two 4GB mallocs => SEGV

[Willem Jan Withagen]

Georgi Guninski wrote:

> On Tue, Oct 26, 2004 at 03:51:22PM +0200, Willem Jan Withagen wrote:
> 
>>And again on a 5.2 i386 box:
>>[/tmp] root at freebee> a.out
>>80000000
> 
> 
> actually on i386 linux i get 0 and i don't crash at all.

facinating, but true...
Willest compiling I get a warning of a overflow that 2G gives a wrap in long, 
but things are interpreted unsigned as far as I see.
And that is because I changed your 4G to 2G, 'cause I only had 2Gb of mem and 
5Gb of swap.

Yours just really overflows to 0, I guess. eg. if the optimizer (or the code) 
calculates the expression from right to left, and *4 is done 'by shiftleft 2'
you're left with an all-zeros...

--WjW