On Sat, Nov 9, 2024 at 12:52 PM Hans de Goede <hdegoede@xxxxxxxxxx> wrote:
> On 9-Nov-24 5:30 AM, Andy Shevchenko wrote:
> > On Sat, Nov 9, 2024 at 1:24 AM Hans de Goede <hdegoede@xxxxxxxxxx> wrote:
...
> >> +static void power_supply_escape_spaces(const char *str, char *buf, size_t bufsize)
> >> +{
> >> + strscpy(buf, str, bufsize);
> >> + strreplace(buf, ' ', '_');
> >> +}
> >
> > The bufsize in all cases here is sizeof(buf), making the above to be a
> > macro we may switch to 2-argument strscpy(). FTR, it embeds the check
> > that buf is an array.
>
> I did think about this already, but using a macro makes this harder
> to read just to save 2 sizeof() calls. So I prefer doing things
> this way.
...
> >> + char escaped_label[32];
> >
> > Even more, the but size seems also the same, can we have buf defined
> > inside the above?
>
> No not really, its address would need to be returned then, requiring
> it to be static, at which point we get race conditions when multiple
> threads use the same function at the same time.
I meant a macro case, but it's not a big deal after all. Perhaps we
can evolve a macro in the generic headers if there are enough users,
like
#define str_copy_and_replace(...) ...
--
With Best Regards,
Andy Shevchenko
