On 15 May 2018 at 15:46, Richard Henderson <r...@twiddle.net> wrote:
> On 05/15/2018 06:25 AM, Peter Maydell wrote:
>> I guess that having the compiler check is better than finding them
>> later with coverity. It's a shame gcc doesn't support the standard
>> mechanism of using /* fallthrough */ to mark these, though.
>
> It does. Apparently not by default anymore, however:
>
> @item @option{-Wimplicit-fallthrough=0} disables the warning altogether.
>
> @item @option{-Wimplicit-fallthrough=1} matches @code{.*} regular
> expression, any comment is used as fallthrough comment.
>
> @item @option{-Wimplicit-fallthrough=2} case insensitively matches
> @code{.*falls?[ \t-]*thr(ough|u).*} regular expression.
>
> @item @option{-Wimplicit-fallthrough=3} case sensitively matches one of the
> following regular expressions:
> ...
>
> I think either =2 or =1 would work for us in QEMU.
1 sounds too broad, we don't want any old comment to count.
2 is probably what we want.
thanks
-- PMM
