[libc-commits] [libc] [libc] Fix 'fgets' test on the GPU for some C libraries (PR #110118)

[via libc-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-libc

Author: Joseph Huber (jhuber6)

<details>
<summary>Changes</summary>

Summary:
The GPU handling for a lot of `FILE *` functions pretty much just
forwards it to the host via RPC. This test checks for implementation
defined behavior, which sometimes passes and sometimes doesn't. We just
disable it here so it works on the standard semantics.

We do this forwarding primarily for interopt w/ the host if the user is
compiling from an offloading language (e.g. CUDA).


---
Full diff: https://github.com/llvm/llvm-project/pull/110118.diff


1 Files Affected:

- (modified) libc/test/src/stdio/fgets_test.cpp (+3) 


``````````diff
diff --git a/libc/test/src/stdio/fgets_test.cpp b/libc/test/src/stdio/fgets_test.cpp
index d005a71710d21a..39337262f1e008 100644
--- a/libc/test/src/stdio/fgets_test.cpp
+++ b/libc/test/src/stdio/fgets_test.cpp
@@ -43,6 +43,8 @@ TEST(LlvmLibcFgetsTest, WriteAndReadCharacters) {
   file = LIBC_NAMESPACE::fopen(FILENAME, "r");
   ASSERT_FALSE(file == nullptr);
 
+  // The GPU build relies on the host C library, so this check may be different.
+#ifndef LIBC_TARGET_ARCH_IS_GPU
   // If we request just 1 byte, it should return just a null byte and not
   // advance the read head. This is implementation defined.
   output = LIBC_NAMESPACE::fgets(buff, 1, file);
@@ -54,6 +56,7 @@ TEST(LlvmLibcFgetsTest, WriteAndReadCharacters) {
   // This is also implementation defined.
   output = LIBC_NAMESPACE::fgets(buff, 0, file);
   ASSERT_TRUE(output == nullptr);
+#endif
 
   const char *output_arr[] = {
       "1234567", "89\n", "1234567", "\n", "123456\n", "1",

``````````

</details>


https://github.com/llvm/llvm-project/pull/110118