It looks like gcj is built with thread support, but the gc is built without.
I forgot about this particular failure mode. Basically it's crashing because the allocator finds itself running on a different stack from the one it's expecting. It's not surprising that that would happen in this configuration.
This build looks extremely unhealthy, in spite of the mixed test results. The tests pass only because they're mostly single-threaded and don't garbage-collect.
Ok, then the result should be, as read on gc-mail list, give OSX the
thread support?