It would be useful to have a swap usage information at the moment of crash with native OOME. With physical memory usage from hs_err file this info will let us distinguish true vm bugs of situations with overloaded hosts.

The easiest way to implement it is to add -XX:onError="collectStatistics".