Fix division truncation in window size calculation for small dtypes in average_pool (#18014)

* Update pooling.h

* Update test_te_create_primfunc.py

* fix lint error

Author: jikechao

include/tvm/topi/nn/pooling.h

@@ -383,7 +383,7 @@ inline Tensor adaptive_pool_impl(const Tensor& x, const Array<PrimExpr>& output_
 
           PrimExpr divide_factor = tvm::cast(x->dtype, 1);
           for (size_t i = 0; i < n_dim; ++i) {
-            divide_factor *= tvm::cast(x->dtype, reduce_axes[i]->dom->extent);
+            divide_factor *= tvm::cast(DataType::Int(32), reduce_axes[i]->dom->extent);
           }
 
           return div(pool_sum(indices), divide_factor);

tests/python/te/test_te_create_primfunc.py

@@ -882,6 +882,14 @@ def te_workload():
     _check_workload(te_workload, tir_workload)
 
 
+def test_global_pool():
+    # fix the issue-17938
+    data = te.placeholder((1, 1, 32, 32), dtype="int8", name="data")
+    op_output = topi.nn.global_pool(data=data, pool_type="avg", layout="NCHW")
+    f = te.create_prim_func([data, op_output])
+    assert f
+
+
 def test_nested_reduce_domain_dependency():
     @T.prim_func
     def tir_workload(