From: ts <decoux@...>
Date: 2004-08-28T22:58:39+09:00
Subject: Re: [PATCH] Subtle bug in bignum.c

>>>>> "M" == Markus  <markus@reality.com> writes:

M> I have found a rather subtle bug in the bitwise operators defined in

 It's really subtle :-)

M> +    RBIGNUM(x)->sign = RBIGNUM(x)->sign?1:0;
M> +    RBIGNUM(y)->sign = RBIGNUM(y)->sign?1:0;
M>      z = bignew(l2, RBIGNUM(x)->sign || RBIGNUM(y)->sign);
M>      zds = BDIGITS(z);

[...]

M> This fixed the problem, at least in so far as the symptoms went away. 
M> It may fix it correctly, if everything else only looks at the lowest bit
M> of RBIGNUM()->sign, or it might be a false step if the real design
M> intent is for RBIGNUM()->sign to always be either 0 or 1.

 it's a GC problem : your patch make the variable `x' (or `y') available
 for the mark phase, without this line `x' can be gc'ed.

 Don't ask me why ruby do this : I've not yet look at the generated
 assembler, and why it can't find the right `x' on the stack.


Guy Decoux