Re: strict aliasing question

[Andrew Haley <aph at redhat dot com>]

Howard Chu writes:
 > I see a lot of APIs (e.g. Cyrus SASL) that have accessor functions 
 > returning values through a void ** argument. As far as I can tell, this 
 > doesn't actually cause any problems, but gcc 4.1 with -Wstrict-aliasing 
 > will complain. For example, take these two separate source files:
 > 
 > alias1.c
 > ####
 > 
 > #include <stdio.h>
 > 
 > extern void getit( void **arg );
 > 
 > main() {
 >         int *foo;
 > 
 >         getit( (void **)&foo);
 >         printf("foo: %x\n", *foo);

This is not legal C.  You are dereferencing a pointer to a type other
than the object that it points to.

This would have been quite legal, albeit implementation defined:

        void **foo;
        getit(foo);
        printf("foo: %x\n", (int)*foo);

 > }
 > ####
 > 
 > 
 > alias2.c
 > ####
 > static short x[] = {16,16};
 > 
 > void getit( void **arg ) {
 >         *arg = x;

This is fine.

 > }
 > ####
 > 
 > gcc -O3 -fstrict-aliasing -Wstrict-aliasing *.c -o alias
 > 
 > The program prints the expected result with both strict-aliasing and 
 > no-strict-aliasing on my x86_64 box.  As such, when/why would I need to 
 > worry about  this warning?

Always.  This isn't C code, and the compiler may not generate the code
you expect for it.

This has been explained countless times before.  It certainly isn't a
suitable topic for the gcc mailing list, which is for development of
gcc, not help using it.  Redirected to gcc-help.

Please read Section 6.3.2.3 of ISO9899:1999, in particular Para. 7.

Andrew.