[Bug tree-optimization/94266] New: aarch64：ICE during GIMPLE pass: forwprop

[qianchao9 at huawei dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94266

            Bug ID: 94266
           Summary: aarch64：ICE during GIMPLE pass: forwprop
           Product: gcc
           Version: 10.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: qianchao9 at huawei dot com
  Target Milestone: ---
            Target: aarch64

hello
I find an ICE when testing GCC10.0 on aarch64 with -march=armv8.2-a+sve
-msve-vector-bits=256 -ftree-loop-vectorize -O2:

Simple test case
---
#include <stdarg.h>
#define N 16
int out[N];
int a[2*N] =
{0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31};

/* Outer-loop vectorization.  */
__attribute__ ((noinline)) void 
foo () 
{
  int i, j;
  int res;
  for (i = 0; i < N; i++)
    {
      res = 1000;

      for (j = 0; j < N; j++) 
        res = res - a[i+j];

      out[i] = res;
    }
}

int main ()
{
  foo ();
  return 0;
}
---
GCC version:
gcc version 10.0.1 20200309 (experimental) (GCC)

Runcommand:
gcc new_test.c -march=armv8.2-a+sve -msve-vector-bits=256 -ftree-loop-vectorize
-O2 -S -o new_test.s

Result:
during GIMPLE pass: forwprop
pr_xxx.c: In function ‘foo’:
pr_xxx.c:10:1: internal compiler error: in maybe_canonicalize_mem_ref_addr, at
gimple-fold.c:4899
   10 | foo ()
      | ^~~
0x97dd5b maybe_canonicalize_mem_ref_addr
        ../../gcc/gimple-fold.c:4899
0x988d97 fold_stmt_1
        ../../gcc/gimple-fold.c:4972
0x98cfa3 fold_stmt_inplace(gimple_stmt_iterator*)
        ../../gcc/gimple-fold.c:5328
0xde455f forward_propagate_addr_expr_1
        ../../gcc/tree-ssa-forwprop.c:878
0xde6513 forward_propagate_addr_expr
        ../../gcc/tree-ssa-forwprop.c:998
0xdea427 execute
        ../../gcc/tree-ssa-forwprop.c:2715