Message-ID: <379cc523-8f08-4c7f-ae20-20f216352a01@googlegroups.com>
Date:	Sat, 12 Jan 2013 11:18:37 -0800 (PST)
From:	antoine.trux@...il.com
To:	fa.linux.kernel@...glegroups.com
Cc:	Johannes Weiner <hannes@...urebad.de>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	clameter@....com, penberg@...helsinki.fi
Subject: Re: Why is the kfree() argument const?

On Wednesday, January 16, 2008 8:39:48 PM UTC+2, Linus Torvalds wrote:

> "const" has *never* been about the thing not being modified. Forget all 
> that claptrap. C does not have such a notion.

I beg your pardon?!

C has had that very notion ever since its first standard (1989). Here is an excerpt from that standard (ISO/IEC 9899:1990, section 6.5.3):

    "If an attempt is made to modify an object defined with a const-qualified type through use of an lvalue with non-const-qualified type, the behavior is undefined."


> "const" is a pointer type issue, and is meant to make certain mis-uses 
> more visible at compile time. It has *no* other meaning, and anybody who 
> thinks it has is just setting himself up for problems.

'const' is also a pointer issue, but not only - see above quote from the C Standard.


Defining an object 'const' can have an impact on optimization (and also on whether the object is placed in read-only memory). Here are trivial examples to illustrate:

<Program1>

    <foo1.c>
    void foo1(const int* pi)
    {
        *(int*)pi = 1;
    }
    </foo1.c>

    <main1.c>
    #include <stdio.h>
    void foo1(const int* pi);
    int main(void)
    {
        int i = 0;
        foo1(&i);
        printf("i = %d\n", i);
        return 0;
    }
    </main1.c>

</Program1>

Program1 defines 'i' non-const, and modifies it through a const pointer, by casting const away in foo1(). This is allowed - although not necessarily wise.

Program1 has well defined behavior: it prints "i = 1". The generated code dutifully retrieves the value of 'i' before passing it to printf().


<Program2>

    <foo2.c>
    void foo2(const int* pi)
    {
    }
    </foo2.c>

    <main2.c>
    #include <stdio.h>
    void foo2(const int* pi);
    int main(void)
    {
        const int i = 0;
        foo2(&i);
        printf("i = %d\n", i);
        return 0;
    }
    </main2.c>

</Program2>

Program2 defines 'i' const. A pointer to 'i' is passed to foo2(), which does not modify 'i'.

Program2 has well defined behavior: it prints "i = 0". When it generates code for main1.c, the compiler can assume that 'i' is not modified, because 'i' is defined const.

When compiling main2.c with gcc 4.4.7 with optimizations turned off (-O0), the generated code retrieves the value of 'i' before passing it to printf(). With optimizations turned on (-O3), it inlines the value of 'i', 0, in the call to printf(). Both versions have the same, correct behavior.


<Program3>

    <foo3.c>
    void foo3(const int* pi)
    {
        *(int*)pi = 1;
    }
    </foo3.c>

    <main3.c>
    #include <stdio.h>
    void foo3(const int* pi);
    int main(void)
    {
        const int i = 0;
        foo3(&i);
        printf("i = %d\n", i);
        return 0;
    }
    </main3.c>

</Program3>

Program3 defines 'i' const, and attempts to modify it through a const pointer, by casting const away in foo3().

On my particular system, when compiling Program3 with gcc 4.4.7 with optimizations turned off (-O0), the program prints "i = 1". With optimizations turned on (-O3), it prints "i = 0".

The question of which of these two behaviors is "correct" would be pointless, since Program3 has undefined behavior.


Antoine
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists