Re: [PATCH] rculist: fix borked __list_for_each_rcu() macro

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Fri, Dec 17, 2010 at 06:10:39PM +0800, Américo Wang wrote:
> On Thu, Dec 16, 2010 at 07:50:54AM -0800, Paul E. McKenney wrote:
> >On Thu, Dec 16, 2010 at 03:38:40PM +0800, Américo Wang wrote:
> >> On Thu, Dec 16, 2010 at 07:02:36AM +0100, Mariusz Kozlowski wrote:
> >> >On Wed, Dec 15, 2010 at 03:20:05PM -0800, Paul E. McKenney wrote:
> >> >> On Wed, Dec 15, 2010 at 11:11:12PM +0100, Mariusz Kozlowski wrote:
> >> >> > This restores parentheses blance.
> >> >> 
> >> >> Good catch, queued!!!
> >> >> 
> >> >> This does not actually appear to be in use anywhere in the kernel any
> >> >> more, so I queued this for 2.6.38 rather than in the 2.6.37 urgent queue.
> >> >> So, just out of curiosity, how did you find this one?
> >> >
> >> >Some years ago I wrote a dumb script that walks trees of () and {}.
> >> >It catches unbalanced trees. It's dumb enough to fail with #ifdef etc,
> >> >but most of the time it does its job. It reaches unreachable code
> >> >and unused one too.
> >> 
> >> gcc will complain about this, however, in this case, it is used.
> >
> >Hello, Américo!
> >
> >I did a "git grep -l __list_for_each_rcu" and its output was only:
> >
> >	include/linux/rculist.h:#define __list_for_each_rcu(pos, head) \
> >
> >This was in Linus's tree.  And gcc certainly would have failed if
> >this macro had been used in any recent build.
> 
> Yeah, my bad, actually I meant to say "unused"... :-(
> Sorry for confusing!

No problem!

> My point is that gcc should do this basic lexical check, no need
> to invent another tool. :)

As an off-by-default warning, this could make a lot of sense, especially
for projects like the Linux kernel that are relatively disciplined in
their use of cpp macros.  Though I am not sure that the recent macros
in the "perf" code would pass such a check.  ;-)

							Thanx, Paul