Merge branch 'develop' into onnx_parse_tests

.clang-tidy

 CheckOptions: 
   - key:             bugprone-unused-return-value.CheckedFunctions
     value:           '::std::async;::std::launder;::std::remove;::std::remove_if;::std::unique;::std::unique_ptr::release;::std::basic_string::empty;::std::vector::empty;::std::find;::std::find_if;::std::find_if_not;::std::all_of;::std::any_of;::std::none_of;::std::count;::std::count_if;::std::mismatch;::std::find_end;::std::find_first_of;::std::adjacent_find;::std::search;::std::search_n;::std::nth_element;::std::lower_bound;::std::upper_bound;::std::binary_search;::std::equal_range;::std::max;::std::max_element;::std::min;::std::min_element;::std::minmax;::std::minmax_element;::std::equal;::std::lexicographical_compare;::std::accumulate;::std::inner_product'
+  - key:             cppcoreguidelines-macro-usage.AllowedRegexp
+    value:           'DEBUG|FALLTHROUGH|_THROW|_REQUIRES|_DECLARE_|_VISIT_|_GENERATE_|_DETAIL_|_MANAGE_PTR|_MATCHER|DEVICE_SHARED'
   - key:             modernize-loop-convert.MinConfidence
     value:           risky   
   - key:             modernize-loop-convert.NamingStyle
     value:           lower_case
+  - key:             performance-unnecessary-copy-initialization.AllowedTypes
+    value:           'shape'
+  - key:             performance-unnecessary-value-param.AllowedTypes
+    value:           'shape'
   - key:             readability-function-size.BranchThreshold
     value:           '15'
   - key:             readability-function-size.LineThreshold

CMakeLists.txt

@@ -52,6 +52,7 @@ rocm_enable_clang_tidy(
         -clang-analyzer-optin.performance.Padding
         -clang-diagnostic-deprecated-declarations
         -clang-diagnostic-extern-c-compat
+        -clang-diagnostic-disabled-macro-expansion
         -clang-diagnostic-unused-command-line-argument
         -cppcoreguidelines-pro-bounds-array-to-pointer-decay
         -cppcoreguidelines-pro-bounds-constant-array-index
@@ -70,13 +71,12 @@ rocm_enable_clang_tidy(
         -hicpp-explicit-conversions
         -hicpp-no-array-decay
         -hicpp-special-member-functions
+        -hicpp-uppercase-literal-suffix
         -hicpp-use-override
         # This check is broken
-        -hicpp-use-auto
         -llvm-header-guard
         -llvm-include-order
         -misc-macro-parentheses
-        -modernize-use-auto
         -modernize-use-override
         -modernize-pass-by-value
         -modernize-use-default-member-init
@@ -84,7 +84,12 @@ rocm_enable_clang_tidy(
         -readability-braces-around-statements
         -readability-else-after-return
         -readability-named-parameter
+        -readability-uppercase-literal-suffix,
+        -*-avoid-c-arrays
         -*-explicit-constructor
+        -*-magic-numbers
+        -*-non-private-member-variables-in-classes
+        -*-use-auto
         -*-use-emplace
         -*-use-equals-default
     ERRORS
@@ -94,6 +99,7 @@ rocm_enable_clang_tidy(
         ".*hpp"
     EXTRA_ARGS
         -DMIGRAPHX_USE_CLANG_TIDY
+        "-Dmain\\\\(...\\\\)=main\\\\(__VA_ARGS__\\\\) // NOLINT"
 
 )
 include(ROCMCppCheck)
@@ -124,6 +130,7 @@ rocm_enable_cppcheck(
         ${CMAKE_CURRENT_SOURCE_DIR}/src/include
         ${CMAKE_CURRENT_SOURCE_DIR}/src/targets/cpu/include
         ${CMAKE_CURRENT_SOURCE_DIR}/src/targets/miopen/include
+        ${CMAKE_CURRENT_SOURCE_DIR}/test/include
     DEFINE
         CPPCHECK=1
 )

