[libc] Fix some tests on NVPTX due to insufficient stack size

[jhuber6]

Summary

A few of these tests were disabled due to failing on NVPTX. After
looking into it the vast majority of these cases were due to
insufficient stack memory. This can be worked around by increasing the
stack size in the loader or by reducing the memory usage in the case of
large string constants.

libc/test/src/__support/CMakeLists.txt

add_libc_test(
   SUITE
     libc-support-tests
   SRCS
     endian_test.cpp
   DEPENDS
     libc.src.__support.common
 )
 
-# These tests fails with an illegal instruction
-if(NOT LIBC_GPU_TARGET_ARCHITECTURE_IS_NVPTX)
-  add_libc_test(
-    high_precision_decimal_test
-    SUITE
-      libc-support-tests
-    SRCS
-    high_precision_decimal_test.cpp
-    DEPENDS
-      libc.src.__support.high_precision_decimal
-      libc.src.__support.uint128
-  )
 
-  add_libc_test(
-    str_to_float_test
-    SUITE
-      libc-support-tests
-    SRCS
-      str_to_float_test.cpp
-    DEPENDS
-      libc.src.__support.str_to_float
-      libc.src.__support.uint128
-      libc.src.errno.errno
-  )
-endif()
 
 add_libc_test(
   integer_to_string_test
   SUITE
     libc-support-tests
   SRCS
     integer_to_string_test.cpp
   DEPENDS
SUITE
       libc-support-tests
     SRCS
       arg_list_test.cpp
     DEPENDS
       libc.src.__support.arg_list
   )
 endif()
 
-# This test fails with an illegal memory access on NVPTX.
-if(NOT LIBC_GPU_TARGET_ARCHITECTURE_IS_NVPTX)
-  add_libc_test(
-    uint_test
-    SUITE
-      libc-support-tests
-    SRCS
-      uint_test.cpp
-    DEPENDS
-      libc.src.__support.uint
-      libc.src.__support.CPP.optional
-  )
-endif()
 
 add_libc_test(
   fixedvector_test
   SUITE
     libc-support-tests
   SRCS
     fixedvector_test.cpp
   DEPENDS
     libc.src.__support.fixedvector
 )
 
-# This test fails with a segmentation fault on NVPTX.
-if(NOT LIBC_GPU_TARGET_ARCHITECTURE_IS_NVPTX)
 add_libc_test(
   char_vector_test
   SUITE
     libc-support-tests
   SRCS
     char_vector_test.cpp
   DEPENDS
     libc.src.__support.char_vector
 )
-endif()
 
 add_executable(
   libc_str_to_float_comparison_test
   str_to_float_comparison_test.cpp
 )
 
 target_link_libraries(libc_str_to_float_comparison_test
   PRIVATE

libc/test/src/__support/CPP/CMakeLists.txt

SUITE
       libc-cpp-utils-tests
     SRCS
       atomic_test.cpp
     DEPENDS
       libc.src.__support.CPP.atomic
   )
 endif()
 
-# This test fails with a segmentation fault on NVPTX.
-if(NOT LIBC_GPU_TARGET_ARCHITECTURE_IS_NVPTX)
-  add_libc_test(
-    stringstream_test
-    SUITE
-      libc-cpp-utils-tests
-    SRCS
-      stringstream_test.cpp
-    DEPENDS
-      libc.src.__support.CPP.span
-      libc.src.__support.CPP.stringstream
-  )
-endif()
 
 add_libc_test(
   optional_test
   SUITE
     libc-cpp-utils-tests
   SRCS
     optional_test.cpp
   DEPENDS

libc/test/src/__support/char_vector_test.cpp

TEST(LlvmLibcCharVectorTest, InitializeCheck) {
   CharVector v;
   ASSERT_EQ(v.length(), size_t(0));
 }
 
 TEST(LlvmLibcCharVectorTest, AppendShort) {
   CharVector v;
   ASSERT_EQ(v.length(), size_t(0));
 
-  constexpr char test_str[] = "1234567890";
+  static constexpr char test_str[] = "1234567890";
   for (size_t i = 0; test_str[i] != '\0'; ++i) {
     v.append(test_str[i]);
   }
   ASSERT_STREQ(v.c_str(), test_str);
 }
 
 TEST(LlvmLibcCharVectorTest, AppendMedium) {
   CharVector v;
   ASSERT_EQ(v.length(), size_t(0));
 
   // 100 characters (each row is 50)
-  constexpr char test_str[] =
+  static constexpr char test_str[] =
       "12345678901234567890123456789012345678901234567890"
       "ABCDEFGHIJKLMNOPQRSTUVWXYabcdefghijklmnopqrstuvwxy";
   for (size_t i = 0; test_str[i] != '\0'; ++i) {
     ASSERT_EQ(v.length(), i);
     v.append(test_str[i]);
   }
   ASSERT_STREQ(v.c_str(), test_str);
   ASSERT_EQ(v.length(), size_t(100));
 }
 
 TEST(LlvmLibcCharVectorTest, AppendLong) {
   CharVector v;
   ASSERT_EQ(v.length(), size_t(0));
 
   // 1000 characters
-  constexpr char test_str[] =
+  static constexpr char test_str[] =
       "12345678901234567890123456789012345678901234567890"
       "ABCDEFGHIJKLMNOPQRSTUVWXYabcdefghijklmnopqrstuvwxy"
       "12345678901234567890123456789012345678901234567890"
       "ABCDEFGHIJKLMNOPQRSTUVWXYabcdefghijklmnopqrstuvwxy"
       "12345678901234567890123456789012345678901234567890"
       "ABCDEFGHIJKLMNOPQRSTUVWXYabcdefghijklmnopqrstuvwxy"
       "12345678901234567890123456789012345678901234567890"
       "ABCDEFGHIJKLMNOPQRSTUVWXYabcdefghijklmnopqrstuvwxy"

libc/utils/gpu/loader/nvptx/Loader.cpp

int load(int argc, char **argv, char **envp, void *image, size_t size,
 
   // Initialize the CUDA context and claim it for this execution.
   CUcontext context;
   if (CUresult err = cuDevicePrimaryCtxRetain(&context, device))
     handle_error(err);
   if (CUresult err = cuCtxSetCurrent(context))
     handle_error(err);
 
+  // Increase the stack size per thread.
+  // TODO: We should allow this to be passed in so only the tests that require a
+  // larger stack can specify it to save on memory usage.
+  if (CUresult err = cuCtxSetLimit(CU_LIMIT_STACK_SIZE, 3 * 1024))
+    handle_error(err);
+
   // Initialize a non-blocking CUDA stream to execute the kernel.
   CUstream stream;
   if (CUresult err = cuStreamCreate(&stream, CU_STREAM_NON_BLOCKING))
     handle_error(err);
 
   // Load the image into a CUDA module.
   CUmodule binary;
   if (CUresult err = cuModuleLoadDataEx(&binary, image, 0, nullptr, nullptr))