This morning's FreeBSD-32 trunk regression finished suspiciously quickly, though it appeared at first glance to have run normally. (I.e., didn't bomb immediately like when bjam can't find a file.)

Sure enough: there's a big gap in matrix, as though a bunch of tests were skipped.

If anyone has a burning desire to troubleshoot, it, I threw bjam.log up on dropbox.  I poked around it a little but nothing jumped out at me. (I didn't scour the entire 21MB.)