divide-by-zero trap handler

[mohit verma]

hi guys,
i have  read in  a paper that divide-by-zero trap handler is not implemented
via system calls ( on old solaris systems).
but if we talk about the interrupt vectors ,as i know that divede-by-zero is
vectored at first in idt.
so why this conflict ?
or is it the case that in new kernels this is the improved corner of kernel?