Question

I’m maintaining and extending the functionality of a diagnostic test suite, and this fragment of code comes up multiple times and I’m not sure what it does:

int ret = 0, i, *reg;
int size = sizeof(Regs)/sizof(Reg);
for(i = 0; i<size; i++) {
    reg= (uint*)Regs[i].Number;
    *reg=0;
}
return ret;

Where i is looping variable, and reg is a pointer to a 32-bit int (Number being the address to that 32-bit int). This particular test clears the registers, but is it meant to change ret?

EDIT: I guess my question was a bit ambiguous, I am wondering specifically what the fragment:

int ret = 0, i, *regs;

Does, or why it is valid.

Answer 1

int ret = 0, i, *reg;

defines three variables: an int called ret, an int called i and an int* called reg. Only ret is initialized (to 0).

The reason is that the grammar for a declaration (slightly simplified) is:

declaration: 
    declaration-specifiers init-declarator-list[_opt_]
init-declarator-list:
    init-declarator
    init-declarator-list , init-declarator
init-declarator:
    declarator
    declarator = initializer

Here the declaration-specifiers are just int, then we have an init-declarator list consisting of three init-declarators. One of them has an initializer and the other two don’t. The declarator *reg defines a “pointer-to-whatever-type-the-declaration-specifiers-say”. See 6.7 “Declarations” in the standard for the gory details (and the bits I left out of the grammar above).