Merge pull request #190 from masterleinad/fix_warnings

FIx warnings in all tools built in the CI

.github/workflows/simple-build.yml

@@ -10,6 +10,6 @@ jobs:
         # The Kokkos Tools systemtap connector requires 'dtrace'
         sudo apt update
         sudo apt --yes --no-install-recommends install systemtap-sdt-dev
-        cmake -B build
+        cmake -B build -DCMAKE_CXX_FLAGS="-Wall -Wextra -Werror"
         cmake --build build
         cmake --install build --prefix ${PWD}/install

common/kernel-filter/kp_kernel_filter.cpp

@@ -129,7 +129,7 @@ extern "C" void kokkosp_init_library(const int loadSeq,
       char* profileLibrary = getenv("KOKKOS_PROFILE_LIBRARY");
       char* envBuffer =
           (char*)malloc(sizeof(char) * (strlen(profileLibrary) + 1));
-      sprintf(envBuffer, "%s", profileLibrary);
+      strcpy(envBuffer, profileLibrary);
 
       char* nextLibrary = strtok(envBuffer, ";");
 

common/kokkos-sampler/kp_sampler_skip.cpp

@@ -9,7 +9,6 @@
 namespace KokkosTools {
 namespace Sampler {
 static uint64_t uniqID           = 0;
-static uint64_t kernelCounter    = 0;
 static uint64_t kernelSampleSkip = 101;
 static int tool_verbosity        = 0;
 static int tool_globFence        = 0;
@@ -65,7 +64,7 @@ void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
   }
 
   char* envBuffer = (char*)malloc(sizeof(char) * (strlen(profileLibrary) + 1));
-  sprintf(envBuffer, "%s", profileLibrary);
+  strcpy(envBuffer, profileLibrary);
 
   char* nextLibrary = strtok(envBuffer, ";");
 
@@ -135,8 +134,7 @@ void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
 
   free(envBuffer);
 
-  kernelCounter = 0;
-  uniqID        = 1;
+  uniqID = 1;
 
   const char* tool_sample = getenv("KOKKOS_TOOLS_SAMPLER_SKIP");
   if (NULL != tool_sample) {
@@ -148,15 +146,18 @@ void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
   }
 }
 
-void kokkosp_finalize_library() {}
+void kokkosp_finalize_library() {
+  if (NULL != finalizeProfileLibrary) (*finalizeProfileLibrary)();
+}
 
 void kokkosp_begin_parallel_for(const char* name, const uint32_t devID,
                                 uint64_t* kID) {
   *kID = uniqID++;
 
   if (((*kID) % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-begin function...\n", *kID);
+      printf("KokkosP: sample %llu calling child-begin function...\n",
+             (unsigned long long)(*kID));
     }
 
     if (NULL != beginForCallee) {
@@ -168,7 +169,8 @@ void kokkosp_begin_parallel_for(const char* name, const uint32_t devID,
 void kokkosp_end_parallel_for(const uint64_t kID) {
   if ((kID % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-end function...\n", kID);
+      printf("KokkosP: sample %llu calling child-end function...\n",
+             (unsigned long long)(kID));
     }
 
     if (NULL != endForCallee) {
@@ -183,7 +185,8 @@ void kokkosp_begin_parallel_scan(const char* name, const uint32_t devID,
 
   if (((*kID) % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-begin function...\n", *kID);
+      printf("KokkosP: sample %llu calling child-begin function...\n",
+             (unsigned long long)(*kID));
     }
 
     if (NULL != beginScanCallee) {
@@ -195,7 +198,8 @@ void kokkosp_begin_parallel_scan(const char* name, const uint32_t devID,
 void kokkosp_end_parallel_scan(const uint64_t kID) {
   if ((kID % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-end function...\n", kID);
+      printf("KokkosP: sample %llu calling child-end function...\n",
+             (unsigned long long)(kID));
     }
 
     if (NULL != endScanCallee) {
@@ -210,7 +214,8 @@ void kokkosp_begin_parallel_reduce(const char* name, const uint32_t devID,
 
   if (((*kID) % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-begin function...\n", *kID);
+      printf("KokkosP: sample %llu calling child-begin function...\n",
+             (unsigned long long)(*kID));
     }
 
     if (NULL != beginReduceCallee) {
@@ -222,7 +227,8 @@ void kokkosp_begin_parallel_reduce(const char* name, const uint32_t devID,
 void kokkosp_end_parallel_reduce(const uint64_t kID) {
   if ((kID % kernelSampleSkip) == 0) {
     if (tool_verbosity > 0) {
-      printf("KokkosP: sample %lu calling child-end function...\n", kID);
+      printf("KokkosP: sample %llu calling child-end function...\n",
+             (unsigned long long)(kID));
     }
 
     if (NULL != endReduceCallee) {
@@ -238,7 +244,7 @@ extern "C" {
 
 namespace impl = KokkosTools::Sampler;
 
