[libc++] Fix initialization-order-fiasco with iostream.cpp constructors

[vitalybuka]

Asan reports it after #124103.

It's know case of false positive for Asan.
https://github.com/google/sanitizers/wiki/AddressSanitizerInitializationOrderFiasco#false-positives

It's can be avoided with constexpr constructors.

In general order global constructors in different
modules is undefined. If global constructor uses
external global, they can be not constructed yet.

However, implementation may contain workaround for
that, or the state of non-constructed global can
be still valid.

Asan will still falsely report such cases, as it
has no machinery to detect correctness of such
cases.

We need to fix/workaround the issue in libc++, as
it will affect many libc++ with Asan users.

[llvmbot]

@llvm/pr-subscribers-libcxx

Author: Vitaly Buka (vitalybuka)

Changes

Asan reports it after #124103.

It's know case of false positive for Asan.
https://github.com/google/sanitizers/wiki/AddressSanitizerInitializationOrderFiasco#false-positives

In general order global constructors in different
modules is undefined. If global constructor uses
external global, they can be not constructed yet.

However, implementation may contain workaround for
that, or the state of non-constructed global can
be still valid.

Asan will still falsely report such cases, as it
has no machinery to detect correctness of such
cases.

We need to fix/workaround the issue in libc++, as
it will affect many libc++ with Asan users.

---

Full diff: https://github.com/llvm/llvm-project/pull/126995.diff

1 Files Affected:

• (added) libcxx/test/std/input.output/iostreams.base/ios.base/ios.types/ios_Init/ios_Init.global.pass.cpp (+27)

diff --git a/libcxx/test/std/input.output/iostreams.base/ios.base/ios.types/ios_Init/ios_Init.global.pass.cpp b/libcxx/test/std/input.output/iostreams.base/ios.base/ios.types/ios_Init/ios_Init.global.pass.cpp
new file mode 100644
index 0000000000000..b1ea8517850f4
--- /dev/null
+++ b/libcxx/test/std/input.output/iostreams.base/ios.base/ios.types/ios_Init/ios_Init.global.pass.cpp
@@ -0,0 +1,27 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include <iostream>
+
+extern "C" const char *__asan_default_options() {
+  return "check_initialization_order=true:strict_init_order=true";
+}
+
+// Test that ios used from globals constructors doesn't trigger
+// Asan initialization-order-fiasco.
+
+struct Global {
+  Global() {
+    std::cout << "Hello!";
+  }
+} global;
+
+int main(int, char**)
+{
+    return 0;
+}

[vitalybuka]

This is just test to see if it triggers any CI
I checked that https://lab.llvm.org/buildbot/#/builders/sanitizer-x86_64-linux-bootstrap-asan will fail this test without the fix.

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[vitalybuka]

Fails with out a fix https://github.com/llvm/llvm-project/actions/runs/13298631288/job/37137221150?pr=126995

[vitalybuka]

can't figure out how to make GCC happy,
all the time: error: ‘constinit’ variable ‘std::__1::cin’ does not have a constant initializer

[philnik777]

Hmm. I'll try to figure something out tomorrow for this. Either a bug against one of the compilers or some code changes should happen here. I guess a CWG issue would also be possible.

[vitalybuka]

Can we land this one in meantime?

[philnik777]

I'm fine with that.