On Sat, Dec 15, 2007 at 02:04:49PM +0800, Herbert Xu wrote:
> On Fri, Dec 14, 2007 at 11:52:18PM -0600, Matt Mackall wrote:
> >
> > No. The code as written above should reduce to:
> > 
> > 	if (val == NULL)
> > 		return -EFAULT;
> > 
> > If I hadn't wanted to return -EFAULT in this case, I would have just written:
> > 
> > 	WARN_ON(val == NULL);
> 
> Well the only reason I introduced
> 
> 	if (WARN_ON)
> 
> is so that what would otherwise be a BUG_ON condition would have
> a chance to get written to disk when invoked from an IRQ handler.
> 
> > I don't want code that was running safely (ie returning -EFAULT) to
> > start crashing the system just because I've, say, disabled printk.
> > That's creating an obnoxious heisenbug.
> 
> I'm disappointed that it has been used in ways that it shouldn't
> have been.
> 
> I suppose we'll have to either introduce a new primitive or just
> go back to using BUG_ON.

Seems we haven't yet reached concensus on what an appropriate use for
BUG_ON is. There's a fairly large camp who think that there are
basically no good reasons to outright crash a machine and that WARN_ON
should replace BUG_ON everywhere.

I tend to agree with this position, except when it comes to handling
filesystems, where panic is often (but not always) the right thing to
do.

-- 
Mathematics is the supreme nostalgia of our time.