-
Notifications
You must be signed in to change notification settings - Fork 13.9k
[libcxx] [test] Fix odr_signature tests with optimizations enabled #144317
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: main
Are you sure you want to change the base?
Conversation
@llvm/pr-subscribers-libcxx Author: Martin Storsjö (mstorsjo) ChangesIf optimization is enabled, the inline Therefore, this test previously actually failed to test what it was supposed to test, if optimization was enabled. Mark the inline functions with Also update the TODO comments to explain why we have an XFAIL for msvc mode here. This avoids these tests unexpectedly passing if building in msvc mode, with optimizations enabled Full diff: https://github.com/llvm/llvm-project/pull/144317.diff 2 Files Affected:
diff --git a/libcxx/test/libcxx/odr_signature.exceptions.sh.cpp b/libcxx/test/libcxx/odr_signature.exceptions.sh.cpp
index 6bf60b5e82d3c..8bc7f6cffbd0c 100644
--- a/libcxx/test/libcxx/odr_signature.exceptions.sh.cpp
+++ b/libcxx/test/libcxx/odr_signature.exceptions.sh.cpp
@@ -6,7 +6,7 @@
//
//===----------------------------------------------------------------------===//
-// TODO: Investigate
+// ABI tags have no effect in MSVC mode.
// XFAIL: msvc
// Test that we encode whether exceptions are supported in an ABI tag to avoid
@@ -21,14 +21,14 @@
// -fno-exceptions
#ifdef TU1
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 1; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 1; }
int tu1() { return f(); }
#endif // TU1
// -fexceptions
#ifdef TU2
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 2; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 2; }
int tu2() { return f(); }
#endif // TU2
diff --git a/libcxx/test/libcxx/odr_signature.hardening.sh.cpp b/libcxx/test/libcxx/odr_signature.hardening.sh.cpp
index 0dc280bf28182..e2ea6bf4a597a 100644
--- a/libcxx/test/libcxx/odr_signature.hardening.sh.cpp
+++ b/libcxx/test/libcxx/odr_signature.hardening.sh.cpp
@@ -6,7 +6,7 @@
//
//===----------------------------------------------------------------------===//
-// TODO: Investigate
+// ABI tags have no effect in MSVC mode.
// XFAIL: msvc
// Test that we encode the hardening mode in an ABI tag to avoid ODR violations
@@ -24,28 +24,28 @@
// fast hardening mode
#ifdef TU1
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 1; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 1; }
int tu1() { return f(); }
#endif // TU1
// extensive hardening mode
#ifdef TU2
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 2; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 2; }
int tu2() { return f(); }
#endif // TU2
// debug hardening mode
#ifdef TU3
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 3; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 3; }
int tu3() { return f(); }
#endif // TU3
// No hardening
#ifdef TU4
# include <__config>
-_LIBCPP_HIDE_FROM_ABI inline int f() { return 4; }
+_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 4; }
int tu4() { return f(); }
#endif // TU4
|
// ABI tags have no effect in MSVC mode. | ||
// XFAIL: msvc |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should probably document that we're not providing our per-TU guarantees on MSVC?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess we should do that somewhere, yes.
@@ -21,14 +21,14 @@ | |||
// -fno-exceptions | |||
#ifdef TU1 | |||
# include <__config> | |||
_LIBCPP_HIDE_FROM_ABI inline int f() { return 1; } | |||
_LIBCPP_HIDE_FROM_ABI __attribute__((noinline)) inline int f() { return 1; } |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's use TEST_NOINLINE
instead.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok, can do. These test files don't include anything before the #ifdef TU*
, where we manually include <__config>
, does it disturb something of that setup if we include "test_macros.h"
globally before this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not sure why it's this way in the first place TBH. I don't think it'll make a difference.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok, changed to TEST_NOINLINE
with an include of "test_macros.h"
now.
If optimization is enabled, the inline `f()` function actually gets inlined, meaning that the functions `tu1()` and `tu2()` trivially return 1 and 2, instead of actually referencing the potentially linker deduplicated function `f()`, which is what the test tries to test. Mark the inline functions with `TEST_NOINLINE` to make sure that they don't get inlined even with optimizations enabled. Also update the TODO comments to explain why we have an XFAIL for msvc mode here. This avoids these tests unexpectedly passing if building in msvc mode, with optimizations enabled (`-DLIBCXX_TEST_PARAMS="optimization=speed"`).
8dc0e86
to
78b157b
Compare
If optimization is enabled, the inline
f()
function actually gets inlined, meaning that the functionstu1()
andtu2()
trivially return 1 and 2, instead of actually referencing the potentially linker deduplicated functionf()
, which is what the test tries to test.Therefore, this test previously actually failed to test what it was supposed to test, if optimization was enabled.
Mark the inline functions with
TEST_NOINLINE
to make sure that they don't get inlined even with optimizations enabled.Also update the TODO comments to explain why we have an XFAIL for msvc mode here.
This avoids these tests unexpectedly passing if building in msvc mode, with optimizations enabled
(
-DLIBCXX_TEST_PARAMS="optimization=speed"
).