-EXPOSE_TOOL_SETTINGS(kokkosp_request_tool_settings)
+EXPOSE_TOOL_SETTINGS(impl::kokkosp_request_tool_settings)
 EXPOSE_INIT(impl::kokkosp_init_library)
 EXPOSE_FINALIZE(impl::kokkosp_finalize_library)
 EXPOSE_BEGIN_PARALLEL_FOR(impl::kokkosp_begin_parallel_for)

debugging/kernel-logger/CMakeLists.txt

@@ -1 +1 @@
-add_library(kp_kernel_logger ${KOKKOSTOOLS_LIBRARY_MODE} kp_kernel_logger.cpp)
+add_library(kp_kernel_logger ${KOKKOSTOOLS_LIBRARY_MODE} kp_kernel_logger.cpp)

debugging/kernel-logger/kp_kernel_logger.cpp

@@ -50,12 +50,12 @@ int kokkosp_print_region_stack() {
 
 extern "C" void kokkosp_init_library(const int loadSeq,
                                      const uint64_t interfaceVer,
-                                     const uint32_t devInfoCount,
-                                     void* deviceInfo) {
+                                     const uint32_t /*devInfoCount*/,
+                                     void* /*deviceInfo*/) {
   printf(
       "KokkosP: Kernel Logger Library Initialized (sequence is %d, version: "
       "%llu)\n",
-      loadSeq, interfaceVer);
+      loadSeq, (unsigned long long)(interfaceVer));
   uniqID = 0;
 }
 
@@ -71,7 +71,7 @@ extern "C" void kokkosp_begin_parallel_for(const char* name,
   printf(
       "KokkosP: Executing parallel-for kernel on device %d with unique "
       "execution identifier %llu\n",
-      devID, *kID);
+      devID, (unsigned long long)(*kID));
 
   int level = kokkosp_print_region_stack();
   kokkosp_print_region_stack_indent(level);
@@ -80,7 +80,8 @@ extern "C" void kokkosp_begin_parallel_for(const char* name,
 }
 
 extern "C" void kokkosp_end_parallel_for(const uint64_t kID) {
-  printf("KokkosP: Execution of kernel %llu is completed.\n", kID);
+  printf("KokkosP: Execution of kernel %llu is completed.\n",
+         (unsigned long long)(kID));
 }
 
 extern "C" void kokkosp_begin_parallel_scan(const char* name,
@@ -91,16 +92,17 @@ extern "C" void kokkosp_begin_parallel_scan(const char* name,
   printf(
       "KokkosP: Executing parallel-scan kernel on device %d with unique "
       "execution identifier %llu\n",
-      devID, *kID);
+      devID, (unsigned long long)(*kID));
 
   int level = kokkosp_print_region_stack();
   kokkosp_print_region_stack_indent(level);
 
   printf("    %s\n", name);
 }
 
-extern "C" void kokkosp_end_parallel_scan(const uint64_t kID) {
-  printf("KokkosP: Execution of kernel %llu is completed.\n", kID);
+extern "C" void kokkospk_end_parallel_scan(const uint64_t kID) {
+  printf("KokkosP: Execution of kernel %llu is completed.\n",
+         (unsigned long long)(kID));
 }
 
 extern "C" void kokkosp_begin_parallel_reduce(const char* name,
@@ -111,7 +113,7 @@ extern "C" void kokkosp_begin_parallel_reduce(const char* name,
   printf(
       "KokkosP: Executing parallel-reduce kernel on device %d with unique "
       "execution identifier %llu\n",
-      devID, *kID);
+      devID, (unsigned long long)(*kID));
 
   int level = kokkosp_print_region_stack();
   kokkosp_print_region_stack_indent(level);
@@ -120,7 +122,8 @@ extern "C" void kokkosp_begin_parallel_reduce(const char* name,
 }
 
 extern "C" void kokkosp_end_parallel_reduce(const uint64_t kID) {
-  printf("KokkosP: Execution of kernel %llu is completed.\n", kID);
+  printf("KokkosP: Execution of kernel %llu is completed.\n",
+         (unsigned long long)(kID));
 }
 
 extern "C" void kokkosp_begin_fence(const char* name, const uint32_t devID,
@@ -139,7 +142,7 @@ extern "C" void kokkosp_begin_fence(const char* name, const uint32_t devID,
     printf(
         "KokkosP: Executing fence on device %d with unique execution "
         "identifier %llu\n",
-        devID, *kID);
+        devID, (unsigned long long)(*kID));
 
     int level = kokkosp_print_region_stack();
     kokkosp_print_region_stack_indent(level);
@@ -153,7 +156,8 @@ extern "C" void kokkosp_end_fence(const uint64_t kID) {
   // dealing with the application's fence, which we filtered out in the callback
   // for fences
   if (kID != std::numeric_limits<uint64_t>::max()) {
-    printf("KokkosP: Execution of fence %llu is completed.\n", kID);
+    printf("KokkosP: Execution of fence %llu is completed.\n",
+           (unsigned long long)(kID));
   }
 }
 
