[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

Reviewed By: zertosh

Differential Revision: D33938055

fbshipit-source-id: 6c0643a18f09854e87e183341f252c66dd6395a6
(cherry picked from commit fd183aedbc0f015bd43a01a28930093ab94ab41e)
diff --git a/test/cpp/jit/test_misc.cpp b/test/cpp/jit/test_misc.cpp
index 641333a..e6a6c08 100644
--- a/test/cpp/jit/test_misc.cpp
+++ b/test/cpp/jit/test_misc.cpp
@@ -2865,9 +2865,13 @@
   auto ref1 = a * (a * b);
   auto ref2 = a * ref1;
   WithCPUFuser g(true);
-   bool fusable_on_device = torch::jit::tensorexpr::getTEMustUseLLVMOnCPU();
+  bool fusable_on_device = torch::jit::tensorexpr::getTEMustUseLLVMOnCPU();
   torch::jit::tensorexpr::getTEMustUseLLVMOnCPU() = false;
-  FuseTensorExprs(graph, /*min_group_size*/2, /*add_composed_op*/true, /*fuse_to_dynamic_shapes*/true);
+  FuseTensorExprs(
+      graph,
+      /*min_group_size*/ 2,
+      /*add_composed_op*/ true,
+      /*fuse_to_dynamic_shapes*/ true);
   Code code(graph, "");
   InterpreterState interpreter{code};
   std::vector<IValue> stack = {a, b};
diff --git a/torch/csrc/jit/passes/symbolic_shape_runtime_fusion.cpp b/torch/csrc/jit/passes/symbolic_shape_runtime_fusion.cpp
index ca55a0a..602f951 100644
--- a/torch/csrc/jit/passes/symbolic_shape_runtime_fusion.cpp
+++ b/torch/csrc/jit/passes/symbolic_shape_runtime_fusion.cpp
@@ -164,7 +164,8 @@
   // otherwise we defer to contiguous
   // TODO: channels last 3d
   // NNC Channels last permutation for outputs causes slowdown, disable
-  if (c10::is_channels_last_strides_2d(sizes, strides) && !tt.device()->is_cpu()) {
+  if (c10::is_channels_last_strides_2d(sizes, strides) &&
+      !tt.device()->is_cpu()) {
     return StrideInput::TENSOR_CONT_CHANNELS_LAST;
   }
   return StrideInput::TENSOR_CONT;
diff --git a/torch/csrc/jit/passes/tensorexpr_fuser.cpp b/torch/csrc/jit/passes/tensorexpr_fuser.cpp
index 3195ddd..3335104 100644
--- a/torch/csrc/jit/passes/tensorexpr_fuser.cpp
+++ b/torch/csrc/jit/passes/tensorexpr_fuser.cpp
@@ -1047,7 +1047,8 @@
       // Allow only if the node has a shape function defined.
       // ListConstruct node is an exception since that is needed to fuse
       // aten::cat, though it does not have a shape function.
-      REQ(node->kind() == prim::ListConstruct || node->kind() == prim::TensorExprGroup ||
+      REQ(node->kind() == prim::ListConstruct ||
+          node->kind() == prim::TensorExprGroup ||
           (node->maybeSchema() && shapeComputeGraphForSchema(node->schema())));
     }
 
@@ -1234,13 +1235,15 @@
   }
 
   if (add_composed_op) {
-    TORCH_INTERNAL_ASSERT(fuse_to_dynamic_shapes, "Fusing static shapes with composed op NYI");
+    TORCH_INTERNAL_ASSERT(
+        fuse_to_dynamic_shapes, "Fusing static shapes with composed op NYI");
   }
 
   // Get rid of dead code so that we don't waste effort fusing it.
   EliminateDeadCode(graph);
 
-  TensorExprFuser fuser(graph, min_group_size, add_composed_op, fuse_to_dynamic_shapes);
+  TensorExprFuser fuser(
+      graph, min_group_size, add_composed_op, fuse_to_dynamic_shapes);
   fuser.run();
 
