Implement -Wswitch-fallthrough

[Marc Glisse]

On Wed, 13 Jul 2016, Marek Polacek wrote:

> On Wed, Jul 13, 2016 at 08:39:28PM +0200, Marc Glisse wrote:
>> On Wed, 13 Jul 2016, Marek Polacek wrote:
>>
>>> Does "__attribute__((fallthrough));" have any advantages over
>>> "__builtin_fallthrough()"?
>>
>> Not a strong argument, but compilers that don't know the construct will give
>> an error on the builtin, and just a warning on the attribute (and ignore the
>> comment, which makes both other versions less relevant).
>
> Hmm, I'd expect that people add something like
>
> #if GCC_VERSION >= 7000
> # define gcc_fallthrough() __builtin_fallthrough ()
> #else
> # define gcc_fallthrough()
> #endif
>
> and not use the builtin directly.

I expect people will mostly use the comment, or maybe the C++17 attribute 
in a few years in C++ code. But accidents happen. I did say it was a weak 
argument ;-)

>> Unrelated question: are there cases where __builtin_fallthrough() has any
>> impact on code generation?
>
> It should not -- my implementation gets rid of all __builtin_fallthrough()
> during gimple-low, before most of the optimizations kick in, so it shouldn't
> make any real difference.

"should", "most", "real"... This is a bit like compare-debug, as a user I 
expect it to only affect warnings, and if it can affect code generation, 
even in small ways, I would like the doc to tell me.

-- 
Marc Glisse