Re: C PATCH to add -Wswitch-unreachable (PR c/49859)

[Marek Polacek <polacek at redhat dot com>]

On Mon, May 23, 2016 at 05:53:38PM -0600, Martin Sebor wrote:
> Sorry I'm a little late with my comments but I noticed one minor
> problem (I raised bug 71249 for it since the patch has already
> been checked in), and have a question about the hunk below:
 
Firstly, thanks for looking into this.

> > @@ -1595,6 +1595,32 @@ gimplify_switch_expr (tree *expr_p, gimple_seq *pre_p)
> >         gimplify_ctxp->case_labels.create (8);
> > 
> >         gimplify_stmt (&SWITCH_BODY (switch_expr), &switch_body_seq);
> > +
> > +      /* Possibly warn about unreachable statements between switch's
> > +	 controlling expression and the first case.  */
> > +      if (warn_switch_unreachable
> > +	  /* This warning doesn't play well with Fortran when optimizations
> > +	     are on.  */
> > +	  && !lang_GNU_Fortran ()
> > +	  && switch_body_seq != NULL)
> > +	{
> > +	  gimple_seq seq = switch_body_seq;
> > +	  if (gimple_code (switch_body_seq) == GIMPLE_BIND)
> > +	    seq = gimple_bind_body (as_a <gbind *> (switch_body_seq));
> > +	  gimple *stmt = gimple_seq_first_stmt (seq);
> > +	  enum gimple_code code = gimple_code (stmt);
> > +	  if (code != GIMPLE_LABEL && code != GIMPLE_TRY)
> 
> Why exempt GIMPLE_TRY?  It suppresses the warning in cases like:
> 
>   switch (i) {
>   try { } catch (...) { }
>   case 1: ;
>   }
> 
> (If excluding GIMPLE_TRY is unavoidable, it might be worthwhile
> to add a comment to the code, and perhaps also mention it in
> the documentation to preempt bug reports by nitpickers like me ;)
 
I think I added that so that we do not warn on
  switch (i)
    {
      int A[3];
      default:
	break;
    }
because at the gimple level that looks like
{
  int A[3];

  try
    {
      <D.1751>:
      goto <D.1752>;
    }
  finally
    {
      A = {CLOBBER};
    }
}
Another problem with try/finally is that it doesn't have a location
so we'd jsut print useless
cc1: warning: statement will never be executed [-Wswitch-unreachable]

Though it seems so improbable that I don't really care about this case.

> Finally, while even this simple warning can be useful, it would
> be even more helpful if it could also point out other unreachable
> statements within the body of the switch statements after
> a break/goto/return and before a subsequent label.  This could
> be especially valuable with optimization to make possible
> diagnosing non-trivial problems like this:
> 
>   switch (i) {
>   case 3:
>     if (i < 3)
>        return 1;
>     i = 8;
>   }
> 
> (I realize this might be outside the scope of the feature request
> and starting to creep into the -Wunreachable-code territory.)

This really sounds like the old -Wunreachable stuff and I don't think
it's limited to switches as this warning is.  Nowadays we have stuff
like gimple_stmt_may_fallthru so maybe that could be useful, but I'm
not about to plunge into this mess anytime soon ;).

	Marek