On Wed, Apr 11, 2018 at 01:30:53PM -0700, Randy Dunlap wrote:
> From: Randy Dunlap <rdunlap@xxxxxxxxxxxxx>
> with help from Linus. (many moons ago)
>
> sparse addition to print all compound/composite global data symbols
> with their sizes and alignment.
Hi,
I gave it some real testing now. All is good, just a few last details.
>
> +static void list_compound_symbols(struct symbol_list *list)
> +{
> + struct symbol *sym, *base;
> +
> + /* Only show arrays, structures, unions */
> + FOR_EACH_PTR(list, sym) {
> + /* Only show symbols that have a non-zero size */
> + if (!sym->bit_size)
This test should be changed to
if (sym->bit_size <= 0)
because VLA have -1 in this field and:
- I don't think you're interested in VLA here;
- the call to show_typename(sym) here under will then
throws an "error: bad constant expression" (but that
could or should be fixed in show_typename() itself).
> + if (is_func_type(sym))
> + continue;
> + if (is_float_type(sym) || is_scalar_type(sym))
> + continue;
These two tests can be removed because they are now redundant with ...
> + switch (base->type) {
> + case SYM_STRUCT: case SYM_UNION: case SYM_ARRAY:
> + break;
> + default:
> + continue;
> + }
this one since only structs, unions & arrays will 'survive' the test
and so functions, floats or integer (and whatever other types) are
automatically rejected. It's what I meant in the previous review.
> + check_symbols(res);
> +
> + if (dbg_compound)
> + list_compound_symbols(res);
> } END_FOR_EACH_PTR_NOTAG(file);
One thing that I think would be good to do is to change slighty
this list_compound_symbols() to not take a list of symbols but
just a single symbol and then do the
if (dbg_compound)
list_compound_symbol(sym);
inside check_symbols() which already loops through the symbol list.
Cheers,
-- Luc
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