Dockerfile

@@ -50,7 +50,7 @@ RUN pip install cget
 RUN pip install https://github.com/pfultz2/rclone/archive/master.tar.gz
 
 # Install hcc
-RUN rclone -b sanitizer1 https://github.com/RadeonOpenCompute/hcc.git /hcc
+RUN rclone -b roc-2.0.x -c 757fb492517b80e7c86338af5fc1a43d63cb25a9 https://github.com/RadeonOpenCompute/hcc.git /hcc
 RUN cget -p $PREFIX install hcc,/hcc
 
 # Use hcc

Jenkinsfile

@@ -8,7 +8,7 @@ def rocmtestnode(variant, name, body) {
             mkdir build
             cd build
             CXX=${compiler} CXXFLAGS='-Werror -Wno-fallback' cmake ${flags} .. 
-            CTEST_PARALLEL_LEVEL=32 make -j32 all doc check
+            CTEST_PARALLEL_LEVEL=32 make -j32 generate all doc check
         """
         echo cmd
         sh cmd
@@ -73,7 +73,7 @@ rocmtest tidy: rocmnode('rocmtest') { cmake_build ->
             rm -rf build
             mkdir build
             cd build
-            CXX='clang++-5.0' cmake .. 
+            CXX=hcc cmake .. 
             make -j8 -k analyze
         '''
     }

cppcheck.rules

@@ -74,7 +74,7 @@
     </message>
 </rule>
 <rule>
-    <pattern>(fclose|free|hipFree) \(</pattern>
+    <pattern>(fclose|free|hipFree|hipHostFree|hipFreeArray|hipMemFree|hipStreamDestroy|hipEventDestroy|hipArrayDestroy|hipCtxDestroy|hipDestroyTextureObject|hipDestroySurfaceObject) \(</pattern>
     <message>
         <id>useManagePointer</id>
         <severity>style</severity>

dev-requirements.txt

 pfultz2/rocm-recipes
 pcre
 danmar/cppcheck@f965e5873 -DHAVE_RULES=1
-ROCm-Developer-Tools/HIP@3c7f5dbce24802ec4237e615038daff2909a2e8e
-# python/cpython@v3.6.6 -X autotools -H sha256:92aa914572c695c0aeb01b0a214813f414da4b51a371234df514a74761f2bb36
+ROCm-Developer-Tools/HIP@fc22ef991ce7cb15821c8ccb4f03cdfc3e7e43cf
+python/cpython@v3.6.6 -X autotools -H sha256:92aa914572c695c0aeb01b0a214813f414da4b51a371234df514a74761f2bb36
 -f requirements.txt

requirements.txt

 google/protobuf -DCMAKE_POSITION_INDEPENDENT_CODE=On
-RadeonOpenCompute/rocm-cmake@6240bb3 --build
+RadeonOpenCompute/rocm-cmake@ac45c6e --build
+ROCmSoftwarePlatform/rocBLAS@30a992ae02fda568688bcd190edd5e277d6674d9
 ROCmSoftwarePlatform/MIOpen@1.6.0
-ROCmSoftwarePlatform/rocBLAS@v14.3.0
 blaze,https://bitbucket.org/blaze-lib/blaze/get/f0755dea0e03.tar.gz -X header -DHEADER_DIR=blaze
 half,https://github.com/pfultz2/half/archive/1.12.0.tar.gz -X header -H sha256:0a08660b68abb176ebc2a0cdf8de46e3182a7f46c66443bb80dbfaaec98cf969

src/constant_propagate.cpp

@@ -13,7 +13,7 @@ struct match_const_add
         return match::name("add")(match::args(match::name("@literal"), match::name("@literal")));
     }
 
-    void apply(program& p, match::matcher_result r) const
+    void apply(program& p, const match::matcher_result& r) const
     {
         auto ins  = r.result;
         auto arg1 = ins->inputs().at(0)->get_literal();

src/include/migraphx/concat_opt.hpp

@@ -235,6 +235,7 @@ inline const ValueType& any_cast(const concat_optimization& x)
 }
 
