On 9/23/20 2:30 PM, Vladimir Oltean wrote: > On Wed, Sep 23, 2020 at 11:51:48PM +0300, Vladimir Oltean wrote: >> On Wed, Sep 23, 2020 at 01:49:54PM -0700, Florian Fainelli wrote: >>> On 9/23/20 1:45 PM, Florian Fainelli wrote: >>> >>> David, Jakub, there is an unnecessary header inclusion in >>> net/dsa/tag_brcm.c in the second patch and the description at the end of >>> the commit was not updated, let me send a v3 right away. >>> -- >>> Florian >> >> Wait a few minutes, I don't think anybody has had a chance to look at it.. > > So I studied a little bit the situation with the MAC header pointer in > the second patch, especially since there are 2 call paths, but it's ok, > since at that point the skb->data points to its final position already, > then the MAC header is naturally 14 bytes behind. > I've been testing with the ocelot driver which resets the MAC header > manually, that explains why I missed that. Anyway, if there are no other > comments I think you can resend.
Yes, I was looking at tag_ocelot.c as well for a reference. At some point I was considering adding a version that let the tagging driver specifying the struct vlan_ethhdr reference directly since the tagging driver knows, but using skb_set_mac_header() works equally well in a more portable way even. Thanks, v3 on its way with your A-b tag. -- Florian