   EliminateCommonSubexpression(graph);
@@ -1250,7 +1253,8 @@
 }
 
 Operation createTensorExprOp(const Node* node) {
-  bool dynamic_shape_fusion_node = node->hasAttribute(attr::striding_inputs_desc);
+  bool dynamic_shape_fusion_node =
+      node->hasAttribute(attr::striding_inputs_desc);
   if (!dynamic_shape_fusion_node) {
     auto kernel =
         std::make_shared<tensorexpr::TensorExprKernel>(node->g(attr::Subgraph));
diff --git a/torch/csrc/jit/python/init.cpp b/torch/csrc/jit/python/init.cpp
index aefb164..d00c56b 100644
--- a/torch/csrc/jit/python/init.cpp
+++ b/torch/csrc/jit/python/init.cpp
@@ -117,13 +117,13 @@
 #include <pybind11/iostream.h>
 #include <pybind11/operators.h>
 
+#include <torch/csrc/jit/runtime/profiling_graph_executor_impl.h>
 #include <memory>
 #include <sstream>
 #include <stdexcept>
 #include <string>
 #include <tuple>
 #include <utility>
-#include <torch/csrc/jit/runtime/profiling_graph_executor_impl.h>
 
 namespace torch {
 namespace jit {
@@ -753,28 +753,38 @@
       .def(
           "_jit_set_bailout_depth",
           [](size_t depth) {
-            TORCH_WARN("Use _jit_set_fusion_strategy, bailout depth is deprecated. Setting to (STATIC, ", depth, ")");
+            TORCH_WARN(
+                "Use _jit_set_fusion_strategy, bailout depth is deprecated. Setting to (STATIC, ",
+                depth,
+                ")");
             size_t old_depth = getBailoutDepth();
             FusionStrategy strat = {{FusionBehavior::STATIC, depth}};
             setFusionStrategy(strat);
             return old_depth;
           })
-      .def("_jit_set_fusion_strategy",
+      .def(
+          "_jit_set_fusion_strategy",
           [](std::vector<std::pair<std::string, size_t>> strategy) {
             FusionStrategy vec_conv;
-            for (const auto& pair: strategy) {
+            for (const auto& pair : strategy) {
               if (pair.first == "STATIC") {
                 vec_conv.emplace_back(FusionBehavior::STATIC, pair.second);
               } else if (pair.first == "DYNAMIC") {
                 vec_conv.emplace_back(FusionBehavior::DYNAMIC, pair.second);
-              } else  {
-                TORCH_INTERNAL_ASSERT("FusionBehavior only supported 'STATIC' or 'DYNAMIC', got: ", pair.first);
+              } else {
+                TORCH_INTERNAL_ASSERT(
+                    "FusionBehavior only supported 'STATIC' or 'DYNAMIC', got: ",
+                    pair.first);
               }
             }
             auto old_strategy = getFusionStrategy();
-            auto strat = fmap(old_strategy, [](std::pair<FusionBehavior, size_t> behav) {
-              return std::pair<std::string, size_t>(behav.first == FusionBehavior::STATIC ? "STATIC" : "DYNAMIC", behav.second);
-            });
+            auto strat =
+                fmap(old_strategy, [](std::pair<FusionBehavior, size_t> behav) {
+                  return std::pair<std::string, size_t>(
+                      behav.first == FusionBehavior::STATIC ? "STATIC"
+                                                            : "DYNAMIC",
+                      behav.second);
+                });
             setFusionStrategy(vec_conv);
             return strat;
           })
diff --git a/torch/csrc/jit/runtime/graph_executor.cpp b/torch/csrc/jit/runtime/graph_executor.cpp
index b24fa49..09f6f7e 100644
--- a/torch/csrc/jit/runtime/graph_executor.cpp
+++ b/torch/csrc/jit/runtime/graph_executor.cpp
@@ -894,7 +894,7 @@
     if (tensorExprFuserEnabled()) {
       auto min_size = getFusionGroupInlining() ? 2 : 1;
       auto dyn_shapes = tensorExprDynamicShapeFusionEnabled();
-      FuseTensorExprs(graph, min_size, /*composed_op*/false, dyn_shapes);
+      FuseTensorExprs(graph, min_size, /*composed_op*/ false, dyn_shapes);
     }
   } else {
     FuseGraph(graph, strict_fuser_check);
diff --git a/torch/csrc/jit/runtime/profiling_graph_executor_impl.cpp b/torch/csrc/jit/runtime/profiling_graph_executor_impl.cpp
index 3b7752e..c0fc02e 100644
--- a/torch/csrc/jit/runtime/profiling_graph_executor_impl.cpp
+++ b/torch/csrc/jit/runtime/profiling_graph_executor_impl.cpp
@@ -74,7 +74,9 @@
 #ifdef FBCODE_CAFFE2
 static FusionStrategy fusion_strategy = {{FusionBehavior::STATIC, 20}};
 #else
-static FusionStrategy fusion_strategy = {{FusionBehavior::STATIC, 2}, {FusionBehavior::DYNAMIC, 10}};
+static FusionStrategy fusion_strategy = {
+    {FusionBehavior::STATIC, 2},
+    {FusionBehavior::DYNAMIC, 10}};
 #endif
 
