Thread: Understanding va_start(ap, v) in STDARG.H

v is the the argument before ...
va_start actually sets the pointer 'ap' to point at the stack memory exactly
below 'v':

Code:

void foo(int a, int b, int c, ...);
...
foo(1,2,3,100,200)

stack:
<return address>
200
100  <-- va_start() sets ap here
3
2
1

va_start has to know the size of last named argument, v, to set up right.

to make the picture more clear:

Code:

stack memory inside foo():
200
100  <-- va_start() sets 'ap' here
3   <-- this is &v. we need to increase to size of v to shift to the first unnamed arg
2
1
<return address>
<foo's prologue>
<foo's local variables>

the _INTSIZEOF(n) macro, I beleive, calculates the size of last named
argument, 'rounded' to sizeof(int), because v, push-ed on stack before
call to foo(), is 'rounded' to size. For example, on IA32,

Code:

void bar(char c);
...
char c = 1;
bar(c);

c is push-ed on stack as 32-bit value, so the size is sizeof(int).
in case of bar(double d), two 32-bit values are push-ed.

I don't get it. valenok.
After all, I know v is a pointer, and &v is a pointer, too, pointing to somewhere in the stack where v is stored. Thanks!

you have to understand how the function call works,
the stack memory layout before and after the call.
If you do not get it, there is no chance you'll understand va_start() macro.
some assembler tutorial on that matter would be in help.

first of all, 'v' may not be a pointer.