Re: [PATCH net 1/1] openvswitch: Fix setting ipv6 fields causing hw csum failure

[Paul Blakey <paulb@nvidia.com>]

On Thu, 10 Feb 2022, Jakub Kicinski wrote:

> On Thu, 10 Feb 2022 10:53:24 +0200 Paul Blakey wrote:
> > > The calls seem a little heavy for single byte replacements.
> > > Can you instead add a helper based on csum_replace4() maybe?
> > > 
> > > BTW doesn't pedit have the same problem?
> > 
> > I don't think they are heavier then csum_replace4,
> 
> csum_replace4 is a handful of instructions all of which will be inlined.
> csum_partial() is a function call and handles variable lengths.
> 
> > but they are more bulletproof in my opinion, since they handle both
> > the COMPLETE and PARTIAL csum cases (in __skb_postpull_rcsum())
> 
> Yes, that's why I said "add a helper based on", a skb helper which
> checks the csum type of the packet but instead of calling csum_partial
> for no reason does the adjustment directly.

Then sure, I will do that and send v2.

> 
> > and resemble what editing of the packet should have done - pull the
> > header, edit, and then push it back.
> 
> That's not what this code is doing, so the argument does not stand IMO.
>