 #endif
+
 } // namespace MIGRAPHX_INLINE_NS
 } // namespace migraphx
 

src/include/migraphx/make_shared_array.hpp

@@ -10,7 +10,7 @@ inline namespace MIGRAPHX_INLINE_NS {
 template <typename T>
 std::shared_ptr<T> make_shared_array(size_t size)
 {
-    return std::shared_ptr<T>(new T[size], std::default_delete<T[]>());
+    return std::shared_ptr<T>(new T[size], std::default_delete<T[]>()); // NOLINT
 }
 
 } // namespace MIGRAPHX_INLINE_NS

src/include/migraphx/operators.hpp

@@ -289,7 +289,7 @@ struct transpose
         }
         std::vector<size_t> output_lens(input_lens.size());
         std::vector<size_t> output_strides(input_lens.size());
-        for(int i = 0; i < output_lens.size(); i++)
+        for(std::size_t i = 0; i < output_lens.size(); i++)
         {
             output_lens[i]    = input_lens[dims[i]];
             output_strides[i] = input_strides[dims[i]];
@@ -326,7 +326,7 @@ struct concat
     std::size_t axis = 0;
     std::string name() const { return "concat"; }
     std::vector<std::size_t> compute_offsets(const shape& output_shape,
-                                             const std::vector<argument> args) const
+                                             const std::vector<argument>& args) const
     {
         std::vector<std::size_t> offsets;
         std::vector<std::size_t> offset(args[0].get_shape().lens().size(), 0);

src/include/migraphx/shape.hpp

@@ -35,22 +35,22 @@ struct shape
     m(uint64_type, uint64_t)
 // clang-format on
 
-#define MIGRAPHX_SHAPE_ENUM_TYPES(x, t) x,
+#define MIGRAPHX_SHAPE_GENERATE_ENUM_TYPES(x, t) x,
     enum type_t
     {
-        MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_ENUM_TYPES)
+        MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_GENERATE_ENUM_TYPES)
     };
-#undef MIGRAPHX_SHAPE_ENUM_TYPES
+#undef MIGRAPHX_SHAPE_GENERATE_ENUM_TYPES
 
     template <class T, class = void>
     struct get_type;
-#define MIGRAPHX_SHAPE_GET_TYPE(x, t)                         \
+#define MIGRAPHX_SHAPE_GENERATE_GET_TYPE(x, t)                \
     template <class T>                                        \
     struct get_type<t, T> : std::integral_constant<type_t, x> \
     {                                                         \
     };
-    MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_GET_TYPE)
-#undef MIGRAPHX_SHAPE_GET_TYPE
+    MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_GENERATE_GET_TYPE)
+#undef MIGRAPHX_SHAPE_GENERATE_GET_TYPE
 
     template <class T>
     struct get_type<const T> : get_type<T>
@@ -148,10 +148,10 @@ struct shape
     {
         switch(this->type())
         {
-#define MIGRAPHX_SHAPE_VISITOR_CASE(x, t) \
+#define MIGRAPHX_SHAPE_GENERATE_VISITOR_CASE(x, t) \
     case x: v(as<t>()); return;
-            MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_VISITOR_CASE)
-#undef MIGRAPHX_SHAPE_VISITOR_CASE
+            MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_GENERATE_VISITOR_CASE)
+#undef MIGRAPHX_SHAPE_GENERATE_VISITOR_CASE
         }
         MIGRAPHX_THROW("Unknown type");
     }

src/include/migraphx/tensor_view.hpp

@@ -164,7 +164,7 @@ bool operator!=(const tensor_view<T>& x, const tensor_view<U>& y)
 }
 
 template <class T>
-tensor_view<T> make_view(shape s, T* data)
+tensor_view<T> make_view(const shape& s, T* data)
 {
     return {s, data};
 }

