)
Tom