Add an error message to the default SIGPIPE handler

UNIX03 conformance requires utilities to flush stdout before exiting and raise
an error if writing fails. Flushing already happens on a call to exit
and thus automatically on a return from main. Write failure is then
detected by LLVM's default SIGPIPE handler. The handler already exits with
a non-zero code, but conformance additionally requires an error message.

Author: TNorthover

llvm/lib/Support/Unix/Signals.inc

@@ -432,6 +432,10 @@ void llvm::sys::SetOneShotPipeSignalFunction(void (*Handler)()) {
 }
 
 void llvm::sys::DefaultOneShotPipeSignalHandler() {
+  // UNIX03 conformance requires a non-zero exit code and an error message
+  // to stderr when writing to a closed stdout fails.
+  errs() << "error: write on a pipe with no reader\n";
+
   // Send a special return code that drivers can check for, from sysexits.h.
   exit(EX_IOERR);
 }

llvm/test/tools/llvm-nm/unix03-sigpipe-exit.test

@@ -0,0 +1,17 @@
+# Test that when nm tries to write to a closed stdout it will finish with
+# a non-zero exit code and an error message on stderr.
+# This is required for UNIX03 conformance.
+
+# UNSUPPORTED: system-windows
+
+# RUN: not %python %s llvm-nm llvm-nm 2>&1 | FileCheck %s
+# CHECK: error: write on a pipe with no reader
+
+import subprocess
+import sys
+
+with subprocess.Popen([sys.argv[1], sys.argv[2]], stdout=subprocess.PIPE) as process:
+  # Read single byte and immediately close pipe to trigger SIGPIPE.
+  process.stdout.read(1)
+  process.stdout.close()
+sys.exit(process.returncode)