|From:||Norberto R. de Goes Jr.|
|Subject:||Re: [lwip-users] TCP - problems with large data|
|Date:||Tue, 8 Jan 2019 11:19:44 -0200|
Am 8. Januar 2019 11:50:48 MEZ schrieb "Norberto R. de Goes Jr." <address@hidden>:
>How are you?
>After a long time spent in this issue with debugging in my network
>(socket raw), lwip configuration tuning (lwipopts.h), my client code,
>etc; I noted (using simhost -d) that large packets caused check-sum
>in the lwip received side. So by searching in the internet, I found
>relating to "offload segmentation".
>Resuming, when I disable the "gro" (generic-receive-offload) of the
>that treats the lwip packets the problem disappears.
>Command (for eth0): > sudo ethtool -K eth0 gro off
>Verification: > ethtool -k eth0 | grep
>Please, does this make sense?
Yes, it does. Sorry for not having found the time to dig into this. It has been on my list...
But it seems you found the reason. We should add this to our documentation somewhere. Where would you have looks for this?
lwip-users mailing list
|[Prev in Thread]||Current Thread||[Next in Thread]|