To: None <email@example.com>
From: Frank van der Linden <firstname.lastname@example.org>
Date: 12/23/1994 23:07:38
After today's update, there seems to be a bug in strtod. The first
symptoms were w(1) dumping core and top doing the same thing, later
cc looped endlessly while recompiling strtod.c itself (presumably
while writing the assembly output file containing a double constant
value) This all happened after the new libc had been installed, of course.
I tried tracking the problem down for a few minutes, and saw that strtod.c
had been cleaned up with regard to endian differences and word sizes.
A diff with the old version did not reveal much; it all seemed like
straightforward substitutions to me.
__dtoa() in strtod.c gets a value awfully wrong somewhere, resulting
in an attempt to allocate a huge amount of Bigint structs -> NULL
pointer b/c of failure -> coredump.
Oh, this is on an i386 system (could be that systems with a different
byteorder are not seeing this problem).