https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111454

            Bug ID: 111454
           Summary: ice in get_nonzero_bits
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

Given the following partially reduced C code

(safe_div_func_int16_t_s_s)(short si1, short si2) {
  return ((si2 == 0) || 0) ? : (si1 / si2);
}
short(safe_rshift_func_int16_t_s_s)();
g_3202 = 0x5FL;
void func_1() __attribute__((,
                             visibility0, optimize("-O2"), aligned0));
void func_1() {
  short __trans_tmp_4;
  int __trans_tmp_3;
  int l_4 __attribute__(()) = 0;
  int l_3120 __attribute__(()) = 0L;
  {
    for (0;; ++l_3120) {
      {
        g_3202 = 5;
        for (g_3202; (g_3202 > 0); g_3202--) {
          (l_4) = ((safe_rshift_func_int16_t_s_s()));
          break;
        }
      }
      __trans_tmp_4 = safe_div_func_int16_t_s_s((0 != (4294967292UL)), (l_4));
      __trans_tmp_3 = safe_lshift_func_int16_t_s_s((__trans_tmp_4), 15);
    }
  }
}

when compiled by recent gcc and -O1, does this:

$ /home/dcb38/gcc/results/bin/gcc -w -c -O1 ~/cvise/bug958B.c
gcc: internal compiler error: Segmentation fault signal terminated program cc1
Please submit a full bug report, with preprocessed source (by using
-freport-bug).

Stack backtrace seems to be:

#0  0x00000000010e502a in get_nonzero_bits (name=name@entry=0x7fffe9fed1b0)
    at ../../trunk.year/gcc/tree-ssanames.cc:494
#1  0x0000000000a452dc in tree_nonzero_bits (t=0x7fffe9fed1b0)
    at ../../trunk.year/gcc/fold-const.cc:16792
#2  0x0000000001deb61f in gimple_zero_one_valued_p (t=0x7fffe9fed1b0, 
    valueize=valueize@entry=0x1084e10 <rpo_vn_valueize(tree_node*)>)
    at gimple-match-3.cc:19
#3  0x0000000001deb728 in gimple_zero_one_valued_p (t=0x7fffe9fedf30, 
    valueize=valueize@entry=0x1084e10 <rpo_vn_valueize(tree_node*)>)
    at gimple-match-3.cc:70
#4  0x0000000001deb728 in gimple_zero_one_valued_p (t=0x7fffe9fed1b0, 
    valueize=valueize@entry=0x1084e10 <rpo_vn_valueize(tree_node*)>)
    at gimple-match-3.cc:70

The bug first seems to occur sometime between g:f73d2d61a5926f42
and gb975c0dc3be28565, a range of 130 commits.

Reply via email to