|
Subject: [PATCH 0/13]: Finish basic TX multiqueue functionality. Newsgroups: gmane.linux.network Date: 2008-07-10 10:55:58 GMT (51 weeks, 3 days, 8 hours and 51 minutes ago) I've respun the rest of the TX multiqueue patches in a tree cloned off of current net-next-2.6: kernel.org:/pub/scm/linux/kernel/git/davem/net-tx-2.6.git Thanks to Johannes Berg I was able to get the WME/QOS mac80211 bits at least basically tested and without the lockups he saw last time. That's a good sign. The most important patch is patch number 5. That's where the bulk of the tree and the existing multiqueue drivers transition completely to the new interfaces. It looks like a lot, but that patch really need to contain all of those bits all at once. There is no other way to sequence it. The full transition request going from: netif_*_queue() flow controls all queues netif_*_subqueue() flow controls individual queues to just: netif_tx_*_queue() flow controls individual queues If you just convert the core networking, the multiqueue drivers break. Whereas if you just convert the drivers, this breaks them too :) During the upheaval I temporarily mark the mac80211 QoS bits as broken, then in patch 9 it all gets converted properly and re-enabled. So, what's left? The qdisc config semantics. I still need to sort that out and pretty much that's what I'll be working on the next few days. This has been allmodconfig build tested on sparc64, basic multiqueue has been tested with NIU driver, non-multiqueue is tested using tg3 on my 2-cpu workstation, and Johannes gave mac80211 a little bit of usage. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo <at> vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html |
|
|