svn commit: r258328 - head/sys/net

[Robert N. M. Watson]

On 20 Nov 2013, at 20:11, Julian Elischer <julian at FreeBSD.ORG> wrote:

>> Currently, it is quite easy to make mistakes regarding individual mbuf chains vs. lists of mbuf chains.  This leads me to wonder whether a new type, perhaps simply constructed on the stack before passing in, should be used for KPIs that accept lists of packets. E.g.,
>> 
>>    /*
>>     * This structure is almost always allocated on a caller stack, so
>>     * cannot itself be queued without memory allocation in most cases.
>>     */
>>    struct mbuf_queue {
>>        struct mbuf    *mq_head;
>>    };
>> 
>> 
> It's hard to believe that we don't have a structure around already that we can't use. With Luigi's comment, I wonder that there isn't an mbuf_list structure already we can just steal. it could almost be the current interface input queue structure.

The exact details don't matter; what does matter is that we make it as easy as possible to detect mistakes using the compiler (e.g., queue passed where mbuf expected, or vice versa) and that those situations we can't check statically, we try to check dynamically (multi-entry queue passed where mbuf expected). In the past we've had bugs along similar lines, where code expects m->m_nextpkt to be NULL when it isn't leading to very unhappy times in socket buffers, etc. Ideally what we did would have no expense at runtime unless debugging features were turned on (INVARIANTS).

Robert