Re: disabling FORTIFY_SOURCE

[Marcus Meissner]

On Sun, Oct 24, 2010 at 09:50:42AM -0600, Vitaliy Margolen wrote:
> On 10/24/2010 12:32 AM, Marcus Meissner wrote:
> > Actually I would like to know if its just more than the dlls/shell32/pidl.c
> > problem...
>
> If you take a look at winternl.h you'll see number of structures
> there look like:
> typedef struct _foo {
>     ULONG length;
>     WCHAR buffer[1];
> } foo, *pfoo;
>
> Or just grep for '[1]' in include directory. Lots and lots of
> declarations in all different places.

As I already wrote, this works.

Here is a sample code which shows the problem dlls/shell32/pidl.c has:

$ cat xx1.c
    #include <string.h>
    #include <stdlib.h>

struct foo {
    	int x;
    	char y[1];
    };
    union bar {
    	struct foo fo;
    	long y;
    	float fl;
    };

struct berk {
    	int t;
    	union bar b;
    };

int
    main(int argc, char **argv) {
    	struct berk *x1;
    	struct foo  *x2;

x1 = malloc (sizeof(struct berk) + 5);
    	x2 = malloc (sizeof(struct foo)  + 5);
    	strcpy(x1->b.fo.y, "hallo");
    	strcpy(x2->y, "hallo");
    }

$ gcc -O2 -Wall -D_FORTIFY_SOURCE=2 -g xx1.c -o xx1
    xx1.c: In function ‘main’:
    xx1.c:28:1: warning: control reaches end of non-void function
    In file included from /usr/include/string.h:640:0,
    		 from xx1.c:1:
    In function ‘strcpy’,
        inlined from ‘main’ at xx1.c:26:8:
    /usr/include/bits/string3.h:107:3: warning: call to __builtin___strcpy_chk will always overflow destination buffer

Only the "strcpy(x1->b.fo.y, "hallo");" with the nested struct is warned
about, while the second strcpy works fine.

It is just nested structs it does not like at this time.

Ciao, Marcus