On Sat, Mar 24, 2012 at 01:52:54PM +0100, Peter Zijlstra wrote:
[...]

quoted

p.s. I know Peter Zijlstra detest the __cond_lock() stuff, but untill
     we have anything better in sparse, let's use it. This particular
     patch helped me to detect one bug that I myself made during
     task->mm fixup series. So, it is useful.

Yeah, so Nacked-by: Peter Zijlstra [off-list ref]

Also, why didn't lockdep catch it?

Because patch authors test their patches on architectures they own
(well, sometimes I do check patches on exotic architectures w/ qemu,
but it is less convenient than just build/sparse-test the patch w/
a cross compiler).

And since lockdep is a runtime checker, it is not very useful.

Sparse is a build-time checker, so it is even better in the sense
that it is able to catch bugs even in code that is executed rarely.

Fix sparse already instead of smearing ugly all over.

Just wonder how do you see the feature implemented?

Something like this?

#define __ret_cond_locked(l, c)	__attribute__((ret_cond_locked(l, c)))
#define __ret_value		__attribute__((ret_value))
#define __ret_locked_nonnull(l)	__ret_cond_locked(l, __ret_value);

extern struct task_struct *find_lock_task_mm(struct task_struct *p)
	__ret_locked_nonnull(&__ret_value->alloc_lock);

Thanks,

-- 
Anton Vorontsov
Email: cbouatmailru@gmail.com