Quoting Christoph Hellwig <[EMAIL PROTECTED]>:
> Btw, the whole sdb_buff code looks a little like it's duplicating the
> network layers sk_buff code. Anyone looking into this higher-level
> thing?
I was looking into the possibility to use list.h to manage queues.
But you are right - there is a similiarity, it might just make sense to use
skbuff.h rather than list.h to manage sdpc_buffs.
Need to think what to do with sdpc_iocb/sdpc_advt then: currently
these are, sometimes, kept on the same queue with sdpc_buff objects.
Worth keeping in mind, and I'll be interested to see such a patch
emerge.
Another option that comes to mind would be to generalize linked-list-with-size
code in skbuff.h, moving it into list.h, sdp would then reuse it.
Something like:
struct slist_entry {
struct slist_entry *next;
struct slist_entry *prev;
struct slist_head *list;
};
Does this make sense to anyone?
Thanks,
--
MST
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
