On Fri, 11 Nov 2011, Felipe Balbi wrote:
> > The proper fix is probably to make ohci_dbg_sw an inline routine
> > instead of a macro.
>
> hmmm, makes sense. Something like below ? I'm not sure about the usage
> of size though...
The idea of this macro is that it writes a formatted message either
into a pre-supplied buffer or to the system log. So *next is the first
free address in the buffer, and *size is the amount of free space
remaining in the buffer -- the macro has to update both of them. If
next is NULL then there is no buffer and the log is used instead.
> -#define ohci_dbg_sw(ohci, next, size, format, arg...) \
> - do { \
> - if (next != NULL) { \
> - unsigned s_len; \
> - s_len = scnprintf (*next, *size, format, ## arg ); \
> - *size -= s_len; *next += s_len; \
> - } else \
> - ohci_dbg(ohci,format, ## arg ); \
> - } while (0);
> -
> +static void ohci_dbg_sw(struct ohci_hcd *ohci, char **next,
> + unsigned *size, const char *fmt, ...)
> +{
> + va_list args;
> +
> + va_start(args, fmt);
> + do {
> + if (next != NULL) {
> + unsigned s_len;
> +
> + s_len = scnprintf(*next, *size, fmt, args);
> + *size -= s_len;
> + *next += s_len;
> + } else {
> + ohci_dbg(ohci, fmt, args);
> + break;
> + }
> + } while(*size);
> + va_end(args);
> +}
No do-while loop is needed. Take that out and it will be more or less
okay, except... Does ohci_dbd() work correctly with a va_list instead
of real arguments? I kind of doubt it. In fact, that may be the
reason this was originally written as a macro.
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
