<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body dir="auto">
<div>My asleep brain is better than the awake one :-)</div>
<div id="AppleMailSignature"><br>
</div>
<div id="AppleMailSignature">Tcp in essence self servos into equilibrium, udp flows (floods) potentially don't. If the protocol doesn't servo then something else should, ideally before buffer full/memory issues make that happen anyway. </div>
<div id="AppleMailSignature"><br>
</div>
<div id="AppleMailSignature">Anyway time to switch brain to dayjob mode :-)</div>
<div id="AppleMailSignature"><br>
<br>
<span style="background-color: rgba(255, 255, 255, 0); font-size: 13pt;">--</span><br>
<div>
<div><span style="background-color: rgba(255, 255, 255, 0);">Cheers,</span></div>
<div><span style="background-color: rgba(255, 255, 255, 0);"><br>
</span></div>
<div><a href="mailto:Kevin@darbyshire-bryant.me.uk" style="background-color: rgba(255, 255, 255, 0);"><font color="#000000">Kevin@Darbyshire-Bryant.me.uk</font></a></div>
<div><span style="background-color: rgba(255, 255, 255, 0);">Sent from my phone, apologies for brevity, spelling & top posting</span></div>
</div>
</div>
<div><br>
On 6 May 2016, at 08:39, Dave Taht <<a href="mailto:dave.taht@gmail.com">dave.taht@gmail.com</a>> wrote:<br>
<br>
</div>
<blockquote type="cite">
<div><span>On Fri, May 6, 2016 at 12:29 AM, Kevin Darbyshire-Bryant</span><br>
<span><<a href="mailto:kevin@darbyshire-bryant.me.uk">kevin@darbyshire-bryant.me.uk</a>> wrote:</span><br>
<blockquote type="cite"><span>Hi All,</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>My brain woke up with this idea rattling around in it this</span><br>
</blockquote>
<blockquote type="cite"><span>morning...obviously the subconscious has been busy. So here it is:</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>Is there any way to use the egress drop signalling at ingress time to drop</span><br>
</blockquote>
<blockquote type="cite"><span>stuff before it gets into the queue so then we don't have to drop it at</span><br>
</blockquote>
<blockquote type="cite"><span>egress?</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>Something like: At enqueue if we've a matching flow check to see if that</span><br>
</blockquote>
<blockquote type="cite"><span>flow had been in egress 'fast dropping' state *and* know how much data in</span><br>
</blockquote>
<blockquote type="cite"><span>terms of time it had to fast drop to get the queue back under the nominal</span><br>
</blockquote>
<blockquote type="cite"><span>time threshold. If say it had to drop 10ms worth of packets to get back to</span><br>
</blockquote>
<blockquote type="cite"><span>the nominal 5ms threshold then it dropped 67% of the packets/data. I'd like</span><br>
</blockquote>
<blockquote type="cite"><span>to think of that as an 'unresponsive flow'...hence could it be possible to</span><br>
</blockquote>
<blockquote type="cite"><span>use that information at ingress time and in essence drop (some? 66%?) of</span><br>
</blockquote>
<blockquote type="cite"><span>them there, we can also signal congestion to the stack at that point to</span><br>
</blockquote>
<blockquote type="cite"><span>(cake already does this signalling when getting to its buffer size limit)</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>Probably a very silly idea.</span><br>
</blockquote>
<span></span><br>
<span>No, actualy, I'd been thinking about the same thing myself for days.</span><br>
<span>:) I've always wanted a way to notify userspace that I was dropping</span><br>
<span>(the heck out of) something on egress to try and stuff it up on</span><br>
<span>ingress, or do something more intelligent like re-route the traffic</span><br>
<span>elsewhere.</span><br>
<span></span><br>
<span>There was a lot of talk about adding the ability to drop stuff really</span><br>
<span>fast in the rx ring recently using the jit BPF stuff.</span><br>
<span></span><br>
<span>I was at a meeting where tom herbert talked about that, and I confess,</span><br>
<span>extremely dubious. Then... well, I see lots of drops on ingress</span><br>
<span>already I don't want, the udp flooding episode made me go look harder</span><br>
<span>at whole system behavior...</span><br>
<span></span><br>
<span>and I can't find the lwn article about it.</span><br>
<span></span><br>
<span></span><br>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>--</span><br>
</blockquote>
<blockquote type="cite"><span>Thanks,</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span><a href="mailto:Kevin@darbyshire-bryant.me.uk">Kevin@Darbyshire-Bryant.me.uk</a></span><br>
</blockquote>
<blockquote type="cite"><span>M: +44 7947 355344 H: +44 1256 478597</span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<blockquote type="cite"><span>_______________________________________________</span><br>
</blockquote>
<blockquote type="cite"><span>Cake mailing list</span><br>
</blockquote>
<blockquote type="cite"><span><a href="mailto:Cake@lists.bufferbloat.net">Cake@lists.bufferbloat.net</a></span><br>
</blockquote>
<blockquote type="cite"><span><a href="https://lists.bufferbloat.net/listinfo/cake">https://lists.bufferbloat.net/listinfo/cake</a></span><br>
</blockquote>
<blockquote type="cite"><span></span><br>
</blockquote>
<span></span><br>
<span></span><br>
<span></span><br>
<span>-- </span><br>
<span>Dave Täht</span><br>
<span>Let's go make home routers and wifi faster! With better software!</span><br>
<span><a href="http://blog.cerowrt.org">http://blog.cerowrt.org</a></span><br>
</div>
</blockquote>
</body>
</html>