Re: [GIT pull] timer updates for 3.4-rc

[Andrew Morton <akpm@linux-foundation.org>]

On Sat, 7 Apr 2012 01:48:51 +0200 (CEST)
Thomas Gleixner <tglx@linutronix.de> wrote:

> On Fri, 6 Apr 2012, David Rientjes wrote:
> 
> > On Sat, 7 Apr 2012, Thomas Gleixner wrote:
> > 
> > > diff --git a/kernel/itimer.c b/kernel/itimer.c
> > > index 22000c3..c70369a 100644
> > > --- a/kernel/itimer.c
> > > +++ b/kernel/itimer.c
> > > @@ -284,8 +284,11 @@ SYSCALL_DEFINE3(setitimer, int, which, struct itimerval __user *, value,
> > >  	if (value) {
> > >  		if(copy_from_user(&set_buffer, value, sizeof(set_buffer)))
> > >  			return -EFAULT;
> > > -	} else
> > > +	} else {
> > >  		memset((char *) &set_buffer, 0, sizeof(set_buffer));
> > > +		WARN_ONCE(1, "setitimer: new_value pointer is NULL."
> > > +			  " Misfeature support will be removed\n");
> > > +	}
> > >  
> > >  	error = do_setitimer(which, &set_buffer, ovalue ? &get_buffer : NULL);
> > >  	if (error || !ovalue)
> > 
> > When I tried doing this for the deprecated /proc/pid/oom_adj there was a 
> > complaint that WARN_ONCE() screws up log parsing scripts that think there 
> > is a kernel issue so I had to change it to printk_once() and include 
> > get_task_comm(current) and current->pid.  See
> > http://lkml.indiana.edu/hypermail/linux/kernel/1108.0/00446.html
> 
> Bah.

It gives you a chance to remove the unneeded cast from the memset call ;)