src/include/migraphx/type_name.hpp

@@ -18,7 +18,7 @@ const std::string& get_type_name()
         name = typeid(PrivateMigraphTypeNameProbe).name();
         name = name.substr(7);
 #else
-        const char parameter_name[] = "PrivateMigraphTypeNameProbe =";
+        const char parameter_name[] = "PrivateMigraphTypeNameProbe ="; // NOLINT
 
         name = __PRETTY_FUNCTION__;
 

src/onnx/mnist.cpp

@@ -14,7 +14,10 @@
 
 auto reverse_int(unsigned int i)
 {
-    unsigned char c1, c2, c3, c4;
+    unsigned char c1;
+    unsigned char c2;
+    unsigned char c3;
+    unsigned char c4;
     c1 = i & 255u;
     c2 = (i >> 8u) & 255u;
     c3 = (i >> 16u) & 255u;
@@ -32,7 +35,9 @@ read_mnist_images(const std::string& full_path, int& number_of_images, int& imag
 
     if(file.is_open())
     {
-        int magic_number = 0, n_rows = 0, n_cols = 0;
+        int magic_number = 0;
+        int n_rows       = 0;
+        int n_cols       = 0;
 
         file.read(reinterpret_cast<char*>(&magic_number), sizeof(magic_number));
         magic_number = reverse_int(magic_number);

src/onnx/verify_onnx.cpp

@@ -116,7 +116,7 @@ void verify_reduced_program(F f, double tolerance = 80)
 {
     migraphx::program p = f();
     auto n              = std::distance(p.begin(), p.end());
-    for(int i = 0; i < n; i++)
+    for(std::size_t i = 0; i < n; i++)
     {
         verify_reduced(f, i, tolerance);
     }

src/shape.cpp

@@ -169,10 +169,10 @@ std::string shape::type_string() const
 {
     switch(this->type())
     {
-#define MIGRAPHX_SHAPE_TYPE_STRING_CASE(x, t) \
+#define MIGRAPHX_SHAPE_GENERATE_TYPE_STRING_CASE(x, t) \
     case x: return #x;
-        MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_TYPE_STRING_CASE)
-#undef MIGRAPHX_SHAPE_TYPE_STRING_CASE
+        MIGRAPHX_SHAPE_VISIT_TYPES(MIGRAPHX_SHAPE_GENERATE_TYPE_STRING_CASE)
+#undef MIGRAPHX_SHAPE_GENERATE_TYPE_STRING_CASE
     }
     MIGRAPHX_THROW("Invalid type");
 }

src/targets/cpu/lowering.cpp

@@ -166,7 +166,8 @@ struct cpu_im2col
             const std::size_t& stride_h = op.stride[0];
             const std::size_t& stride_w = op.stride[1];
 
-            int kdiv2_h, kdiv2_w;
+            int kdiv2_h;
+            int kdiv2_w;
             kdiv2_h = kernel_h / 2;
             kdiv2_w = kernel_w / 2;
             // calculate output sizes

src/targets/gpu/abs.cpp

@@ -18,7 +18,8 @@ argument miopen_abs::compute(context& ctx,
                              const shape& output_shape,
                              const std::vector<argument>& args) const
 {
-    float alpha = 1, beta = 0;
+    float alpha = 1;
+    float beta  = 0;
     auto x_desc = make_tensor(args[0].get_shape());
     auto y_desc = make_tensor(output_shape);
     miopenActivationForward(ctx.get_stream().get_miopen(),

src/targets/gpu/batchnorm.cpp

@@ -22,7 +22,8 @@ argument miopen_batch_norm_inference::compute(context& ctx,
     auto y_desc  = make_tensor(output_shape);
     auto bn_desc = make_tensor(args[3].get_shape());
 
-    float alpha = 1.0, beta = 0.0f;
+    float alpha = 1.0;
+    float beta  = 0.0f;
 
     miopenBatchNormalizationForwardInference(ctx.get_stream().get_miopen(),
                                              miopenBatchNormMode_t(op.bn_mode),