While this bug has not been present in any release version, the latest commit to git-master has fixed a buffer overflow in floating point printf introduced in this release cycle: https://git.musl-libc.org/cgit/musl/commit/?id=0ccaf0572e9cccda2cced0f7ee659af4c1c6679a
Only archs with IEEE-quad long double seem to be affected, and only when using the %Le, %Lf, or %Lg format specifier with particular combinations of exponent and mantissa value.
If you are using musl from git master, it's recommended either to upgrade to latest or apply the patch from the above commit.