Comment
Comment by lgg
There is a bug here... Clearly the author intended to cache the value of nextmalloc to avoid calling dlsym() on every malloc. The correct code should be:
Somehow the fact that the optimization is incorrectly missed here feels appropriate ;-)