I can circumvent this issue by linking to libgmp instead of libbf (license is not an issue), but I wonder if this a problem at all, since a bit below, the text says “HINT: this may be a false positive if your program uses some custom stack unwind mechanism, swapcontext or vfork”.
Unfortunately, I cannot provide anything more than reporting this issue. Without ASAN, the bf_trig_alloc test runs fine, and one can query plunit_bigint:exponential_div(3,16000,R) without obvious problems.
This reproduces. The fact that it does not occur with GMP is probably just luck. The same, apparently broken, code is used both for GMP and LibBF, using a small stack memory buffer so we can avoid trashing malloc() for the common rather small bigint operations.