fix segfault in `cat` on CPU with tensors that can't be indexed with 32-bit ints. (#21530)

Summary: Should be self-explanatory. This `int` variable is overflowing. Reported in #21526 Pull Request resolved: https://github.com/pytorch/pytorch/pull/21530 Differential Revision: D15719275 Pulled By: umanwizard fbshipit-source-id: 24e917a00a5b78bc3af29ef3b8b72eea7e89d5d5

fix segfault in `cat` on CPU with tensors that can't be indexed with 32-bit ints. (#21530)
Summary: Should be self-explanatory. This `int` variable is overflowing. Reported in #21526 Pull Request resolved: https://github.com/pytorch/pytorch/pull/21530 Differential Revision: D15719275 Pulled By: umanwizard fbshipit-source-id: 24e917a00a5b78bc3af29ef3b8b72eea7e89d5d5
74828be4 · Brennan Vincent · Facebook Github Bot · 40637465 · 74828be4 · 74828be4
--- a/aten/src/TH/generic/THTensor.cpp
+++ b/aten/src/TH/generic/THTensor.cpp
@@ -799,7 +799,7 @@ void THTensor_(catArray)(THTensor *result, THTensor **inputs, int numInputs, int
        if (!should_skip(inputs[j])) {
          THTensor* input0 = inputs[j];
          scalar_t* input0_data = THStorage_(data)(THTensor_getStoragePtr(input0)) + input0->storage_offset();
-          int local_inner = inner * input0->size(dimension);
+          int64_t local_inner = inner * input0->size(dimension);
          if (local_inner != 0) {
            memcpy(result_data + offset, input0_data + o*local_inner, local_inner*sizeof(scalar_t));
          } // input0_size != 0

--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -4948,6 +4948,16 @@ class _TestTorchMixin(object):
    def test_cat_empty(self):
        self._test_cat_empty(self)

+    @slowTest
+    def test_cat_big(self):
+        SIZE1 = 6500
+        SIZE2 = 4500
+        concat_list = []
+        concat_list.append(torch.ones((SIZE1, 1024 * 512), dtype=torch.uint8))
+        concat_list.append(torch.ones((SIZE2, 1024 * 512), dtype=torch.uint8))
+        result = torch.cat(concat_list)
+        self.assertEqual(result.size(0), SIZE1 + SIZE2)
+
    def test_narrow(self):
        x = torch.Tensor([[0, 1, 2], [3, 4, 5], [6, 7, 8]])
        self.assertEqual(x.narrow(0, 0, 1), torch.Tensor([[0, 1, 2]]))