[Bug optimization/11379] ICE with optimization option -fnew-ra

["spigel at olvs dot miee dot ru" <gcc-bugzilla at gcc dot gnu dot org>]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11379



------- Additional Comments From spigel at olvs dot miee dot ru  2003-07-18 13:02 -------
Subject: Re:  ICE with optimization option -fnew-ra

I can confirm this on 3.3.1 (20030715) with "-O1 -march=athlon-xp -fnew-ra".
It can be the good code for testing new compiler versions.

pinskia at physics dot uc dot edu wrote:

>------- Additional Comments From pinskia at physics dot uc dot edu  2003-07-15 14:42 -------
>Here is a reduce test case (down to 36 lines):
>typedef unsigned char Byte;
>typedef unsigned int uInt;
>typedef uInt uIntf;
>typedef struct inflate_huft_s inflate_huft;
>struct inflate_huft_s
>{
>  Byte Exop;
>  Byte Bits;
>  uInt base;
>};
>int huft_build (uInt n, uInt s, const uIntf * d,
>            const uIntf * e)
>{
>
>  uInt a = n;
>  uInt c[15 + 1];
>  uInt i = n;
>  uInt j = n;
>  uInt k = n;
>  register uIntf *p;
>  inflate_huft *q = 0;
>  struct inflate_huft_s r;
>  p = c;
>  for (; k <= n; k++)
>    { 
>      while (a--)
>        { 
>          r.Exop = (Byte) (e[*p - s] + 16 + 64);
>          r.base = d[*p++ - s];
>          for (j = i >> n; j < n; j += 1 << (k - n))
>            q[j] = r;
>        }
>    }
>  return 0;
>}
>