Re: [PATCH v7 07/10] vsprintf: Consolidate handling of unknown pointer specifiers

[Sergey Senozhatsky]

On (04/17/19 13:53), Petr Mladek wrote:
[..]
>  char *ip_addr_string(char *buf, char *end, const void *ptr,
>  		     struct printf_spec spec, const char *fmt)
>  {
> +	char *err_fmt_msg;
> +
>  	switch (fmt[1]) {
>  	case '6':
>  		return ip6_addr_string(buf, end, ptr, spec, fmt);
> @@ -1457,7 +1459,8 @@ char *ip_addr_string(char *buf, char *end, const void *ptr,
>  		}}
>  	}
>  
> -	return ptr_to_id(buf, end, ptr, spec);
> +	err_fmt_msg = fmt[0] == 'i' ? "(%pi?)" : "(%pI?)";
> +	return string_nocheck(buf, end, err_fmt_msg, spec);
>  }

We cannot err on fmt[0] here, can we. Both %pi and %pI are OK
and handled by pointer():

	switch (fmt[0]) {
	case 'I':
	case 'i':
		return ip_addr_string(buf, end, ptr, spec, fmt);


It's fmt[1] which can be unrecognized:

	switch (fmt[1]) {
	case '6':
	case '4':
	case 'S':
	}

	-ss