[openib-general] ipoib lockdep warning

[Zach Brown]

Roland Dreier wrote:
> Hmm, good point.
> 
> It sort of seems to me like the idr interfaces are broken by design.
> Internally, lib/idr.c uses bare spin_lock(&idp->lock) with no
> interrupt disabling or anything in both the idr_pre_get() and
> idr_get_new() code paths.

I wasn't thrilled to see that either.  We seem to have a fair precedent
(list.h, rbtree, etc) for leaving serialization to callers.

> So, ugh... maybe the best thing to do is change lib/idr.c to use
> spin_lock_irqsave() internally?

I dunno, it seems to have had _irq() locking in the past?  From the
comment at the top:

 * Modified by George Anzinger to reuse immediately and to use
 * find bit instructions.  Also removed _irq on spinlocks.

- z