chore: import upstream snapshot with attribution

source/backend/cpu/CPULinSpace.cpp

@@ -0,0 +1,54 @@
+//
+//  CPULinSpace.cpp
+//  MNN
+//
+//  Created by MNN on 2019/12/11.
+//  Copyright © 2018, Alibaba Group Holding Limited
+//
+
+#include "backend/cpu/CPULinSpace.hpp"
+#include "backend/cpu/CPUBackend.hpp"
+
+namespace MNN {
+ErrorCode CPULinSpace::onExecute(const std::vector<Tensor*>& inputs, const std::vector<Tensor*>& outputs) {
+    MNN_ASSERT(inputs.size() == 3);
+    MNN_ASSERT(outputs.size() == 1);
+    const float start = inputs[0]->host<float>()[0];
+    const float stop = inputs[1]->host<float>()[0];
+    const int num = inputs[2]->host<int32_t>()[0];
+    MNN_ASSERT(num > 0);
+
+    float* outputData = outputs[0]->host<float>();
+
+    if (num == 1) {
+        outputData[0] = start;
+        return NO_ERROR;
+    }
+
+    if (num == 2) {
+        outputData[0] = start;
+        outputData[1] = stop;
+        return NO_ERROR;
+    }
+    
+    // make sure that start with the first and end with the last.
+    outputData[0] = start;
+    outputData[num - 1] = stop;
+    const float step = (stop - start) / (num - 1);
+    for (int i = 1; i < num - 1; ++i) {
+        outputData[i] = start + i * step;
+    }
+
+    return NO_ERROR;
+}
+
+class CPULinSpaceCreator : public CPUBackend::Creator {
+public:
+    virtual Execution* onCreate(const std::vector<Tensor*>& inputs, const std::vector<Tensor*>& outputs,
+                                const MNN::Op* op, Backend* backend) const {
+        return new CPULinSpace(backend);
+    }
+};
+
+REGISTER_CPU_OP_CREATOR(CPULinSpaceCreator, OpType_LinSpace);
+} // namespace MNN