> > On Fri, Oct 28, 2022 at 09:55:57AM +0000, Neal Liu wrote: > > > > > > > Thanks for your feedback. > > > > > > > I tried to reproduce it on my side, and it cannot be reproduce it. > > > > > > > Here are my test sequences: > > > > > > > 1. emulate one of the vhub port to usb ethernet through > > > > > > > Linux gadget > > > > > > > (ncm) > > > > > > > > > > > > We are using rndis instead of ncm. > > > > > > > > > > > > > 2. connect BMC vhub to Host > > > > > > > 3. BMC & Host can ping each other (both usb eth dev default > > > > > > > mtu is > > > > > > > 1500) 4. Set BMC mtu to 1000 (Host OS cannot set usb eth dev > > > > > > > mtu to 2000, it's maxmtu is 1500) > > > > > > > > > > > > Not sure if it's related, but in my case (USB rndis, Debian 10 > > > > > > OS) it should be able to set MTU to 2000. > > > > > > > > > > Using rndis is able to set MTU to 2000, and the issue can be > reproduced. > > USB ecm is also tested and it is possible to set MTU to 2000, and could > reproduce the issue. > So I think this patch is needed anyway. > > @Neal Liu Could you kindly help to verify the USB ECM case? How to set MTU to 2000 on USB ECM case? I remember last time I cannot set by using "ifconfig ..." Regardless ECM or RNDIS, I agree this patch is still needed. > > > > > > > > Please NEVER use rndis anymore. I need to go just delete that > > > > driver from the tree. > > > > > > > > It is insecure-by-design and will cause any system that runs it to > > > > be instantly compromised and it can not be fixed. Never trust it. > > > > > > > > Even for data throughput tests, I wouldn't trust it as it does odd > > > > things with packet sizes as you show here. > > > > > > Thanks for the info, Greg. > > > If rndis will no longer be supported, how to use usb-ethernet on Windows > OS? > > > For my understanding, ncm/ecm cannot work on Windows OS. > > > > rndis should ONLY be there for Windows XP, which is long out-of-support. > > Newer versions of windows have more sane usb protocols built into it > > and this driver is not needed. > > > > As proof of this, Android devices removed this from their kernel > > configuration a few years ago and no one has complained :) > > > > thanks, > > > > greg k-h
