fix(frontend): 填充边信息时不能接受输入边不存在

Signed-off-by: YdrMaster <ydrml@hotmail.com>
InfiniTensor · Jan 31, 2024 · 27a8ad6 · 27a8ad6
1 parent 35dc6c8
commit 27a8ad6
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 22 deletions.
diff --git a/src/00common/include/common/error_handler.h b/src/00common/include/common/error_handler.h
@@ -30,9 +30,9 @@ namespace refactor {
     std::abort()
 
 #ifndef DISABLE_ASSERT
-#define ASSERT(CONDITION, F, ...)                                                      \
-    {                                                                                  \
-        if (!(CONDITION)) RUNTIME_ERROR(fmt::format("Assertion: " #F, ##__VA_ARGS__)); \
+#define ASSERT(CONDITION, F, ...)                                                     \
+    {                                                                                 \
+        if (!(CONDITION)) RUNTIME_ERROR(fmt::format("Assertion: " F, ##__VA_ARGS__)); \
     }
 #else
 #define ASSERT(CONDITION, F)

diff --git a/src/06frontend/src/graph.cc b/src/06frontend/src/graph.cc
@@ -98,27 +98,21 @@ namespace refactor::frontend {
         auto const startTime = high_resolution_clock::now();
         // 拓扑遍历
         for (auto [nodeIdx, inputs, outputs] : _internal.topology) {
-            auto unknownEdge = false, inputChanged = false;
-            for (auto i : inputs) {
-                auto const &input = _internal.edges[i].tensor;
-                if (!input) {// 有入边未知
-                    unknownEdge = true;
-                    break;
-                }
-                auto checked = edgeChanged[2 * i];    // NOTICE `std::vector<bool>::operator[]` 产生常引用！！！
-                auto changed = edgeChanged[2 * i + 1];// NOTICE `std::vector<bool>::operator[]` 产生常引用！！！
+            auto inputChanged = false;
+            for (auto i : range0_(inputs.size())) {
+                auto j = inputs[i];
+                auto const &input = _internal.edges[j].tensor;
+                ASSERT(input, "The {}th input of \"{}\" is nullptr", i, _internal.nodes[nodeIdx].name);
+                auto checked = edgeChanged[2 * j];    // NOTICE `std::vector<bool>::operator[]` 产生常引用！！！
+                auto changed = edgeChanged[2 * j + 1];// NOTICE `std::vector<bool>::operator[]` 产生常引用！！！
                 if (!checked) {
                     checked = true;
-                    if (changed = _edgeSnapshot[i] != *input) {
-                        _edgeSnapshot[i] = input->snapshot();
+                    if (changed = _edgeSnapshot[j] != *input) {
+                        _edgeSnapshot[j] = input->snapshot();
                     }
                 }
                 inputChanged |= changed;
             }
-            // 有入边未知，跳过节点
-            if (unknownEdge) {
-                continue;
-            }
             if (!inputChanged && std::all_of(outputs.begin(), outputs.end(),
                                              [this](auto i) { return _internal.edges[i].tensor; })) {
                 // 入边未发生变化，且出边已推导

diff --git a/src/07onnx/src/operators/split.cc b/src/07onnx/src/operators/split.cc
@@ -12,8 +12,8 @@ namespace refactor::onnx {
           numOutputs(numOutputs_) {}
 
     auto Op::build(ModelContext const &, std::string_view, Attributes attributes) -> OpBox {
-        auto axis = attributes.getOrInsert( "axis", {0}).int_();
-        auto numOutputs = attributes.getOrInsert( "num_outputs", {0}).int_();
+        auto axis = attributes.getOrInsert("axis", {0}).int_();
+        auto numOutputs = attributes.getOrInsert("num_outputs", {0}).int_();
         return OpBox(std::make_unique<Op>(axis, numOutputs));
     }
     auto Op::typeId() -> size_t {
@@ -45,7 +45,7 @@ namespace refactor::onnx {
                     ans[i] = Tensor::share(input.dataType, input.shape, dependencies);
                     ans[i]->shape[axis_] = DimExpr(each);
                 } else {
-                    ASSERT(i == numOutputs - 1, ERROR_MSG("Split error"));
+                    ASSERT(i == numOutputs - 1, "Split error");
                     ans[i] = Tensor::share(input.dataType, input.shape, dependencies);
                     ans[i]->shape[axis_] = DimExpr(total);
                 }

diff --git a/src/07onnx/src/operators/tile.cc b/src/07onnx/src/operators/tile.cc
@@ -29,7 +29,7 @@ namespace refactor::onnx {
             return Err(InferError(ERROR_MSG("repeats not support")));
         }
         EXPECT_VAL(repeats.shape[0], repeatsSize)
-        ASSERT(repeatsSize == rank, ERROR_MSG("repeats size error"));
+        ASSERT(repeatsSize == rank, "repeats size error");
 
         auto repeats_ = repeats.data->get<int64_t>();
         Shape output(rank, DimExpr(1));