site_archiver@lists.apple.com Delivered-To: darwin-kernel@lists.apple.com On Apr 10, 2008, at 11:44 AM, Justin C. Walker wrote: On Apr 10, 2008, at 06:45 , Michael Cashwell wrote: Fragmentation is fun. :-) You are certainly correct, but the driver isn't involved in that. Justin -- Justin C. Walker, Curmudgeon-At-Large Institute for the Absorption of Federal Funds -------- If you're not confused, You're not paying attention -------- _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.a... On Apr 10, 2008, at 10:35 , Michael Cashwell wrote: The point in rimas' case is that just because the data are UDP does not give the Ethernet driver more permission to drop them than if they were TCP [corrected] nor does it mean they are less important. You are correct that an ethernet driver does not have permission to drop one kind of packet over another, I don't think you are correct in your implication that it cares about network-layer packet types Did I imply it does? I didn't intend to. You did say that the driver should not have "more permission" to drop UDP over TCP packets. I don't know of any driver that pays attention to IP frame type (in particular because they can't always tell; as you say, fragmentation is fun). : these are all one to the driver (it's possible for the driver to obey QOS indicators of some type, but it can't know about the IP frame type: in particular, it may not know that a particular packet is in fact (part of) a UDP frame). After all, it's ethernet. Even if the packet gets to the media layer, there's no guarantee that (a) it will make it out the door; or (b) even if it gets out the door, it may not survive the trip across the local link :-}). I don't think TCP/UDP or QoS is involved in rimas' problem. That really does look like a driver issue. I don't think so, although I suppose it's worth looking at. The driver can only do so much; it's part of a bucket brigade, and if it doesn't have buffers, that's it for the outgoing packets. The driver has no clue where the packets come from; it's just all bits to it. My point was only to underscore that there are different dimensions to the question of what delivery characteristics are most important. If you're transferring live video having data arrive late is no different from it not arrive at all. In fact, if "late" requires retries and resends (and this consumed even more bandwidth and resources) only to just be dropped anyway then late is actually worse than not at all. Assuming that someone's data is less important just because they used UDP is flawed. I don't think there is any assumption regarding that except in the UDP code, where it's assumed that if there's no space for the packet, it's OK to drop it. This email sent to site_archiver@lists.apple.com