[Bug] Missing range check for index parameter in mindspore.ops.scatter.

[Redmept1on]

Environment

Hardware Environment(Ascend/GPU/CPU):

device cpu

Software Environment:

• MindSpore version (source or binary): v2.2.13
• Python version (e.g., Python 3.7.5): 3.8.18
• OS platform and distribution (e.g., Linux Ubuntu 16.04): Linux Ubuntu 18.04

Describe the current behavior

when index as a negative or positive number exceeds the range of size, no error is reported.

Describe the expected behavior

RuntimeError: index 10000 is out of bounds for dimension 1 with size 5

Steps to reproduce the issue

import mindspore as mind
import numpy as np

input = mind.ops.ones((3,5), mind.int64)*2
index = mind.tensor(np.array([[0,-10000,2],[0,1,4]],))
src = mind.tensor(np.array([[0,10,20],[50,60,70]]))
out = mind.ops.scatter(input, 1, index, src)
print(out)

pytorch v2.0.0+cu117

import torch
import numpy as np

input = torch.ones((3,5), dtype=torch.int64)*2
index = torch.tensor(np.array([[0,-10000,2],[0,1,4]],))
src = torch.tensor(np.array([[0,10,20],[50,60,70]]))
out = torch.scatter(input, 1, index, src)
print(out)

Special notes for this issue

when set the number in index at 10000, the bug exists the same.

[Redmept1on]

mindspore.ops.tensor_scatter_elements exist the same bug, when number in indices parameter exceed the range of size, there's no error message.

import mindspore
from mindspore import Tensor, ops
from mindspore import Parameter
import numpy as np
input_x = Parameter(Tensor(np.array([[1, 2, 3, 4, 5]]), mindspore.int32), name="x")
indices = Tensor(np.array([[10000, 4]]), mindspore.int32)
updates = Tensor(np.array([[8, 8]]), mindspore.int32)
axis = 1
reduction = "none"
output = ops.tensor_scatter_elements(input_x, indices, updates, axis, reduction)
print(output)