On 5/31/19 7:54 AM, Herbert Xu wrote:

On Fri, May 31, 2019 at 07:50:06AM -0700, Eric Dumazet wrote:

quoted

What do you mean by should ?

Are they currently already linearized before the function is called,
or is it missing and a bug needs to be fixed ?

AFAICS this is the code-path for locally generated outbound packets.
Under what circumstances can the IPv6 header be not in the head?

I guess this means we had yet another random submission from Young Xiao :/

Thanks.