@@ -173,14 +177,14 @@ extern "C" void kokkosp_pop_profile_region() {
 
 extern "C" void kokkosp_allocate_data(SpaceHandle handle, const char* name,
                                       void* ptr, uint64_t size) {
-  printf("KokkosP: Allocate<%s> name: %s pointer: %p size: %lu\n", handle.name,
-         name, ptr, size);
+  printf("KokkosP: Allocate<%s> name: %s pointer: %p size: %llu\n", handle.name,
+         name, ptr, (unsigned long long)(size));
 }
 
 extern "C" void kokkosp_deallocate_data(SpaceHandle handle, const char* name,
                                         void* ptr, uint64_t size) {
-  printf("KokkosP: Deallocate<%s> name: %s pointer: %p size: %lu\n",
-         handle.name, name, ptr, size);
+  printf("KokkosP: Deallocate<%s> name: %s pointer: %p size: %llu\n",
+         handle.name, name, ptr, (unsigned long long)(size));
 }
 
 extern "C" void kokkosp_begin_deep_copy(SpaceHandle dst_handle,
@@ -191,7 +195,7 @@ extern "C" void kokkosp_begin_deep_copy(SpaceHandle dst_handle,
                                         const void* src_ptr, uint64_t size) {
   printf(
       "KokkosP: DeepCopy<%s,%s> DST(name: %s pointer: %p) SRC(name: %s pointer "
-      "%p) Size: %lu\n",
+      "%p) Size: %llu\n",
       dst_handle.name, src_handle.name, dst_name, dst_ptr, src_name, src_ptr,
-      size);
+      (unsigned long long)(size));
 }

profiling/chrome-tracing/kp_chrome_tracing.cpp

@@ -127,8 +127,8 @@ struct State {
     gethostname(hostname, 256);
 
     char *fileOutput = (char *)malloc(sizeof(char) * 256);
-    sprintf(fileOutput, "%s-%d-%s.json", hostname, (int)getpid(),
-            (NULL == mpi_rank) ? "0" : mpi_rank);
+    snprintf(fileOutput, 256, "%s-%d-%s.json", hostname, (int)getpid(),
+             (NULL == mpi_rank) ? "0" : mpi_rank);
 #if defined(USE_MPI) && USE_MPI
     MPI_Comm_rank(MPI_COMM_WORLD, &my_mpi_rank);
 #else

profiling/memory-events/kp_memory_events.cpp

@@ -51,13 +51,13 @@ double max_mem_usage() {
 }
 
 void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
-                          const uint32_t devInfoCount,
-                          Kokkos_Profiling_KokkosPDeviceInfo* deviceInfo) {
+                          const uint32_t /*devInfoCount*/,
+                          Kokkos_Profiling_KokkosPDeviceInfo* /*deviceInfo*/) {
   num_spaces = 0;
   for (int i = 0; i < 16; i++) space_size[i] = 0;
 
   printf("KokkosP: MemoryEvents loaded (sequence: %d, version: %llu)\n",
-         loadSeq, interfaceVer);
+         loadSeq, (unsigned long long)(interfaceVer));
 
   timer.reset();
 }
@@ -69,7 +69,7 @@ void kokkosp_finalize_library() {
 
   {
     char* fileOutput = (char*)malloc(sizeof(char) * 256);
-    sprintf(fileOutput, "%s-%d.mem_events", hostname, pid);
+    snprintf(fileOutput, 256, "%s-%d.mem_events", hostname, pid);
 
     FILE* ofile = fopen(fileOutput, "wb");
     free(fileOutput);
@@ -78,14 +78,15 @@ void kokkosp_finalize_library() {
     fprintf(ofile,
             "# Time     Ptr                  Size        MemSpace      Op      "
             "   Name\n");
-    for (int i = 0; i < events.size(); i++) events[i].print_record(ofile);
+    for (unsigned int i = 0; i < events.size(); i++)
+      events[i].print_record(ofile);
     fclose(ofile);
   }
 
   for (int s = 0; s < num_spaces; s++) {
     char* fileOutput = (char*)malloc(sizeof(char) * 256);
-    sprintf(fileOutput, "%s-%d-%s.memspace_usage", hostname, pid,
-            space_name[s]);
+    snprintf(fileOutput, 256, "%s-%d-%s.memspace_usage", hostname, pid,
+             space_name[s]);
 
