Internal compiler error in `float_signal'

[Herbert Xu]

reassign 33786 gcc
quit

On Sat, Jul 17, 1999 at 08:12:14AM -0300, Alexandre Oliva wrote:
> 
> This bug occurs in an `idiv' instruction, dividing 0x80000000 by -1,
> just like in the following assembly code:

Firstly the bit of assembly that you showed didn't do what you meant.
This is closer to the division you had in mind

.text
        .align 4
.globl main
        .type    main,@function
main:
        movl $-1,%edx
        movl $-2147483648,%eax
        idiv %edx
        ret

And the CPU has every right to generate a fault because the quotient
2147483648 is too big for %eax.  It is clearly documented that all
divide overflows cause divide exceptions rather than setting the
overflow bit.

So there is no difference between this case and the compiler trying to
divide by zero.  It should be fixed in egcs.
-- 
Debian GNU/Linux 2.1 is out! ( http://www.debian.org/ )
Email:  Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt