[PATCH v3] malloc: send freed small chunks to smallbin

[DJ Delorie]

The patch looks mechanically OK and the logic is sound, but I wonder if
this patch affects the RSS of running applications any.  In theory, it
should reduce RSS by giving more time to coalesce chunks and thus
possibly either reuse those or shrink the top chunk.  But, RSS logic in
real world applications is complicated.  Have you run any such
benchmarks or tests?