 FusionStrategy getFusionStrategy() {
@@ -113,7 +115,7 @@
 size_t getBailoutDepth() {
   // Initialize bailout_depth from command-line flag.
   size_t depth = 0;
-  for (const auto& pair: getFusionStrategy()) {
+  for (const auto& pair : getFusionStrategy()) {
     depth += pair.second;
   }
   return depth;
@@ -375,7 +377,7 @@
 FusionBehavior getCurrentBehavior(size_t remaining_depth) {
   size_t curr_depth = 0;
   auto curr_strategy = getFusionStrategy();
-  for (int i = static_cast<int>(curr_strategy.size()) -1; i >= 0; i--) {
+  for (int i = static_cast<int>(curr_strategy.size()) - 1; i >= 0; i--) {
     curr_depth += curr_strategy[i].second;
     if (remaining_depth <= curr_depth) {
       return curr_strategy[i].first;
@@ -386,7 +388,9 @@
   return FusionBehavior::STATIC;
 }
 
-void runNoGradOptimizations(std::shared_ptr<Graph>& graph, size_t remaining_bailout_depth) {
+void runNoGradOptimizations(
+    std::shared_ptr<Graph>& graph,
+    size_t remaining_bailout_depth) {
   GRAPH_DEBUG(
       "After customPostPasses (beginning of runNoGradOptimizations)\n", *graph);
   // runNondiffOptimization
@@ -419,8 +423,9 @@
       BatchMM(graph);
       GRAPH_DEBUG("After BatchMM, before Fusion\n", *graph);
       auto min_size = getFusionGroupInlining() ? 2 : 1;
-      bool dyn_shapes = getCurrentBehavior(remaining_bailout_depth) == FusionBehavior::DYNAMIC;
-      FuseTensorExprs(graph, min_size, /*composed_op*/false, dyn_shapes);
+      bool dyn_shapes = getCurrentBehavior(remaining_bailout_depth) ==
+          FusionBehavior::DYNAMIC;
+      FuseTensorExprs(graph, min_size, /*composed_op*/ false, dyn_shapes);
       GRAPH_DEBUG(
           "After Fusion, before RemoveTensorTypeSpecializations\n", *graph);
 
