Re: [PATCH iproute2] iproute: make clang happy

[Stephen Hemminger <stephen@networkplumber.org>]

On Mon, 20 Aug 2018 16:44:28 -0700
Mahesh Bandewar (महेश बंडेवार) <maheshb@google.com> wrote:

> On Mon, Aug 20, 2018 at 4:38 PM, Mahesh Bandewar (महेश बंडेवार)
> <maheshb@google.com> wrote:
> > On Mon, Aug 20, 2018 at 3:52 PM, Stephen Hemminger
> > <stephen@networkplumber.org> wrote:  
> >> On Mon, 20 Aug 2018 14:42:15 -0700
> >> Mahesh Bandewar <mahesh@bandewar.net> wrote:
> >>  
> >>> diff --git a/tc/m_ematch.c b/tc/m_ematch.c
> >>> index ace4b3dd738b..a524b520b276 100644
> >>> --- a/tc/m_ematch.c
> >>> +++ b/tc/m_ematch.c
> >>> @@ -277,6 +277,7 @@ static int flatten_tree(struct ematch *head, struct ematch *tree)
> >>>       return count;
> >>>  }
> >>>
> >>> +__attribute__((format(printf, 5, 6)))
> >>>  int em_parse_error(int err, struct bstr *args, struct bstr *carg,
> >>>                  struct ematch_util *e, char *fmt, ...)  
> >>
> >> I think the printf attribute needs to go on the function prototype
> >> here:
> >> tc/m_ematch.h:extern int em_parse_error(int err, struct bstr *args, struct bstr *carg,
> >>  
> > The attributes are attached to the definitions only and not prototype
> > declarations. Please see the definition/declaration for jsonw_printf()
> > in the same patch.  
> I take that back. Seems like it's fine either way.

The reason to put the attributes in the .h file is that then the compiler
can test for misuse in other files.  For example if em_parse_error had
a bad format in em_u32.c, then the warning would not happen unless
the attribute was on the function prototype.