[ofa-general] Re: [PATCH for-2.6.22] IB/cm: improve local id allocation

[Michael S. Tsirkin]

> Quoting Or Gerlitz <ogerlitz at voltaire.com>:
> Subject: Re: [PATCH for-2.6.22] IB/cm: improve local id allocation
> 
> Michael S. Tsirkin wrote:
> >IB/cm uses idr for local id allocations, with a running counter
> >as start_id. This fails to generate distinct ids 
> 
> >static int cm_alloc_id(struct cm_id_private *cm_id_priv)
> >{
> >        unsigned long flags;
> >        int ret, id;
> >        static int next_id;
> >
> >        do {
> >                spin_lock_irqsave(&cm.lock, flags);
> >                ret = idr_get_new_above(&cm.local_id_table, cm_id_priv,
> >                                        next_id++, &id);
> >                spin_unlock_irqrestore(&cm.lock, flags);
> >        } while( (ret == -EAGAIN) && idr_pre_get(&cm.local_id_table, 
> >        GFP_KERNEL) );
> >
> >        cm_id_priv->id.local_id = (__force __be32) (id ^ 
> >        cm.random_id_operand);
> 
> Doesn't this Xor of the resulted ID with a random value done after the 
> idr allocation causes the cm to --always-- produce distinct ids???
> 
> Or.

No - the "cm.random_id_operand" is initialized at module load time.
That's why we have the static next_id iterator.

-- 
MST