@@ -448,7 +453,8 @@
 }
 
 void ProfilingGraphExecutorImpl::runProfilingOptimizations(
-    std::shared_ptr<Graph>& copy, size_t remaining_bailout_depth) {
+    std::shared_ptr<Graph>& copy,
+    size_t remaining_bailout_depth) {
   GRAPH_DEBUG("Before runProfilingOptimizations:\n", *copy);
   if (!getGraphExecutorOptimize()) {
     runNooptPassPipeline(copy);
diff --git a/torch/csrc/jit/runtime/profiling_graph_executor_impl.h b/torch/csrc/jit/runtime/profiling_graph_executor_impl.h
index a3c6e0b..af9d1a2 100644
--- a/torch/csrc/jit/runtime/profiling_graph_executor_impl.h
+++ b/torch/csrc/jit/runtime/profiling_graph_executor_impl.h
@@ -26,7 +26,8 @@
 //   [3,4,5] are observed, then it is assumed that future inputs will have
 //   shapes [a,b,c] and [b,c,d] for some values of a,b,c,d.
 //
-//   In both cases, we also recompile on new striding behavior, device, or dtype.
+//   In both cases, we also recompile on new striding behavior, device, or
+//   dtype.
 //
 // Behavior - fallback functions & depth:
 //   When an input doesn't match the format required by the specialized compiled
@@ -56,7 +57,6 @@
 // returns previous strategy
 TORCH_API FusionStrategy setFusionStrategy(FusionStrategy& fusion_strategy);
 
-
 struct TORCH_API ProfilingGraphExecutorImpl : public GraphExecutorImplBase {
   ProfilingGraphExecutorImpl(
       const std::shared_ptr<Graph>& graph,
@@ -75,7 +75,7 @@
     optimized_plan_.reset();
     // prevent memory leaks
     fallback_functions_.clear();
-  remaining_bailout_depth_.reset();
+    remaining_bailout_depth_.reset();
   }
 
   bool isOptimized() const override {
@@ -87,7 +87,9 @@
       Stack& stack,
       size_t remaining_bailout_depth);
   void runProfilingInsensitiveOptimizations(std::shared_ptr<Graph>& graph);
-  void runProfilingOptimizations(std::shared_ptr<Graph>& graph, size_t remaining_depth);
+  void runProfilingOptimizations(
+      std::shared_ptr<Graph>& graph,
+      size_t remaining_depth);
   void replaceFallbackGraphWithFallbackFunction(Block* b);
   std::unique_ptr<ProfilingRecord> pr_;
   c10::optional<ExecutionPlan>
diff --git a/torch/csrc/jit/runtime/static/fusion.cpp b/torch/csrc/jit/runtime/static/fusion.cpp
index b89c658..d9c2837 100644
--- a/torch/csrc/jit/runtime/static/fusion.cpp
+++ b/torch/csrc/jit/runtime/static/fusion.cpp
@@ -334,7 +334,7 @@
       traced_graph,
       /*min_group_size*/ 2,
       /*add_composed_op*/ true,
-      /*fuse_to_dynamic_shapes*/true);
+      /*fuse_to_dynamic_shapes*/ true);
   graph->block()->clear();
   graph->block()->cloneFrom(traced_graph->block(), nullptr);
   GRAPH_DUMP("Graph after fusion: ", graph);
diff --git a/torch/csrc/jit/tensorexpr/cuda_codegen.cpp b/torch/csrc/jit/tensorexpr/cuda_codegen.cpp
index f532514..e398333 100644
--- a/torch/csrc/jit/tensorexpr/cuda_codegen.cpp
+++ b/torch/csrc/jit/tensorexpr/cuda_codegen.cpp
@@ -1160,7 +1160,7 @@
   std::vector<void*> extent_args;
   size_t raw_args_size = raw_args.size();
   extent_args.reserve(raw_args_size);
-  for (size_t i = 0 ; i < raw_args_size; ++i) {
+  for (size_t i = 0; i < raw_args_size; ++i) {
     if (arg_pos_in_extents_[i]) {
       extent_args.push_back(raw_args[i]);
     }
@@ -1177,7 +1177,8 @@
       gpu_thread_extents_v[i] = immediateAs<int64_t>(gpu_thread_extents[i]);
       continue;
     }
-    gpu_thread_extents_v[i] = thread_extents_eval_[i].value<int64_t>(extent_args);
+    gpu_thread_extents_v[i] =
+        thread_extents_eval_[i].value<int64_t>(extent_args);
   }
 
   // Skip launching the kernel if there are no elements to process.