<div dir="ltr"><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">"I *think* that what Eric means is </span><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">that the GSO logic should automatically size the GSO superpackets so the </span><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">latency cost is negligible for the actual link rate."</span><div><br></div><div>Something like this?</div><div><br></div><div><a href="https://lwn.net/Articles/564979/">https://lwn.net/Articles/564979/</a></div><div><br></div><div><a href="https://lwn.net/Articles/564978/">https://lwn.net/Articles/564978/</a></div><div><br></div><div>/Jonas<br><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Apr 26, 2018 at 9:34 AM, Toke Høiland-Jørgensen <span dir="ltr"><<a href="mailto:toke@toke.dk" target="_blank">toke@toke.dk</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Kevin Darbyshire-Bryant <<a href="mailto:kevin@darbyshire-bryant.me.uk">kevin@darbyshire-bryant.me.uk</a><wbr>> writes:<br>
<span class="gmail-"><br>
>> On 25 Apr 2018, at 21:45, Toke Høiland-Jørgensen <<a href="mailto:toke@toke.dk">toke@toke.dk</a>> wrote:<br>
>> <br>
>> For those who have not been following the discussion on the upstreaming<br>
>> patches, here's an update:<br>
>> <br>
</span>>> <snip><br>
<span class="gmail-">>> <br>
>> So please do test the current git version (cobalt branch, still). I'm<br>
>> planning to resubmit on Friday.<br>
><br>
</span>> The two routers running that latest code survived the night which is a<br>
> good sign.<br>
<br>
Awesome!<br>
<br>
> I’ve sort of half been following the ‘discussion’, as ever the<br>
> reaction from the kernel people makes it a place I never wish to<br>
> look/contribute, ….. this from Eric has me disturbed "If you keep<br>
> saying this old urban legend, I will NACK your patch.I am tired of<br>
> people pretending GSO/TSO are bad for latencies.”<br>
<br>
Heh, yeah, the tone on kernel lists can be a bit... abrasive... just<br>
smile and wave and ignore the vitriol is my approach. But I can totally<br>
understand why some people don't want to put up with it... :)<br>
<span class="gmail-"><br>
> Genuine question: I have a superpacket circa 64K, this is a lump of<br>
> data in a tcp flow. I have another small VOIP packet, it’s latency<br>
> sensitive. If I split the super packet into individual 1.5K packets<br>
> as they would be on the wire, I can insert my VOIP packet at suitable<br>
> place in time such that jitter targets are not exceeded. If I don’t<br>
> split the super packet, surely I have to wait till the end of the<br>
> superpacket’s queue (for want of a better word) and possibly exceed my<br>
> latency target. That looks to me like ‘GSO/TSO’ is potentially bad<br>
> for interflow latencies. What don’t I understand here?<br>
<br>
</span>You are right in principle, of course. I *think* that what Eric means is<br>
that the GSO logic should automatically size the GSO superpackets so the<br>
latency cost is negligible for the actual link rate. I was actually<br>
thinking I would do some measurements at some point to test this at<br>
various rates; since we have a nice piece of code that can adaptively<br>
split GSO packets that should be pretty straight-forward :)<br>
<div class="gmail-HOEnZb"><div class="gmail-h5"><br>
-Toke<br>
______________________________<wbr>_________________<br>
Cake mailing list<br>
<a href="mailto:Cake@lists.bufferbloat.net">Cake@lists.bufferbloat.net</a><br>
<a href="https://lists.bufferbloat.net/listinfo/cake" rel="noreferrer" target="_blank">https://lists.bufferbloat.net/<wbr>listinfo/cake</a><br>
</div></div></blockquote></div><br></div></div></div>