Communities

Writing
Writing
Codidact Meta
Codidact Meta
The Great Outdoors
The Great Outdoors
Photography & Video
Photography & Video
Scientific Speculation
Scientific Speculation
Cooking
Cooking
Electrical Engineering
Electrical Engineering
Judaism
Judaism
Languages & Linguistics
Languages & Linguistics
Software Development
Software Development
Mathematics
Mathematics
Christianity
Christianity
Code Golf
Code Golf
Music
Music
Physics
Physics
Linux Systems
Linux Systems
Power Users
Power Users
Tabletop RPGs
Tabletop RPGs
Community Proposals
Community Proposals
tag:snake search within a tag
answers:0 unanswered questions
user:xxxx search by author id
score:0.5 posts with 0.5+ score
"snake oil" exact phrase
votes:4 posts with 4+ votes
created:<1w created < 1 week ago
post_type:xxxx type of post
Search help
Notifications
Mark all as read See all your notifications »
Q&A

Welcome to Software Development on Codidact!

Will you help us build our independent community of developers helping developers? We're small and trying to grow. We welcome questions about all aspects of software development, from design to code to QA and more. Got questions? Got answers? Got code you'd like someone to review? Please join us.

Comments on Assert that some code is not present in the final binary, at compile or link time.

Parent

Assert that some code is not present in the final binary, at compile or link time.

+3
−0

I'd like to assert that some code can be optimized out, and is not present in the final binary object.

#define CONSTANT  0

#if (!CONSTANT)
[[landmine_A]]
#endif
static int foo(void);

void bar(void)
{
    if (CONSTANT) {
        foo();
    }
}

static int foo(void)
{
    if (!CONSTANT)
        landmine_B();
}

Attributes, builtins, expressions, ..., everything is fair play, as long as it guarantees that the program is not built with foo(), unless #define CONSTANT 1.

I'd like the compiler (or linker, but preferably the compiler) to warn/error if foo() is used, but not if it's inside an if (0).

In the past, this could probably be achieved by __builtin_unreachable();, and the corresponding warning, but it's ignored nowadays...

An option (the one in use, which I'm trying to improve), is to build conditionally foo, but then I also need to use preprocessor stuff at call site, which I don't entirely like, because it hides code to the compiler, so I need to test multiple configurations.

History
Why does this post require attention from curators or moderators?
You might want to add some details to your flag.
Why should this post be closed?

0 comment threads

Post
+2
−0

Calling an undefined function will have that behavior at link time:

landmine.c:

#ifndef CONSTANT
#define CONSTANT  0
#endif

#define assert_not_in_binary_if(e)  do \
{ \
	if (e) \
		undefined_function(); \
} while (0)

void undefined_function(void);
[[gnu::noipa]] static void foo0(void);
[[gnu::noipa]] static void foo1(void);

int main(void)
{
	CONSTANT ? foo1() : foo0();
}

static void foo0(void)
{
	assert_not_in_binary_if(!CONSTANT);
}

static void foo1(void)
{
	assert_not_in_binary_if(!CONSTANT);
}
alx@asus5775:~/tmp$ cc -Wall -Wextra -O3 -DCONSTANT=1 landmine.c
alx@asus5775:~/tmp$ cc -Wall -Wextra -O3 -DCONSTANT=0 landmine.c
/usr/bin/ld: /tmp/ccmIFcBz.o: in function `foo0':
landmine.c:(.text+0x1): undefined reference to `undefined_function'
collect2: error: ld returned 1 exit status

I used [[gnu::noipa]] just to check which function is triggering the assertion, and it is foo0() as expected, and only when CONSTANT == 0.

It would be nicer to have a compile-time assertion, but link-time is good enough.

EDITED: renamed landmine() to assert_not_in_binary_if(), per Dirk's suggestion.

History
Why does this post require attention from curators or moderators?
You might want to add some details to your flag.

2 comment threads

Why not use static assert? (4 comments)
Nice trick, would suggest different name for readability (2 comments)
Why not use static assert?
Lundin‭ wrote over 2 years ago · edited over 2 years ago

Not entirely sure what problem you are trying to solve, but is there any reason why you can't do if (e) { _Static_assert(!e,"Called but not in binary."); } \ instead, something like that? To get a readable compile-time error instead of a mysterious link-time one?

alx‭ wrote over 2 years ago · edited over 2 years ago

Okay, so the problem (the current code I'm dealing with) is:

A program might be compiled with support for feature A, or without support for it (I guess for having a slimmer/faster binary). The feature is basically some A_*() functions, and some calls to those functions from core functions.

Right now this is handled by adding #if (A) around the function declarations, function definitions, and function calls, and having #else at the call sites. I hate this approach, because I would need to compile many combinations of ./configureations to analyze all of the code.

I want the compiler to see all of the code, even if it's not being used at all. But I still want to make sure that the code is not present in the binary (otherwise I'd just remove the ability to remove support for A, and make my life easier).

...

alx‭ wrote over 2 years ago

...

The static assert would need to be at call site, but that would not make sure the function is not referenced anywhere else in the code. And a static assert at the A_* function definitions would always be triggered, as long as the functions are being compiled.

I only want to trigger it if the function is being called under certain conditions. [[deprecated]] or [[gnu::unavailable]] would be the closest thing, but it would be triggered by any call; I only want it to be triggered conditionally. I don't want to trigger the attribute within if (0) { A_foo(); }, since that's not a real use (with some sane optimization flags).

Lundin‭ wrote over 2 years ago

Ok, well... conditional compilation is always messy. I would consider solving this through version control and program design instead. As in, include a foo.h with the common interface, then link either A_foo.c or B_foo.c depending on version.