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
