On Mon, Oct 29, 2018 at 03:39:50PM +0000, Ben Dooks wrote:
> Add code to parse the __attribute__((format)) used to indicate that
> a variadic function takes a printf-style format string and where
> those are. Save the data in ctype ready for checking when such an
> function is encoutered.
>
> Signed-off-by: Ben Dooks <ben.dooks@xxxxxxxxxxxxxxx>
> ---
> Fixes since v1:
> - moved to using ctype in base_type to store infromation
> - fixed formatting issues
> - updated check for bad format arguments
> - reduced the line count to unsigned short to save space
>
> Notes:
> - What to do when base_type is not set... current code doesn't work
Yes, I saw this. See here below.
> ---
> parse.c | 74 ++++++++++++++++++++++++++++++++++++++++++++++++++++++--
> symbol.h | 2 ++
> 2 files changed, 74 insertions(+), 2 deletions(-)
>
> diff --git a/parse.c b/parse.c
> index 02a55a7..9b0d40e 100644
> --- a/parse.c
> +++ b/parse.c
> @@ -1051,6 +1061,67 @@ static struct token *attribute_address_space(struct token *token, struct symbol
> return token;
> }
>
> +static struct token *attribute_format(struct token *token, struct symbol *attr, struct decl_state *ctx)
> +{
> + struct expression *args[3];
> + struct symbol *fmt_sym = NULL;
> + int argc = 0;
> +
> + /* expecting format ( type, start, va_args at) */
> +
> + token = expect(token, '(', "after format attribute");
> + while (!match_op(token, ')')) {
> + struct expression *expr = NULL;
> +
> + if (argc == 0) {
> + if (token_type(token) == TOKEN_IDENT)
> + fmt_sym = lookup_keyword(token->ident, NS_KEYWORD);
> +
> + if (!fmt_sym || !fmt_sym->op ||
> + fmt_sym->op != &attr_printf_op) {
> + sparse_error(token->pos,
> + "unknown format type '%s'\n",
> + show_ident(token->ident));
> + fmt_sym = NULL;
> + }
> + }
> +
> + token = conditional_expression(token, &expr);
> + if (!expr)
> + break;
> + if (argc < 3)
> + args[argc++] = expr;
> + if (!match_op(token, ','))
> + break;
> + token = token->next;
> + }
> +
> + if (argc != 3 || !fmt_sym) {
> + warning(token->pos, "incorrect format attribute");
> + } else {
> + long long start, at;
> +
> + start = get_expression_value(args[2]);
> + at = get_expression_value(args[1]);
> +
> + if (start <= 0 || at <= 0 || (start == at && start > 0)) {
> + warning(token->pos, "bad format positions");
> + } else if (start < at) {
> + warning(token->pos, "format cannot be after va_args");
> + } else if (!ctx->ctype.base_type) {
> + warning(token->pos, "no base type to set");
> + ctx->ctype.printf_va_start = start;
> + ctx->ctype.printf_msg = at;
> + } else {
> + ctx->ctype.base_type->ctype.printf_va_start = start;
> + ctx->ctype.base_type->ctype.printf_msg = at;
> + }
This makes me realize that only a few attributes are handled and can be
handled. The last lines should simply be written as:
+ } else if (start < at) {
+ warning(token->pos, "format cannot be after va_args");
+ } else {
+ ctx->ctype.printf_va_start = start;
+ ctx->ctype.printf_msg = at;
+ }
In other words, this parsing should not poke into the base type. But as-is
the two args will not be propagated into the function because ctx is for the
whole declaration. You will need to add something like:
diff --git a/parse.c b/parse.c
index 83bca24b3..4e273b743 100644
--- a/parse.c
+++ b/parse.c
@@ -2979,6 +2979,10 @@ struct token *external_declaration(struct token *token, struct symbol_list **lis
if (!(decl->ctype.modifiers & MOD_STATIC))
decl->ctype.modifiers |= MOD_EXTERN;
+
+ // apply non-modifier function attributes
+ base_type->ctype.printf_msg = decl->ctype.printf_msg;
+ base_type->ctype.printf_va_start = decl->ctype.printf_va_start;
} else if (base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
sparse_error(token->pos, "void declaration");
}
and at some later stage, I'll make it more generic.
Kind regards,
-- Luc
