On Tue, Oct 20, 2020 at 04:08:39PM +0800, Leo Yan wrote:

SNIP

> > 
> > please update man page with this
> > 
> > >   else {
> > >           pr_err("failed: unknown display type: %s\n", str);
> > >           return -1;
> > > @@ -2766,9 +2795,10 @@ static int build_cl_output(char *cl_sort, bool 
> > > no_source)
> > >   }
> > >  
> > >   if (asprintf(&c2c.cl_output,
> > > -         "%s%s%s%s%s%s%s%s%s%s",
> > > +         "%s%s%s%s%s%s%s%s%s%s%s",
> > 
> > why is there extra '%s' when we did not add new argument.. ?
> 
> This is deliberate.  The change is as below:
> 
>         if (asprintf(&c2c.cl_output,
> -               "%s%s%s%s%s%s%s%s%s%s",
> +               "%s%s%s%s%s%s%s%s%s%s%s",
>                 c2c.use_stdio ? "cl_num_empty," : "",
> -               "percent_rmt_hitm,"
> +               c2c.display == DISPLAY_LLC ? "percent_llc_hit," :
> +                                            "percent_rmt_hitm,",
>                 "percent_lcl_hitm,"
> 
> In the old code the string "percent_rmt_hitm," is merged with later
> lines (the second string is "percent_lcl_hitm,") into single string.
> 
> In this patch, it needs to check condition c2c.display and pass
> different string ("percent_llc_hit," vs "percent_rmt_hitm,"), thus
> the string ("percent_llc_hit," or "percent_rmt_hitm,") is passed
> independently, it's _NOT_ jointed with sequnetial lines.

ah right, it's now 2 arguments instead of one

thanks,
jirka

Reply via email to