Path: utzoo!utgpu!cs.utexas.edu!usc!sdd.hp.com!zaphod.mps.ohio-state.edu!rpi!uupsi!sunic!zyx!arndt
From: arndt@zyx.ZYX.SE (Arndt Jonasson)
Newsgroups: alt.sys.sun
Subject: cc optimizer bug (arithmetic shifts).
Message-ID: <1991Feb07.132335.13531@ZYX.SE>
Date: 7 Feb 91 13:23:35 GMT
Sender: arndt@ZYX.SE (Arndt Jonasson)
Organization: ZYX Sweden AB, Stockholm, Sweden
Lines: 46


The version of SunOS is 4.1.

-------------------- start of script --------------------
% cat mc.c
#include <stdio.h>

main()
{
   int i;

   for (i=0; i<8; i++)
      test(i<<29);
}

test(j)
int j;
{
   printf("%d\n", j);
}
% cc -o mc mc.c
% mc
0
536870912
1073741824
1610612736
-2147483648
-1610612736
-1073741824
-536870912
% cc -O -o mc mc.c
% mc
0
% uname -srm
SunOS 4.1 sun4c
% 
-------------------- end of script --------------------

What seems to happen is that the optimizer sees a chance to avoid
shifting 'i' all the time, and therefore keeps 'i' shifted left 29
steps, modifying the index arithmetic accordingly. It's basically a
good idea, but it only works if the maximum shifted value of 'i'
doesn't overflow, which it does here.
-- 
Arndt Jonasson, ZYX AB, Styrmansgatan 6, 114 54 Stockholm, Sweden
email address:   arndt@zyx.SE   or      <backbone>!mcsun!sunic!zyx!arndt