     FILE* ofile = fopen(fileOutput, "wb");
     free(fileOutput);
@@ -94,7 +95,7 @@ void kokkosp_finalize_library() {
     fprintf(ofile,
             "# Time(s)  Size(MB)   HighWater(MB)   HighWater-Process(MB)\n");
     uint64_t maxvalue = 0;
-    for (int i = 0; i < space_size_track[s].size(); i++) {
+    for (unsigned int i = 0; i < space_size_track[s].size(); i++) {
       if (std::get<1>(space_size_track[s][i]) > maxvalue)
         maxvalue = std::get<1>(space_size_track[s][i]);
       fprintf(ofile, "%lf %.1lf %.1lf %.1lf\n",
@@ -126,7 +127,6 @@ void kokkosp_allocate_data(const SpaceHandle space, const char* label,
   space_size_track[space_i].push_back(
       std::make_tuple(time, space_size[space_i], max_mem_usage()));
 
-  int i = events.size();
   events.push_back(
       EventRecord(ptr, size, MEMOP_ALLOCATE, space_i, time, label));
 }
@@ -151,7 +151,6 @@ void kokkosp_deallocate_data(const SpaceHandle space, const char* label,
         std::make_tuple(time, space_size[space_i], max_mem_usage()));
   }
 
-  int i = events.size();
   events.push_back(
       EventRecord(ptr, size, MEMOP_DEALLOCATE, space_i, time, label));
 }

profiling/memory-hwm/kp_hwm.cpp

@@ -33,15 +33,13 @@
 namespace KokkosTools {
 namespace HighwaterMark {
 
-static uint64_t uniqID = 0;
-
 void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
-                          const uint32_t devInfoCount,
-                          Kokkos_Profiling_KokkosPDeviceInfo* deviceInfo) {
+                          const uint32_t /*devInfoCount*/,
+                          Kokkos_Profiling_KokkosPDeviceInfo* /*deviceInfo*/) {
   printf(
       "KokkosP: High Water Mark Library Initialized (sequence is %d, version: "
       "%llu)\n",
-      loadSeq, interfaceVer);
+      loadSeq, (unsigned long long)(interfaceVer));
 }
 
 // darwin report rusage.ru_maxrss in bytes

profiling/memory-usage/kp_memory_usage.cpp

@@ -47,9 +47,10 @@ double max_mem_usage() {
   return max_rssKB * 1024;
 }
 
-void kokkosp_init_library(const int loadSeq, const uint64_t interfaceVer,
-                          const uint32_t devInfoCount,
-                          Kokkos_Profiling_KokkosPDeviceInfo* deviceInfo) {
+void kokkosp_init_library(const int /*loadSeq*/,
+                          const uint64_t /*interfaceVer*/,
+                          const uint32_t /*devInfoCount*/,
+                          Kokkos_Profiling_KokkosPDeviceInfo* /*deviceInfo*/) {
   num_spaces = 0;
   for (int i = 0; i < 16; i++) space_size[i] = 0;
 
@@ -73,7 +74,7 @@ void kokkosp_finalize_library() {
     fprintf(ofile,
             "# Time(s)  Size(MB)   HighWater(MB)   HighWater-Process(MB)\n");
     uint64_t maxvalue = 0;
-    for (int i = 0; i < space_size_track[s].size(); i++) {
+    for (unsigned int i = 0; i < space_size_track[s].size(); i++) {
       if (std::get<1>(space_size_track[s][i]) > maxvalue)
         maxvalue = std::get<1>(space_size_track[s][i]);
       fprintf(ofile, "%lf %.1lf %.1lf %.1lf\n",
@@ -87,8 +88,8 @@ void kokkosp_finalize_library() {
   free(hostname);
 }
 
-void kokkosp_allocate_data(const SpaceHandle space, const char* label,
-                           const void* const ptr, const uint64_t size) {
+void kokkosp_allocate_data(const SpaceHandle space, const char* /*label*/,
+                           const void* const /*ptr*/, const uint64_t size) {
   std::lock_guard<std::mutex> lock(m);
 
   double time = timer.seconds();
@@ -106,8 +107,8 @@ void kokkosp_allocate_data(const SpaceHandle space, const char* label,
       std::make_tuple(time, space_size[space_i], max_mem_usage()));
 }
 
-void kokkosp_deallocate_data(const SpaceHandle space, const char* label,
-                             const void* const ptr, const uint64_t size) {
+void kokkosp_deallocate_data(const SpaceHandle space, const char* /*label*/,
+                             const void* const /*ptr*/, const uint64_t size) {
   std::lock_guard<std::mutex> lock(m);
 
   double time = timer.seconds();