Re: mmotm 2013-06-27-16-36 uploaded (wait event common)

[Oleg Nesterov <oleg@redhat.com>]

On 06/28, Oleg Nesterov wrote:
>
> On 06/28, Stephen Rothwell wrote:
> >
> > On Thu, 27 Jun 2013 23:06:43 -0700 Randy Dunlap <rdunlap@infradead.org> wrote:
> > >
> > > >> My builds are littered with hundreds of warnings like this one:
> > > >>
> > > >> drivers/tty/tty_ioctl.c:220:6: warning: the omitted middle operand in ?: will always be 'true', suggest explicit middle operand [-Wparentheses]
> > > >>
> > > >> I guess due to this line from wait_event_common():
> > > >>
> > > >> +		__ret = __wait_no_timeout(tout) ?: (tout) ?: 1;
> > > >>
> > I added the following to linux-next today:
> > (sorry Randy, I forgot the Reported-by:, Andrew please add)
> >
> > From: Stephen Rothwell <sfr@canb.auug.org.au>
> > Date: Fri, 28 Jun 2013 16:52:58 +1000
> > Subject: [PATCH] fix warnings from ?: operator in wait.h
>
> Argh. This patch strikes again.
>
> Thanks, and sorry. And please help!
>
> I am not sure I understand. Since when gcc dislikes '?:' ?
> /bin/grep shows a lot of users of 'X ?: Y' shortcut?

OK, I have found the machine with the newer gcc.

	#define test_1(tout) (!tout ?: 1)

	int func_1(long timeout)
	{
		return test_1(timeout);
	}

	#define test_2(tout) (tout ?: 1)

	int func_2(long timeout)
	{
		return test_2(timeout);
	}

test_1() triggers the same warning, test_2() doesn't. So it doesn't
like "computed-boolean ?: long".

> > Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
> > ---
> >  include/linux/wait.h | 18 ++++++++++++++----
> >  1 file changed, 14 insertions(+), 4 deletions(-)
> > 
> > diff --git a/include/linux/wait.h b/include/linux/wait.h
> > index 1c08a6c..f3b793d 100644
> > --- a/include/linux/wait.h
> > +++ b/include/linux/wait.h
> > @@ -197,7 +197,12 @@ wait_queue_head_t *bit_waitqueue(void *, int);
> >  	for (;;) {							\
> >  		__ret = prepare_to_wait_event(&wq, &__wait, state);	\
> >  		if (condition) {					\
> > -			__ret = __wait_no_timeout(tout) ?: __tout ?: 1;	\
> > +			__ret = __wait_no_timeout(tout);		\
> > +			if (!__ret) {					\
> > +				__ret = __tout;				\
> > +				if (!__ret)				\
> > +					__ret = 1;			\
> > +			}						\
> >  			break;						\
> >  		}							\
> >  									\
> > @@ -218,9 +223,14 @@ wait_queue_head_t *bit_waitqueue(void *, int);
> >  #define wait_event_common(wq, condition, state, tout)			\
> >  ({									\
> >  	long __ret;							\
> > -	if (condition)							\
> > -		__ret = __wait_no_timeout(tout) ?: (tout) ?: 1;		\
> > -	else								\
> > +	if (condition) {						\
> > +		__ret = __wait_no_timeout(tout);			\
> > +		if (!__ret) {						\
> > +			__ret = (tout);					\
> > +			if (!__ret)					\
> > +				__ret = 1;				\
> > +		}							\
> > +	} else								\
> >  		__ret = __wait_event_common(wq, condition, state, tout);\
> >  	__ret;								\
> >  })

Thanks. This should fix the isssue.

I'll try to send the cleanup patch later, this doesn't look very nice...

Oleg.