Just some notes: 0) probably counting number of live maps is the best approach at least when map space is too big (that shouldn't happen often), otherwise it's not clear when map compact should be triggered (but we can introduce some heuristics though); 1) I should probably merge map pointers update and sweeping not to traverse most of the spaces twice; 2) populating free list in MapSpace::AdjustAllocationInfo is a bad idea; 3) there is a test failure I can reproduce even further reducing number of bits in page index field, investigating.
yours, anton. On Wed, Dec 23, 2009 at 10:37 AM, <[email protected]> wrote: > That should have been Søren instead of Kevic C. > > Sorry for noise. > > http://codereview.chromium.org/509035 > -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
