Re: [PATCH][ATM] make atm (and clip) modular + try_module_get()

[chas williams]

In message <[e-mail]>,Werner Almesberger writes:
>see getting moved to net/core/skbuff.c, because it seems to provide
>a reasonably generic function.

it has been suggested to me that the locking in skb_migrate might not be
completely correct.  any comments on the following?

        spinlock_t *first, *second;
        if ((unsigned long)from < (unsigned long) to)) {
                first = &from->lock;
                second = &to->lock;
        } else {
                first = &to->lock;
                second = &from->lock;
        }

        local_irq_save(flags);
        spin_lock(&first);
        spin_lock(&second);

i imagine this is to prevent deadlocks when you do something silly like
skb_migrate(a,b) and then skb_migrate(b,a) elsewhere.